Compare commits

..
Author SHA1 Message Date
sysadminandClaude Opus 4.8 b7a63a5579 feat(webui): unified session/lease/lock/worktree inventory API (Closes #636)
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
2026-07-22 21:13:54 -04:00
sysadmin 53c2c92782 Merge pull request 'feat(webui): versioned project registry API (Closes #635)' (#819) from feat/issue-635-project-registry-api into master 2026-07-22 20:04:26 -05:00
sysadmin 14c9c4d702 Merge pull request 'fix(mcp): forward worktree_path into publication preflight (Closes #815)' (#817) from fix/issue-815-preflight-worktree-forwarding into master 2026-07-22 18:34:31 -05:00
jcwalker3andClaude Opus 4.8 95a5eb254f fix(mcp): forward worktree_path into publication preflight (Closes #815)
gitea_publish_unpublished_issue_branch accepted a required worktree_path
but resolved it only after verify_preflight_purity had already run, so the
#618 branches-only guard and every other workspace-resolution layer behind
preflight received None and fell back to the MCP process root. A daemon
rooted at the stable control checkout therefore refused a valid registered
issue worktree the caller had explicitly supplied, before the publication
assessor could use it — the sole verify_preflight_purity call site that
accepted a worktree argument and dropped it.

Resolve the workspace once, before preflight, and forward it. A blank or
absent path forwards None and keeps the ordinary #618 fail-closed fallback,
so guard strictness is unchanged for missing, empty, unregistered, foreign,
or control-checkout worktrees. Public tool contract, ownership, cleanliness,
hash, ancestry, and read-after-write protections from PR #814 are untouched.

Adds tests/test_issue_815_preflight_worktree_forwarding.py: a #735-style
capture proving the argument reaches verify_preflight_purity, a faithful
production reproduction (control-rooted daemon, no session lock, explicit
worktree) that clears #618 on the fixed source and is trapped at #618 on the
unpatched source, an end-to-end control-rooted publication in the real
topology (PROJECT_ROOT is the stable control checkout, the issue worktree is
a distinct registered path, production guards forced on), and negative
coverage keeping every #618 and assessor refusal intact. The prior #812
suite masked the defect by patching PROJECT_ROOT to equal the issue worktree.

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
2026-07-22 17:27:49 -05:00
sysadmin 910b6edbdc Merge pull request 'feat(mcp): publish an unpublished local commit on a registered issue worktree (Closes #812)' (#814) from feat/issue-812-publish-unpublished-commit into master 2026-07-22 16:40:54 -05:00
jcwalker3andClaude Opus 4.8 (1M context) &lt;[email protected]&gt; 99fda93bcc feat(mcp): publish an unpublished local commit on a registered issue worktree (Closes #812)
Entry point B of #812 is the state where an author's work has already advanced to a local commit: the worktree is registered, clean, on the issue branch, and carries the only copy of the implementation, but the branch has never been published. Two individually correct predicates close a cycle around it: exact-owner lease renewal refuses without an observable remote head, and every publication path is lock-derived under #618, so nothing can create that remote head without first holding the lock renewal would grant.

This adds the missing operation. gitea_publish_unpublished_issue_branch publishes an already-committed local head to its remote branch, so exact-owner renewal has the evidence its model requires. Publication is the whole of its authority: it renews, reclaims, rebinds, and clears nothing.

Why this is not a lock bypass: the operation can only publish a branch whose durable issue-lock record already names the caller as claimant. Ownership is read from the lock file, never asserted by the caller. Guard strictness is unchanged (AC15): a dirty tree, an untracked file the commit does not carry, an unregistered worktree, a non-issue or stable branch, a changed local HEAD, a remote head that is not an ancestor of the commit, a competing open PR for the same issue on another branch, and any declared-hash mismatch each fail closed. The refspec names the commit SHA explicitly and never forces.

Record separation (AC23): the durable issue-lock file and the control-plane workflow lease are distinct records. This reads the former as ownership evidence and writes neither.

Truthful process evidence (AC24): the recorded owner pid's liveness is never consulted or asserted. A regression pins the recorded pid to a live process, proves publication still succeeds, and proves expired-lock reclaim still refuses for that same pid.

Scope is AC20 only. AC21 cannot unblock on its own, so the two are separable and only the smaller one is implemented here.

Tests: 36 new cases against synthetic fixtures only, using a real git repository with a real local bare remote so publication and read-after-write verification are genuinely executed rather than mocked. AC17 is honoured and a regression asserts the protected worktree is never referenced.

Full suite: 11 failed, 4354 passed, 6 skipped, 533 subtests passed. The 11 failures are the documented pre-existing drift baseline at 9eb0f29, unchanged in count and identity.

Co-Authored-By: Claude Opus 4.8 (1M context) &lt;[email protected]&gt;
2026-07-22 16:18:00 -05:00
sysadminandClaude Opus 4.8 b2f6e9a6dc feat(webui): versioned project registry API (Closes #635)
Evolve the MVP project registry (#427) into a versioned, fail-closed
project registry API for the console (Phase 1, read-only).

- Add schema version 2 with project `status`, per-step onboarding
  `state`/`required`, optional redacted `last_seen_health`, and
  `remote_name`. Version 1 files stay loadable and are normalized with
  explicit defaults.
- Serve `/api/v1/projects` and `/api/v1/projects/{project_id}` with API
  provenance (`api_version`, `schema_version`, `source`). `/api/projects`
  is retained as an unversioned Phase 1 alias.
- Replace bare `ValueError` with `RegistryError`, carrying an operator
  `remediation` and `field_path`; invalid registries fail closed as a
  500 JSON payload or a dedicated HTML error page instead of a traceback.
- Reject credential-shaped keys before any DTO is built, reusing
  `registry_safety.is_forbidden_key` as the single source of truth
  shared with the worker registry (#798).
- Render HTML views from `project_to_dict`, so the console and the JSON
  API cannot disagree about status or onboarding progress.
- Document the contract in docs/webui-project-registry-api.md.

Tests: registry load/validate (valid, missing project, schema
validation, credential rejection) and API route coverage.

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
2026-07-22 16:12:07 -04:00
sysadmin 9eb0f29cef Merge pull request 'feat(webui): worker registry and configuration schema (Closes #798)' (#810) from feat/issue-798-worker-registry-schema into master 2026-07-22 06:48:29 -05:00
sysadmin 0b29404031 Merge pull request 'docs(webui): MCP Control Plane Web Console architecture ADR (Closes #632)' (#796) from docs/issue-632-web-console-architecture into master 2026-07-22 06:22:11 -05:00
jcwalker3andClaude Opus 4.8 e33b8d3712 docs(webui): add MCP Control Plane Web Console architecture ADR (Closes #632)
Adds the durable architecture record epic #631 lacked: layer contract,
authority boundaries, the redaction boundary, /api/v1 versioning with a
compatibility rule for the existing unversioned MVP exports, a target page
map, per-child component ownership for all twenty children (#632-#651),
phase gates, and explicit forbidden paths.

Documentation only. No UI, API, or deployment change.

- docs/architecture/webui-control-plane-console-architecture-adr.md (new)
- docs/webui-local-dev.md: cross-link to the ADR
- tests/test_webui_architecture_docs.py: doc gate for the acceptance
  criteria and the cross-link (12 cases)

