feat(6243): a directory moved in the tree keeps its shape judgments
CI & Build / Python lint (push) Successful in 3s
CI & Build / Plugin hooks (push) Failing after 15s
CI & Build / TypeScript typecheck (push) Successful in 55s
CI & Build / integration (push) Successful in 1m15s
CI & Build / Python tests (push) Successful in 2m9s
CI & Build / Build & push image (push) Successful in 30s

Ledger rows are keyed by path, so after a directory moves the next sync
marks every judged row vanished and lists the same shapes again, unjudged.
move_shapes(project_id, from_path, to_path) re-keys the rows under one path
to the other, in every repo of the project, keeping each row's judgment,
uses edges and history and adding a `moved` event that records the path it
left. A row already at the target is replaced when nobody judged it; a
judged one stands, and the old row is reported in `kept`.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
2026-10-22 15:03:03 -04:00
co-authored by Claude Opus 5.5
parent 5ad1df8e0f
commit 47b52197aa
8 changed files with 259 additions and 9 deletions
+1 -1
View File
@@ -200,7 +200,7 @@ _WRITE_TOOLS = frozenset({
"merge_snippets", "unmerge_snippet",
"create_process", "update_process", "delete_process",
"classify_shapes", "classify_shapes_by_rule", "confirm_shape_proposals",
"refresh_pattern_coverage",
"refresh_pattern_coverage", "move_shapes",
# design systems
"create_design_system", "update_design_system", "delete_design_system",
"create_design_token", "update_design_token", "delete_design_token",
+31 -2
View File
@@ -263,7 +263,8 @@ async def shape_history(
snippet_id, who, why — one per judgment, so a shape that was an instance
of #N and later a variant of #M shows both), `vanished`, `reappeared`,
`drifted` (the body moved under a judgment; see list_shapes flag=
"recheck"). Each event carries the commit the tree was read at.
"recheck"), `moved` (move_shapes carried the row; the event's `path`
is the one it left). Each event carries the commit the tree was read at.
Read it as a timeline: "instance of #N from <first classified at>,
re-judged variant of #M at <at> because <reason>, vanished at <commit>".
@@ -398,10 +399,38 @@ async def refresh_pattern_coverage(project_id: int) -> dict:
}
async def move_shapes(project_id: int, from_path: str, to_path: str) -> dict:
"""Carry the judgments of a file or directory that moved in the tree.
Rows are keyed by path. After `git mv src/old src/new`, the next coverage
refresh would mark every row under src/old vanished and list the same
shapes again under src/new, unclassified, with every judgment left
behind. Run this once the move is committed, before that refresh:
each row under `from_path` is re-keyed to the same place under
`to_path`, in every repo bound to the project, keeping its judgment,
uses edges and history, and gaining a `moved` event (shape_history).
A row already at the new path is replaced when nobody judged it (the
refresh ran first, or the sync stamped it `scoped`); a judged one stands,
and the old row is left behind and listed in `kept`. The refresh that
follows compares bodies as always: a judged shape whose body changed in
the move comes back flagged `recheck`.
Args:
project_id: The project whose ledger moves. Requires write access.
from_path: The file or directory the shapes were found at.
to_path: Where it lives now. Neither path may contain the other.
Returns {"moved": N, "replaced": N, "kept": [{path, symbol, kind}]}.
"""
uid = current_user_id()
return await shape_ledger_svc.move_shapes(uid, project_id, from_path, to_path)
def register(mcp) -> None:
for fn in (
classify_shapes, classify_shapes_by_rule, list_shapes,
refresh_pattern_coverage, confirm_shape_proposals, shape_history,
stamps_to_review,
stamps_to_review, move_shapes,
):
mcp.tool(name=fn.__name__)(fn)
+4 -3
View File
@@ -311,8 +311,9 @@ class CodeShapeConsumer(Base):
# What a shape's history records. Not "appeared" — first_seen and
# created_at already say that on the row; history is for what CHANGED:
SHAPE_EVENTS = ("classified", "vanished