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
2026-07-22 04:52:21 -05:00
19 changed files with 5195 additions and 94 deletions
+1
View File
@@ -87,6 +87,7 @@ MUTATION_TASKS = frozenset({
"edit_pr", "edit_pr",
"commit_files", "commit_files",
"gitea_commit_files", "gitea_commit_files",
"publish_unpublished_branch",
"delete_branch", "delete_branch",
"cleanup_merged_pr_branch", "cleanup_merged_pr_branch",
"cleanup_stale_claims", "cleanup_stale_claims",
+591
View File
@@ -0,0 +1,591 @@
"""Publish an unpublished local commit on a registered issue worktree (#812 AC20).
Entry point B of #812 is the state where an author's work has already advanced
to a local commit: the worktree is registered, clean, on the issue branch, and
carries the only copy of the implementation, but the branch has never been
published. That state deadlocks, because two individually correct predicates
close a cycle:
* ``issue_lock_renewal.assess_exact_owner_lease_renewal`` refuses to renew an
expired lease without an observable remote head — an unpublished branch has
none.
* Every publication path (``gitea_commit_files``, ``gitea_create_pr``) derives
its workspace from the author issue lock under #618, so nothing can create
that remote head without first holding the lock.
This module supplies the missing operation: it publishes an *already committed*
local head to the remote branch, so exact-owner renewal has the evidence it
requires. It deliberately does **not** renew, reclaim, rebind, or clear any
lock. Publication is the whole of its authority.
Why this is not a lock bypass
-----------------------------
The operation can only publish a branch whose **durable issue-lock record
already names the caller as claimant**. Ownership is read from the lock file on
disk (``issue_lock_store``), never from a caller-supplied flag, so the tool
cannot manufacture a claim it does not already hold. Nothing here weakens the
#510/#618/#713 guards: a dirty tree, an unregistered worktree, a foreign
claimant, a changed HEAD, or a divergent remote head each refuse, exactly as
they do today. The only thing this adds is the ability to make an existing,
owned, committed, clean branch observable on the remote.
Separation of records (#812 AC23)
---------------------------------
The durable **issue-lock file** and the control-plane **workflow lease** are
distinct records. This module reads the former as ownership evidence and writes
neither. Publishing changes remote git state only; no lock is renewed,
abandoned, reclaimed, or generation-bumped here.
Process evidence (#812 AC24)
----------------------------
Liveness of the lock's recorded pid is **not consulted**. That is deliberate:
the recorded pid routinely belongs to the long-running MCP daemon rather than to
an active author client, and the existing reclaim predicate
(``assess_expired_lock_reclaim``) can never be satisfied while that daemon runs.
Publication does not require the recording process to be dead, so this module
never asserts, infers, or depends on a process being dead. Ownership is proven
by identity and profile match against the recorded claimant instead.
"""
from __future__ import annotations
import hashlib
import os
import re
import subprocess
from reviewer_worktree import parse_dirty_tracked_files
from stable_branch_push_guard import is_stable_ref, redact_command
# Assessment outcomes.
PUBLISH_SANCTIONED = "publish_sanctioned"
ALREADY_PUBLISHED = "already_published"
REFUSED = "refused"
#: Implementation branches must stay traceable to their issue (#713 lineage).
ISSUE_BRANCH_RE = re.compile(r"^(fix|feat|docs|chore)/issue-(\d+)-.+$")
_SHA_RE = re.compile(r"^[0-9a-f]{40}$")
def _text(value: object) -> str:
return value.strip() if isinstance(value, str) else ""
def _realpath(value: str | None) -> str | None:
path = _text(value)
return os.path.realpath(path) if path else None
def parse_untracked_files(porcelain_status: str) -> list[str]:
"""Return untracked paths from ``git status --porcelain`` output.
``reviewer_worktree.parse_dirty_tracked_files`` deliberately skips ``??``
entries. Publication needs both halves: an untracked file in the worktree is
unpublished content that the commit does not carry, so publishing would
silently leave it behind.
"""
untracked: list[str] = []
for line in (porcelain_status or "").splitlines():
if not line.startswith("??"):
continue
path = line[2:].strip()
if path:
untracked.append(path)
return untracked
def hash_worktree_files(worktree_path: str, paths) -> dict[str, str | None]:
"""SHA-256 each path under *worktree_path*; ``None`` when unreadable."""
root = _text(worktree_path)
hashes: dict[str, str | None] = {}
for rel in paths or ():
rel_text = _text(rel)
if not rel_text:
continue
full = os.path.join(root, rel_text)
try:
with open(full, "rb") as handle:
digest = hashlib.sha256()
for chunk in iter(lambda: handle.read(65536), b""):
digest.update(chunk)
hashes[rel_text] = digest.hexdigest()
except OSError:
hashes[rel_text] = None
return hashes
def read_remote_branch_head(
worktree_path: str, remote_name: str, branch_name: str
) -> dict:
"""Observe the remote head for *branch_name*, read-only.
``probe_ok`` False means git could not answer at all. That is kept distinct
from "the branch does not exist": an unobservable remote must fail closed
rather than be mistaken for an absent branch, because the two lead to
opposite dispositions.
"""
path = _text(worktree_path)
remote = _text(remote_name)
branch = _text(branch_name)
result: dict = {
"probe_ok": False,
"remote_branch_exists": False,
"remote_head_sha": None,
"reasons": [],
}
if not (path and remote and branch):
result["reasons"].append(
"remote head probe requires a worktree path, remote name, and branch"
)
return result
try:
res = subprocess.run(
["git", "-C", path, "ls-remote", remote, f"refs/heads/{branch}"],
capture_output=True,
text=True,
check=False,
)
except OSError as exc: # git unavailable — fail closed, never assume absent
result["reasons"].append(f"remote head probe could not run: {exc}")
return result
if res.returncode != 0:
result["reasons"].append(
f"remote head probe failed for '{branch}' on remote '{remote}'"
)
return result
result["probe_ok"] = True
for line in (res.stdout or "").splitlines():
parts = line.split()
if len(parts) >= 2 and parts[1] == f"refs/heads/{branch}":
result["remote_branch_exists"] = True
result["remote_head_sha"] = parts[0].strip()
break
return result
def read_is_ancestor(
worktree_path: str, ancestor_sha: str, descendant_sha: str
) -> dict:
"""Observe whether *ancestor_sha* is an ancestor of *descendant_sha*."""
path = _text(worktree_path)
ancestor = _text(ancestor_sha)
descendant = _text(descendant_sha)
result: dict = {"probe_ok": False, "is_ancestor": False, "reasons": []}
if not (path and ancestor and descendant):
result["reasons"].append(
"ancestry probe requires a worktree path and both commit SHAs"
)
return result
try:
present = subprocess.run(
["git", "-C", path, "rev-parse", "--verify", "--quiet",
f"{ancestor}^{{commit}}"],
capture_output=True, text=True, check=False,
)
if present.returncode != 0:
result["reasons"].append(
f"remote head {ancestor} is not present locally, so it cannot be "
"proven an ancestor of the commit being published"
)
return result
res = subprocess.run(
["git", "-C", path, "merge-base", "--is-ancestor", ancestor, descendant],
capture_output=True, text=True, check=False,
)
except OSError as exc:
result["reasons"].append(f"ancestry probe could not run: {exc}")
return result
result["probe_ok"] = res.returncode in (0, 1)
result["is_ancestor"] = res.returncode == 0
return result
def assess_unpublished_commit_publication(
existing_lock,
*,
issue_number: int,
branch_name: str,
worktree_path: str,
expected_head: str,
remote: str,
org: str,
repo: str,
identity: str | None,
profile: str | None,
worktree_state,
worktree_registered: bool | None = None,
remote_probe=None,
ancestry=None,
competing_open_prs=(),
expected_file_hashes=None,
observed_file_hashes=None,
) -> dict:
"""Decide whether an unpublished local commit may be published (#812 AC20).
Pure predicate. Every input is either a caller-declared expectation that
must be *matched* against observation, or a server-side observation. No
caller-supplied boolean is accepted as proof of ownership, liveness, or
eligibility: ``existing_lock`` comes from the durable lock file and the
git/PR state is observed by the server.
The single mutating disposition it can return is "publish this exact commit
to this exact branch". It never sanctions renewal, reclamation, force
updates, history rewriting, or publication of uncommitted content.
"""
reasons: list[str] = []
branch = _text(branch_name)
head = _text(expected_head).lower()
workspace = _realpath(worktree_path)
state = worktree_state if isinstance(worktree_state, dict) else {}
lock = existing_lock if isinstance(existing_lock, dict) else None
evidence: dict = {
"issue_number": issue_number,
"branch_name": branch or None,
"worktree_path": workspace,
"expected_head": head or None,
"remote": _text(remote) or None,
"org": _text(org) or None,
"repo": _text(repo) or None,
"identity": _text(identity) or None,
"profile": _text(profile) or None,
"lock_record_present": lock is not None,
"recorded_claimant": None,
"recorded_branch": None,
"recorded_worktree": None,
"lock_generation": None,
"local_head_sha": _text(state.get("head_sha")) or None,
"current_branch": _text(state.get("current_branch")) or None,
"dirty_tracked_files": [],
"untracked_files": [],
"worktree_registered": worktree_registered,
"remote_branch_exists": None,
"remote_head_sha": None,
"fast_forward_from_remote": None,
"competing_open_prs": [],
"file_hashes_verified": None,
"hash_mismatches": [],
# Recorded explicitly so no reader mistakes silence for a liveness
# claim, and so the audit shows which records were left alone (AC23/AC24).
"owner_pid_liveness_consulted": False,
"workflow_lease_touched": False,
"issue_lock_record_mutated": False,
}
# ── declared shape ────────────────────────────────────────────────────
if not branch:
reasons.append("branch name not declared; fail closed")
if not head:
reasons.append(
"expected_head not declared; publication must name the exact commit"
)
elif not _SHA_RE.match(head):
reasons.append(
f"expected_head '{head}' is not a full 40-character commit SHA; "
"abbreviated or symbolic revisions are refused"
)
if not workspace:
reasons.append("worktree path not declared; fail closed")
if branch:
match = ISSUE_BRANCH_RE.match(branch)
if not match:
reasons.append(
f"branch '{branch}' is not an issue-linked implementation branch "
"((fix|feat|docs|chore)/issue-<number>-<description>); fail closed"
)
elif int(match.group(2)) != int(issue_number):
reasons.append(
f"branch '{branch}' does not carry issue number {issue_number}; "
"fail closed"
)
if is_stable_ref(branch):
reasons.append(
f"refusing to publish stable branch '{branch}'; this operation "
"publishes issue branches only"
)
# ── ownership: durable issue-lock record only (AC8, AC20, AC24) ───────
if lock is None:
reasons.append(
"no durable issue-lock record for this issue; publication requires an "
"existing recorded claim naming the caller, so this operation cannot "
"be used to bypass the author lock"
)
else:
lease = lock.get("work_lease")
lease = lease if isinstance(lease, dict) else {}
claimant = lease.get("claimant")
claimant = claimant if isinstance(claimant, dict) else {}
recorded_user = _text(claimant.get("username"))
recorded_profile = _text(claimant.get("profile"))
recorded_branch = _text(lock.get("branch_name")) or _text(lease.get("branch"))
recorded_worktree = _realpath(
_text(lock.get("worktree_path")) or _text(lease.get("worktree_path"))
)
evidence["recorded_claimant"] = {
"username": recorded_user or None,
"profile": recorded_profile or None,
}
evidence["recorded_branch"] = recorded_branch or None
evidence["recorded_worktree"] = recorded_worktree
try:
evidence["lock_generation"] = int(lock.get("lock_generation") or 0)
except (TypeError, ValueError):
evidence["lock_generation"] = 0
try:
recorded_issue = int(lock.get("issue_number") or 0)
except (TypeError, ValueError):
recorded_issue = 0
if recorded_issue != int(issue_number):
reasons.append(
f"durable lock records issue {lock.get('issue_number')}, not "
f"{issue_number}; ambiguous ownership, fail closed"
)
for field, declared in (
("remote", _text(remote)),
("org", _text(org)),
("repo", _text(repo)),
):
recorded = _text(lock.get(field))
if recorded and declared and recorded != declared:
reasons.append(
f"durable lock records {field} '{recorded}' but the request "
f"declares '{declared}'; repository mismatch, fail closed"
)
if recorded_branch and branch and recorded_branch != branch:
reasons.append(
f"durable lock records branch '{recorded_branch}' but the request "
f"declares '{branch}'; fail closed"
)
if recorded_worktree and workspace and recorded_worktree != workspace:
reasons.append(
f"durable lock records worktree '{recorded_worktree}' but the "
f"request declares '{workspace}'; fail closed"
)
if not recorded_user or not recorded_profile:
reasons.append(
"durable lock does not record a claimant username and profile; "
"ownership cannot be proven, fail closed"
)
else:
if recorded_user != _text(identity):
reasons.append(
f"durable lock claimant '{recorded_user}' is not the acting "
f"identity '{_text(identity) or '(unknown)'}'; foreign claim, "
"fail closed"
)
if recorded_profile != _text(profile):
reasons.append(
f"durable lock claimant profile '{recorded_profile}' is not "
f"the active profile '{_text(profile) or '(unknown)'}'; "
"fail closed"
)
# ── worktree: registered, on-branch, clean, at the expected commit ────
if worktree_registered is False:
reasons.append(
f"worktree '{workspace}' is not listed in git worktree list; #713 "
"requires a genuinely registered worktree, fail closed"
)
current_branch = _text(state.get("current_branch"))
if not current_branch:
reasons.append("worktree branch could not be observed; fail closed")
elif branch and current_branch != branch:
reasons.append(
f"worktree is on branch '{current_branch}', not '{branch}'; fail closed"
)
porcelain = state.get("porcelain_status") or ""
dirty_tracked = parse_dirty_tracked_files(porcelain)
untracked = parse_untracked_files(porcelain)
evidence["dirty_tracked_files"] = dirty_tracked
evidence["untracked_files"] = untracked
if dirty_tracked:
reasons.append(
"worktree has dirty tracked files, so the commit is not the whole of "
f"the work: {', '.join(dirty_tracked)}. This operation publishes an "
"existing clean commit only; uncommitted content is out of scope"
)
if untracked:
reasons.append(
"worktree has untracked files that the commit does not carry: "
f"{', '.join(untracked)}. Publishing would silently leave them "
"behind; fail closed"
)
local_head = _text(state.get("head_sha")).lower()
if not local_head:
reasons.append("local HEAD could not be observed; fail closed")
elif head and local_head != head:
reasons.append(
f"worktree HEAD is {local_head} but the request declares {head}; the "
"local commit changed since it was recorded, fail closed"
)
# ── remote state ──────────────────────────────────────────────────────
probe = remote_probe if isinstance(remote_probe, dict) else {}
already_published = False
if not probe.get("probe_ok"):
reasons.append(
"remote branch head could not be observed; publication must not "
"proceed against an unknown remote state, fail closed"
)
reasons.extend(probe.get("reasons") or [])
else:
remote_exists = bool(probe.get("remote_branch_exists"))
remote_head = _text(probe.get("remote_head_sha")).lower() or None
evidence["remote_branch_exists"] = remote_exists
evidence["remote_head_sha"] = remote_head
if remote_exists and remote_head and head:
if remote_head == head:
already_published = True
evidence["fast_forward_from_remote"] = True
else:
anc = ancestry if isinstance(ancestry, dict) else {}
is_anc = bool(anc.get("probe_ok")) and bool(anc.get("is_ancestor"))
evidence["fast_forward_from_remote"] = is_anc
if not is_anc:
reasons.append(
f"remote branch '{branch}' already exists at {remote_head}, "
f"which is not an ancestor of {head}; publishing would "
"discard or rewrite published history, fail closed"
)
reasons.extend(anc.get("reasons") or [])
elif remote_exists and not remote_head:
reasons.append(
f"remote branch '{branch}' exists but its head could not be read; "
"fail closed"
)
# ── competing claims ──────────────────────────────────────────────────
competing = [p for p in (competing_open_prs or ()) if p]
evidence["competing_open_prs"] = list(competing)
if competing:
reasons.append(
f"open pull request(s) {competing} already claim issue {issue_number} "
"or this branch; ambiguous ownership, fail closed"
)
# ── content verification before publication ───────────────────────────
if expected_file_hashes:
observed = (
observed_file_hashes if isinstance(observed_file_hashes, dict) else {}
)
mismatches: list[str] = []
for path, expected_digest in dict(expected_file_hashes).items():
actual = observed.get(path)
if actual is None:
mismatches.append(f"{path}: missing or unreadable in the worktree")
elif _text(actual).lower() != _text(expected_digest).lower():
mismatches.append(
f"{path}: expected {expected_digest}, observed {actual}"
)
evidence["hash_mismatches"] = mismatches
evidence["file_hashes_verified"] = not mismatches
if mismatches:
reasons.append(
"declared content hashes do not match the worktree: "
+ "; ".join(mismatches)
+ ". Refusing to publish content that is not what was recorded"
)
if reasons:
return {
"outcome": REFUSED,
"publish_sanctioned": False,
"already_published": False,
"reasons": reasons,
"evidence": evidence,
}
return {
"outcome": ALREADY_PUBLISHED if already_published else PUBLISH_SANCTIONED,
# Idempotent retry: a remote head that already equals the assessed commit
# needs no second push, so the caller verifies instead of acting.
"publish_sanctioned": not already_published,
"already_published": already_published,
"reasons": [],
"evidence": evidence,
}
def publish_commit_to_remote_branch(
*,
worktree_path: str,
remote_name: str,
branch_name: str,
expected_head: str,
) -> dict:
"""Send exactly *expected_head* to ``refs/heads/<branch_name>``.
The refspec names the commit SHA explicitly rather than ``HEAD`` or the
local branch, so what lands is the commit that was assessed and nothing
else. No force, no lease, no ``+`` prefix: a non-fast-forward is rejected by
git itself, the last of several independent guards against overwriting
published history.
"""
path = _text(worktree_path)
remote = _text(remote_name)
branch = _text(branch_name)
head = _text(expected_head)
result: dict = {
"success": False,
"pushed_ref": f"refs/heads/{branch}" if branch else None,
"pushed_sha": head or None,
"stderr": None,
"reasons": [],
}
if not (path and remote and branch and head):
result["reasons"].append(
"publication requires a worktree path, remote, branch, and commit SHA"
)
return result
refspec = f"{head}:refs/heads/{branch}"
try:
res = subprocess.run(
["git", "-C", path, "push", remote, refspec],
capture_output=True,
text=True,
check=False,
)
except OSError as exc:
result["reasons"].append(f"publication could not run: {exc}")
return result
if res.returncode != 0:
# Redact before surfacing: failures can echo credentialed remote URLs.
result["stderr"] = redact_command(res.stderr or "")
result["reasons"].append(
f"publication of {head} to '{branch}' on remote '{remote}' failed"
)
return result
result["success"] = True
return result
def verify_published_head(
*, worktree_path: str, remote_name: str, branch_name: str, expected_head: str
) -> dict:
"""Read-after-write: confirm the remote head equals *expected_head* (AC20)."""
probe = read_remote_branch_head(worktree_path, remote_name, branch_name)
head = _text(expected_head).lower()
observed = _text(probe.get("remote_head_sha")).lower() or None
verified = bool(head) and bool(probe.get("probe_ok")) and observed == head
reasons: list[str] = list(probe.get("reasons") or [])
if probe.get("probe_ok") and not verified:
reasons.append(
"read-after-write verification failed: remote head is "
f"{observed or '(absent)'}, expected {head}"
)
return {
"verified": verified,
"remote_head_sha": observed,
"expected_head": head or None,
"reasons": reasons,
}
@@ -0,0 +1,201 @@
# ADR: MCP Control Plane Web Console architecture and information architecture
- **Status:** Proposed (documentation only; blocks no code, gates every #631 child)
- **Date:** 2026-07-22
- **Tracking issue:** [#632](https://gitea.prgs.cc/Scaled-Tech-Consulting/Gitea-Tools/issues/632) — architecture and information architecture (Phase 1)
- **Parent epic:** [#631](https://gitea.prgs.cc/Scaled-Tech-Consulting/Gitea-Tools/issues/631) — MCP Control Plane Web Console
- **Foundation (closed, extend — do not recreate):** #425 tracker and children #426 skeleton, #427 projects, #428 prompts, #429 queue, #430 runtime, #431 audit paste, #432 worktrees, #433 leases, #434 gated actions, #435 auth/deployment boundary, #436 tests/CI
- **Related:** `mcp-allocator-control-plane-observability-adr.md`, `mcp-stable-control-runtime-policy-adr.md`, `control-plane-db-substrate.md`, `../safety-model.md`, `../tool-boundaries.md`, `../credential-isolation.md`, `../webui-local-dev.md`, `../webui-deployment.md`
## 1. Context
The MVP web UI shipped under `webui/` as a read-only Starlette application with ten operator routes and a JSON export beside most of them. It is a working foundation, not the console product described by epic #631, and it carries no durable architecture record: no layer contract, no authority boundary, no API versioning rule, no page map, and no statement of which phase may open a write path.
Twenty children (#632#651) hang off #631. Without one architecture document each implementer re-derives boundaries, and the most likely failure is not a bad view — it is a privileged action wired into the browser before the authorization and audit model of #633 exists.
This ADR is the single retrievable design source for the console. It decides structure only. It implements no UI, no API, and no change to deployment topology.
## 2. Decision summary (core)
| Layer | Owns | Must not |
|-------|------|----------|
| **Browser UI** | Rendering, navigation, operator affordances | Hold tokens, call Gitea/providers directly, or execute an action the server did not gate |
| **HTTP route layer** (`webui/app.py`) | Versioned routing, authentication, authorization, redaction boundary, audit emission | Contain domain logic or reach past a loader to a raw credential |
| **Domain loaders** (`webui/*_loader.py`, `*_scanner.py`, `runtime_health.py`, `project_registry.py`) | Assembling read models from authoritative sources | Mutate anything, or emit unredacted secrets across the boundary |
| **Gitea** | Durable work record: issues, PRs, comments, reviews, labels, merges | Be the concurrency lock under multi-session load |
| **Control-plane DB** | Sessions, assignment, leases, heartbeats, events | Replace Gitea history |
| **MCP tools / capability gates** | Mutation authorization | Be re-implemented, mirrored, or bypassed by console code |
| **External providers** (Sentry/GlitchTip, AI providers) | Incident and usage data | Assign work or mutate Gitea outside the #612 bridge |
**One-liner:** **Gitea records. The DB coordinates. MCP tools authorize. The console projects state and executes only capability-checked, audited actions. Providers observe.**
## 3. Console surface today versus target
`webui/app.py` currently registers these routes (see `../webui-local-dev.md` for the operator-facing table): `/`, `/health`, `/queue`, `/projects`, `/projects/{id}`, `/prompts`, `/runtime`, `/audit`, `/worktrees`, `/leases`, `/actions`, and the unversioned exports `/api/queue`, `/api/projects`, `/api/prompts`, `/api/runtime`, `/api/audit`, `/api/worktrees`, `/api/leases`, `/api/actions`, `/api/actions/{id}/preview`, `/api/actions/{id}/attempt`.
Every one of these is **retained and evolved**. No child issue may recreate a route from scratch; each states in its PR which MVP surface it extends and what it changes.
## 4. Authority boundaries
### 4.1 Gitea (durable record)
Authoritative for issue and PR identity and state, comments, reviews and verdicts, labels, merges, and branch refs. When the console and Gitea disagree about durable state, Gitea wins and the console view is refreshed — never the reverse.
### 4.2 Control-plane DB (coordination)
Authoritative for live coordination: which session holds which assignment or lease, heartbeat freshness, expiry, and the allocation event log. The console reads it; only allocator and lease tools write it.
### 4.3 MCP capability gates (authorization)
`task_capability_map.py` and `gitea_resolve_task_capability` remain the only authority that decides whether a mutation may run. The console asks; it never answers. A console action that cannot name the MCP tool it delegates to is not an action — it is a defect.
### 4.4 Filesystem and git (local state)
Issue lock files, `branches/` worktrees, and registered git worktrees are read through existing scanners. The console never deletes, rebinds, or force-clears local state outside a Phase 2 gated action.
### 4.5 Providers (observe only)
Sentry/GlitchTip and AI providers are read surfaces. The #612 incident bridge is the only path that turns an observation into Gitea work.
## 5. Request flow and the redaction boundary
```text
browser ──HTTP──> route layer ──> domain loader ──> Gitea REST
│ ├──> control-plane DB
│ ├──> filesystem / git
│ └──> providers
[redaction boundary]
audit event
```
| Stage | May hold credentials | Emits |
|-------|----------------------|-------|
| Loader → route layer | yes (server-side, via `gitea_auth`) | domain objects |
| Route layer → browser | **no** | redacted DTOs, HTML |
Two invariants govern the boundary and are non-negotiable for every child:
1. **No secrets to the browser.** Tokens, keychain identifiers, Authorization headers, raw provider endpoints, and credential-bearing URLs are redacted by default, consistent with `../safety-model.md` §3 and `../credential-isolation.md`. Serializers redact; templates do not sanitize after the fact.
2. **No ungated mutations.** A write reaches an authoritative system only by delegating to an MCP tool that passed its own capability gate. HTML forms and JSON endpoints are transport, never authority.
## 6. API naming and versioning
**Decision:** all console APIs added from Phase 1 onward are served under `/api/v1/...`.
- Nouns are plural and hierarchical: `/api/v1/inventory/leases`, `/api/v1/system/health`.
- Read endpoints are `GET` and side-effect free.
- Phase 2 action endpoints are `POST /api/v1/actions/{action_id}/preview` and `POST /api/v1/actions/{action_id}/execute`; `preview` stays side-effect free and returns a mutation ledger.
- The existing unversioned MVP exports remain as **compatibility aliases** for the whole of Phase 1 so the current operator flow never breaks. They may be retired no earlier than Phase 2, and only after the replacing `v1` route ships and `../webui-local-dev.md` records the swap.
- A breaking change to a `v1` payload requires `/api/v2/...`, not an in-place edit.
- Every JSON payload carries enough provenance for an auditor to tell where the data came from — at minimum the source system and whether the inventory was complete, matching the pagination-proof habit the MVP queue export already established.
## 7. Page map
| Page | Purpose | Owning child | Evolves |
|------|---------|--------------|---------|
| `/` | Console shell, navigation, next-safe-action summary | #638 | MVP `/` (#426) |
| `/system` | System-health dashboard | #639 | new, backed by #634 |
| `/traffic` | Workflow traffic control, queues, blockers | #640 | MVP `/queue` (#429) |
| `/runtime` | Runtime and session view | #641 | MVP `/runtime` (#430) |
| `/projects`, `/projects/{id}` | Project registry and onboarding | #635 | MVP `/projects` (#427) |
| `/inventory` | Sessions, leases, locks, worktrees in one surface | #636 | MVP `/leases` (#433) + `/worktrees` (#432) |
| `/timeline` | Workflow events and conversation timeline | #637 | new |
| `/actions` | Gated action registry, preview, execution | #642, #643, #644 | MVP `/actions` (#434) |
| `/gitea` | Issue and PR linkage console | #645 | new |
| `/policy` | Guardrail visibility, then versioned editing | #646, #647 | new |
| `/notifications` | Human-attention routing | #648 | new |
| `/observability` | Sentry/GlitchTip correlation and durable issue creation | #649 | new |
| `/providers` | AI-provider connections and insights | #650 | new |
| `/analytics` | Usage, token cost, latency, workflow performance | #651 | new |
| `/audit` | Final-report validator preview and audit log | #431 foundation, extended by #633 | MVP `/audit` (#431) |
| `/prompts`, `/prompts/{id}` | Canonical prompt library | #638 | MVP `/prompts` (#428) |
| `/health` | Liveness and deployment metadata | #634 | MVP `/health` (#435) |
## 8. Component ownership for every epic child
Each #631 child maps to at least one architectural component defined above.
| Child | Capability area | Primary component | Phase |
|-------|-----------------|-------------------|-------|
| #632 | Architecture and information architecture | this ADR | 1 |
| #633 | Authorization, RBAC, secret redaction, audit and retention | route layer + redaction boundary (§5) | 1 |
| #634 | Read-only system-health API | `/api/v1/system/health` + health loader | 1 |
| #635 | Project registry API evolution | `/api/v1/projects` + `project_registry.py` | 1 |
| #636 | Session, lease, lock, worktree inventory API | `/api/v1/inventory/*` + `lease_loader.py`, `worktree_scanner.py` | 1 |
| #637 | Workflow-event and conversation timeline model | `/api/v1/events` + control-plane DB event log | 1 |
| #638 | Application shell evolution | browser UI layer + `layout.py` | 1 |
| #639 | System-health dashboard | `/system` page over #634 | 1 |
| #640 | Workflow traffic-control view | `/traffic` page over the queue loader | 1 |
| #641 | Runtime and session view | `/runtime` page over `runtime_health.py` | 1 |
| #642 | Sanctioned restart and graceful reload controls | gated action framework, restart class | 2 |
| #643 | Requests, intent preview, authorization, workflow initiation | `/api/v1/actions/*` execute path | 2 |
| #644 | Stale-runtime recovery, worktree rebinding, reconciliation controls | gated actions over filesystem/git authority | 2 |
| #645 | Gitea issue and PR linkage console | `/gitea` page over Gitea authority | 3 |
| #646 | Workflow policy and guardrail visibility | `/policy` read view over the capability map | 3 |
| #647 | Versioned policy editing, validation, simulation, approval, rollback | `/policy` write path, gated | 3 |
| #648 | Notifications and human-attention routing | notification component over the event model | 3 |
| #649 | Sentry/GlitchTip connections, correlation, durable issue creation | provider layer + #612 incident bridge | 4 |
| #650 | AI-provider connections and operational insights | provider layer | 4 |
| #651 | Model usage, token cost, latency, workflow analytics | analytics component over the event model | 4 |
Related but **outside** this epic: #667 (restart status, impact preview, and approval controls) belongs to the #655 restart-governance umbrella and must reuse the #642 action class rather than adding a second restart surface.
## 9. Phase gates
| Phase | May ship | Entry condition |
|-------|----------|-----------------|
| **1 — read-only visibility** | `GET` pages and `GET /api/v1/...` | this ADR accepted |
| **2 — controlled actions** | gated `POST` action execution | #633 authorization, RBAC, and audit model landed |
| **3 — orchestration and policy** | linkage, policy visibility, versioned policy editing | Phase 1 inventory plus the Phase 2 action framework |
| **4 — insights** | provider correlation, analytics | evidence-backed sources from Phases 13 |
Phase 1 must not open a mutation endpoint, and the read-only guard that returns `405 read-only-mvp` stays in force until the Phase 2 entry condition is met. A phase is not entered by exception; if a control is urgent, the entry condition is what gets prioritized.
## 10. Security and workflow safety
- **Fail closed** on unknown authentication, missing RBAC mapping, or ambiguous lease ownership. An unknown state renders as blocked, never as permitted.
- **Redact by default**, per §5.
- **Every privileged action** requires a resolved capability, an explicit operator confirmation, and a durable audit event naming actor, action, target, and outcome.
- **Contamination surfaces.** Session contamination — including a manually killed MCP daemon (#630) — must be shown and must block clean claims rather than being silently repaired.
- **Deployment boundary unchanged.** Loopback by default, with the existing refusal of public binds (#435). This ADR documents that target; it does not widen it.
## 11. Forbidden paths
These are rejected designs, not preferences:
1. **Raw provider incidents as work.** The allocator never receives an unclassified Sentry/GlitchTip incident; only the #612 bridge turns an observation into a Gitea issue.
2. **Browser-held tokens.** No credential, keychain identifier, or Authorization header is ever sent to the browser or embedded in a client bundle.
3. **Process-kill recovery.** The console must not expose `pkill`, process-identifier termination, or any host process kill as a recovery affordance (#630). Restart is the sanctioned, operator-owned path of #642 and the #655 umbrella.
4. **Ungated browser mutations.** No review, approval, merge, close, or comment may originate from the browser without passing an MCP capability gate.
5. **Policy invented in the console.** The console projects policy from the capability map and canonical workflows; it never encodes a second copy.
6. **Recreating MVP scope.** Re-implementing a #426#436 surface without an explicit evolve-or-extend statement is out of bounds.
## 12. Approval checklist (readable without chat history)
A controller can accept or reject this ADR against these six points alone:
1. Layers and their owners are defined (§2) and each authority is named (§4).
2. The redaction boundary and the two invariants are stated (§5).
3. API versioning is decided, including what happens to the existing unversioned routes (§6).
4. A page map exists and names an owning child for every page (§7).
5. Every #631 child maps to at least one component and one phase (§8).
6. Phase gates and forbidden paths are explicit (§9, §11).
## 13. Open questions and follow-ups
Unresolved choices are recorded here rather than settled by implication. Each needs its own durable issue before the phase that depends on it:
- **Authentication mechanism.** Whether the console authenticates via an access proxy (Cloudflare Access or equivalent) or an application-level session is deferred to #633. This ADR requires only that it fail closed.
- **Event model substrate.** Whether the #637 timeline reads the control-plane event log directly or through a projection is deferred to #637.
- **CI path filter coverage.** `webui/ci_paths.py` triggers the web UI suite on `webui/`, `tests/test_webui_*`, and `docs/webui*`. This ADR lives under `docs/architecture/`, so editing it alone does not trigger that gate; the accompanying `tests/test_webui_architecture_docs.py` does run in the full suite. Widening the filter is a small follow-up, deliberately not bundled into a documentation-only change.
- **Retention.** Audit-event retention duration is owned by #633.
## 14. Acceptance
Accepting this ADR means:
- Phase 1 children may proceed against the layers, page map, and API rules above.
- Phase 2 children may not open a write path until #633 lands.
- Any deviation is recorded as an amendment to this file with its own issue reference, not as an undocumented divergence in code.
+1
View File
@@ -120,6 +120,7 @@ that gates each call, not which tools exist.
- `gitea_observability_list_projects` - `gitea_observability_list_projects`
- `gitea_observability_reconcile_incident` - `gitea_observability_reconcile_incident`
- `gitea_post_heartbeat` - `gitea_post_heartbeat`
- `gitea_publish_unpublished_issue_branch`
- `gitea_quarantine_contaminated_review` - `gitea_quarantine_contaminated_review`
- `gitea_reclaim_expired_workflow_lease` - `gitea_reclaim_expired_workflow_lease`
- `gitea_reconcile_already_landed_pr` - `gitea_reconcile_already_landed_pr`
+61 -2
View File
@@ -37,6 +37,16 @@ Optional environment variables:
See [webui-deployment.md](webui-deployment.md) for internal-only serving, See [webui-deployment.md](webui-deployment.md) for internal-only serving,
Cloudflare Access/WARP/VPN guidance, and unsafe bind overrides (#435). Cloudflare Access/WARP/VPN guidance, and unsafe bind overrides (#435).
See
[architecture/webui-control-plane-console-architecture-adr.md](architecture/webui-control-plane-console-architecture-adr.md)
for the console architecture: layer and authority boundaries, the redaction
boundary, `/api/v1/...` versioning, the target page map, and the phase gates
that govern when a write path may open (#632, epic #631).
See [webui-project-registry-api.md](webui-project-registry-api.md) for the
versioned project registry contract: registry schema versions 1 and 2, project
status, onboarding checklist state, and the fail-closed error payloads (#635).
## Routes (MVP) ## Routes (MVP)
| Path | Description | | Path | Description |
@@ -45,9 +55,11 @@ Cloudflare Access/WARP/VPN guidance, and unsafe bind overrides (#435).
| `/health` | JSON liveness (`status`, `service`, `mode`, `timestamp`) | | `/health` | JSON liveness (`status`, `service`, `mode`, `timestamp`) |
| `/queue` | Live PR and issue queue dashboard (#429) | | `/queue` | Live PR and issue queue dashboard (#429) |
| `/api/queue` | JSON queue export with pagination metadata | | `/api/queue` | JSON queue export with pagination metadata |
| `/projects` | Project registry list (#427) | | `/projects` | Project registry list with status and onboarding progress (#427, #635) |
| `/projects/{id}` | Project detail + onboarding checklist | | `/projects/{id}` | Project detail + onboarding checklist |
| `/api/projects` | JSON registry export | | `/api/v1/projects` | Versioned JSON registry export (#635) |
| `/api/v1/projects/{id}` | Versioned JSON project detail (#635) |
| `/api/projects` | JSON registry export — unversioned Phase 1 alias of `/api/v1/projects` |
| `/prompts` | Prompt library with per-prompt copy buttons (#428) | | `/prompts` | Prompt library with per-prompt copy buttons (#428) |
| `/api/prompts` | JSON prompt export with workflow hashes | | `/api/prompts` | JSON prompt export with workflow hashes |
| `/runtime` | MCP runtime health and stale detection (#430) | | `/runtime` | MCP runtime health and stale detection (#430) |
@@ -200,6 +212,53 @@ health, workflow/schema SHA-256 hashes, and stale-runtime warnings when the
checkout is behind merged safety-gate changes. Restart guidance links to #420; checkout is behind merged safety-gate changes. Restart guidance links to #420;
no tokens or MCP restart actions are exposed. no tokens or MCP restart actions are exposed.
## Inventory API (#636)
`GET /api/v1/inventory` returns one versioned, read-only snapshot that unifies
what the lease (#433), worktree (#432), and runtime (#430) MVP views each show
separately, so traffic-control and recovery consumers read the same source.
`GET /api/v1/inventory/{section}` returns a single section under the identical
schema (`sessions`, `leases`, `locks`, `worktrees`, `namespaces`); an unknown
section is a `404` with `error: unknown_section`. Both routes are `GET`-only.
Each section carries its own `status` (`ok` / `degraded` / `unavailable`), a
`reason` when not `ok`, and a `scan_ms`. A subsystem that cannot be read
degrades to a reasoned section; it never raises and never emits an empty list
that would read as "nothing is there".
### Field authority
Every section names where its rows came from; authorities are never blended.
| Section | Authority | Source |
|---|---|---|
| `sessions` | `control_plane_db` | #613 control-plane DB (`mode=ro`), authoritative for exclusive ownership (#600/#601) |
| `leases` | `control_plane_db` | #613 control-plane DB; degrades if the `work_items` table is absent |
| `locks` | `filesystem` | durable per-issue lock files (`issue_lock_store`) |
| `worktrees` | `filesystem` | registered git worktrees via the #432 hygiene scanner |
| `namespaces` | `filesystem` | the active profile serving this web process (others are not enumerable) |
The payload restates this map under `field_authority` for machine consumers.
### Ownership safety
`ownership_authority_complete` is true only when every ownership-bearing section
(`sessions`, `leases`, `locks`) read cleanly. While it is false, nothing is
reported as unowned and no collision is asserted from a degraded source —
absence of evidence is reported as absence of evidence, never as free work.
`collisions` surfaces detectable conflicts, each with a `kind` and `severity`:
`lock-without-worktree`, `duplicate-live-lock`, `live-lock-dead-owner` (unexpired
lease, dead pid — a #753 recovery candidate that would read as live to a naive
timestamp check), `stale-lock-dead-owner`, `expired-lock-live-owner` (the
#635/#760 daemon-pid deadlock), `concurrent-active-lease`, `active-lease-past-expiry`,
and `orphan-lease`. Collisions are emitted only from sections that read cleanly.
The control-plane DB is opened through a `mode=ro` URI so a read never creates
or migrates it; paths are collapsed against `$HOME`, URLs lose userinfo and
query strings, and credential-shaped values are redacted at the boundary. Lease
steal/release and worktree deletion are Phase 2+ and have no representation here.
## Tests ## Tests
```bash ```bash
+213
View File
@@ -0,0 +1,213 @@
# Project registry API (#635)
Phase 1 of the [console architecture ADR](architecture/webui-control-plane-console-architecture-adr.md)
gives the project registry a versioned, read-only API. This document is the
field-by-field contract for that API and for the registry file behind it.
Everything here is **read-only**. The console never writes the registry; an
operator edits the JSON file, and an invalid file fails closed rather than
rendering a partial inventory.
## Routes
| Route | Method | Description |
|-------|--------|-------------|
| `/api/v1/projects` | GET | Versioned registry export: all projects, with provenance |
| `/api/v1/projects/{project_id}` | GET | Single project; `404` with `project_not_found` when unknown |
| `/api/projects` | GET | Unversioned MVP alias (#427), retained for all of Phase 1 |
| `/projects` | GET | HTML list — status and onboarding progress per project |
| `/projects/{project_id}` | GET | HTML detail — identity, profiles, paths, checklist |
Per ADR section 6 the unversioned alias may be retired no earlier than Phase 2,
and only after this document and `webui-local-dev.md` record the swap. The alias
returns the same payload as `/api/v1/projects`, including the legacy `version`
and `source_path` keys #427 consumers already read.
The HTML views render from the same DTO the JSON routes serialize
(`project_to_dict`), so the console and the API cannot disagree about a
project's status or onboarding progress.
## Registry file
Default location: `webui/data/projects.registry.json`. Override with the
`WEBUI_PROJECT_REGISTRY` environment variable.
Schema versions: **1** and **2** are accepted; **2** is current. A version 1
file loads unchanged and is normalized with the documented defaults, so an
existing operator registry keeps working without edits.
### Root
| Field | Type | Required | Notes |
|-------|------|----------|-------|
| `version` | int | yes | `1` or `2`. Anything else fails closed |
| `projects` | array | yes | Must be non-empty |
### Project
| Field | Type | Required | Default | Notes |
|-------|------|----------|---------|-------|
| `id` | string | yes | — | Stable registry id used in URLs |
| `repo_name` | string | yes | — | Gitea repository name |
| `gitea_owner` | string | yes | — | Owning org or user |
| `remote_host` | string | yes | — | Instance base URL, no credentials |
| `remote_name` | string | no | `null` | Logical remote label, e.g. `prgs` (v2) |
| `default_branch` | string | yes | — | Stable branch name |
| `local_checkout_path` | string | yes | — | Control checkout path |
| `status` | string | no | `active` | `active`, `onboarding`, `paused`, `archived` (v2) |
| `profiles` | object | yes | — | Must map `author`, `reviewer`, `reconciler` |
| `workflow_paths` | object | yes | — | Non-empty; label to repo-relative path |
| `schema_paths` | object | no | `{}` | Label to repo-relative path |
| `onboarding_checklist` | array | no | `[]` | See below |
| `last_seen_health` | object | no | `null` | Redacted health only (v2) |
### Onboarding step
| Field | Type | Required | Default | Notes |
|-------|------|----------|---------|-------|
| `id` | string | yes | — | Stable step id |
| `title` | string | yes | — | Short operator-facing label |
| `description` | string | yes | — | Self-contained; assumes no chat history |
| `state` | string | no | `pending` | `complete`, `pending`, `blocked`, `not_applicable` (v2) |
| `required` | bool | no | `true` | Optional steps never block readiness (v2) |
### Last-seen health
| Field | Type | Required | Notes |
|-------|------|----------|-------|
| `status` | string | no (default `unknown`) | `healthy`, `degraded`, `unreachable`, `unknown` |
| `checked_at` | string | no | ISO-8601 UTC timestamp, e.g. `2026-01-01T00:00:00Z` |
| `detail` | string | no | Short redacted note |
Health is recorded metadata, not a live probe: Phase 1 performs no outbound
health checks. Endpoints, tokens, and keychain identifiers must never appear
here.
## Response shape
`GET /api/v1/projects`:
```json
{
"api_version": "v1",
"schema_version": 2,
"version": 2,
"source_path": "/path/to/webui/data/projects.registry.json",
"source": {
"kind": "file",
"path": "/path/to/webui/data/projects.registry.json",
"inventory_complete": true
},
"project_count": 1,
"projects": [
{
"id": "example",
"repo_name": "Example",
"gitea_owner": "Org",
"repo_full_name": "Org/Example",
"remote_host": "https://gitea.example.invalid",
"remote_name": "example-remote",
"default_branch": "main",
"local_checkout_path": ".",
"status": "active",
"profiles": {"author": "...", "reviewer": "...", "reconciler": "..."},
"workflow_paths": {"skill": "skills/..."},
"schema_paths": {},
"onboarding_checklist": [
{
"id": "profiles",
"title": "Configure execution profiles",
"description": "...",
"state": "complete",
"required": true
}
],
"onboarding_summary": {
"total": 1,
"complete": 1,
"pending": 0,
"blocked": 0,
"not_applicable": 0,
"required_outstanding": 0,
"onboarding_complete": true
},
"last_seen_health": null
}
]
}
```
`GET /api/v1/projects/{project_id}` returns `api_version`, `schema_version`,
`source`, and a single `project` object with the same fields.
The `source` block satisfies the ADR section 6 provenance rule: every payload
states where the data came from and whether the inventory is complete. A
file-backed registry is always complete — there is no pagination to truncate it.
`onboarding_summary` is derived, never stored. `required_outstanding` counts
steps that are `required` **and** in state `pending` or `blocked`;
`onboarding_complete` is true when that count is zero.
## Fail-closed errors
Validation failures raise `RegistryError`, which routes render instead of a
traceback.
`404` — unknown project id on `/api/v1/projects/{project_id}`:
```json
{
"error": "project_not_found",
"project_id": "not-registered",
"known_project_ids": ["example"],
"remediation": "Request one of the known project ids, or add the project ...",
"source": {"kind": "file", "path": "...", "inventory_complete": true}
}
```
`500` — invalid registry, on both the versioned route and the alias:
```json
{
"error": "registry_invalid",
"detail": "unsupported registry version: 42",
"remediation": "Set 'version' to one of 1, 2 (current schema is 2) ...",
"field_path": "version",
"source_path": "/path/to/registry.json"
}
```
`field_path` points at the offending location (`projects[0].profiles.reconciler`,
`projects[0].onboarding_checklist[2].state`, and so on). The HTML routes render
the same detail, field, source, and remediation on a "Project registry
unavailable" page.
Conditions that fail closed:
* file missing or unreadable;
* invalid JSON (the remediation names line and column);
* root not an object, or `projects` missing/empty;
* unsupported `version`;
* a credential-shaped key anywhere in the file (`token`, `*_secret`, `auth_*`, and similar);
* a project missing a required field, or missing an `author`/`reviewer`/`reconciler` profile;
* an unknown `status`, onboarding `state`, or health `status`.
## Credential rule
The registry stores redacted metadata only. Credential-shaped keys are
rejected at load time, before any DTO is built, consistent with
[safety-model.md](safety-model.md) and
[credential-isolation.md](credential-isolation.md). Tokens live in the keychain
and are resolved server-side by `gitea_auth`.
## Migrating a version 1 registry
1. Set `"version": 2`.
2. Optionally add `"status"` per project (omitted means `active`).
3. Optionally add `"remote_name"` per project.
4. Optionally add `"state"` and `"required"` to each onboarding step (omitted
means `pending` and `true`).
5. Optionally add `"last_seen_health"`.
No step is mandatory: a version 1 file keeps loading. Bumping the version only
declares that the file may use the v2 fields.
+248
View File
@@ -2024,6 +2024,7 @@ import review_quarantine # noqa: E402 # #695 contaminated formal-review quaran
import mcp_daemon_guard # noqa: E402 # #695 native transport provenance import mcp_daemon_guard # noqa: E402 # #695 native transport provenance
import already_landed_reconcile # noqa: E402 import already_landed_reconcile # noqa: E402
import author_mutation_worktree # noqa: E402 import author_mutation_worktree # noqa: E402
import branch_publish # noqa: E402 # #812 AC20 unpublished-commit publication
import root_checkout_guard # noqa: E402 import root_checkout_guard # noqa: E402
import workflow_scope_guard # noqa: E402 # #683 production scope / force-on guards import workflow_scope_guard # noqa: E402 # #683 production scope / force-on guards
import stable_branch_push_guard # noqa: E402 import stable_branch_push_guard # noqa: E402
@@ -9082,6 +9083,253 @@ def gitea_commit_files(
} }
def _publication_block(reasons: list[str], **extra) -> dict:
"""Uniform fail-closed shape for publication refusals (#812 AC20)."""
payload = {
"success": False,
"performed": False,
"published": False,
"verified": False,
"outcome": branch_publish.REFUSED,
"reasons": reasons,
# State every record this operation left alone, so a refusal can never
# be misread as a lock mutation (#812 AC23).
"issue_lock_record_mutated": False,
"workflow_lease_touched": False,
}
payload.update(extra)
return payload
@mcp.tool()
def gitea_publish_unpublished_issue_branch(
issue_number: int,
branch_name: str,
worktree_path: str,
expected_head: str,
remote: str = "dadeschools",
host: str | None = None,
org: str | None = None,
repo: str | None = None,
git_remote_name: str | None = None,
expected_file_hashes: dict | None = None,
dry_run: bool = False,
) -> dict:
"""Publish an already-committed, unpublished issue branch (#812 AC20).
Creates the remote head for a branch whose work is *already* a local commit
on a registered, clean worktree, so exact-owner lease renewal
(``issue_lock_renewal``) has the published head its evidence model requires.
This is the one step of the entry point B deadlock that no existing tool can
perform: publication is otherwise lock-derived under #618, and the lock
itself is withheld until a remote head exists.
Not a lock bypass. The branch's **durable issue-lock record must already
name the caller as claimant** ownership is read from the lock file, never
asserted by the caller so this can only publish work the caller already
owns. It refuses a dirty or untracked-carrying worktree, an unregistered
worktree, a changed local HEAD, a remote head that is not an ancestor of the
commit, a competing open PR on another branch for the same issue, and any
declared-hash mismatch. It renews, reclaims, and clears nothing: the durable
issue-lock file and the control-plane workflow lease are both left untouched
(#812 AC23), and the recorded owner pid's liveness is never consulted or
asserted (#812 AC24).
Args:
issue_number: The issue whose recorded claim authorizes publication.
branch_name: Issue branch to publish, ``(fix|feat|docs|chore)/issue-N-``.
worktree_path: Registered worktree holding the commit.
expected_head: Full 40-character SHA of the commit to publish. Required:
publication names the exact commit, and a mismatch fails closed.
remote: Known instance 'dadeschools' or 'prgs'.
host: Override the Gitea host.
org: Override the owner/organization.
repo: Override the repository name.
git_remote_name: Git remote to publish to; defaults to *remote*.
expected_file_hashes: Optional ``{path: sha256}`` verified against the
worktree before publication. Any mismatch or missing file refuses.
dry_run: Report the decision and evidence, mutate nothing.
Returns:
dict with 'success', 'performed', 'published', 'verified', 'outcome',
'remote_head_sha', 'reasons', and 'evidence'.
"""
task = "publish_unpublished_branch"
ok, block_reasons = role_session_router.check_author_mutation_after_reviewer_stop(
task
)
if not ok:
return _publication_block(block_reasons)
blocked = _namespace_mutation_block(task, remote=remote)
if blocked:
return blocked
blocked = _profile_permission_block(
task_capability_map.required_permission(task),
remote=remote, host=host, org=org, repo=repo,
org_explicit=org is not None,
repo_explicit=repo is not None,
)
if blocked:
return blocked
# #815: resolve the caller's worktree *before* preflight and forward it, so
# every workspace-resolution layer behind verify_preflight_purity — including
# the #618 branches-only guard — judges the registered issue worktree this
# publication actually operates on. Resolving it afterwards let preflight
# fall back to the MCP process root, so a daemon rooted at the stable control
# checkout refused a valid explicit worktree before the assessor ever ran.
# A caller supplying nothing usable forwards None and keeps the ordinary
# fail-closed fallback.
explicit_worktree = (worktree_path or "").strip()
workspace = os.path.realpath(os.path.abspath(explicit_worktree or "."))
verify_preflight_purity(
remote,
worktree_path=workspace if explicit_worktree else None,
task=task,
org=org,
repo=repo,
)
h, o, r = _resolve(remote, host, org, repo)
git_remote = (git_remote_name or remote or "").strip()
existing_lock = issue_lock_store.load_issue_lock(
remote=remote, org=o, repo=r, issue_number=int(issue_number)
)
git_state = issue_lock_worktree.read_worktree_git_state(workspace)
registered = author_mutation_worktree.path_in_git_worktree_list(
workspace, PROJECT_ROOT
)
remote_probe = branch_publish.read_remote_branch_head(
workspace, git_remote, branch_name
)
ancestry = None
probe_head = (remote_probe.get("remote_head_sha") or "").strip()
if probe_head and probe_head.lower() != (expected_head or "").strip().lower():
ancestry = branch_publish.read_is_ancestor(workspace, probe_head, expected_head)
# A PR on this very branch is this work's own PR, not a rival claim. Only an
# open PR for the same issue on a *different* branch is a competing claim.
competing: list = []
for pull in _list_open_pulls(h, o, r, _auth(h)):
ref = str((pull.get("head") or {}).get("ref") or "")
if not ref or ref == branch_name:
continue
if issue_lock_adoption.branch_carries_issue_marker(ref, int(issue_number)):
competing.append(pull.get("number"))
observed_hashes = None
if expected_file_hashes:
observed_hashes = branch_publish.hash_worktree_files(
workspace, list(expected_file_hashes.keys())
)
claimant = _work_lease_claimant(h)
assessment = branch_publish.assess_unpublished_commit_publication(
existing_lock,
issue_number=int(issue_number),
branch_name=branch_name,
worktree_path=workspace,
expected_head=expected_head,
remote=remote,
org=o,
repo=r,
identity=claimant.get("username"),
profile=claimant.get("profile"),
worktree_state=git_state,
worktree_registered=registered,
remote_probe=remote_probe,
ancestry=ancestry,
competing_open_prs=competing,
expected_file_hashes=expected_file_hashes,
observed_file_hashes=observed_hashes,
)
if assessment["outcome"] == branch_publish.REFUSED:
return _publication_block(
assessment["reasons"], evidence=assessment["evidence"]
)
if dry_run:
return {
"success": True,
"performed": False,
"published": False,
"verified": False,
"dry_run": True,
"outcome": assessment["outcome"],
"would_publish": assessment["publish_sanctioned"],
"remote_head_sha": assessment["evidence"].get("remote_head_sha"),
"reasons": [],
"evidence": assessment["evidence"],
"issue_lock_record_mutated": False,
"workflow_lease_touched": False,
}
performed = False
if assessment["publish_sanctioned"]:
with _audited(
task, host=h, remote=remote, org=o, repo=r,
target_branch=branch_name,
request_metadata={
"issue_number": int(issue_number),
"expected_head": expected_head,
"git_remote": git_remote,
},
):
push = branch_publish.publish_commit_to_remote_branch(
worktree_path=workspace,
remote_name=git_remote,
branch_name=branch_name,
expected_head=expected_head,
)
if not push.get("success"):
return _publication_block(
push.get("reasons") or ["publication failed"],
evidence=assessment["evidence"],
stderr=push.get("stderr"),
)
performed = True
verification = branch_publish.verify_published_head(
worktree_path=workspace,
remote_name=git_remote,
branch_name=branch_name,
expected_head=expected_head,
)
if not verification.get("verified"):
return _publication_block(
verification.get("reasons") or ["read-after-write verification failed"],
evidence=assessment["evidence"],
performed=performed,
published=performed,
remote_head_sha=verification.get("remote_head_sha"),
)
return {
"success": True,
"performed": performed,
"published": True,
"verified": True,
"outcome": assessment["outcome"],
"remote_head_sha": verification.get("remote_head_sha"),
"reasons": [],
"evidence": assessment["evidence"],
# Publication is the whole of this operation's authority (#812 AC23).
"issue_lock_record_mutated": False,
"workflow_lease_touched": False,
"exact_next_action": (
f"Remote head for '{branch_name}' is now observable. Call "
f"gitea_lock_issue(issue_number={int(issue_number)}, "
f"branch_name='{branch_name}', worktree_path='{workspace}') to renew "
"the exact-owner lease, then gitea_create_pr."
),
}
# Merge methods supported by the Gitea merge API. # Merge methods supported by the Gitea merge API.
_MERGE_METHODS = ("merge", "squash", "rebase") _MERGE_METHODS = ("merge", "squash", "rebase")
+9
View File
@@ -62,6 +62,14 @@ TASK_CAPABILITY_MAP: dict[str, dict[str, str]] = {
"permission": "gitea.branch.push", "permission": "gitea.branch.push",
"role": "author", "role": "author",
}, },
# #812 AC20: publish an already-committed, unpublished local head so
# exact-owner lease renewal has an observable remote head to reason about.
# Same authority as any other author push — deliberately not a new
# operation name, so it cannot widen an already-configured author profile.
"publish_unpublished_branch": {
"permission": "gitea.branch.push",
"role": "author",
},
"create_pr": { "create_pr": {
"permission": "gitea.pr.create", "permission": "gitea.pr.create",
"role": "author", "role": "author",
@@ -500,6 +508,7 @@ ROLE_EXCLUSIVE_TASKS: frozenset[str] = frozenset(
"gitea_release_merger_pr_lease", "gitea_release_merger_pr_lease",
"create_branch", "create_branch",
"push_branch", "push_branch",
"publish_unpublished_branch",
"create_pr", "create_pr",
"commit_files", "commit_files",
"gitea_commit_files", "gitea_commit_files",
@@ -0,0 +1,650 @@
"""Publication of an unpublished local commit (#812 AC20).
Entry point B of #812: a registered worktree, clean, on its issue branch,
holding a local commit that has never been published. Exact-owner lease renewal
refuses such a claim for want of an observable remote head, and every existing
publication path is lock-derived, so the two predicates close a cycle around
work that is otherwise complete.
These tests exercise the disposition through its *evidence*, never through any
particular issue number: every case uses an arbitrary issue number against a
synthetic repository, and the same assertions hold for any other. Nothing here
reads, writes, or references the live protected worktree named in #812 AC17 —
that content is preserved evidence for the duration of this work, so the
fixtures below build their own repositories from scratch.
The remote is a local bare repository, so publication and read-after-write
verification are genuinely executed rather than mocked.
"""
from __future__ import annotations
import os
import subprocess
import sys
import tempfile
import unittest
from datetime import datetime, timedelta, timezone
from unittest.mock import patch
sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
import branch_publish # noqa: E402
import issue_lock_provenance # noqa: E402
import issue_lock_renewal # noqa: E402
import issue_lock_store # noqa: E402
import mcp_server # noqa: E402
from mutation_profile_fixture import shared_mutation_env # noqa: E402
ISSUE = 9812
BRANCH = f"feat/issue-{ISSUE}-publish-fixture"
IDENTITY = "example-user"
PROFILE = "test-author-prgs"
ORG = "Scaled-Tech-Consulting"
REPO = "Gitea-Tools"
GIT_REMOTE = "prgs"
def _ts(hours: int) -> str:
return (
(datetime.now(timezone.utc) + timedelta(hours=hours))
.isoformat()
.replace("+00:00", "Z")
)
class _PublishBase(unittest.TestCase):
"""Real git repo + real bare remote + durable lock naming the caller.
The recorded owner pid is deliberately **this live process**. That mirrors
the production shape #812 documents, where the pid belongs to a long-running
MCP daemon rather than to a dead author client, and it proves publication
never depends on a dead process (#812 AC24).
"""
def setUp(self):
self.lock_dir = tempfile.TemporaryDirectory()
self.addCleanup(self.lock_dir.cleanup)
self.origin = tempfile.mkdtemp(prefix="issue812-origin-")
self.repo = tempfile.mkdtemp(prefix="issue812-work-")
for path in (self.origin, self.repo):
self.addCleanup(
lambda p=path: subprocess.run(["rm", "-rf", p], check=False)
)
self._init_repos()
self.remotes = patch.dict(
mcp_server.REMOTES,
{"prgs": {"host": "gitea.prgs.cc", "org": ORG, "repo": REPO}},
)
self.remotes.start()
self.addCleanup(patch.stopall)
mcp_server._IDENTITY_CACHE.clear()
# ── fixture construction ─────────────────────────────────────────────
def _git(self, *args, cwd=None):
return subprocess.run(
["git", "-C", cwd or self.repo, *args],
capture_output=True,
text=True,
check=True,
)
def _init_repos(self):
subprocess.run(
["git", "init", "-q", "--bare", "-b", "master", self.origin], check=True
)
self._git("init", "-q", "-b", "master")
self._git("config", "user.email", "[email protected]")
self._git("config", "user.name", "Test")
self._git("remote", "add", GIT_REMOTE, self.origin)
with open(os.path.join(self.repo, "seed.txt"), "w") as fh:
fh.write("seed\n")
self._git("add", "seed.txt")
self._git("commit", "-q", "-m", "seed")
self.base_sha = self._git("rev-parse", "HEAD").stdout.strip()
self._git("push", "-q", GIT_REMOTE, "master")
self._git("checkout", "-q", "-b", BRANCH)
with open(os.path.join(self.repo, "work.txt"), "w") as fh:
fh.write("unpublished implementation\n")
self._git("add", "work.txt")
self._git("commit", "-q", "-m", "unpublished implementation")
self.head_sha = self._git("rev-parse", "HEAD").stdout.strip()
self.worktree = os.path.realpath(self.repo)
def lock_path(self):
return issue_lock_store.lock_file_path(
remote="prgs", org=ORG, repo=REPO, issue_number=ISSUE,
lock_dir=self.lock_dir.name,
)
def write_lock(self, **overrides):
path = self.lock_path()
claimant = overrides.pop(
"claimant", {"username": IDENTITY, "profile": PROFILE}
)
pid = overrides.pop("session_pid", os.getpid())
lease = {
"operation_type": issue_lock_store.AUTHOR_ISSUE_WORK_LEASE,
"issue_number": ISSUE,
"pr_number": None,
"branch": overrides.get("branch_name", BRANCH),
"worktree_path": overrides.get("worktree_path", self.worktree),
"claimant": claimant,
"created_at": _ts(-2),
"last_heartbeat_at": _ts(-2),
# Expired: entry point B's lease has lapsed, which is precisely why
# renewal — and therefore a published head — is needed.
"expires_at": _ts(-1),
}
lease.update(overrides.pop("work_lease", {}))
data = {
"issue_number": ISSUE,
"branch_name": BRANCH,
"remote": "prgs",
"org": ORG,
"repo": REPO,
"worktree_path": self.worktree,
"session_pid": pid,
"pid": pid,
"lock_generation": 1,
"work_lease": lease,
"lock_provenance": issue_lock_provenance.build_sanctioned_lock_provenance(
tool="gitea_lock_issue", claimant=claimant
),
}
data.update(overrides)
data["lock_file_path"] = path
issue_lock_store.save_lock_file(path, data)
return path
def _tool_env(self):
env = shared_mutation_env(
PROFILE, include_example_repo=True,
GITEA_ISSUE_LOCK_DIR=self.lock_dir.name,
)
env["GITEA_ISSUE_LOCK_DIR"] = self.lock_dir.name
# These tests repoint PROJECT_ROOT at a synthetic repository so the
# registered-worktree proof runs for real. Pin the parity gate to the
# server's own startup head so that repointing does not read as a stale
# daemon; the gate itself stays live and enforced.
startup_head = mcp_server._STARTUP_PARITY.get("startup_head") or ""
env["GITEA_TEST_CURRENT_HEAD"] = startup_head
env["GITEA_TEST_LIVE_REMOTE_HEAD"] = startup_head
return env
# ── tool driver ──────────────────────────────────────────────────────
def run_publish(self, *, open_prs=None, expected_head=None, **kwargs):
"""Drive the public publication tool against the synthetic fixture."""
env = self._tool_env()
with patch(
"mcp_server._list_open_pulls", return_value=list(open_prs or [])
), patch(
"mcp_server._auth", return_value="token x"
), patch(
"mcp_server.get_auth_header", return_value="token x"
), patch(
"mcp_server._work_lease_claimant",
return_value={"username": IDENTITY, "profile": PROFILE},
), patch.object(
mcp_server, "PROJECT_ROOT", self.repo
), patch.dict(os.environ, env, clear=True):
os.environ["GITEA_ISSUE_LOCK_DIR"] = self.lock_dir.name
return mcp_server.gitea_publish_unpublished_issue_branch(
issue_number=kwargs.pop("issue_number", ISSUE),
branch_name=kwargs.pop("branch_name", BRANCH),
worktree_path=kwargs.pop("worktree_path", self.worktree),
expected_head=expected_head or self.head_sha,
remote="prgs",
git_remote_name=kwargs.pop("git_remote_name", GIT_REMOTE),
**kwargs,
)
def remote_head(self, branch=BRANCH):
res = subprocess.run(
["git", "-C", self.origin, "rev-parse", "--verify", "--quiet", branch],
capture_output=True, text=True, check=False,
)
return (res.stdout or "").strip() or None
class TestSuccessfulPublication(_PublishBase):
"""AC20 — the branch becomes observable and is verified after the write."""
def test_publishes_clean_unpublished_commit(self):
self.write_lock()
self.assertIsNone(self.remote_head(), "fixture must start unpublished")
result = self.run_publish()
self.assertTrue(result["success"], result.get("reasons"))
self.assertTrue(result["performed"])
self.assertTrue(result["published"])
self.assertTrue(result["verified"], "read-after-write must be proven")
self.assertEqual(result["remote_head_sha"], self.head_sha)
self.assertEqual(self.remote_head(), self.head_sha)
def test_publication_does_not_rewrite_the_commit(self):
self.write_lock()
self.run_publish()
# The published object is the same commit, not a copy or a rewrite.
self.assertEqual(self.remote_head(), self.head_sha)
self.assertEqual(
self._git("rev-parse", "HEAD").stdout.strip(), self.head_sha
)
def test_exact_next_action_names_the_lock_call(self):
self.write_lock()
result = self.run_publish()
self.assertIn("gitea_lock_issue", result["exact_next_action"])
class TestFailsClosed(_PublishBase):
"""AC20/AC9 — each refusal reason, exercised independently."""
def test_changed_local_head_refuses(self):
self.write_lock()
stale = self.base_sha # a real commit, but not the declared head
result = self.run_publish(expected_head=stale)
self.assertFalse(result["success"])
self.assertTrue(
any("local commit changed" in r for r in result["reasons"]),
result["reasons"],
)
self.assertIsNone(self.remote_head(), "refusal must not publish")
def test_abbreviated_sha_refuses(self):
self.write_lock()
result = self.run_publish(expected_head=self.head_sha[:8])
self.assertFalse(result["success"])
self.assertTrue(
any("40-character" in r for r in result["reasons"]), result["reasons"]
)
def test_dirty_tracked_worktree_refuses(self):
self.write_lock()
with open(os.path.join(self.repo, "work.txt"), "a") as fh:
fh.write("uncommitted edit\n")
result = self.run_publish()
self.assertFalse(result["success"])
self.assertTrue(
any("dirty tracked files" in r for r in result["reasons"]),
result["reasons"],
)
self.assertIn("work.txt", result["evidence"]["dirty_tracked_files"])
self.assertIsNone(self.remote_head())
def test_untracked_file_refuses(self):
self.write_lock()
with open(os.path.join(self.repo, "stray.txt"), "w") as fh:
fh.write("not committed\n")
result = self.run_publish()
self.assertFalse(result["success"])
self.assertTrue(
any("untracked files" in r for r in result["reasons"]), result["reasons"]
)
self.assertIn("stray.txt", result["evidence"]["untracked_files"])
self.assertIsNone(self.remote_head())
def test_unexpected_remote_head_refuses(self):
"""A remote head that is not an ancestor must never be overwritten."""
self.write_lock()
# Publish a divergent commit to the branch from a separate line.
self._git("checkout", "-q", "-b", "divergent", self.base_sha)
with open(os.path.join(self.repo, "other.txt"), "w") as fh:
fh.write("someone else's work\n")
self._git("add", "other.txt")
self._git("commit", "-q", "-m", "divergent")
divergent = self._git("rev-parse", "HEAD").stdout.strip()
self._git("push", "-q", GIT_REMOTE, f"{divergent}:refs/heads/{BRANCH}")
self._git("checkout", "-q", BRANCH)
result = self.run_publish()
self.assertFalse(result["success"])
self.assertTrue(
any("not an ancestor" in r for r in result["reasons"]), result["reasons"]
)
self.assertEqual(
self.remote_head(), divergent, "the other head must survive intact"
)
def test_fast_forward_remote_head_is_allowed(self):
"""An ancestor head is an honest fast-forward, not a conflict."""
self.write_lock()
self._git("push", "-q", GIT_REMOTE, f"{self.base_sha}:refs/heads/{BRANCH}")
result = self.run_publish()
self.assertTrue(result["success"], result.get("reasons"))
self.assertTrue(result["evidence"]["fast_forward_from_remote"])
self.assertEqual(self.remote_head(), self.head_sha)
def test_content_hash_mismatch_refuses(self):
self.write_lock()
wrong = {"work.txt": "0" * 64}
result = self.run_publish(expected_file_hashes=wrong)
self.assertFalse(result["success"])
self.assertTrue(
any("declared content hashes" in r for r in result["reasons"]),
result["reasons"],
)
self.assertFalse(result["evidence"]["file_hashes_verified"])
self.assertIsNone(self.remote_head())
def test_matching_content_hashes_publish(self):
self.write_lock()
digests = branch_publish.hash_worktree_files(self.worktree, ["work.txt"])
result = self.run_publish(expected_file_hashes=digests)
self.assertTrue(result["success"], result.get("reasons"))
self.assertTrue(result["evidence"]["file_hashes_verified"])
def test_missing_declared_file_refuses(self):
self.write_lock()
result = self.run_publish(expected_file_hashes={"absent.txt": "0" * 64})
self.assertFalse(result["success"])
self.assertTrue(
any("missing or unreadable" in r for r in result["reasons"]),
result["reasons"],
)
def test_foreign_claimant_refuses(self):
"""Ownership comes from the durable record, not from the caller."""
self.write_lock(claimant={"username": "someone-else", "profile": PROFILE})
result = self.run_publish()
self.assertFalse(result["success"])
self.assertTrue(
any("foreign claim" in r for r in result["reasons"]), result["reasons"]
)
self.assertIsNone(self.remote_head())
def test_foreign_profile_refuses(self):
self.write_lock(
claimant={"username": IDENTITY, "profile": "test-reviewer-prgs"}
)
result = self.run_publish()
self.assertFalse(result["success"])
self.assertTrue(
any("claimant profile" in r for r in result["reasons"]), result["reasons"]
)
def test_absent_lock_record_refuses(self):
"""No recorded claim means this cannot be used to bypass the lock."""
result = self.run_publish() # no write_lock()
self.assertFalse(result["success"])
self.assertTrue(
any("no durable issue-lock record" in r for r in result["reasons"]),
result["reasons"],
)
self.assertIsNone(self.remote_head())
def test_branch_mismatch_against_lock_refuses(self):
self.write_lock(branch_name=f"feat/issue-{ISSUE}-different")
result = self.run_publish()
self.assertFalse(result["success"])
self.assertTrue(
any("records branch" in r for r in result["reasons"]), result["reasons"]
)
def test_worktree_mismatch_against_lock_refuses(self):
self.write_lock(worktree_path="/tmp/some/other/worktree")
result = self.run_publish()
self.assertFalse(result["success"])
self.assertTrue(
any("records worktree" in r for r in result["reasons"]), result["reasons"]
)
def test_competing_open_pr_on_another_branch_refuses(self):
self.write_lock()
competing = [{"number": 4242, "head": {"ref": f"fix/issue-{ISSUE}-rival"}}]
result = self.run_publish(open_prs=competing)
self.assertFalse(result["success"])
self.assertTrue(
any("already claim issue" in r for r in result["reasons"]),
result["reasons"],
)
self.assertIsNone(self.remote_head())
def test_open_pr_on_the_same_branch_is_not_competing(self):
"""This branch's own PR is not a rival claim against itself."""
self.write_lock()
own = [{"number": 77, "head": {"ref": BRANCH}}]
result = self.run_publish(open_prs=own)
self.assertTrue(result["success"], result.get("reasons"))
class TestGuardStrictnessPreserved(_PublishBase):
"""AC15 — publication is an operation, never a weakening of the guards."""
def test_non_issue_branch_refuses(self):
self._git("checkout", "-q", "-b", "scratch/not-issue-linked")
self.write_lock(branch_name="scratch/not-issue-linked")
result = self.run_publish(branch_name="scratch/not-issue-linked")
self.assertFalse(result["success"])
self.assertTrue(
any("issue-linked" in r for r in result["reasons"]), result["reasons"]
)
def test_stable_branch_refuses(self):
self.write_lock(branch_name="master")
result = self.run_publish(branch_name="master")
self.assertFalse(result["success"])
self.assertTrue(
any("issue-linked" in r or "stable branch" in r for r in result["reasons"]),
result["reasons"],
)
def test_branch_number_must_match_the_issue(self):
other = "feat/issue-7777-mismatched"
self._git("checkout", "-q", "-b", other)
self.write_lock(branch_name=other)
result = self.run_publish(branch_name=other)
self.assertFalse(result["success"])
self.assertTrue(
any("does not carry issue number" in r for r in result["reasons"]),
result["reasons"],
)
def test_unregistered_worktree_refuses(self):
"""#713 — an improvised directory is not a registered worktree."""
path = self.write_lock()
assessment = branch_publish.assess_unpublished_commit_publication(
issue_lock_store.read_lock_file(path),
issue_number=ISSUE, branch_name=BRANCH, worktree_path=self.worktree,
expected_head=self.head_sha, remote="prgs", org=ORG, repo=REPO,
identity=IDENTITY, profile=PROFILE,
worktree_state={
"current_branch": BRANCH, "porcelain_status": "",
"head_sha": self.head_sha,
},
worktree_registered=False,
remote_probe={"probe_ok": True, "remote_branch_exists": False},
)
self.assertEqual(assessment["outcome"], branch_publish.REFUSED)
self.assertTrue(
any("not listed in git worktree list" in r
for r in assessment["reasons"]),
assessment["reasons"],
)
def test_unobservable_remote_refuses(self):
"""An unknown remote state must not be mistaken for an absent branch."""
self.write_lock()
result = self.run_publish(git_remote_name="no-such-remote")
self.assertFalse(result["success"])
self.assertTrue(
any("could not be observed" in r for r in result["reasons"]),
result["reasons"],
)
class TestRecordSeparation(_PublishBase):
"""AC23 — the durable issue lock and the workflow lease are distinct."""
def test_publication_leaves_the_issue_lock_byte_identical(self):
path = self.write_lock()
with open(path, "rb") as fh:
before = fh.read()
result = self.run_publish()
self.assertTrue(result["success"], result.get("reasons"))
with open(path, "rb") as fh:
after = fh.read()
self.assertEqual(before, after, "publication must not mutate the lock record")
self.assertFalse(result["issue_lock_record_mutated"])
self.assertFalse(result["workflow_lease_touched"])
def test_refusal_also_reports_untouched_records(self):
result = self.run_publish() # refuses: no lock record
self.assertFalse(result["issue_lock_record_mutated"])
self.assertFalse(result["workflow_lease_touched"])
def test_lock_generation_is_not_advanced(self):
path = self.write_lock()
self.run_publish()
lock = issue_lock_store.read_lock_file(path)
self.assertEqual(lock["lock_generation"], 1)
class TestTruthfulProcessEvidence(_PublishBase):
"""AC24 — a live daemon pid is never represented as a dead process."""
def test_live_recorded_pid_does_not_block_publication(self):
# The recorded pid is this live process, standing in for the live MCP
# daemon. Reclaim would refuse here; publication legitimately does not.
path = self.write_lock(session_pid=os.getpid())
lock = issue_lock_store.read_lock_file(path)
self.assertEqual(lock["pid"], os.getpid())
result = self.run_publish()
self.assertTrue(result["success"], result.get("reasons"))
self.assertEqual(self.remote_head(), self.head_sha)
def test_liveness_is_not_consulted_as_evidence(self):
self.write_lock(session_pid=os.getpid())
result = self.run_publish()
self.assertFalse(result["evidence"]["owner_pid_liveness_consulted"])
def test_reclaim_still_refuses_for_the_same_live_pid(self):
"""Publication does not soften the reclaim predicate it routes around."""
path = self.write_lock(session_pid=os.getpid())
lock = issue_lock_store.read_lock_file(path)
reclaim = issue_lock_store.assess_expired_lock_reclaim(lock)
self.assertFalse(reclaim["reclaim_allowed"])
class TestIdempotentRetry(_PublishBase):
"""AC20 — retry is safe and read-after-write is proven every time."""
def test_second_publication_reports_already_published(self):
self.write_lock()
first = self.run_publish()
self.assertTrue(first["performed"])
second = self.run_publish()
self.assertTrue(second["success"], second.get("reasons"))
self.assertFalse(second["performed"], "no second push is needed")
self.assertTrue(second["published"])
self.assertTrue(second["verified"])
self.assertEqual(second["outcome"], branch_publish.ALREADY_PUBLISHED)
self.assertEqual(self.remote_head(), self.head_sha)
class TestDryRun(_PublishBase):
"""AC12 — dry run reports the decision and mutates nothing."""
def test_dry_run_reports_intent_without_publishing(self):
self.write_lock()
result = self.run_publish(dry_run=True)
self.assertTrue(result["success"])
self.assertTrue(result["dry_run"])
self.assertTrue(result["would_publish"])
self.assertFalse(result["performed"])
self.assertIsNone(self.remote_head(), "dry run must not publish")
def test_dry_run_and_apply_agree_on_a_refusal(self):
"""AC11 — the reported decision does not depend on which mode ran."""
self.write_lock(claimant={"username": "someone-else", "profile": PROFILE})
dry = self.run_publish(dry_run=True)
applied = self.run_publish()
self.assertFalse(dry["success"])
self.assertFalse(applied["success"])
self.assertEqual(dry["reasons"], applied["reasons"])
class TestRenewalUnblocked(_PublishBase):
"""AC20/AC21 — renewal is permitted only after verified publication."""
def _renewal(self, remote_head):
return issue_lock_renewal.assess_exact_owner_lease_renewal(
issue_lock_store.read_lock_file(self.lock_path()),
issue_number=ISSUE, branch_name=BRANCH, worktree_path=self.worktree,
remote="prgs", org=ORG, repo=REPO,
identity=IDENTITY, profile=PROFILE,
current_branch=BRANCH, porcelain_status="", worktree_exists=True,
head_sha=self.head_sha, remote_head_sha=remote_head,
)
def test_renewal_refuses_before_publication(self):
self.write_lock()
decision = self._renewal(None)
self.assertFalse(decision["renewal_sanctioned"])
self.assertTrue(
any("unpublished branch" in r for r in decision["reasons"]),
decision["reasons"],
)
def test_renewal_is_sanctioned_after_publication(self):
self.write_lock()
result = self.run_publish()
self.assertTrue(result["verified"], result.get("reasons"))
decision = self._renewal(self.remote_head())
self.assertTrue(decision["renewal_sanctioned"], decision["reasons"])
class TestProtectedAssetUntouched(unittest.TestCase):
"""AC17 — no test or fixture may reference the protected worktree."""
def test_no_reference_to_the_protected_worktree(self):
here = os.path.dirname(os.path.abspath(__file__))
root = os.path.dirname(here)
needle = "issue-635-project-registry" + "-api"
for path in (
os.path.join(here, "test_issue_812_publish_unpublished_commit.py"),
os.path.join(root, "branch_publish.py"),
):
with open(path, "r", encoding="utf-8") as fh:
body = fh.read()
self.assertNotIn(needle, body)
if __name__ == "__main__": # pragma: no cover
unittest.main()
@@ -0,0 +1,622 @@
"""Publication preflight must receive the caller's worktree (#815).
``gitea_publish_unpublished_issue_branch`` takes a **required** ``worktree_path``
but resolved it only *after* ``verify_preflight_purity`` had already run. Every
workspace-resolution layer behind that preflight — canonical root, root checkout,
create-issue bootstrap, the #618 branches-only guard, issue scope, and anti-stomp
— therefore received ``None`` and fell back to the MCP process root. A daemon
rooted at the stable control checkout refused a valid registered issue worktree
that the caller had explicitly supplied, before the publication assessor ever ran.
The #812 suite could not see this. Its fixture sets ``self.worktree =
os.path.realpath(self.repo)`` and patches ``PROJECT_ROOT`` to that same path, so
the fallback resolved to the very worktree the argument named. The production
topology — control checkout on a stable branch, issue worktree somewhere else —
was never constructed, and preflight additionally no-ops under pytest unless
production guards are forced on.
These tests build that topology honestly:
* ``PROJECT_ROOT`` is a control checkout sitting on ``master``;
* the registered issue worktree is a genuinely separate path under ``branches/``;
* ``GITEA_TEST_FORCE_PRODUCTION_GUARDS`` is set so the #618 guard really runs;
* no patch makes the issue worktree appear to be ``PROJECT_ROOT``.
Nothing here reads, writes, or references the protected worktree named in #812
AC17 and #815 AC9. Every fixture is built from scratch against a local bare
remote, so publication and read-after-write verification genuinely execute.
"""
from __future__ import annotations
import os
import subprocess
import sys
import tempfile
import unittest
from datetime import datetime, timedelta, timezone
from unittest.mock import patch
sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
import issue_lock_provenance # noqa: E402
import issue_lock_store # noqa: E402
import mcp_server # noqa: E402
from mutation_profile_fixture import shared_mutation_env # noqa: E402
ISSUE = 9815
BRANCH = f"feat/issue-{ISSUE}-forwarding-fixture"
WORKTREE_DIRNAME = BRANCH.replace("/", "-")
IDENTITY = "example-user"
PROFILE = "test-author-prgs"
ORG = "Scaled-Tech-Consulting"
REPO = "Gitea-Tools"
GIT_REMOTE = "prgs"
AUTHOR_PROFILE = {
"profile_name": "prgs-author",
"role": "author",
"allowed_operations": [
"gitea.read", "gitea.issue.create", "gitea.issue.comment",
"gitea.pr.create", "gitea.repo.commit", "gitea.branch.push",
],
"forbidden_operations": [],
"audit_label": "prgs-author",
}
def _ts(hours: int) -> str:
return (
(datetime.now(timezone.utc) + timedelta(hours=hours))
.isoformat()
.replace("+00:00", "Z")
)
class TestPreflightReceivesTheWorktree(unittest.TestCase):
"""AC1 — the supplied path reaches ``verify_preflight_purity`` itself.
Follows the #735 capture pattern: replace preflight with a recorder that
raises, so the argument can be proven forwarded without performing the
mutation. This is the direct unit-level statement of the defect.
"""
def _capture_preflight(self, **kwargs):
captured: dict = {}
def _capture(*a, **kw):
captured.update(kw)
captured["_args"] = a
raise RuntimeError("capture-only")
with patch.object(
mcp_server, "verify_preflight_purity", side_effect=_capture
), patch.object(
mcp_server, "get_profile", return_value=AUTHOR_PROFILE
), patch.object(
mcp_server, "_resolve",
return_value=("gitea.prgs.cc", ORG, REPO),
), patch.object(
mcp_server, "_auth", return_value="token fake",
), patch.object(
mcp_server.role_session_router,
"check_author_mutation_after_reviewer_stop",
return_value=(True, []),
), patch.object(
mcp_server, "_namespace_mutation_block", return_value=None
), patch.object(
mcp_server, "_profile_permission_block", return_value=None
):
try:
mcp_server.gitea_publish_unpublished_issue_branch(**kwargs)
except RuntimeError as exc:
if "capture-only" not in str(exc) and not captured:
raise
self.assertTrue(
captured,
"gitea_publish_unpublished_issue_branch never called "
"verify_preflight_purity",
)
return captured
def _base_kwargs(self, **overrides):
kwargs = {
"issue_number": ISSUE,
"branch_name": BRANCH,
"worktree_path": "/tmp/issue-815-explicit-worktree",
"expected_head": "a" * 40,
"remote": "prgs",
"org": ORG,
"repo": REPO,
"git_remote_name": GIT_REMOTE,
}
kwargs.update(overrides)
return kwargs
def test_explicit_worktree_path_reaches_preflight(self):
captured = self._capture_preflight(**self._base_kwargs())
self.assertEqual(
captured.get("worktree_path"),
os.path.realpath(os.path.abspath("/tmp/issue-815-explicit-worktree")),
"the authoritative worktree_path must be forwarded into preflight",
)
def test_forwarded_path_is_the_one_publication_uses(self):
"""AC4 — preflight and publication must judge the same resolved path."""
raw = "/tmp/issue-815-explicit-worktree/./"
captured = self._capture_preflight(**self._base_kwargs(worktree_path=raw))
expected = os.path.realpath(os.path.abspath(raw.strip()))
self.assertEqual(captured.get("worktree_path"), expected)
def test_blank_worktree_path_forwards_none(self):
"""AC5/AC8 — nothing usable supplied keeps the fail-closed fallback."""
for blank in ("", " "):
with self.subTest(blank=repr(blank)):
captured = self._capture_preflight(
**self._base_kwargs(worktree_path=blank)
)
self.assertIsNone(
captured.get("worktree_path"),
"a blank worktree must not resolve to the process cwd",
)
def test_org_repo_and_task_forwarding_are_not_regressed(self):
"""AC6 — #735's org/repo forwarding and the task name still hold."""
captured = self._capture_preflight(**self._base_kwargs())
self.assertEqual(captured.get("org"), ORG)
self.assertEqual(captured.get("repo"), REPO)
self.assertEqual(captured.get("task"), "publish_unpublished_branch")
class _ProductionTopologyBase(unittest.TestCase):
"""Control checkout on master + a distinct registered issue worktree.
This is the shape the production daemon runs in and the shape the #812
fixture never built. ``PROJECT_ROOT`` is the control checkout; the issue
worktree is a real registered worktree at a different path; production
guards are forced on so the #618 branches-only guard genuinely evaluates.
"""
def setUp(self):
self.lock_dir = tempfile.TemporaryDirectory()
self.addCleanup(self.lock_dir.cleanup)
self.origin = tempfile.mkdtemp(prefix="issue815-origin-")
self.control = tempfile.mkdtemp(prefix="issue815-control-")
for path in (self.origin, self.control):
self.addCleanup(
lambda p=path: subprocess.run(["rm", "-rf", p], check=False)
)
self._init_repos()
self.remotes = patch.dict(
mcp_server.REMOTES,
{"prgs": {"host": "gitea.prgs.cc", "org": ORG, "repo": REPO}},
)
self.remotes.start()
self.addCleanup(patch.stopall)
mcp_server._IDENTITY_CACHE.clear()
def _git(self, *args, cwd=None):
return subprocess.run(
["git", "-C", cwd or self.control, *args],
capture_output=True, text=True, check=True,
)
def _init_repos(self):
subprocess.run(
["git", "init", "-q", "--bare", "-b", "master", self.origin], check=True
)
self._git("init", "-q", "-b", "master")
self._git("config", "user.email", "[email protected]")
self._git("config", "user.name", "Test")
self._git("remote", "add", GIT_REMOTE, self.origin)
with open(os.path.join(self.control, "seed.txt"), "w") as fh:
fh.write("seed\n")
# The real repository gitignores branches/, so a registered worktree
# living there does not dirty the stable control checkout. Mirror that,
# or the #615 dirty-runtime block fires on the worktree we just created.
with open(os.path.join(self.control, ".gitignore"), "w") as fh:
fh.write("branches/\n")
self._git("add", "seed.txt", ".gitignore")
self._git("commit", "-q", "-m", "seed")
self.base_sha = self._git("rev-parse", "HEAD").stdout.strip()
self._git("push", "-q", GIT_REMOTE, "master")
# The control checkout STAYS on master. This is the whole point: the
# daemon's process root is the stable control checkout, never the
# worktree the publication targets.
self.worktree = os.path.realpath(
os.path.join(self.control, "branches", WORKTREE_DIRNAME)
)
self._git("worktree", "add", "-q", "-b", BRANCH, self.worktree, "master")
with open(os.path.join(self.worktree, "work.txt"), "w") as fh:
fh.write("unpublished implementation\n")
self._git("add", "work.txt", cwd=self.worktree)
self._git("commit", "-q", "-m", "unpublished implementation", cwd=self.worktree)
self.head_sha = self._git("rev-parse", "HEAD", cwd=self.worktree).stdout.strip()
self.control_branch = self._git(
"rev-parse", "--abbrev-ref", "HEAD"
).stdout.strip()
# ── durable lock naming the caller and the issue worktree ────────────
def lock_path(self):
return issue_lock_store.lock_file_path(
remote="prgs", org=ORG, repo=REPO, issue_number=ISSUE,
lock_dir=self.lock_dir.name,
)
def write_lock(self, *, bind_session=True, **overrides):
path = self.lock_path()
claimant = overrides.pop(
"claimant", {"username": IDENTITY, "profile": PROFILE}
)
pid = overrides.pop("session_pid", os.getpid())
lease = {
"operation_type": issue_lock_store.AUTHOR_ISSUE_WORK_LEASE,
"issue_number": ISSUE,
"pr_number": None,
"branch": overrides.get("branch_name", BRANCH),
"worktree_path": overrides.get("worktree_path", self.worktree),
"claimant": claimant,
"created_at": _ts(-2),
"last_heartbeat_at": _ts(-2),
"expires_at": _ts(-1),
}
lease.update(overrides.pop("work_lease", {}))
data = {
"issue_number": ISSUE,
"branch_name": BRANCH,
"remote": "prgs",
"org": ORG,
"repo": REPO,
"worktree_path": self.worktree,
"session_pid": pid,
"pid": pid,
"lock_generation": 1,
"work_lease": lease,
"lock_provenance": issue_lock_provenance.build_sanctioned_lock_provenance(
tool="gitea_lock_issue", claimant=claimant
),
}
data.update(overrides)
data["lock_file_path"] = path
issue_lock_store.save_lock_file(path, data)
# Bind the session pointer so the #683 issue-scope guard resolves an
# owning issue for this author session. In real production the publish
# task does not require a session lock — require_author_lock is keyed on
# the test-only production_guards_forced() flag, which this suite must
# set to make preflight run at all — so this pointer is fixture
# scaffolding to clear a guard production would not apply here, never a
# softening of the worktree-forwarding behaviour under test. The
# preflight-negative cases below leave it unbound precisely so the #618
# guard is reached with no session fallback to rescue a bad worktree.
if bind_session:
pointer = {
"pid": os.getpid(),
"lock_file_path": path,
"issue_number": ISSUE,
"branch_name": data["branch_name"],
"remote": "prgs",
"org": ORG,
"repo": REPO,
}
issue_lock_store.save_lock_file(
issue_lock_store.session_pointer_path(self.lock_dir.name), pointer
)
return path
def _tool_env(self):
env = shared_mutation_env(
PROFILE, include_example_repo=True,
GITEA_ISSUE_LOCK_DIR=self.lock_dir.name,
)
env["GITEA_ISSUE_LOCK_DIR"] = self.lock_dir.name
# The defect only exists where preflight actually runs. Under pytest the
# production root/branches/scope guards are skipped unless forced on, so
# force them: this test exists to exercise the #618 guard, not to bypass
# it. Parity is pinned to the server's own startup head so repointing
# PROJECT_ROOT does not read as a stale daemon.
env["GITEA_TEST_FORCE_PRODUCTION_GUARDS"] = "1"
# Production is a promoted stable-control runtime. The pytest process
# itself runs from a branches/ worktree, which the #615 runtime-mode
# gate correctly classifies as dev-test; declaring the sanctioned mode
# models the production daemon rather than defeating the gate. Without
# this, forcing production guards on would trip the *runtime-mode* block
# for a reason unrelated to the #815 worktree-forwarding defect.
env["GITEA_MCP_RUNTIME_MODE"] = "stable-control"
startup_head = mcp_server._STARTUP_PARITY.get("startup_head") or ""
env["GITEA_TEST_CURRENT_HEAD"] = startup_head
env["GITEA_TEST_LIVE_REMOTE_HEAD"] = startup_head
return env
def run_publish(self, *, open_prs=None, expected_head=None, **kwargs):
"""Drive the public tool with PROJECT_ROOT pinned to the CONTROL checkout."""
env = self._tool_env()
with patch(
"mcp_server._list_open_pulls", return_value=list(open_prs or [])
), patch(
"mcp_server._auth", return_value="token x"
), patch(
"mcp_server.get_auth_header", return_value="token x"
), patch(
"mcp_server._work_lease_claimant",
return_value={"username": IDENTITY, "profile": PROFILE},
), patch.object(
# NOTE: the control checkout — deliberately NOT self.worktree.
mcp_server, "PROJECT_ROOT", self.control
), patch.dict(os.environ, env, clear=True):
os.environ["GITEA_ISSUE_LOCK_DIR"] = self.lock_dir.name
return mcp_server.gitea_publish_unpublished_issue_branch(
issue_number=kwargs.pop("issue_number", ISSUE),
branch_name=kwargs.pop("branch_name", BRANCH),
worktree_path=kwargs.pop("worktree_path", self.worktree),
expected_head=expected_head or self.head_sha,
remote="prgs",
git_remote_name=kwargs.pop("git_remote_name", GIT_REMOTE),
**kwargs,
)
def remote_head(self, branch=BRANCH):
res = subprocess.run(
["git", "-C", self.origin, "rev-parse", "--verify", "--quiet", branch],
capture_output=True, text=True, check=False,
)
return (res.stdout or "").strip() or None
class TestForwardingClearsThe618Guard(_ProductionTopologyBase):
"""AC2 — the faithful production reproduction, and the sharpest fix proof.
The production recovery worker had **no** session issue lock — acquiring one
was the very thing the deadlock prevented — so preflight had nothing but the
explicit ``worktree_path`` argument to resolve the workspace from. This class
reproduces exactly that: no session pointer is bound, so there is no
author-lock fallback to rescue a dropped argument.
With the argument forwarded (fixed source) the #618 branches-only guard
accepts the registered issue worktree and the call advances to the next
guard. With the argument dropped (the buggy source this issue reports)
preflight falls back to ``PROJECT_ROOT`` — the stable control checkout — and
the #618 guard traps the call there. The two outcomes are told apart by the
guard that fired, on its own error text.
This test therefore *fails* against the unpatched source (the call is trapped
at #618 instead of clearing it), which is what makes it a regression rather
than a smoke test.
"""
_CONTROL_CHECKOUT_MARKERS = ("stable control checkout", "#618")
def test_explicit_worktree_clears_618_without_a_session_lock(self):
# No write_lock(): the session is deliberately unbound, as in production.
with self.assertRaises(RuntimeError) as ctx:
self.run_publish()
message = str(ctx.exception)
# The workspace guard is satisfied — the failure is the *later* scope
# guard (no owning issue), never the control-checkout refusal. If the
# argument were dropped, this call would be trapped at #618 instead.
for marker in self._CONTROL_CHECKOUT_MARKERS:
self.assertNotIn(
marker, message,
f"the explicit worktree must clear #618; got a control-checkout "
f"refusal instead: {message}",
)
self.assertIn(
"owning issue", message,
f"expected the downstream scope guard to fire, got: {message}",
)
self.assertIsNone(self.remote_head())
def test_dropped_argument_would_be_trapped_at_618(self):
# Simulate the buggy call shape directly: no session lock, and preflight
# given no worktree, exactly as the unpatched source left it. This pins
# the control-checkout refusal that the fix eliminates, so the pair of
# tests brackets the defect from both sides regardless of which source
# version is loaded.
env = self._tool_env()
with patch(
"mcp_server._list_open_pulls", return_value=[]
), patch(
"mcp_server._auth", return_value="token x"
), patch(
"mcp_server.get_auth_header", return_value="token x"
), patch(
"mcp_server._work_lease_claimant",
return_value={"username": IDENTITY, "profile": PROFILE},
), patch.object(
mcp_server, "PROJECT_ROOT", self.control
), patch.dict(os.environ, env, clear=True):
os.environ["GITEA_ISSUE_LOCK_DIR"] = self.lock_dir.name
with self.assertRaises(RuntimeError) as ctx:
# Drive verify_preflight_purity the way the buggy body did:
# no worktree_path forwarded at all.
mcp_server.verify_preflight_purity(
"prgs",
task="publish_unpublished_branch",
org=ORG,
repo=REPO,
)
message = str(ctx.exception)
self.assertTrue(
any(m in message for m in self._CONTROL_CHECKOUT_MARKERS),
f"a dropped worktree must trap at the control checkout: {message}",
)
self.assertIsNone(self.remote_head())
class TestProductionTopologyPublishes(_ProductionTopologyBase):
"""AC2/AC4/AC7 — the explicit registered worktree is what preflight validates."""
def test_fixture_is_genuinely_the_production_topology(self):
"""Guard the guard: if this drifts, the regression stops meaning anything."""
self.assertNotEqual(
os.path.realpath(self.control), self.worktree,
"the issue worktree must not be PROJECT_ROOT",
)
self.assertEqual(
self.control_branch, "master",
"the control checkout must sit on a stable branch",
)
self.assertTrue(
os.path.realpath(self.worktree).startswith(
os.path.realpath(os.path.join(self.control, "branches")) + os.sep
),
"the issue worktree must live under branches/",
)
listed = subprocess.run(
["git", "-C", self.control, "worktree", "list"],
capture_output=True, text=True, check=True,
).stdout
self.assertIn(
self.worktree, listed, "the issue worktree must be genuinely registered"
)
def test_publishes_from_a_control_rooted_daemon(self):
"""The exact production failure: this refused with #618 before the fix."""
self.write_lock()
self.assertIsNone(self.remote_head(), "fixture must start unpublished")
res = self.run_publish()
self.assertTrue(res.get("success"), res)
self.assertTrue(res.get("performed"), res)
self.assertEqual(self.remote_head(), self.head_sha)
def test_dry_run_uses_the_explicit_worktree(self):
"""AC4 — dry-run reaches the same decision without publishing."""
self.write_lock()
res = self.run_publish(dry_run=True)
self.assertTrue(res.get("success"), res)
self.assertFalse(res.get("performed"), res)
self.assertTrue(res.get("would_publish"), res)
self.assertIsNone(self.remote_head(), "dry-run must not publish")
def test_dry_run_and_apply_agree_on_the_same_worktree(self):
"""AC4 — both paths resolve the same workspace, so both succeed."""
self.write_lock()
dry = self.run_publish(dry_run=True)
self.assertTrue(dry.get("would_publish"), dry)
applied = self.run_publish()
self.assertTrue(applied.get("performed"), applied)
self.assertEqual(self.remote_head(), self.head_sha)
def test_read_after_write_verification_still_runs(self):
"""AC6 — PR #814's post-publication verification is unchanged."""
self.write_lock()
res = self.run_publish()
self.assertTrue(res.get("verified"), res)
self.assertEqual(res.get("remote_head_sha"), self.head_sha)
class TestProductionTopologyFailsClosed(_ProductionTopologyBase):
"""AC3/AC5/AC8 — the fix does not weaken any refusal.
A refusal reaches the caller by one of two mechanisms, and this class holds
them apart deliberately. A bad *workspace* is caught by the #618 preflight
guard, which raises before the assessor is built. A bad *content/ownership*
fact passes preflight (the worktree itself is fine) and is then refused by
the publication assessor, which returns ``success: False``. Both are
fail-closed; asserting the wrong mechanism would hide a regression.
"""
# ── #618 preflight refusals: no session lock, so nothing rescues a bad
# workspace and the guard fires exactly as it does in production ──────
def _assert_preflight_raises(self, **kwargs):
with self.assertRaises(RuntimeError) as ctx:
self.run_publish(**kwargs)
self.assertIsNone(
self.remote_head(), "a blocked publication must not reach the remote"
)
return str(ctx.exception)
def test_blank_worktree_path_fails_closed_via_618(self):
"""AC5 — a blank path forwards None, so preflight sees the control root."""
for blank in ("", " "):
with self.subTest(blank=repr(blank)):
message = self._assert_preflight_raises(worktree_path=blank)
self.assertIn("618", message)
def test_control_checkout_as_worktree_fails_closed_via_618(self):
"""AC5 — naming the stable control checkout explicitly is still refused."""
message = self._assert_preflight_raises(worktree_path=self.control)
self.assertIn("618", message)
def test_unregistered_directory_fails_closed(self):
"""AC3 — a plain directory under branches/ is not a registered worktree."""
bogus = os.path.join(self.control, "branches", "not-a-worktree")
os.makedirs(bogus, exist_ok=True)
self._assert_preflight_raises(worktree_path=bogus)
def test_missing_worktree_path_fails_closed(self):
"""AC3 — a path that does not exist is refused, not silently replaced."""
missing = os.path.join(self.control, "branches", "absent-worktree")
self._assert_preflight_raises(worktree_path=missing)
# ── assessor refusals: preflight passes on a valid worktree, then the
# publication assessor refuses on content/ownership evidence ──────────
def _assert_assessor_refuses(self, **kwargs):
res = self.run_publish(**kwargs)
self.assertFalse(res.get("success"), res)
self.assertFalse(res.get("performed"), res)
self.assertIsNone(self.remote_head())
return res
def test_changed_local_head_still_refuses(self):
"""AC6 — the declared expected_head remains authoritative."""
self.write_lock()
self._assert_assessor_refuses(expected_head="b" * 40)
def test_foreign_claimant_still_refuses(self):
"""AC6 — ownership still comes from the durable lock record."""
self.write_lock(claimant={"username": "someone-else", "profile": PROFILE})
self._assert_assessor_refuses()
def test_dirty_worktree_still_refuses(self):
"""AC6 — cleanliness enforcement survives the forwarding change."""
self.write_lock()
with open(os.path.join(self.worktree, "work.txt"), "a") as fh:
fh.write("uncommitted drift\n")
self._assert_assessor_refuses()
def test_competing_open_pr_still_refuses(self):
"""AC6 — a rival claim on another branch still blocks."""
self.write_lock()
self._assert_assessor_refuses(
open_prs=[{"number": 4242, "head": {"ref": f"fix/issue-{ISSUE}-rival"}}]
)
def test_issue_lock_record_is_not_mutated_by_a_refusal(self):
"""AC6 — record separation (#812 AC23) is unaffected by this change."""
path = self.write_lock()
with open(path, "rb") as fh:
before = fh.read()
self._assert_assessor_refuses(expected_head="c" * 40)
with open(path, "rb") as fh:
self.assertEqual(before, fh.read())
class TestProtectedFixtureNotReferenced(unittest.TestCase):
"""AC9 — this regression never names the protected #635 fixture.
The forbidden tokens are reconstructed from fragments so this assertion
file does not itself contain them and produce a false positive.
"""
def test_no_reference_to_the_protected_worktree(self):
forbidden = [
"issue-635-" + "project-registry-api",
"b2f6e9a6dc40e9651ef8" + "76f322dd0a68bddebfd8",
]
here = os.path.abspath(__file__)
with open(here, "r", encoding="utf-8") as fh:
text = fh.read()
for token in forbidden:
self.assertNotIn(
token, text,
f"the protected #635 fixture must not be referenced: {token}",
)
if __name__ == "__main__":
unittest.main()
@@ -139,6 +139,9 @@ EXPECTED_ROLE_EXCLUSIVE_TASKS = frozenset(
"gitea_release_merger_pr_lease", "gitea_release_merger_pr_lease",
"create_branch", "create_branch",
"push_branch", "push_branch",
# #812 AC20: publishing an unpublished local head is author-only for the
# same reason every other push is — it writes a branch to the remote.
"publish_unpublished_branch",
"create_pr", "create_pr",
"commit_files", "commit_files",
"gitea_commit_files", "gitea_commit_files",
+149
View File
@@ -0,0 +1,149 @@
"""Documentation acceptance for the web console architecture ADR (#632 / epic #631).
Enforces the acceptance criteria of issue #632:
* AC1 — the ADR exists and covers layers, authority, phases, API versioning,
and a page map.
* AC2 — every #631 child (#632#651) maps to at least one architectural
component.
* AC3 — the closed MVP (#425#436) is stated as foundation, not recreated.
* AC4 — forbidden paths are explicit: raw provider incidents as work,
browser-held tokens, process-kill recovery.
* AC5 — a controller can approve the document without reading chat history.
Plus the linkage requirement: ``docs/webui-local-dev.md`` cross-links the ADR.
"""
from pathlib import Path
REPO_ROOT = Path(__file__).resolve().parent.parent
ADR = (
REPO_ROOT
/ "docs"
/ "architecture"
/ "webui-control-plane-console-architecture-adr.md"
)
ADR_BASENAME = "webui-control-plane-console-architecture-adr.md"
LOCAL_DEV = REPO_ROOT / "docs" / "webui-local-dev.md"
# Epic #631 children, phases 1-4 (twenty capability areas).
EPIC_CHILDREN = tuple(f"#{number}" for number in range(632, 652))
def _read(path: Path) -> str:
assert path.is_file(), f"missing {path.relative_to(REPO_ROOT)}"
return path.read_text(encoding="utf-8")
def test_ac1_adr_exists_with_required_sections():
text = _read(ADR)
lower = text.lower()
assert text.lstrip().startswith("#"), "ADR lacks a title"
assert "#631" in text and "#632" in text
for heading in (
"## 2. Decision summary",
"## 4. Authority boundaries",
"## 5. Request flow and the redaction boundary",
"## 6. API naming and versioning",
"## 7. Page map",
"## 8. Component ownership",
"## 9. Phase gates",
"## 11. Forbidden paths",
):
assert heading in text, f"ADR must contain section {heading!r}"
assert "browser ui" in lower and "domain loader" in lower
assert "control-plane db" in lower and "capability gate" in lower
def test_ac1_api_versioning_is_decided_including_legacy_routes():
text = _read(ADR)
assert "/api/v1/" in text, "ADR must decide the versioned API prefix"
assert "/api/v2/" in text, "ADR must state how breaking changes are handled"
lower = text.lower()
assert "compatibility alias" in lower, (
"ADR must say what happens to the existing unversioned MVP exports"
)
def test_ac1_page_map_covers_mvp_routes():
text = _read(ADR)
for route in ("`/`", "`/health`", "`/projects`", "`/prompts`", "`/runtime`",
"`/audit`", "`/actions`"):
assert route in text, f"page map must account for MVP route {route}"
def test_ac2_every_epic_child_maps_to_a_component():
text = _read(ADR)
ownership = text.split("## 8. Component ownership", 1)[-1].split("## 9.", 1)[0]
missing = [child for child in EPIC_CHILDREN if child not in ownership]
assert not missing, (
f"epic #631 children without an architectural component: {missing}"
)
def test_ac2_every_child_row_declares_a_phase():
text = _read(ADR)
ownership = text.split("## 8. Component ownership", 1)[-1].split("## 9.", 1)[0]
for child in EPIC_CHILDREN:
row = next(
(line for line in ownership.splitlines() if line.startswith(f"| {child} ")),
None,
)
assert row is not None, f"no ownership row for {child}"
assert row.rstrip().endswith(("| 1 |", "| 2 |", "| 3 |", "| 4 |")), (
f"ownership row for {child} must end with its phase: {row!r}"
)
def test_ac3_mvp_is_foundation_not_recreated():
text = _read(ADR)
assert "#425" in text and "#436" in text
lower = text.lower()
assert "do not recreate" in lower or "recreating mvp scope" in lower
assert "retained and evolved" in lower
def test_ac4_forbidden_paths_are_explicit():
text = _read(ADR)
forbidden = text.split("## 11. Forbidden paths", 1)[-1].split("## 12.", 1)[0]
lower = forbidden.lower()
assert "raw provider incidents" in lower and "#612" in forbidden
assert "browser-held tokens" in lower
assert "process-kill recovery" in lower and "#630" in forbidden
assert "ungated browser mutations" in lower
def test_ac5_approval_checklist_is_self_contained():
text = _read(ADR)
assert "## 12. Approval checklist" in text
checklist = text.split("## 12. Approval checklist", 1)[-1].split("## 13.", 1)[0]
for marker in ("1.", "2.", "3.", "4.", "5.", "6."):
assert marker in checklist, f"approval checklist missing item {marker}"
def test_adr_states_the_two_boundary_invariants():
text = _read(ADR)
lower = text.lower()
assert "no secrets to the browser" in lower
assert "no ungated mutations" in lower
def test_open_questions_are_recorded_not_implied():
text = _read(ADR)
assert "## 13. Open questions and follow-ups" in text
section = text.split("## 13. Open questions and follow-ups", 1)[-1]
assert "#633" in section, "deferred authorization work must name its issue"
def test_local_dev_doc_cross_links_the_adr():
text = _read(LOCAL_DEV)
assert ADR_BASENAME in text, (
"docs/webui-local-dev.md must cross-link the console architecture ADR "
"(issue #632 scope)"
)
def test_docs_do_not_embed_secrets():
for path in (ADR, LOCAL_DEV):
text = _read(path)
for marker in ("ghp_", "BEGIN PRIVATE KEY", "Authorization: Bearer"):
assert marker not in text, f"{path.name} contains {marker!r}"
+468
View File
@@ -0,0 +1,468 @@
"""Tests for the unified web-console inventory API (#636).
Covers the four cases the issue names — empty, populated, partial failure, and
the no-false-unowned invariant — plus redaction, collision detection, the
resource-split routes, and read-only guarantees against a real control-plane
database and real durable lock files.
"""
import json
import os
import sys
import tempfile
import unittest
from datetime import datetime, timedelta, timezone
from pathlib import Path
sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
from starlette.testclient import TestClient
import control_plane_db
from webui.app import create_app
from webui import inventory
def _iso(dt: datetime) -> str:
return dt.astimezone(timezone.utc).isoformat()
def _write_lock(lock_dir: str, name: str, payload: dict) -> str:
path = os.path.join(lock_dir, name)
with open(path, "w", encoding="utf-8") as handle:
json.dump(payload, handle)
return path
def _live_lock_payload(
*,
issue_number: int,
branch: str,
worktree_path: str,
pid: int,
username: str = "jcwalker3",
profile: str = "prgs-author",
) -> dict:
now = datetime.now(timezone.utc)
future = now + timedelta(hours=2)
return {
"branch_name": branch,
"issue_number": issue_number,
"org": "Scaled-Tech-Consulting",
"repo": "Gitea-Tools",
"remote": "prgs",
"pid": pid,
"session_pid": pid,
"lock_generation": 1,
"worktree_path": worktree_path,
"claimant": {"username": username, "profile": profile},
"work_lease": {
"branch": branch,
"issue_number": issue_number,
"operation_type": "author_issue_work",
"created_at": _iso(now),
"expires_at": _iso(future),
"last_heartbeat_at": _iso(now),
"claimant": {"username": username, "profile": profile},
},
}
class _FixtureMixin(unittest.TestCase):
def setUp(self) -> None:
self._tmp = tempfile.TemporaryDirectory()
self.tmp = self._tmp.name
self.lock_dir = os.path.join(self.tmp, "locks")
os.makedirs(self.lock_dir, mode=0o700)
self.db_path = os.path.join(self.tmp, "control_plane.db")
self.addCleanup(self._tmp.cleanup)
def _seed_db(self) -> control_plane_db.ControlPlaneDB:
db = control_plane_db.ControlPlaneDB(self.db_path)
db.upsert_session(
session_id="prgs-author-1",
role="author",
profile="prgs-author",
namespace="gitea-author",
pid=os.getpid(),
)
db.upsert_work_item(
remote="prgs",
org="Scaled-Tech-Consulting",
repo="Gitea-Tools",
kind="issue",
number=636,
)
db.assign_and_lease(
session_id="prgs-author-1",
role="author",
remote="prgs",
org="Scaled-Tech-Consulting",
repo="Gitea-Tools",
kind="issue",
number=636,
)
return db
class TestRedaction(unittest.TestCase):
def test_redact_path_collapses_home(self):
home = os.path.expanduser("~")
self.assertEqual(
inventory.redact_path(f"{home}/Development/Gitea-Tools"),
"~/Development/Gitea-Tools",
)
def test_redact_url_strips_userinfo_and_query(self):
self.assertEqual(
inventory.redact_url("https://user:[email protected]/api?token=abc"),
"https://gitea.prgs.cc/api",
)
def test_scrub_drops_credential_keys(self):
scrubbed = inventory.scrub(
{"token": "abc123", "api_key": "k", "profile": "prgs-author"}
)
self.assertEqual(scrubbed["token"], "[redacted]")
self.assertEqual(scrubbed["api_key"], "[redacted]")
self.assertEqual(scrubbed["profile"], "prgs-author")
def test_scrub_is_recursive_and_never_raises(self):
class Weird:
def __repr__(self) -> str:
return "weird-obj"
out = inventory.scrub({"nested": [{"password": "p", "obj": Weird()}]})
self.assertEqual(out["nested"][0]["password"], "[redacted]")
self.assertEqual(out["nested"][0]["obj"], "weird-obj")
class TestEmptyInventory(_FixtureMixin):
def test_empty_db_and_locks_degrade_without_raising(self):
# No DB file, no locks: sessions/leases unavailable, locks ok+empty.
snap = inventory.load_inventory_snapshot(
db_path=self.db_path,
lock_dir=self.lock_dir,
load_hygiene=lambda: _StubHygiene(entries=()),
)
sessions = snap.section("sessions")
leases = snap.section("leases")
locks = snap.section("locks")
self.assertEqual(sessions.status, inventory.STATUS_UNAVAILABLE)
self.assertEqual(leases.status, inventory.STATUS_UNAVAILABLE)
self.assertEqual(locks.status, inventory.STATUS_OK)
self.assertEqual(len(locks.items), 0)
# Ownership authority is incomplete because the DB is missing.
self.assertFalse(snap.ownership_authority_complete)
self.assertEqual(snap.collisions, ())
def test_empty_db_present_but_unpopulated(self):
control_plane_db.ControlPlaneDB(self.db_path) # creates schema, no rows
snap = inventory.load_inventory_snapshot(
db_path=self.db_path,
lock_dir=self.lock_dir,
load_hygiene=lambda: _StubHygiene(entries=()),
)
self.assertEqual(snap.section("sessions").status, inventory.STATUS_OK)
self.assertEqual(len(snap.section("sessions").items), 0)
self.assertEqual(snap.section("leases").status, inventory.STATUS_OK)
self.assertTrue(snap.ownership_authority_complete)
class TestPopulatedInventory(_FixtureMixin):
def test_sections_populated_and_correlated(self):
self._seed_db()
wt = f"{self.tmp}/branches/issue-636-inventory-api"
_write_lock(
self.lock_dir,
"prgs-Scaled-Tech-Consulting-Gitea-Tools-636.json",
_live_lock_payload(
issue_number=636,
branch="feat/issue-636-inventory-api",
worktree_path=wt,
pid=os.getpid(),
),
)
hygiene = _StubHygiene(
entries=(
_StubEntry(
rel_path="branches/issue-636-inventory-api",
branch="feat/issue-636-inventory-api",
classification="active-issue",
),
)
)
snap = inventory.load_inventory_snapshot(
db_path=self.db_path,
lock_dir=self.lock_dir,
load_hygiene=lambda: hygiene,
)
self.assertTrue(snap.ownership_authority_complete)
self.assertEqual(len(snap.section("sessions").items), 1)
self.assertEqual(len(snap.section("leases").items), 1)
self.assertEqual(len(snap.section("locks").items), 1)
self.assertEqual(len(snap.section("worktrees").items), 1)
# The lease, lock, and worktree for #636 correlate onto one row.
row = next(r for r in snap.correlations if r["issue_number"] == 636)
self.assertEqual(row["branch"], "feat/issue-636-inventory-api")
self.assertTrue(row["lock_live"])
self.assertEqual(row["worktree_classification"], "active-issue")
self.assertEqual(len(row["lease_ids"]), 1)
# No collision: live lock, live pid, matching worktree.
self.assertEqual(snap.collisions, ())
def test_serialized_payload_declares_field_authority(self):
self._seed_db()
snap = inventory.load_inventory_snapshot(
db_path=self.db_path,
lock_dir=self.lock_dir,
load_hygiene=lambda: _StubHygiene(entries=()),
)
payload = inventory.snapshot_to_dict(snap)
self.assertEqual(payload["api_version"], "v1")
self.assertEqual(payload["schema_version"], 1)
self.assertEqual(payload["field_authority"]["sessions"], "control_plane_db")
self.assertEqual(payload["field_authority"]["locks"], "filesystem")
self.assertIn("sessions", payload["sections"])
class TestPartialFailure(_FixtureMixin):
def test_worktree_scan_failure_degrades_only_that_section(self):
self._seed_db()
def _boom():
raise RuntimeError("git worktree list exploded")
snap = inventory.load_inventory_snapshot(
db_path=self.db_path,
lock_dir=self.lock_dir,
load_hygiene=_boom,
)
self.assertEqual(
snap.section("worktrees").status, inventory.STATUS_UNAVAILABLE
)
self.assertIn("exploded", snap.section("worktrees").reason)
# DB-backed sections still healthy.
self.assertEqual(snap.section("sessions").status, inventory.STATUS_OK)
self.assertIn("worktrees", snap.degraded_sections)
def test_degraded_ownership_suppresses_unowned_claim(self):
# DB absent → sessions/leases unavailable → ownership incomplete even
# though a lock exists and could look "unclaimed" by the DB alone.
_write_lock(
self.lock_dir,
"prgs-Scaled-Tech-Consulting-Gitea-Tools-636.json",
_live_lock_payload(
issue_number=636,
branch="feat/issue-636-inventory-api",
worktree_path=f"{self.tmp}/wt",
pid=os.getpid(),
),
)
snap = inventory.load_inventory_snapshot(
db_path=self.db_path,
lock_dir=self.lock_dir,
load_hygiene=lambda: _StubHygiene(entries=()),
)
self.assertFalse(snap.ownership_authority_complete)
payload = inventory.snapshot_to_dict(snap)
self.assertIn("may be treated as unowned", payload["ownership_note"])
class TestCollisionDetection(_FixtureMixin):
def test_live_lock_dead_owner_flagged(self):
_write_lock(
self.lock_dir,
"prgs-Scaled-Tech-Consulting-Gitea-Tools-700.json",
_live_lock_payload(
issue_number=700,
branch="feat/issue-700-x",
worktree_path=f"{self.tmp}/wt700",
pid=999_999_999, # not a running pid
),
)
control_plane_db.ControlPlaneDB(self.db_path) # empty but present
snap = inventory.load_inventory_snapshot(
db_path=self.db_path,
lock_dir=self.lock_dir,
load_hygiene=lambda: _StubHygiene(entries=()),
)
kinds = {c.kind for c in snap.collisions}
self.assertIn("live-lock-dead-owner", kinds)
# Also lock-without-worktree, since no worktree carries the branch.
self.assertIn("lock-without-worktree", kinds)
def test_duplicate_live_lock_on_same_branch(self):
for issue in (800, 801):
_write_lock(
self.lock_dir,
f"prgs-Scaled-Tech-Consulting-Gitea-Tools-{issue}.json",
_live_lock_payload(
issue_number=issue,
branch="feat/issue-800-shared",
worktree_path=f"{self.tmp}/wt{issue}",
pid=os.getpid(),
),
)
control_plane_db.ControlPlaneDB(self.db_path)
snap = inventory.load_inventory_snapshot(
db_path=self.db_path,
lock_dir=self.lock_dir,
load_hygiene=lambda: _StubHygiene(entries=()),
)
self.assertIn(
"duplicate-live-lock", {c.kind for c in snap.collisions}
)
def test_no_collision_when_sections_degraded(self):
# locks ok but worktrees unavailable → lock-without-worktree must NOT
# be asserted (a missing scan is not a missing worktree).
_write_lock(
self.lock_dir,
"prgs-Scaled-Tech-Consulting-Gitea-Tools-636.json",
_live_lock_payload(
issue_number=636,
branch="feat/issue-636-inventory-api",
worktree_path=f"{self.tmp}/wt",
pid=os.getpid(),
),
)
control_plane_db.ControlPlaneDB(self.db_path)
def _boom():
raise RuntimeError("scan down")
snap = inventory.load_inventory_snapshot(
db_path=self.db_path,
lock_dir=self.lock_dir,
load_hygiene=_boom,
)
self.assertNotIn(
"lock-without-worktree", {c.kind for c in snap.collisions}
)
class TestSectionInclude(_FixtureMixin):
def test_include_restricts_scanned_sections(self):
self._seed_db()
snap = inventory.load_inventory_snapshot(
db_path=self.db_path,
lock_dir=self.lock_dir,
include=("locks",),
)
self.assertIsNotNone(snap.section("locks"))
self.assertIsNone(snap.section("sessions"))
self.assertIsNone(snap.section("worktrees"))
class TestRoutes(_FixtureMixin):
def setUp(self) -> None:
super().setUp()
# Point the loaders at the fixture DB and lock dir via env, and stub
# the worktree scan so the route does not shell out to git.
self._prev_env = {
"GITEA_CONTROL_PLANE_DB": os.environ.get("GITEA_CONTROL_PLANE_DB"),
"GITEA_ISSUE_LOCK_DIR": os.environ.get("GITEA_ISSUE_LOCK_DIR"),
"WEBUI_TEST_OFFLINE": os.environ.get("WEBUI_TEST_OFFLINE"),
}
os.environ["GITEA_CONTROL_PLANE_DB"] = self.db_path
os.environ["GITEA_ISSUE_LOCK_DIR"] = self.lock_dir
os.environ["WEBUI_TEST_OFFLINE"] = "1"
self._seed_db()
self.client = TestClient(create_app())
def tearDown(self) -> None:
for key, value in self._prev_env.items():
if value is None:
os.environ.pop(key, None)
else:
os.environ[key] = value
def test_inventory_route_returns_versioned_payload(self):
resp = self.client.get("/api/v1/inventory")
self.assertEqual(resp.status_code, 200)
body = resp.json()
self.assertEqual(body["api_version"], "v1")
self.assertIn("sessions", body["sections"])
self.assertIn("field_authority", body)
def test_section_route_restricts_and_labels(self):
resp = self.client.get("/api/v1/inventory/locks")
self.assertEqual(resp.status_code, 200)
body = resp.json()
self.assertEqual(body["requested_section"], "locks")
self.assertIn("locks", body["sections"])
self.assertNotIn("sessions", body["sections"])
def test_unknown_section_is_404(self):
resp = self.client.get("/api/v1/inventory/bogus")
self.assertEqual(resp.status_code, 404)
self.assertEqual(resp.json()["error"], "unknown_section")
def test_inventory_route_rejects_post(self):
resp = self.client.post("/api/v1/inventory")
self.assertEqual(resp.status_code, 405)
class TestReadOnly(_FixtureMixin):
def test_snapshot_does_not_create_db_file(self):
missing = os.path.join(self.tmp, "does-not-exist.db")
inventory.load_inventory_snapshot(
db_path=missing,
lock_dir=self.lock_dir,
load_hygiene=lambda: _StubHygiene(entries=()),
)
self.assertFalse(os.path.exists(missing))
def test_readonly_connection_refuses_write(self):
self._seed_db()
conn = inventory._open_readonly(self.db_path)
try:
with self.assertRaises(Exception):
conn.execute(
"INSERT INTO sessions(session_id, role, started_at, "
"last_heartbeat_at, status) VALUES ('x','author',"
"'t','t','active')"
)
conn.commit()
finally:
conn.close()
# ── lightweight stand-ins for the #432 hygiene snapshot ──────────────────────
class _StubEntry:
def __init__(
self,
*,
rel_path: str,
branch: str | None = None,
classification: str = "stale-clean",
head_sha: str | None = "abc123",
dirty_tracked: int = 0,
dirty_untracked: bool = False,
detached: bool = False,
registered_worktree: bool = True,
notes: str = "",
) -> None:
self.rel_path = rel_path
self.folder_name = rel_path.split("/", 1)[-1]
self.branch = branch
self.classification = classification
self.head_sha = head_sha
self.dirty_tracked = dirty_tracked
self.dirty_untracked = dirty_untracked
self.detached = detached
self.registered_worktree = registered_worktree
self.notes = notes
class _StubHygiene:
def __init__(self, *, entries=(), scan_error=None) -> None:
self.entries = tuple(entries)
self.scan_error = scan_error
if __name__ == "__main__":
unittest.main()
+335 -30
View File
@@ -1,4 +1,4 @@
"""Tests for web UI project registry (#427).""" """Tests for web UI project registry (#427) and its API evolution (#635)."""
import json import json
import sys import sys
import tempfile import tempfile
@@ -11,57 +11,246 @@ from starlette.testclient import TestClient
from webui.app import create_app from webui.app import create_app
from webui.project_registry import ( from webui.project_registry import (
CURRENT_SCHEMA_VERSION,
REGISTRY_API_VERSION,
SUPPORTED_SCHEMA_VERSIONS,
RegistryError,
default_registry_path, default_registry_path,
load_registry, load_registry,
onboarding_summary,
project_to_dict, project_to_dict,
) )
from webui.registry_safety import is_forbidden_key
_REPO_ROOT = Path(__file__).resolve().parent.parent
_API_DOC = _REPO_ROOT / "docs" / "webui-project-registry-api.md"
class TestProjectRegistryLoader(unittest.TestCase): def _valid_project(**overrides):
project = {
"id": "example",
"repo_name": "Example",
"gitea_owner": "Org",
"remote_host": "https://gitea.example.invalid",
"default_branch": "main",
"local_checkout_path": ".",
"profiles": {"author": "a", "reviewer": "r", "reconciler": "c"},
"workflow_paths": {"skill": "skills/x.md"},
}
project.update(overrides)
return project
def _write_registry(payload) -> Path:
with tempfile.NamedTemporaryFile("w", suffix=".json", delete=False) as handle:
json.dump(payload, handle)
return Path(handle.name)
class RegistryFileCase(unittest.TestCase):
"""Base class that cleans up temporary registry files."""
def setUp(self):
self._temp_paths: list[Path] = []
def tearDown(self):
for path in self._temp_paths:
path.unlink(missing_ok=True)
def write_registry(self, payload) -> Path:
path = _write_registry(payload)
self._temp_paths.append(path)
return path
class TestProjectRegistryLoader(RegistryFileCase):
def test_default_registry_loads_gitea_tools(self): def test_default_registry_loads_gitea_tools(self):
registry = load_registry() registry = load_registry()
self.assertEqual(registry.version, 1) self.assertEqual(registry.version, CURRENT_SCHEMA_VERSION)
self.assertEqual(registry.schema_version, CURRENT_SCHEMA_VERSION)
self.assertEqual(registry.api_version, REGISTRY_API_VERSION)
self.assertEqual(len(registry.projects), 1) self.assertEqual(len(registry.projects), 1)
project = registry.projects[0] project = registry.projects[0]
self.assertEqual(project.id, "gitea-tools") self.assertEqual(project.id, "gitea-tools")
self.assertEqual(project.repo_name, "Gitea-Tools") self.assertEqual(project.repo_name, "Gitea-Tools")
self.assertEqual(project.gitea_owner, "Scaled-Tech-Consulting") self.assertEqual(project.gitea_owner, "Scaled-Tech-Consulting")
self.assertEqual(project.repo_full_name, "Scaled-Tech-Consulting/Gitea-Tools")
self.assertEqual(project.remote_host, "https://gitea.prgs.cc") self.assertEqual(project.remote_host, "https://gitea.prgs.cc")
self.assertEqual(project.remote_name, "prgs")
self.assertEqual(project.status, "active")
self.assertEqual(project.profiles["author"], "prgs-author") self.assertEqual(project.profiles["author"], "prgs-author")
self.assertEqual(project.profiles["reviewer"], "prgs-reviewer") self.assertEqual(project.profiles["reviewer"], "prgs-reviewer")
self.assertEqual(project.profiles["reconciler"], "prgs-reconciler") self.assertEqual(project.profiles["reconciler"], "prgs-reconciler")
self.assertIn("skill", project.workflow_paths) self.assertIn("skill", project.workflow_paths)
self.assertGreaterEqual(len(project.onboarding_checklist), 4) self.assertGreaterEqual(len(project.onboarding_checklist), 4)
def test_registry_rejects_credential_keys(self): def test_default_registry_onboarding_summary_is_complete(self):
payload = { summary = onboarding_summary(load_registry().projects[0])
self.assertEqual(summary.total, summary.complete)
self.assertEqual(summary.required_outstanding, 0)
self.assertTrue(summary.onboarding_complete)
def test_version_1_registry_still_loads_with_defaults(self):
path = self.write_registry({
"version": 1, "version": 1,
"projects": [ "projects": [
{ _valid_project(
"id": "bad", onboarding_checklist=[
"repo_name": "Bad", {"id": "step", "title": "Step", "description": "Do it"}
"gitea_owner": "Org", ]
"remote_host": "https://gitea.example.invalid", )
"default_branch": "main",
"local_checkout_path": ".",
"profiles": {
"author": "a",
"reviewer": "r",
"reconciler": "c",
},
"workflow_paths": {"skill": "skills/x.md"},
"api_token": "secret",
}
], ],
})
registry = load_registry(path)
self.assertEqual(registry.schema_version, 1)
self.assertIn(1, SUPPORTED_SCHEMA_VERSIONS)
project = registry.projects[0]
self.assertEqual(project.status, "active")
self.assertIsNone(project.remote_name)
self.assertIsNone(project.last_seen_health)
step = project.onboarding_checklist[0]
self.assertEqual(step.state, "pending")
self.assertTrue(step.required)
self.assertFalse(onboarding_summary(project).onboarding_complete)
def test_onboarding_summary_counts_states(self):
path = self.write_registry({
"version": 2,
"projects": [
_valid_project(
onboarding_checklist=[
{"id": "a", "title": "A", "description": "d", "state": "complete"},
{"id": "b", "title": "B", "description": "d", "state": "blocked"},
{
"id": "c",
"title": "C",
"description": "d",
"state": "pending",
"required": False,
},
{
"id": "d",
"title": "D",
"description": "d",
"state": "not_applicable",
},
]
)
],
})
summary = onboarding_summary(load_registry(path).projects[0])
self.assertEqual(summary.total, 4)
self.assertEqual(summary.complete, 1)
self.assertEqual(summary.blocked, 1)
self.assertEqual(summary.pending, 1)
self.assertEqual(summary.not_applicable, 1)
# Only the blocked step is both required and outstanding.
self.assertEqual(summary.required_outstanding, 1)
self.assertFalse(summary.onboarding_complete)
def test_last_seen_health_is_parsed_when_present(self):
path = self.write_registry({
"version": 2,
"projects": [
_valid_project(
last_seen_health={
"status": "degraded",
"checked_at": "2026-01-01T00:00:00Z",
"detail": "daemon restart pending",
} }
with tempfile.NamedTemporaryFile("w", suffix=".json", delete=False) as handle: )
json.dump(payload, handle) ],
path = Path(handle.name) })
try: health = load_registry(path).projects[0].last_seen_health
with self.assertRaises(ValueError): self.assertIsNotNone(health)
self.assertEqual(health.status, "degraded")
self.assertEqual(health.checked_at, "2026-01-01T00:00:00Z")
def test_registry_rejects_credential_keys(self):
path = self.write_registry({
"version": 1,
"projects": [_valid_project(id="bad", api_token="redacted-placeholder")],
})
with self.assertRaises(RegistryError) as ctx:
load_registry(path) load_registry(path)
finally: self.assertIn("credential", ctx.exception.remediation.lower())
path.unlink(missing_ok=True) self.assertEqual(ctx.exception.field_path, "projects[0].api_token")
def test_unsupported_version_fails_closed_with_remediation(self):
path = self.write_registry({"version": 99, "projects": [_valid_project()]})
with self.assertRaises(RegistryError) as ctx:
load_registry(path)
self.assertIn("unsupported registry version", ctx.exception.message)
self.assertIn(str(CURRENT_SCHEMA_VERSION), ctx.exception.remediation)
self.assertEqual(ctx.exception.field_path, "version")
def test_missing_required_field_fails_closed(self):
broken = _valid_project()
del broken["default_branch"]
path = self.write_registry({"version": 2, "projects": [broken]})
with self.assertRaises(RegistryError) as ctx:
load_registry(path)
self.assertIn("default_branch", ctx.exception.message)
self.assertEqual(ctx.exception.field_path, "projects[0]")
def test_unknown_status_fails_closed(self):
path = self.write_registry({
"version": 2,
"projects": [_valid_project(status="mystery")],
})
with self.assertRaises(RegistryError) as ctx:
load_registry(path)
self.assertEqual(ctx.exception.field_path, "projects[0].status")
self.assertIn("active", ctx.exception.remediation)
def test_unknown_onboarding_state_fails_closed(self):
path = self.write_registry({
"version": 2,
"projects": [
_valid_project(
onboarding_checklist=[
{"id": "a", "title": "A", "description": "d", "state": "almost"}
]
)
],
})
with self.assertRaises(RegistryError) as ctx:
load_registry(path)
self.assertEqual(
ctx.exception.field_path,
"projects[0].onboarding_checklist[0].state",
)
def test_missing_profile_role_fails_closed(self):
path = self.write_registry({
"version": 2,
"projects": [_valid_project(profiles={"author": "a", "reviewer": "r"})],
})
with self.assertRaises(RegistryError) as ctx:
load_registry(path)
self.assertEqual(ctx.exception.field_path, "projects[0].profiles.reconciler")
def test_empty_projects_fails_closed(self):
path = self.write_registry({"version": 2, "projects": []})
with self.assertRaises(RegistryError) as ctx:
load_registry(path)
self.assertEqual(ctx.exception.field_path, "projects")
def test_invalid_json_fails_closed_with_location(self):
with tempfile.NamedTemporaryFile("w", suffix=".json", delete=False) as handle:
handle.write("{not json")
path = Path(handle.name)
self._temp_paths.append(path)
with self.assertRaises(RegistryError) as ctx:
load_registry(path)
self.assertIn("not valid JSON", ctx.exception.message)
self.assertIn("line", ctx.exception.remediation)
def test_missing_file_fails_closed(self):
missing = Path(tempfile.gettempdir()) / "webui-registry-does-not-exist.json"
with self.assertRaises(RegistryError) as ctx:
load_registry(missing)
self.assertIn("could not be read", ctx.exception.message)
def test_default_registry_path_points_at_packaged_data(self): def test_default_registry_path_points_at_packaged_data(self):
path = default_registry_path() path = default_registry_path()
@@ -81,30 +270,146 @@ class TestProjectRegistryRoutes(unittest.TestCase):
self.assertIn("prgs-author", response.text) self.assertIn("prgs-author", response.text)
self.assertNotIn("child issue", response.text.lower()) self.assertNotIn("child issue", response.text.lower())
def test_projects_page_shows_status_and_progress(self):
response = self.client.get("/projects")
self.assertIn("Status", response.text)
self.assertIn("Onboarding", response.text)
self.assertIn("4/4 complete", response.text)
def test_project_detail_renders_checklist(self): def test_project_detail_renders_checklist(self):
response = self.client.get("/projects/gitea-tools") response = self.client.get("/projects/gitea-tools")
self.assertEqual(response.status_code, 200) self.assertEqual(response.status_code, 200)
self.assertIn("Onboarding checklist", response.text) self.assertIn("Onboarding checklist", response.text)
self.assertIn("Configure execution profiles", response.text) self.assertIn("Configure execution profiles", response.text)
self.assertIn("branches/", response.text) self.assertIn("branches/", response.text)
self.assertIn("Complete", response.text)
self.assertIn("required outstanding 0", response.text)
def test_project_detail_404(self): def test_project_detail_404(self):
response = self.client.get("/projects/unknown-repo") response = self.client.get("/projects/unknown-repo")
self.assertEqual(response.status_code, 404) self.assertEqual(response.status_code, 404)
def test_api_projects_json(self): def test_api_projects_alias_stays_compatible(self):
response = self.client.get("/api/projects") response = self.client.get("/api/projects")
self.assertEqual(response.status_code, 200) self.assertEqual(response.status_code, 200)
data = response.json() data = response.json()
self.assertEqual(data["version"], 1) # #427 consumers keep these keys.
self.assertEqual(data["version"], CURRENT_SCHEMA_VERSION)
self.assertIn("source_path", data)
self.assertEqual(len(data["projects"]), 1) self.assertEqual(len(data["projects"]), 1)
self.assertEqual(data["projects"][0]["id"], "gitea-tools") self.assertEqual(data["projects"][0]["id"], "gitea-tools")
self.assertIn("onboarding_checklist", data["projects"][0]) self.assertIn("onboarding_checklist", data["projects"][0])
def test_api_v1_projects_payload(self):
response = self.client.get("/api/v1/projects")
self.assertEqual(response.status_code, 200)
data = response.json()
self.assertEqual(data["api_version"], REGISTRY_API_VERSION)
self.assertEqual(data["schema_version"], CURRENT_SCHEMA_VERSION)
self.assertEqual(data["project_count"], 1)
self.assertEqual(data["source"]["kind"], "file")
self.assertTrue(data["source"]["inventory_complete"])
project = data["projects"][0]
self.assertEqual(project["status"], "active")
self.assertEqual(project["remote_name"], "prgs")
self.assertEqual(
project["repo_full_name"], "Scaled-Tech-Consulting/Gitea-Tools"
)
self.assertTrue(project["onboarding_summary"]["onboarding_complete"])
self.assertEqual(project["onboarding_checklist"][0]["state"], "complete")
self.assertIsNone(project["last_seen_health"])
def test_api_v1_project_detail(self):
response = self.client.get("/api/v1/projects/gitea-tools")
self.assertEqual(response.status_code, 200)
data = response.json()
self.assertEqual(data["api_version"], REGISTRY_API_VERSION)
self.assertEqual(data["project"]["id"], "gitea-tools")
self.assertEqual(data["source"]["kind"], "file")
def test_api_v1_project_detail_missing_fails_closed(self):
response = self.client.get("/api/v1/projects/not-registered")
self.assertEqual(response.status_code, 404)
data = response.json()
self.assertEqual(data["error"], "project_not_found")
self.assertEqual(data["project_id"], "not-registered")
self.assertIn("gitea-tools", data["known_project_ids"])
self.assertIn("remediation", data)
def test_api_v1_projects_is_read_only(self):
response = self.client.post("/api/v1/projects", json={})
self.assertEqual(response.status_code, 405)
self.assertEqual(response.json()["error"], "read-only-mvp")
def test_project_to_dict_is_json_safe(self): def test_project_to_dict_is_json_safe(self):
registry = load_registry() registry = load_registry()
encoded = json.dumps(project_to_dict(registry.projects[0])) dto = project_to_dict(registry.projects[0])
encoded = json.dumps(dto)
self.assertIn("gitea-tools", encoded) self.assertIn("gitea-tools", encoded)
# Prose may mention tokens; no serialized *key* may look like a secret.
for key in dto:
with self.subTest(key=key):
self.assertFalse(is_forbidden_key(key))
class TestInvalidRegistryFailsClosedOverHttp(RegistryFileCase):
def setUp(self):
super().setUp()
self.path = self.write_registry({"version": 42, "projects": []})
self.client = TestClient(create_app())
def _with_bad_registry(self, url: str):
import os
from unittest import mock
with mock.patch.dict(
os.environ, {"WEBUI_PROJECT_REGISTRY": str(self.path)}, clear=False
):
return self.client.get(url)
def test_api_v1_reports_actionable_error(self):
response = self._with_bad_registry("/api/v1/projects")
self.assertEqual(response.status_code, 500)
data = response.json()
self.assertEqual(data["error"], "registry_invalid")
self.assertIn("unsupported registry version", data["detail"])
self.assertTrue(data["remediation"])
self.assertEqual(data["field_path"], "version")
def test_unversioned_alias_reports_actionable_error(self):
response = self._with_bad_registry("/api/projects")
self.assertEqual(response.status_code, 500)
self.assertEqual(response.json()["error"], "registry_invalid")
def test_html_page_reports_actionable_error(self):
response = self._with_bad_registry("/projects")
self.assertEqual(response.status_code, 500)
self.assertIn("Project registry unavailable", response.text)
self.assertIn("Remediation", response.text)
class TestProjectRegistryApiDocs(unittest.TestCase):
def test_api_contract_is_documented(self):
self.assertTrue(_API_DOC.is_file(), f"missing {_API_DOC}")
text = _API_DOC.read_text(encoding="utf-8")
for token in (
"/api/v1/projects",
"/api/v1/projects/{project_id}",
"/api/projects",
"onboarding_summary",
"last_seen_health",
"registry_invalid",
"#635",
):
with self.subTest(token=token):
self.assertIn(token, text)
def test_route_table_lists_versioned_routes(self):
local_dev = (_REPO_ROOT / "docs" / "webui-local-dev.md").read_text(
encoding="utf-8"
)
self.assertIn("/api/v1/projects", local_dev)
self.assertIn("webui-project-registry-api.md", local_dev)
if __name__ == "__main__": if __name__ == "__main__":
+107 -5
View File
@@ -11,8 +11,20 @@ from starlette.routing import Route
from webui.deployment_boundary import deployment_snapshot from webui.deployment_boundary import deployment_snapshot
from webui.layout import render_page from webui.layout import render_page
from webui.project_registry import find_project, load_registry, registry_to_dict from webui.project_registry import (
from webui.project_views import render_project_detail, render_projects_list ProjectRegistry,
RegistryError,
find_project,
known_project_ids,
load_registry,
project_detail_to_dict,
registry_to_dict,
)
from webui.project_views import (
render_project_detail,
render_projects_list,
render_registry_error,
)
from webui.prompt_library import find_prompt, library_to_dict from webui.prompt_library import find_prompt, library_to_dict
from webui.prompt_views import render_prompt_detail, render_prompts_page from webui.prompt_views import render_prompt_detail, render_prompts_page
from final_report_validator import FINAL_REPORT_TASK_KINDS from final_report_validator import FINAL_REPORT_TASK_KINDS
@@ -29,6 +41,11 @@ from webui.worktree_scanner import load_hygiene_snapshot, snapshot_to_dict as wo
from webui.worktree_views import render_worktrees_page from webui.worktree_views import render_worktrees_page
from webui.runtime_health import load_runtime_snapshot, snapshot_to_dict as runtime_snapshot_to_dict from webui.runtime_health import load_runtime_snapshot, snapshot_to_dict as runtime_snapshot_to_dict
from webui.runtime_views import render_runtime_page from webui.runtime_views import render_runtime_page
from webui.inventory import (
SECTION_NAMES as _INVENTORY_SECTIONS,
load_inventory_snapshot,
snapshot_to_dict as inventory_snapshot_to_dict,
)
_READ_ONLY_METHODS = frozenset({"GET", "HEAD", "OPTIONS"}) _READ_ONLY_METHODS = frozenset({"GET", "HEAD", "OPTIONS"})
_AUDIT_MUTATION_PATHS = frozenset({"/audit", "/api/audit"}) _AUDIT_MUTATION_PATHS = frozenset({"/audit", "/api/audit"})
@@ -81,14 +98,26 @@ async def api_queue(_request: Request) -> JSONResponse:
return JSONResponse(queue_snapshot_to_dict(load_queue_snapshot())) return JSONResponse(queue_snapshot_to_dict(load_queue_snapshot()))
def _load_project_registry() -> tuple[ProjectRegistry | None, RegistryError | None]:
"""Load the registry, converting validation failure into a fail-closed pair."""
try:
return load_registry(), None
except RegistryError as exc:
return None, exc
async def projects(_request: Request) -> HTMLResponse: async def projects(_request: Request) -> HTMLResponse:
registry = load_registry() registry, error = _load_project_registry()
if error is not None:
return HTMLResponse(render_registry_error(error), status_code=500)
return HTMLResponse(render_projects_list(registry)) return HTMLResponse(render_projects_list(registry))
async def project_detail(request: Request) -> HTMLResponse: async def project_detail(request: Request) -> HTMLResponse:
project_id = request.path_params["project_id"] project_id = request.path_params["project_id"]
registry = load_registry() registry, error = _load_project_registry()
if error is not None:
return HTMLResponse(render_registry_error(error), status_code=500)
project = find_project(registry, project_id) project = find_project(registry, project_id)
if project is None: if project is None:
return HTMLResponse( return HTMLResponse(
@@ -106,10 +135,47 @@ async def project_detail(request: Request) -> HTMLResponse:
async def api_projects(_request: Request) -> JSONResponse: async def api_projects(_request: Request) -> JSONResponse:
registry = load_registry() """Unversioned MVP alias, retained through Phase 1 (#632 section 6)."""
registry, error = _load_project_registry()
if error is not None:
return JSONResponse(error.to_dict(), status_code=500)
return JSONResponse(registry_to_dict(registry)) return JSONResponse(registry_to_dict(registry))
async def api_v1_projects(_request: Request) -> JSONResponse:
registry, error = _load_project_registry()
if error is not None:
return JSONResponse(error.to_dict(), status_code=500)
return JSONResponse(registry_to_dict(registry))
async def api_v1_project_detail(request: Request) -> JSONResponse:
project_id = request.path_params["project_id"]
registry, error = _load_project_registry()
if error is not None:
return JSONResponse(error.to_dict(), status_code=500)
project = find_project(registry, project_id)
if project is None:
return JSONResponse(
{
"error": "project_not_found",
"project_id": project_id,
"known_project_ids": known_project_ids(registry),
"remediation": (
"Request one of the known project ids, or add the project to the "
"registry file named in 'source'."
),
"source": {
"kind": "file",
"path": str(registry.source_path),
"inventory_complete": True,
},
},
status_code=404,
)
return JSONResponse(project_detail_to_dict(registry, project))
async def prompts(_request: Request) -> HTMLResponse: async def prompts(_request: Request) -> HTMLResponse:
return HTMLResponse(render_prompts_page()) return HTMLResponse(render_prompts_page())
@@ -241,6 +307,30 @@ async def api_action_attempt(request: Request) -> JSONResponse:
return JSONResponse(result, status_code=status) return JSONResponse(result, status_code=status)
async def api_inventory(_request: Request) -> JSONResponse:
"""Unified read-only session/lease/lock/worktree inventory (#636)."""
snapshot = load_inventory_snapshot()
return JSONResponse(inventory_snapshot_to_dict(snapshot))
async def api_inventory_section(request: Request) -> JSONResponse:
"""Resource-split view: one inventory section under the shared schema."""
section = request.path_params["section"]
if section not in _INVENTORY_SECTIONS:
return JSONResponse(
{
"error": "unknown_section",
"detail": f"no inventory section named {section!r}",
"available": sorted(_INVENTORY_SECTIONS),
},
status_code=404,
)
snapshot = load_inventory_snapshot(include=(section,))
payload = inventory_snapshot_to_dict(snapshot)
payload["requested_section"] = section
return JSONResponse(payload)
async def method_not_allowed(request: Request, _exc: Exception) -> Response: async def method_not_allowed(request: Request, _exc: Exception) -> Response:
path = request.url.path path = request.url.path
if path in _AUDIT_MUTATION_PATHS and request.method == "POST": if path in _AUDIT_MUTATION_PATHS and request.method == "POST":
@@ -268,6 +358,12 @@ def create_app(*, bind_host: str | None = None) -> Starlette:
Route("/projects", projects, methods=["GET"]), Route("/projects", projects, methods=["GET"]),
Route("/projects/{project_id}", project_detail, methods=["GET"]), Route("/projects/{project_id}", project_detail, methods=["GET"]),
Route("/api/projects", api_projects, methods=["GET"]), Route("/api/projects", api_projects, methods=["GET"]),
Route("/api/v1/projects", api_v1_projects, methods=["GET"]),
Route(
"/api/v1/projects/{project_id}",
api_v1_project_detail,
methods=["GET"],
),
Route("/prompts", prompts, methods=["GET"]), Route("/prompts", prompts, methods=["GET"]),
Route("/prompts/{prompt_id}", prompt_detail, methods=["GET"]), Route("/prompts/{prompt_id}", prompt_detail, methods=["GET"]),
Route("/api/prompts", api_prompts, methods=["GET"]), Route("/api/prompts", api_prompts, methods=["GET"]),
@@ -291,6 +387,12 @@ def create_app(*, bind_host: str | None = None) -> Starlette:
methods=["POST"], methods=["POST"],
), ),
Route("/api/leases", api_leases, methods=["GET"]), Route("/api/leases", api_leases, methods=["GET"]),
Route("/api/v1/inventory", api_inventory, methods=["GET"]),
Route(
"/api/v1/inventory/{section}",
api_inventory_section,
methods=["GET"],
),
], ],
exception_handlers={405: method_not_allowed}, exception_handlers={405: method_not_allowed},
) )
+15 -5
View File
@@ -1,13 +1,15 @@
{ {
"version": 1, "version": 2,
"projects": [ "projects": [
{ {
"id": "gitea-tools", "id": "gitea-tools",
"repo_name": "Gitea-Tools", "repo_name": "Gitea-Tools",
"gitea_owner": "Scaled-Tech-Consulting", "gitea_owner": "Scaled-Tech-Consulting",
"remote_name": "prgs",
"remote_host": "https://gitea.prgs.cc", "remote_host": "https://gitea.prgs.cc",
"default_branch": "master", "default_branch": "master",
"local_checkout_path": ".", "local_checkout_path": ".",
"status": "active",
"profiles": { "profiles": {
"author": "prgs-author", "author": "prgs-author",
"reviewer": "prgs-reviewer", "reviewer": "prgs-reviewer",
@@ -26,22 +28,30 @@
{ {
"id": "profiles", "id": "profiles",
"title": "Configure execution profiles", "title": "Configure execution profiles",
"description": "Install author, reviewer, and reconciler MCP profiles (prgs-author, prgs-reviewer, prgs-reconciler) in separate namespaces. Tokens stay in keychain — never in this registry." "description": "Install author, reviewer, and reconciler MCP profiles (prgs-author, prgs-reviewer, prgs-reconciler) in separate namespaces. Tokens stay in keychain — never in this registry.",
"state": "complete",
"required": true
}, },
{ {
"id": "mcp_config", "id": "mcp_config",
"title": "Wire MCP v2 contexts", "title": "Wire MCP v2 contexts",
"description": "Copy and customize gitea-mcp.v2-contexts.example.json for your machine. Map this repo path under projects with default_owner Scaled-Tech-Consulting and default_repo Gitea-Tools." "description": "Copy and customize gitea-mcp.v2-contexts.example.json for your machine. Map this repo path under projects with default_owner Scaled-Tech-Consulting and default_repo Gitea-Tools.",
"state": "complete",
"required": true
}, },
{ {
"id": "wiki_gate", "id": "wiki_gate",
"title": "Wiki publication readiness", "title": "Wiki publication readiness",
"description": "For wiki-tracked work, satisfy the live Gitea Wiki proof gate (#224) before closing issues. See docs/wiki/Safety-and-Gates.md." "description": "For wiki-tracked work, satisfy the live Gitea Wiki proof gate (#224) before closing issues. See docs/wiki/Safety-and-Gates.md.",
"state": "complete",
"required": true
}, },
{ {
"id": "branches_layout", "id": "branches_layout",
"title": "Isolate work under branches/", "title": "Isolate work under branches/",
"description": "All LLM task edits happen in worktrees under branches/. Main checkout stays clean; use skills/llm-project-workflow templates for start-issue and review flows." "description": "All LLM task edits happen in worktrees under branches/. Main checkout stays clean; use skills/llm-project-workflow templates for start-issue and review flows.",
"state": "complete",
"required": true
} }
] ]
} }
+952
View File
@@ -0,0 +1,952 @@
"""Unified session/lease/lock/worktree inventory for the web console (#636).
Leases (#433), worktrees (#432), and runtime (#430) each ship their own MVP
view, each with its own shape and its own idea of what "owned" means. A
traffic-control or recovery operator has to read all three and correlate them
by hand, which is exactly the step that goes wrong under collision pressure.
This module aggregates them into one versioned, read-only snapshot so the
console, and any worker asking "what is safe to do next", read the same
inventory from the same authority.
Field authority is explicit and never blended. Every section declares where its
rows came from:
* ``control_plane_db`` — the #613 substrate: sessions, leases, assignments.
Authoritative for *exclusive ownership* (#600/#601).
* ``filesystem`` — durable per-issue lock files (:mod:`issue_lock_store`) and
registered git worktrees. Authoritative for *what exists on this machine*.
* ``gitea`` — remote issue/PR state, reached only through existing loaders.
Safety invariants:
* **Read-only.** The control-plane database is opened through a ``mode=ro``
URI. :class:`control_plane_db.ControlPlaneDB` creates directories and runs
migrations in its constructor, which an inventory read must never do, so this
module talks to sqlite directly rather than through that class.
* **Fail-soft, never fail-silent.** A subsystem that cannot be read degrades to
a section carrying ``status`` and ``reason``. It never raises, and it never
produces an empty list that reads like "nothing is there".
* **Never invent active ownership.** This is the invariant that matters most.
A degraded ownership source sets ``ownership_authority_complete`` false, and
while that flag is false no work item is reported unowned and no collision is
asserted. Absence of evidence is reported as absence of evidence.
* **Redaction at the boundary.** Absolute paths are collapsed against the home
directory, URLs lose userinfo and query strings, and no credential-shaped
value is emitted. No session token exists in these sources and none is read.
Phase 1 is read-only. Lease steal/release and worktree deletion are Phase 2+
and deliberately have no representation here, not even a disabled one.
"""
from __future__ import annotations
import os
import re
import sqlite3
import time
from dataclasses import dataclass, field
from datetime import datetime, timezone
from typing import Any, Callable
from urllib.parse import urlparse
import control_plane_db
import issue_lock_store
SCHEMA_VERSION = 1
API_VERSION = "v1"
#: Sections whose absence would make an ownership claim unprovable. If any of
#: these is not ``ok``, the snapshot refuses to describe anything as unowned.
OWNERSHIP_SECTIONS = ("sessions", "leases", "locks")
SECTION_NAMES = ("sessions", "leases", "locks", "worktrees", "namespaces")
STATUS_OK = "ok"
STATUS_DEGRADED = "degraded"
STATUS_UNAVAILABLE = "unavailable"
AUTHORITY_CONTROL_PLANE_DB = "control_plane_db"
AUTHORITY_FILESYSTEM = "filesystem"
AUTHORITY_GITEA = "gitea"
_CREDENTIAL_KEY_RE = re.compile(
r"(token|secret|password|passwd|api[_-]?key|authorization|bearer|credential)",
re.IGNORECASE,
)
_REDACTED = "[redacted]"
@dataclass(frozen=True)
class InventorySection:
"""One subsystem's contribution, with its authority and health."""
name: str
authority: str
status: str
items: tuple[dict[str, Any], ...] = ()
reason: str | None = None
scan_ms: float | None = None
@property
def ok(self) -> bool:
return self.status == STATUS_OK
def to_dict(self) -> dict[str, Any]:
return {
"name": self.name,
"authority": self.authority,
"status": self.status,
"count": len(self.items),
"reason": self.reason,
"scan_ms": self.scan_ms,
"items": [dict(item) for item in self.items],
}
@dataclass(frozen=True)
class CollisionSignal:
"""A detected conflict between two ownership records."""
kind: str
message: str
severity: str = "warning"
issue_number: int | None = None
branch: str | None = None
worktree_path: str | None = None
session_ids: tuple[str, ...] = ()
def to_dict(self) -> dict[str, Any]:
return {
"kind": self.kind,
"severity": self.severity,
"message": self.message,
"issue_number": self.issue_number,
"branch": self.branch,
"worktree_path": self.worktree_path,
"session_ids": list(self.session_ids),
}
@dataclass(frozen=True)
class InventorySnapshot:
"""Versioned aggregate of every inventory section."""
generated_at: str
sections: tuple[InventorySection, ...]
collisions: tuple[CollisionSignal, ...] = ()
correlations: tuple[dict[str, Any], ...] = ()
schema_version: int = SCHEMA_VERSION
api_version: str = API_VERSION
scan_ms: float | None = None
_section_index: dict[str, InventorySection] = field(
default_factory=dict, repr=False, compare=False
)
def section(self, name: str) -> InventorySection | None:
return self._section_index.get(name)
@property
def degraded_sections(self) -> tuple[str, ...]:
return tuple(s.name for s in self.sections if not s.ok)
@property
def ownership_authority_complete(self) -> bool:
"""True only when every ownership-bearing section read cleanly.
While this is false the snapshot must not describe any work item as
unowned: a lease the reader could not load is not an absent lease.
"""
for name in OWNERSHIP_SECTIONS:
section = self._section_index.get(name)
if section is None or not section.ok:
return False
return True
@property
def status(self) -> str:
if all(s.ok for s in self.sections):
return STATUS_OK
return STATUS_DEGRADED
# ── redaction ────────────────────────────────────────────────────────────────
def redact_path(path: str | None) -> str | None:
"""Collapse an absolute path against ``$HOME`` for browser display."""
if not path:
return path
text = str(path)
home = os.path.expanduser("~")
if home and home != "/" and text.startswith(home):
return "~" + text[len(home) :]
return text
def redact_url(value: str | None) -> str | None:
"""Strip userinfo and query string from a URL."""
if not value:
return value
text = str(value)
try:
parsed = urlparse(text)
except ValueError:
return _REDACTED
if not parsed.scheme or not parsed.netloc:
return text
netloc = parsed.hostname or ""
if parsed.port:
netloc = f"{netloc}:{parsed.port}"
rebuilt = f"{parsed.scheme}://{netloc}{parsed.path}"
return rebuilt.rstrip("/") or rebuilt
def scrub(value: Any, *, key: str | None = None) -> Any:
"""Recursively drop credential-shaped values and redact paths/URLs.
Never raises: an unexpected object degrades to its ``repr`` rather than
propagating out of a read-only view.
"""
if key and _CREDENTIAL_KEY_RE.search(key):
return _REDACTED
if isinstance(value, dict):
return {str(k): scrub(v, key=str(k)) for k, v in value.items()}
if isinstance(value, (list, tuple)):
return [scrub(v, key=key) for v in value]
if isinstance(value, str):
if value.startswith(("http://", "https://")):
return redact_url(value)
if value.startswith("/") or value.startswith("~"):
return redact_path(value)
return value
if isinstance(value, (int, float, bool)) or value is None:
return value
return repr(value)
# ── control-plane database (read-only) ───────────────────────────────────────
def _open_readonly(db_path: str) -> sqlite3.Connection:
"""Open the control-plane DB without creating or migrating anything."""
conn = sqlite3.connect(f"file:{db_path}?mode=ro", uri=True, timeout=5)
conn.row_factory = sqlite3.Row
return conn
def _table_names(conn: sqlite3.Connection) -> set[str]:
rows = conn.execute(
"SELECT name FROM sqlite_master WHERE type = 'table'"
).fetchall()
return {str(row[0]) for row in rows}
def _load_cp_db_sections(
*,
db_path: str | None = None,
limit: int = 200,
) -> tuple[InventorySection, InventorySection]:
"""Return the ``sessions`` and ``leases`` sections from the #613 DB."""
path = (db_path or control_plane_db.default_db_path()).strip()
def _both_unavailable(reason: str) -> tuple[InventorySection, InventorySection]:
return (
InventorySection(
name="sessions",
authority=AUTHORITY_CONTROL_PLANE_DB,
status=STATUS_UNAVAILABLE,
reason=reason,
),
InventorySection(
name="leases",
authority=AUTHORITY_CONTROL_PLANE_DB,
status=STATUS_UNAVAILABLE,
reason=reason,
),
)
if not path:
return _both_unavailable("control-plane database path is not configured")
if not os.path.exists(path):
return _both_unavailable(
f"control-plane database not present at {redact_path(path)}; "
"no session or lease authority available"
)
started = time.perf_counter()
try:
conn = _open_readonly(path)
except sqlite3.Error as exc:
return _both_unavailable(f"control-plane database could not be opened: {exc}")
try:
tables = _table_names(conn)
if "sessions" not in tables or "leases" not in tables:
missing = sorted({"sessions", "leases"} - tables)
return _both_unavailable(
"control-plane database is missing required tables: "
+ ", ".join(missing)
)
session_rows = [
dict(row)
for row in conn.execute(
"SELECT session_id, role, profile, namespace, pid, started_at,"
" last_heartbeat_at, status FROM sessions"
" ORDER BY last_heartbeat_at DESC LIMIT ?",
(max(1, int(limit)),),
).fetchall()
]
has_work_items = "work_items" in tables
if has_work_items:
lease_sql = (
"SELECT l.lease_id, l.session_id, l.role, l.phase, l.status,"
" l.expires_at, w.remote, w.org, w.repo, w.kind AS work_kind,"
" w.number AS work_number, w.state AS work_state,"
" s.pid AS session_pid, s.profile AS session_profile,"
" s.namespace AS session_namespace, s.status AS session_status"
" FROM leases l"
" JOIN work_items w ON w.work_item_id = l.work_item_id"
" LEFT JOIN sessions s ON s.session_id = l.session_id"
" ORDER BY l.expires_at DESC LIMIT ?"
)
else:
lease_sql = (
"SELECT l.lease_id, l.session_id, l.role, l.phase, l.status,"
" l.expires_at FROM leases l"
" ORDER BY l.expires_at DESC LIMIT ?"
)
lease_rows = [
dict(row)
for row in conn.execute(lease_sql, (max(1, int(limit)),)).fetchall()
]
except sqlite3.Error as exc:
return _both_unavailable(f"control-plane database read failed: {exc}")
finally:
conn.close()
elapsed = (time.perf_counter() - started) * 1000.0
now = datetime.now(timezone.utc)
sessions = tuple(
scrub(
{
"session_id": row.get("session_id"),
"role": row.get("role"),
"profile": row.get("profile"),
"namespace": row.get("namespace"),
"pid": row.get("pid"),
"pid_alive": issue_lock_store.is_process_alive(row.get("pid")),
"started_at": row.get("started_at"),
"last_heartbeat_at": row.get("last_heartbeat_at"),
"status": row.get("status"),
}
)
for row in session_rows
)
leases = tuple(
scrub(
{
"lease_id": row.get("lease_id"),
"session_id": row.get("session_id"),
"role": row.get("role"),
"phase": row.get("phase"),
"status": row.get("status"),
"expires_at": row.get("expires_at"),
"expired": _is_expired(row.get("expires_at"), now=now),
"remote": row.get("remote"),
"org": row.get("org"),
"repo": row.get("repo"),
"work_kind": row.get("work_kind"),
"work_number": row.get("work_number"),
"work_state": row.get("work_state"),
"session_pid": row.get("session_pid"),
"session_profile": row.get("session_profile"),
"session_namespace": row.get("session_namespace"),
"session_status": row.get("session_status"),
}
)
for row in lease_rows
)
degraded_reason = (
None
if has_work_items
else "work_items table absent; lease rows carry no work linkage"
)
lease_status = STATUS_OK if has_work_items else STATUS_DEGRADED
return (
InventorySection(
name="sessions",
authority=AUTHORITY_CONTROL_PLANE_DB,
status=STATUS_OK,
items=sessions,
scan_ms=round(elapsed, 3),
),
InventorySection(
name="leases",
authority=AUTHORITY_CONTROL_PLANE_DB,
status=lease_status,
items=leases,
reason=degraded_reason,
scan_ms=round(elapsed, 3),
),
)
def _is_expired(expires_at: str | None, *, now: datetime) -> bool | None:
if not expires_at:
return None
text = str(expires_at).strip().replace("Z", "+00:00")
try:
parsed = datetime.fromisoformat(text)
except ValueError:
return None
if parsed.tzinfo is None:
parsed = parsed.replace(tzinfo=timezone.utc)
return parsed <= now
# ── durable issue locks (filesystem) ─────────────────────────────────────────
def _load_locks_section(*, lock_dir: str | None = None) -> InventorySection:
started = time.perf_counter()
try:
paths = issue_lock_store.iter_lock_files(lock_dir)
except OSError as exc:
return InventorySection(
name="locks",
authority=AUTHORITY_FILESYSTEM,
status=STATUS_UNAVAILABLE,
reason=f"issue lock directory could not be listed: {exc}",
)
items: list[dict[str, Any]] = []
unreadable = 0
for path in paths:
try:
record = issue_lock_store.read_lock_file(path)
except (OSError, ValueError):
unreadable += 1
continue
if not record:
unreadable += 1
continue
try:
freshness = issue_lock_store.assess_lock_freshness(record)
except Exception: # noqa: BLE001 — a read-only view never raises
freshness = {"status": "unknown", "live": False, "stale": False}
claimant = record.get("claimant") or (
(record.get("work_lease") or {}).get("claimant") or {}
)
items.append(
scrub(
{
"issue_number": record.get("issue_number"),
"branch_name": record.get("branch_name"),
"remote": record.get("remote"),
"org": record.get("org"),
"repo": record.get("repo"),
"worktree_path": record.get("worktree_path"),
"pid": record.get("session_pid") or record.get("pid"),
"pid_alive": issue_lock_store.is_process_alive(
record.get("session_pid") or record.get("pid")
),
"claimant_username": (claimant or {}).get("username"),
"claimant_profile": (claimant or {}).get("profile"),
"lock_generation": record.get("lock_generation"),
"freshness_status": freshness.get("status"),
"live": bool(freshness.get("live")),
"stale": bool(freshness.get("stale")),
"freshness_reason": freshness.get("reason"),
"lock_path": record.get("lock_file_path") or path,
}
)
)
elapsed = (time.perf_counter() - started) * 1000.0
reason = (
f"{unreadable} lock file(s) were unreadable and are not represented"
if unreadable
else None
)
return InventorySection(
name="locks",
authority=AUTHORITY_FILESYSTEM,
status=STATUS_DEGRADED if unreadable else STATUS_OK,
items=tuple(items),
reason=reason,
scan_ms=round(elapsed, 3),
)
# ── worktrees (filesystem, via the #432 scanner) ─────────────────────────────
def _load_worktrees_section(
*, load_hygiene: Callable[[], Any] | None = None
) -> InventorySection:
started = time.perf_counter()
try:
loader = load_hygiene
if loader is None:
from webui.worktree_scanner import load_hygiene_snapshot
loader = load_hygiene_snapshot
snapshot = loader()
except Exception as exc: # noqa: BLE001 — fail soft, never fail the request
return InventorySection(
name="worktrees",
authority=AUTHORITY_FILESYSTEM,
status=STATUS_UNAVAILABLE,
reason=f"worktree scan failed: {exc}",
)
items = tuple(
scrub(
{
"rel_path": entry.rel_path,
"folder_name": entry.folder_name,
"classification": entry.classification,
"branch": entry.branch,
"head_sha": entry.head_sha,
"dirty_tracked": entry.dirty_tracked,
"dirty_untracked": entry.dirty_untracked,
"detached": entry.detached,
"registered_worktree": entry.registered_worktree,
"notes": entry.notes,
}
)
for entry in snapshot.entries
)
scan_error = getattr(snapshot, "scan_error", None)
elapsed = (time.perf_counter() - started) * 1000.0
return InventorySection(
name="worktrees",
authority=AUTHORITY_FILESYSTEM,
status=STATUS_DEGRADED if scan_error else STATUS_OK,
items=items,
reason=scan_error,
scan_ms=round(elapsed, 3),
)
# ── namespaces / capability summary ──────────────────────────────────────────
def _load_namespaces_section() -> InventorySection:
started = time.perf_counter()
try:
from gitea_auth import get_profile
profile = get_profile() or {}
except Exception as exc: # noqa: BLE001
return InventorySection(
name="namespaces",
authority=AUTHORITY_FILESYSTEM,
status=STATUS_UNAVAILABLE,
reason=f"active profile could not be resolved: {exc}",
)
allowed = list(profile.get("allowed_operations") or [])
forbidden = list(profile.get("forbidden_operations") or [])
profile_name = str(profile.get("profile_name") or "")
namespace = None
try:
import role_namespace_gate
namespace = role_namespace_gate.infer_mcp_namespace(profile_name)
except Exception: # noqa: BLE001 — namespace inference is advisory
namespace = None
item = scrub(
{
"profile_name": profile_name,
"role": profile.get("role"),
"mcp_namespace": namespace,
"allowed_operations": sorted(allowed),
"forbidden_operations": sorted(forbidden),
"capability_summary": {
"can_author": "gitea.pr.create" in allowed,
"can_review": "gitea.pr.approve" in allowed,
"can_merge": "gitea.pr.merge" in allowed,
"can_close_pr": "gitea.pr.close" in allowed,
},
"active": True,
}
)
elapsed = (time.perf_counter() - started) * 1000.0
return InventorySection(
name="namespaces",
authority=AUTHORITY_FILESYSTEM,
status=STATUS_OK,
items=(item,),
reason=(
"only the profile serving this web process is observable; other "
"namespaces are not enumerable from here"
),
scan_ms=round(elapsed, 3),
)
# ── correlation and collision detection ──────────────────────────────────────
def _issue_from_branch(branch: str | None) -> int | None:
match = re.search(r"issue-(\d+)", str(branch or ""), re.IGNORECASE)
return int(match.group(1)) if match else None
def correlate(
*,
leases: InventorySection,
locks: InventorySection,
worktrees: InventorySection,
sessions: InventorySection,
) -> tuple[tuple[dict[str, Any], ...], tuple[CollisionSignal, ...]]:
"""Join lease owner ↔ lock ↔ worktree ↔ namespace where evidence allows.
Correlation rows are emitted from whatever sections did load. Collision
signals are only emitted from sections that are ``ok``: a conflict inferred
from a partially-read source would be a false accusation.
"""
correlations: list[dict[str, Any]] = []
collisions: list[CollisionSignal] = []
worktree_by_branch: dict[str, dict[str, Any]] = {}
for entry in worktrees.items:
branch = (entry.get("branch") or "").strip()
if branch:
worktree_by_branch.setdefault(branch, entry)
session_by_id = {
str(s.get("session_id")): s for s in sessions.items if s.get("session_id")
}
# Lock-centred rows: a durable lock names an issue, a branch, and a worktree.
for lock in locks.items:
branch = (lock.get("branch_name") or "").strip()
worktree = worktree_by_branch.get(branch)
matching_leases = [
lease
for lease in leases.items
if lease.get("work_kind") == "issue"
and lease.get("work_number") == lock.get("issue_number")
]
correlations.append(
{
"issue_number": lock.get("issue_number"),
"branch": branch or None,
"lock_live": bool(lock.get("live")),
"lock_claimant": lock.get("claimant_profile"),
"lock_pid": lock.get("pid"),
"lock_pid_alive": lock.get("pid_alive"),
"worktree_rel_path": (worktree or {}).get("rel_path"),
"worktree_classification": (worktree or {}).get("classification"),
"worktree_registered": (worktree or {}).get("registered_worktree"),
"lease_ids": [
lease.get("lease_id")
for lease in matching_leases
if lease.get("lease_id")
],
"lease_sessions": [
lease.get("session_id")
for lease in matching_leases
if lease.get("session_id")
],
}
)
if locks.ok and worktrees.ok:
# A claim whose lease window is still open but has no registered
# worktree is an anomaly regardless of whether its pid is alive; a
# fully time-expired lease is on its way out and is not flagged.
if (
lock.get("freshness_status") != "expired"
and branch
and worktree is None
):
collisions.append(
CollisionSignal(
kind="lock-without-worktree",
severity="warning",
issue_number=lock.get("issue_number"),
branch=branch,
worktree_path=lock.get("worktree_path"),
message=(
f"Live lock on issue #{lock.get('issue_number')} names "
f"branch {branch!r} but no registered worktree carries "
"that branch (#404)"
),
)
)
if locks.ok:
# A lock whose recorded pid is gone is held by nobody: a clean #753
# dead-session recovery candidate. Subclassify by the lease window,
# because the two cases need different operator urgency. When the
# window is still open the lock would read as live to a naive
# timestamp check even though the owner is dead — the more dangerous
# case — so it is flagged distinctly from a fully time-expired lease.
if lock.get("pid_alive") is False and lock.get("stale"):
if lock.get("freshness_status") == "expired":
collisions.append(
CollisionSignal(
kind="stale-lock-dead-owner",
severity="warning",
issue_number=lock.get("issue_number"),
branch=branch or None,
message=(
f"Lock on issue #{lock.get('issue_number')} is stale "
f"and its recorded pid {lock.get('pid')} is not running "
"(#753 dead-session recovery candidate)"
),
)
)
else:
collisions.append(
CollisionSignal(
kind="live-lock-dead-owner",
severity="warning",
issue_number=lock.get("issue_number"),
branch=branch or None,
message=(
f"Lock on issue #{lock.get('issue_number')} has an "
"unexpired lease but its recorded pid "
f"{lock.get('pid')} is not running; it would read as "
"live to a timestamp check (#753 dead-session recovery "
"candidate)"
),
)
)
# The #635 trap: the lease has expired but the recorded pid is a
# still-running daemon, so neither dead-pid reclaim nor exact-owner
# renewal applies. This is the collision an operator must see.
elif (
lock.get("freshness_status") == "expired"
and lock.get("pid_alive") is True
):
collisions.append(
CollisionSignal(
kind="expired-lock-live-owner",
severity="error",
issue_number=lock.get("issue_number"),
branch=branch or None,
message=(
f"Lock on issue #{lock.get('issue_number')} has an expired "
f"lease but its recorded pid {lock.get('pid')} is still "
"running (daemon-pid deadlock; needs an operator decision, "
"#635/#760)"
),
)
)
# Two live locks on one branch, or two active leases on one work item.
if locks.ok:
by_branch: dict[str, list[dict[str, Any]]] = {}
for lock in locks.items:
if not lock.get("live"):
continue
branch = (lock.get("branch_name") or "").strip()
if branch:
by_branch.setdefault(branch, []).append(lock)
for branch, entries in sorted(by_branch.items()):
if len(entries) > 1:
collisions.append(
CollisionSignal(
kind="duplicate-live-lock",
severity="error",
branch=branch,
message=(
f"{len(entries)} live locks name branch {branch!r}: "
"issues "
+ ", ".join(
f"#{e.get('issue_number')}" for e in entries
)
),
)
)
if leases.ok:
by_work: dict[tuple[str, int], list[dict[str, Any]]] = {}
for lease in leases.items:
if str(lease.get("status") or "").lower() != "active":
continue
kind = str(lease.get("work_kind") or "").strip().lower()
number = lease.get("work_number")
if not kind or number is None:
continue
by_work.setdefault((kind, int(number)), []).append(lease)
for (kind, number), entries in sorted(by_work.items()):
sessions_held = {
str(e.get("session_id")) for e in entries if e.get("session_id")
}
if len(sessions_held) > 1:
collisions.append(
CollisionSignal(
kind="concurrent-active-lease",
severity="error",
issue_number=number if kind == "issue" else None,
session_ids=tuple(sorted(sessions_held)),
message=(
f"{len(sessions_held)} sessions hold an active lease on "
f"{kind} #{number}"
),
)
)
for entry in entries:
if entry.get("expired") is True:
collisions.append(
CollisionSignal(
kind="active-lease-past-expiry",
severity="warning",
issue_number=number if kind == "issue" else None,
session_ids=(
(str(entry.get("session_id")),)
if entry.get("session_id")
else ()
),
message=(
f"Lease {entry.get('lease_id')} on {kind} #{number} "
"is still marked active past its expiry"
),
)
)
# A lease whose owning session is gone is an orphan, not free work.
if leases.ok and sessions.ok:
for lease in leases.items:
if str(lease.get("status") or "").lower() != "active":
continue
session_id = str(lease.get("session_id") or "")
if session_id and session_id not in session_by_id:
collisions.append(
CollisionSignal(
kind="orphan-lease",
severity="error",
session_ids=(session_id,),
message=(
f"Active lease {lease.get('lease_id')} names session "
f"{session_id}, which has no session record"
),
)
)
return tuple(correlations), tuple(collisions)
# ── snapshot assembly ────────────────────────────────────────────────────────
def load_inventory_snapshot(
*,
db_path: str | None = None,
lock_dir: str | None = None,
load_hygiene: Callable[[], Any] | None = None,
include: tuple[str, ...] | None = None,
) -> InventorySnapshot:
"""Build the unified inventory snapshot.
Every section is loaded independently and fails soft. *include* restricts
the sections that are scanned; omitted sections are simply absent rather
than reported as empty, so a resource-split request cannot be mistaken for
a whole-inventory answer.
"""
started = time.perf_counter()
wanted = tuple(include) if include else SECTION_NAMES
sections: list[InventorySection] = []
sessions_section: InventorySection | None = None
leases_section: InventorySection | None = None
if "sessions" in wanted or "leases" in wanted:
sessions_section, leases_section = _load_cp_db_sections(db_path=db_path)
if "sessions" in wanted:
sections.append(sessions_section)
if "leases" in wanted:
sections.append(leases_section)
locks_section = (
_load_locks_section(lock_dir=lock_dir)
if "locks" in wanted
else _empty_section("locks", AUTHORITY_FILESYSTEM)
)
if "locks" in wanted:
sections.append(locks_section)
worktrees_section = (
_load_worktrees_section(load_hygiene=load_hygiene)
if "worktrees" in wanted
else _empty_section("worktrees", AUTHORITY_FILESYSTEM)
)
if "worktrees" in wanted:
sections.append(worktrees_section)
if "namespaces" in wanted:
sections.append(_load_namespaces_section())
correlations, collisions = correlate(
leases=leases_section or _empty_section("leases", AUTHORITY_CONTROL_PLANE_DB),
locks=locks_section,
worktrees=worktrees_section,
sessions=sessions_section
or _empty_section("sessions", AUTHORITY_CONTROL_PLANE_DB),
)
elapsed = (time.perf_counter() - started) * 1000.0
index = {section.name: section for section in sections}
return InventorySnapshot(
generated_at=datetime.now(timezone.utc).isoformat(),
sections=tuple(sections),
collisions=collisions,
correlations=correlations,
scan_ms=round(elapsed, 3),
_section_index=index,
)
def _empty_section(name: str, authority: str) -> InventorySection:
"""A section that was not requested — never a claim that it is empty."""
return InventorySection(
name=name,
authority=authority,
status=STATUS_UNAVAILABLE,
reason="section not requested in this scan",
)
def snapshot_to_dict(snapshot: InventorySnapshot) -> dict[str, Any]:
"""Serialize the snapshot for the versioned API."""
return {
"api_version": snapshot.api_version,
"schema_version": snapshot.schema_version,
"generated_at": snapshot.generated_at,
"status": snapshot.status,
"scan_ms": snapshot.scan_ms,
"ownership_authority_complete": snapshot.ownership_authority_complete,
"ownership_note": (
"Every ownership source read cleanly; an item absent from leases "
"and locks is genuinely unclaimed."
if snapshot.ownership_authority_complete
else "One or more ownership sources are degraded; nothing in this "
"snapshot may be treated as unowned. Collisions are reported only "
"from sections that read cleanly."
),
"degraded_sections": list(snapshot.degraded_sections),
"field_authority": {
"sessions": AUTHORITY_CONTROL_PLANE_DB,
"leases": AUTHORITY_CONTROL_PLANE_DB,
"locks": AUTHORITY_FILESYSTEM,
"worktrees": AUTHORITY_FILESYSTEM,
"namespaces": AUTHORITY_FILESYSTEM,
},
"sections": {section.name: section.to_dict() for section in snapshot.sections},
"correlations": [dict(row) for row in snapshot.correlations],
"collisions": [signal.to_dict() for signal in snapshot.collisions],
}
+457 -22
View File
@@ -1,4 +1,16 @@
"""Load and validate the web UI project registry (#427).""" """Load and validate the web UI project registry (#427, evolved for #635).
Phase 1 of the console architecture ADR keeps this loader read-only. It owns
the versioned project registry contract served at ``/api/v1/projects``:
* the on-disk file carries a ``version`` (schema version 1 or 2);
* version 1 files stay loadable and are normalized with explicit defaults, so
an operator registry written for #427 keeps working;
* every validation failure raises :class:`RegistryError`, which carries an
actionable ``remediation`` string instead of leaking a traceback;
* serialization never emits credentials — credential-shaped keys are rejected
at load time, before any DTO is built.
"""
from __future__ import annotations from __future__ import annotations
@@ -8,7 +20,59 @@ from dataclasses import dataclass
from pathlib import Path from pathlib import Path
from typing import Any from typing import Any
from webui.registry_safety import reject_credential_keys as _reject_credential_keys from webui.registry_safety import is_forbidden_key
#: Version of the JSON contract served under ``/api/v1/...``.
REGISTRY_API_VERSION = "v1"
#: Schema version written by this repository's packaged registry.
CURRENT_SCHEMA_VERSION = 2
#: Schema versions this loader accepts. Version 1 is normalized on load.
SUPPORTED_SCHEMA_VERSIONS = (1, 2)
#: Lifecycle state of a registered project.
PROJECT_STATUSES = ("active", "onboarding", "paused", "archived")
_DEFAULT_PROJECT_STATUS = "active"
#: Completion state of a single onboarding step.
ONBOARDING_STATES = ("complete", "pending", "blocked", "not_applicable")
_DEFAULT_ONBOARDING_STATE = "pending"
#: Redacted, last-seen health of a project's control plane.
HEALTH_STATUSES = ("healthy", "degraded", "unreachable", "unknown")
class RegistryError(ValueError):
"""A registry file could not be loaded or failed validation.
Carries an operator-facing ``remediation`` so routes can fail closed with
an actionable message rather than a stack trace.
"""
def __init__(
self,
message: str,
*,
remediation: str,
source_path: Path | None = None,
field_path: str | None = None,
) -> None:
super().__init__(message)
self.message = message
self.remediation = remediation
self.source_path = source_path
self.field_path = field_path
def to_dict(self) -> dict[str, Any]:
"""Serialize for a fail-closed JSON error response."""
return {
"error": "registry_invalid",
"detail": self.message,
"remediation": self.remediation,
"field_path": self.field_path,
"source_path": str(self.source_path) if self.source_path else None,
}
_REQUIRED_PROJECT_FIELDS = ( _REQUIRED_PROJECT_FIELDS = (
"id", "id",
@@ -29,6 +93,30 @@ class OnboardingStep:
id: str id: str
title: str title: str
description: str description: str
state: str = _DEFAULT_ONBOARDING_STATE
required: bool = True
@dataclass(frozen=True)
class OnboardingSummary:
"""Aggregate onboarding progress for a single project."""
total: int
complete: int
pending: int
blocked: int
not_applicable: int
required_outstanding: int
onboarding_complete: bool
@dataclass(frozen=True)
class ProjectHealth:
"""Redacted last-seen health. Never carries endpoints or credentials."""
status: str
checked_at: str | None
detail: str | None
@dataclass(frozen=True) @dataclass(frozen=True)
@@ -43,6 +131,13 @@ class ProjectRecord:
workflow_paths: dict[str, str] workflow_paths: dict[str, str]
schema_paths: dict[str, str] schema_paths: dict[str, str]
onboarding_checklist: tuple[OnboardingStep, ...] onboarding_checklist: tuple[OnboardingStep, ...]
status: str = _DEFAULT_PROJECT_STATUS
remote_name: str | None = None
last_seen_health: ProjectHealth | None = None
@property
def repo_full_name(self) -> str:
return f"{self.gitea_owner}/{self.repo_name}"
@dataclass(frozen=True) @dataclass(frozen=True)
@@ -51,6 +146,15 @@ class ProjectRegistry:
projects: tuple[ProjectRecord, ...] projects: tuple[ProjectRecord, ...]
source_path: Path source_path: Path
@property
def schema_version(self) -> int:
"""Alias of :attr:`version` — the schema version read from disk."""
return self.version
@property
def api_version(self) -> str:
return REGISTRY_API_VERSION
def default_registry_path() -> Path: def default_registry_path() -> Path:
override = os.environ.get("WEBUI_PROJECT_REGISTRY", "").strip() override = os.environ.get("WEBUI_PROJECT_REGISTRY", "").strip()
@@ -59,40 +163,221 @@ def default_registry_path() -> Path:
return (Path(__file__).resolve().parent / "data" / "projects.registry.json").resolve() return (Path(__file__).resolve().parent / "data" / "projects.registry.json").resolve()
def _parse_onboarding(raw: list[dict[str, Any]] | None) -> tuple[OnboardingStep, ...]: def _reject_credential_keys(obj: Any, *, path: str = "", source: Path | None = None) -> None:
if not raw: """Recursive credential-key guard that reports an actionable ``field_path``.
Key *shape* is decided by :func:`webui.registry_safety.is_forbidden_key`, the
single source of truth shared with the worker registry (#798).
"""
if isinstance(obj, dict):
for key, value in obj.items():
key_path = f"{path}.{key}" if path else key
if is_forbidden_key(key):
raise RegistryError(
f"registry must not store credentials ({key_path})",
remediation=(
f"Remove the credential-shaped key '{key_path}' from the registry. "
"Tokens live in the keychain and are resolved server-side by "
"gitea_auth; the registry is redacted metadata only."
),
source_path=source,
field_path=key_path,
)
_reject_credential_keys(value, path=key_path, source=source)
elif isinstance(obj, list):
for index, item in enumerate(obj):
_reject_credential_keys(item, path=f"{path}[{index}]", source=source)
def _require_enum(
value: Any,
*,
allowed: tuple[str, ...],
field_path: str,
source: Path | None,
) -> str:
text = str(value)
if text not in allowed:
raise RegistryError(
f"{field_path} must be one of {', '.join(allowed)} (got {text!r})",
remediation=(
f"Set {field_path} to one of: {', '.join(allowed)}. "
"Unknown values fail closed so the console never renders an "
"unverified state."
),
source_path=source,
field_path=field_path,
)
return text
def _parse_onboarding(
raw: Any,
*,
project_path: str,
source: Path | None,
) -> tuple[OnboardingStep, ...]:
if raw is None:
return () return ()
if not isinstance(raw, list):
raise RegistryError(
f"{project_path}.onboarding_checklist must be an array",
remediation=(
f"Rewrite {project_path}.onboarding_checklist as a JSON array of "
"steps with id, title, description, and optional state."
),
source_path=source,
field_path=f"{project_path}.onboarding_checklist",
)
steps: list[OnboardingStep] = [] steps: list[OnboardingStep] = []
for item in raw: for index, item in enumerate(raw):
step_path = f"{project_path}.onboarding_checklist[{index}]"
if not isinstance(item, dict):
raise RegistryError(
f"{step_path} must be an object",
remediation=f"Rewrite {step_path} as an object with id, title, description.",
source_path=source,
field_path=step_path,
)
missing = [field for field in ("id", "title", "description") if field not in item]
if missing:
raise RegistryError(
f"{step_path} missing required fields: {', '.join(missing)}",
remediation=(
f"Add {', '.join(missing)} to {step_path}. Every onboarding step "
"must be self-describing for an operator who has no chat history."
),
source_path=source,
field_path=step_path,
)
state = _require_enum(
item.get("state", _DEFAULT_ONBOARDING_STATE),
allowed=ONBOARDING_STATES,
field_path=f"{step_path}.state",
source=source,
)
steps.append( steps.append(
OnboardingStep( OnboardingStep(
id=str(item["id"]), id=str(item["id"]),
title=str(item["title"]), title=str(item["title"]),
description=str(item["description"]), description=str(item["description"]),
state=state,
required=bool(item.get("required", True)),
) )
) )
return tuple(steps) return tuple(steps)
def _parse_project(raw: dict[str, Any]) -> ProjectRecord: def _parse_health(
raw: Any,
*,
project_path: str,
source: Path | None,
) -> ProjectHealth | None:
if raw is None:
return None
if not isinstance(raw, dict):
raise RegistryError(
f"{project_path}.last_seen_health must be an object when present",
remediation=(
f"Rewrite {project_path}.last_seen_health as an object with status "
f"(one of {', '.join(HEALTH_STATUSES)}), optional checked_at and detail, "
"or remove it. Never store endpoints or credentials here."
),
source_path=source,
field_path=f"{project_path}.last_seen_health",
)
status = _require_enum(
raw.get("status", "unknown"),
allowed=HEALTH_STATUSES,
field_path=f"{project_path}.last_seen_health.status",
source=source,
)
checked_at = raw.get("checked_at")
detail = raw.get("detail")
return ProjectHealth(
status=status,
checked_at=str(checked_at) if checked_at is not None else None,
detail=str(detail) if detail is not None else None,
)
def _parse_project(raw: Any, *, index: int, source: Path | None) -> ProjectRecord:
project_path = f"projects[{index}]"
if not isinstance(raw, dict):
raise RegistryError(
f"{project_path} must be an object",
remediation=f"Rewrite {project_path} as a JSON object describing one project.",
source_path=source,
field_path=project_path,
)
missing = [field for field in _REQUIRED_PROJECT_FIELDS if field not in raw] missing = [field for field in _REQUIRED_PROJECT_FIELDS if field not in raw]
if missing: if missing:
raise ValueError(f"project missing required fields: {', '.join(missing)}") raise RegistryError(
f"{project_path} missing required fields: {', '.join(missing)}",
remediation=(
f"Add {', '.join(missing)} to {project_path}. See "
"docs/webui-project-registry-api.md for the field-by-field contract."
),
source_path=source,
field_path=project_path,
)
profiles = raw["profiles"] profiles = raw["profiles"]
if not isinstance(profiles, dict): if not isinstance(profiles, dict):
raise ValueError("profiles must be an object") raise RegistryError(
f"{project_path}.profiles must be an object",
remediation=(
f"Rewrite {project_path}.profiles as an object mapping "
f"{', '.join(_REQUIRED_PROFILE_ROLES)} to MCP profile names."
),
source_path=source,
field_path=f"{project_path}.profiles",
)
for role in _REQUIRED_PROFILE_ROLES: for role in _REQUIRED_PROFILE_ROLES:
if role not in profiles or not profiles[role]: if role not in profiles or not profiles[role]:
raise ValueError(f"profiles.{role} is required") raise RegistryError(
f"{project_path}.profiles.{role} is required",
remediation=(
f"Set {project_path}.profiles.{role} to the configured MCP profile "
"name for that role. Role separation is a workflow-safety invariant."
),
source_path=source,
field_path=f"{project_path}.profiles.{role}",
)
workflow_paths = raw["workflow_paths"] workflow_paths = raw["workflow_paths"]
if not isinstance(workflow_paths, dict) or not workflow_paths: if not isinstance(workflow_paths, dict) or not workflow_paths:
raise ValueError("workflow_paths must be a non-empty object") raise RegistryError(
f"{project_path}.workflow_paths must be a non-empty object",
remediation=(
f"Add at least a 'skill' entry to {project_path}.workflow_paths pointing "
"at the project's canonical workflow skill."
),
source_path=source,
field_path=f"{project_path}.workflow_paths",
)
schema_paths = raw.get("schema_paths") or {} schema_paths = raw.get("schema_paths") or {}
if not isinstance(schema_paths, dict): if not isinstance(schema_paths, dict):
raise ValueError("schema_paths must be an object when present") raise RegistryError(
f"{project_path}.schema_paths must be an object when present",
remediation=(
f"Rewrite {project_path}.schema_paths as an object of label to repo path, "
"or remove it."
),
source_path=source,
field_path=f"{project_path}.schema_paths",
)
status = _require_enum(
raw.get("status", _DEFAULT_PROJECT_STATUS),
allowed=PROJECT_STATUSES,
field_path=f"{project_path}.status",
source=source,
)
remote_name = raw.get("remote_name")
return ProjectRecord( return ProjectRecord(
id=str(raw["id"]), id=str(raw["id"]),
@@ -104,61 +389,211 @@ def _parse_project(raw: dict[str, Any]) -> ProjectRecord:
profiles={role: str(profiles[role]) for role in _REQUIRED_PROFILE_ROLES}, profiles={role: str(profiles[role]) for role in _REQUIRED_PROFILE_ROLES},
workflow_paths={key: str(value) for key, value in workflow_paths.items()}, workflow_paths={key: str(value) for key, value in workflow_paths.items()},
schema_paths={key: str(value) for key, value in schema_paths.items()}, schema_paths={key: str(value) for key, value in schema_paths.items()},
onboarding_checklist=_parse_onboarding(raw.get("onboarding_checklist")), onboarding_checklist=_parse_onboarding(
raw.get("onboarding_checklist"),
project_path=project_path,
source=source,
),
status=status,
remote_name=str(remote_name) if remote_name else None,
last_seen_health=_parse_health(
raw.get("last_seen_health"),
project_path=project_path,
source=source,
),
) )
def load_registry(path: Path | None = None) -> ProjectRegistry: def load_registry(path: Path | None = None) -> ProjectRegistry:
"""Load the versioned project registry from disk.""" """Load the versioned project registry from disk.
Raises:
RegistryError: whenever the file is unreadable, is not valid JSON, or
fails schema validation. The error carries an operator remediation.
"""
source = (path or default_registry_path()).resolve() source = (path or default_registry_path()).resolve()
try:
raw_text = source.read_text(encoding="utf-8") raw_text = source.read_text(encoding="utf-8")
except OSError as exc:
raise RegistryError(
f"registry file could not be read: {exc.strerror or exc}",
remediation=(
f"Create a readable registry at {source}, or point "
"WEBUI_PROJECT_REGISTRY at an existing file."
),
source_path=source,
) from exc
try:
payload = json.loads(raw_text) payload = json.loads(raw_text)
except json.JSONDecodeError as exc:
raise RegistryError(
f"registry is not valid JSON: {exc.msg} (line {exc.lineno}, column {exc.colno})",
remediation=(
f"Fix the JSON syntax in {source} at line {exc.lineno}, column {exc.colno}."
),
source_path=source,
) from exc
if not isinstance(payload, dict): if not isinstance(payload, dict):
raise ValueError("registry root must be an object") raise RegistryError(
"registry root must be an object",
remediation=(
"Wrap the registry in a JSON object with 'version' and 'projects' keys."
),
source_path=source,
)
version = payload.get("version") version = payload.get("version")
if version != 1: if version not in SUPPORTED_SCHEMA_VERSIONS:
raise ValueError(f"unsupported registry version: {version!r}") supported = ", ".join(str(item) for item in SUPPORTED_SCHEMA_VERSIONS)
raise RegistryError(
f"unsupported registry version: {version!r}",
remediation=(
f"Set 'version' to one of {supported} (current schema is "
f"{CURRENT_SCHEMA_VERSION}). Migration notes live in "
"docs/webui-project-registry-api.md."
),
source_path=source,
field_path="version",
)
_reject_credential_keys(payload) _reject_credential_keys(payload, source=source)
projects_raw = payload.get("projects") projects_raw = payload.get("projects")
if not isinstance(projects_raw, list) or not projects_raw: if not isinstance(projects_raw, list) or not projects_raw:
raise ValueError("projects must be a non-empty array") raise RegistryError(
"projects must be a non-empty array",
remediation=(
"Add at least one project object to 'projects'. An empty console "
"registry fails closed rather than rendering a blank inventory."
),
source_path=source,
field_path="projects",
)
projects = tuple(_parse_project(item) for item in projects_raw) projects = tuple(
return ProjectRegistry(version=version, projects=projects, source_path=source) _parse_project(item, index=index, source=source)
for index, item in enumerate(projects_raw)
)
return ProjectRegistry(version=int(version), projects=projects, source_path=source)
def onboarding_summary(project: ProjectRecord) -> OnboardingSummary:
"""Aggregate a project's onboarding checklist state."""
steps = project.onboarding_checklist
counts = {state: 0 for state in ONBOARDING_STATES}
for step in steps:
counts[step.state] += 1
required_outstanding = sum(
1
for step in steps
if step.required and step.state in ("pending", "blocked")
)
return OnboardingSummary(
total=len(steps),
complete=counts["complete"],
pending=counts["pending"],
blocked=counts["blocked"],
not_applicable=counts["not_applicable"],
required_outstanding=required_outstanding,
onboarding_complete=required_outstanding == 0,
)
def project_to_dict(project: ProjectRecord) -> dict[str, Any]: def project_to_dict(project: ProjectRecord) -> dict[str, Any]:
"""Serialize a project for JSON API responses.""" """Serialize a project for JSON API responses and HTML views.
The HTML views render from this same DTO, so the console and the API can
never disagree about a project's status or onboarding progress.
"""
summary = onboarding_summary(project)
health = project.last_seen_health
return { return {
"id": project.id, "id": project.id,
"repo_name": project.repo_name, "repo_name": project.repo_name,
"gitea_owner": project.gitea_owner, "gitea_owner": project.gitea_owner,
"repo_full_name": project.repo_full_name,
"remote_host": project.remote_host, "remote_host": project.remote_host,
"remote_name": project.remote_name,
"default_branch": project.default_branch, "default_branch": project.default_branch,
"local_checkout_path": project.local_checkout_path, "local_checkout_path": project.local_checkout_path,
"status": project.status,
"profiles": dict(project.profiles), "profiles": dict(project.profiles),
"workflow_paths": dict(project.workflow_paths), "workflow_paths": dict(project.workflow_paths),
"schema_paths": dict(project.schema_paths), "schema_paths": dict(project.schema_paths),
"onboarding_checklist": [ "onboarding_checklist": [
{"id": step.id, "title": step.title, "description": step.description} {
"id": step.id,
"title": step.title,
"description": step.description,
"state": step.state,
"required": step.required,
}
for step in project.onboarding_checklist for step in project.onboarding_checklist
], ],
"onboarding_summary": {
"total": summary.total,
"complete": summary.complete,
"pending": summary.pending,
"blocked": summary.blocked,
"not_applicable": summary.not_applicable,
"required_outstanding": summary.required_outstanding,
"onboarding_complete": summary.onboarding_complete,
},
"last_seen_health": (
None
if health is None
else {
"status": health.status,
"checked_at": health.checked_at,
"detail": health.detail,
}
),
} }
def registry_to_dict(registry: ProjectRegistry) -> dict[str, Any]: def registry_to_dict(registry: ProjectRegistry) -> dict[str, Any]:
"""Serialize the whole registry, including API provenance (#632 section 6)."""
return { return {
"api_version": registry.api_version,
"schema_version": registry.schema_version,
# Retained for the unversioned MVP alias consumers (#427).
"version": registry.version, "version": registry.version,
"source_path": str(registry.source_path), "source_path": str(registry.source_path),
"source": {
"kind": "file",
"path": str(registry.source_path),
"inventory_complete": True,
},
"project_count": len(registry.projects),
"projects": [project_to_dict(project) for project in registry.projects], "projects": [project_to_dict(project) for project in registry.projects],
} }
def project_detail_to_dict(
registry: ProjectRegistry,
project: ProjectRecord,
) -> dict[str, Any]:
"""Serialize a single project for ``/api/v1/projects/{project_id}``."""
return {
"api_version": registry.api_version,
"schema_version": registry.schema_version,
"source": {
"kind": "file",
"path": str(registry.source_path),
"inventory_complete": True,
},
"project": project_to_dict(project),
}
def find_project(registry: ProjectRegistry, project_id: str) -> ProjectRecord | None: def find_project(registry: ProjectRegistry, project_id: str) -> ProjectRecord | None:
for project in registry.projects: for project in registry.projects:
if project.id == project_id: if project.id == project_id:
return project return project
return None return None
def known_project_ids(registry: ProjectRegistry) -> list[str]:
return [project.id for project in registry.projects]
+107 -25
View File
@@ -1,34 +1,65 @@
"""HTML views for project registry pages (#427).""" """HTML views for project registry pages (#427, evolved for #635).
Every view renders from :func:`webui.project_registry.project_to_dict`, the
same DTO the ``/api/v1/projects`` JSON responses use, so the HTML console and
the API can never disagree about status or onboarding progress.
"""
from __future__ import annotations from __future__ import annotations
import html import html
from typing import Any
from webui.layout import render_page from webui.layout import render_page
from webui.project_registry import ProjectRecord, ProjectRegistry from webui.project_registry import (
ProjectRecord,
ProjectRegistry,
RegistryError,
project_to_dict,
)
_STATE_LABELS = {
"complete": "Complete",
"pending": "Pending",
"blocked": "Blocked",
"not_applicable": "Not applicable",
}
def _escape(text: str) -> str: def _escape(text: str) -> str:
return html.escape(text, quote=True) return html.escape(text, quote=True)
def _progress_label(summary: dict[str, Any]) -> str:
total = summary["total"]
if not total:
return "no steps"
label = f"{summary['complete']}/{total} complete"
if summary["blocked"]:
label += f", {summary['blocked']} blocked"
return label
def render_projects_list(registry: ProjectRegistry) -> str: def render_projects_list(registry: ProjectRegistry) -> str:
rows = [] rows = []
for project in registry.projects: for project in registry.projects:
dto = project_to_dict(project)
rows.append( rows.append(
"<tr>" "<tr>"
f"<td><a href=\"/projects/{_escape(project.id)}\">{_escape(project.repo_name)}</a></td>" f"<td><a href=\"/projects/{_escape(dto['id'])}\">{_escape(dto['repo_name'])}</a></td>"
f"<td>{_escape(project.gitea_owner)}</td>" f"<td>{_escape(dto['gitea_owner'])}</td>"
f"<td>{_escape(project.remote_host)}</td>" f"<td>{_escape(dto['remote_host'])}</td>"
f"<td>{_escape(project.default_branch)}</td>" f"<td>{_escape(dto['default_branch'])}</td>"
f"<td><code>{_escape(project.profiles['author'])}</code></td>" f"<td><code>{_escape(dto['status'])}</code></td>"
f"<td>{_escape(_progress_label(dto['onboarding_summary']))}</td>"
f"<td><code>{_escape(dto['profiles']['author'])}</code></td>"
"</tr>" "</tr>"
) )
table = ( table = (
"<table class=\"registry\">" "<table class=\"registry\">"
"<thead><tr>" "<thead><tr>"
"<th>Repository</th><th>Owner</th><th>Remote</th>" "<th>Repository</th><th>Owner</th><th>Remote</th>"
"<th>Branch</th><th>Author profile</th>" "<th>Branch</th><th>Status</th><th>Onboarding</th><th>Author profile</th>"
"</tr></thead>" "</tr></thead>"
f"<tbody>{''.join(rows)}</tbody></table>" f"<tbody>{''.join(rows)}</tbody></table>"
) )
@@ -36,32 +67,39 @@ def render_projects_list(registry: ProjectRegistry) -> str:
"<h2>Projects</h2>" "<h2>Projects</h2>"
"<p>Configured repositories managed by the MCP Control Plane.</p>" "<p>Configured repositories managed by the MCP Control Plane.</p>"
f"<p class=\"meta\">Registry: <code>{_escape(str(registry.source_path))}</code> " f"<p class=\"meta\">Registry: <code>{_escape(str(registry.source_path))}</code> "
f"(version {registry.version})</p>" f"(schema version {registry.schema_version}, "
f"API {_escape(registry.api_version)})</p>"
f"{table}" f"{table}"
"<p><a href=\"/api/projects\">JSON API</a></p>" "<p><a href=\"/api/v1/projects\">JSON API</a> "
"(<a href=\"/api/projects\">unversioned alias</a>)</p>"
) )
return render_page(title="Projects", body_html=body) return render_page(title="Projects", body_html=body)
def render_project_detail(project: ProjectRecord) -> str: def render_project_detail(project: ProjectRecord) -> str:
dto = project_to_dict(project)
profile_rows = "".join( profile_rows = "".join(
f"<tr><th>{_escape(role)}</th><td><code>{_escape(name)}</code></td></tr>" f"<tr><th>{_escape(role)}</th><td><code>{_escape(name)}</code></td></tr>"
for role, name in project.profiles.items() for role, name in dto["profiles"].items()
) )
workflow_rows = "".join( workflow_rows = "".join(
f"<tr><th>{_escape(key)}</th><td><code>{_escape(path)}</code></td></tr>" f"<tr><th>{_escape(key)}</th><td><code>{_escape(path)}</code></td></tr>"
for key, path in project.workflow_paths.items() for key, path in dto["workflow_paths"].items()
) )
schema_rows = "".join( schema_rows = "".join(
f"<tr><th>{_escape(key)}</th><td><code>{_escape(path)}</code></td></tr>" f"<tr><th>{_escape(key)}</th><td><code>{_escape(path)}</code></td></tr>"
for key, path in project.schema_paths.items() for key, path in dto["schema_paths"].items()
) )
checklist_items = [] checklist_items = []
for index, step in enumerate(project.onboarding_checklist, start=1): for index, step in enumerate(dto["onboarding_checklist"], start=1):
state_label = _STATE_LABELS.get(step["state"], step["state"])
requirement = "required" if step["required"] else "optional"
checklist_items.append( checklist_items.append(
"<li>" f"<li class=\"step-{_escape(step['state'])}\">"
f"<strong>{index}. {_escape(step.title)}</strong>" f"<strong>{index}. {_escape(step['title'])}</strong>"
f"<p>{_escape(step.description)}</p>" f" <span class=\"badge\">{_escape(state_label)}</span>"
f" <span class=\"meta\">({_escape(requirement)})</span>"
f"<p>{_escape(step['description'])}</p>"
"</li>" "</li>"
) )
checklist_html = ( checklist_html = (
@@ -69,16 +107,38 @@ def render_project_detail(project: ProjectRecord) -> str:
if checklist_items if checklist_items
else "<p>No onboarding steps defined.</p>" else "<p>No onboarding steps defined.</p>"
) )
summary = dto["onboarding_summary"]
summary_html = (
"<p class=\"meta\">Onboarding: "
f"{_escape(_progress_label(summary))}; required outstanding "
f"{summary['required_outstanding']}.</p>"
)
health = dto["last_seen_health"]
health_html = (
"<p class=\"meta\">No health probe recorded (Phase 1 is read-only).</p>"
if health is None
else (
"<table class=\"detail\">"
f"<tr><th>Status</th><td><code>{_escape(health['status'])}</code></td></tr>"
f"<tr><th>Checked at</th><td>{_escape(str(health['checked_at'] or 'unknown'))}</td></tr>"
f"<tr><th>Detail</th><td>{_escape(str(health['detail'] or ''))}</td></tr>"
"</table>"
)
)
remote_name = dto["remote_name"] or "unset"
body = ( body = (
f"<h2>{_escape(project.repo_name)}</h2>" f"<h2>{_escape(dto['repo_name'])}</h2>"
"<p><a href=\"/projects\">← All projects</a></p>" "<p><a href=\"/projects\">← All projects</a></p>"
"<h3>Identity</h3>" "<h3>Identity</h3>"
"<table class=\"detail\">" "<table class=\"detail\">"
f"<tr><th>Registry id</th><td><code>{_escape(project.id)}</code></td></tr>" f"<tr><th>Registry id</th><td><code>{_escape(dto['id'])}</code></td></tr>"
f"<tr><th>Gitea owner</th><td>{_escape(project.gitea_owner)}</td></tr>" f"<tr><th>Status</th><td><code>{_escape(dto['status'])}</code></td></tr>"
f"<tr><th>Remote host</th><td>{_escape(project.remote_host)}</td></tr>" f"<tr><th>Gitea owner</th><td>{_escape(dto['gitea_owner'])}</td></tr>"
f"<tr><th>Default branch</th><td><code>{_escape(project.default_branch)}</code></td></tr>" f"<tr><th>Repository</th><td><code>{_escape(dto['repo_full_name'])}</code></td></tr>"
f"<tr><th>Local checkout</th><td><code>{_escape(project.local_checkout_path)}</code></td></tr>" f"<tr><th>Remote name</th><td><code>{_escape(remote_name)}</code></td></tr>"
f"<tr><th>Remote host</th><td>{_escape(dto['remote_host'])}</td></tr>"
f"<tr><th>Default branch</th><td><code>{_escape(dto['default_branch'])}</code></td></tr>"
f"<tr><th>Local checkout</th><td><code>{_escape(dto['local_checkout_path'])}</code></td></tr>"
"</table>" "</table>"
"<h3>Profiles</h3>" "<h3>Profiles</h3>"
f"<table class=\"detail\">{profile_rows}</table>" f"<table class=\"detail\">{profile_rows}</table>"
@@ -86,8 +146,30 @@ def render_project_detail(project: ProjectRecord) -> str:
f"<table class=\"detail\">{workflow_rows}</table>" f"<table class=\"detail\">{workflow_rows}</table>"
"<h3>Schema paths</h3>" "<h3>Schema paths</h3>"
f"<table class=\"detail\">{schema_rows}</table>" f"<table class=\"detail\">{schema_rows}</table>"
"<h3>Last seen health</h3>"
f"{health_html}"
"<h3>Onboarding checklist</h3>" "<h3>Onboarding checklist</h3>"
"<p class=\"meta\">Read-only MVP — complete these steps outside the UI.</p>" "<p class=\"meta\">Read-only — complete these steps outside the UI.</p>"
f"{summary_html}"
f"{checklist_html}" f"{checklist_html}"
f"<p><a href=\"/api/v1/projects/{_escape(dto['id'])}\">JSON detail</a></p>"
) )
return render_page(title=project.repo_name, body_html=body) return render_page(title=dto["repo_name"], body_html=body)
def render_registry_error(error: RegistryError) -> str:
"""Render a fail-closed page for an invalid registry."""
source = str(error.source_path) if error.source_path else "unknown"
field = error.field_path or "n/a"
body = (
"<h2>Project registry unavailable</h2>"
"<p>The registry failed validation, so the console refuses to render a "
"partial inventory.</p>"
"<table class=\"detail\">"
f"<tr><th>Detail</th><td>{_escape(error.message)}</td></tr>"
f"<tr><th>Field</th><td><code>{_escape(field)}</code></td></tr>"
f"<tr><th>Source</th><td><code>{_escape(source)}</code></td></tr>"
f"<tr><th>Remediation</th><td>{_escape(error.remediation)}</td></tr>"
"</table>"
)
return render_page(title="Project registry unavailable", body_html=body)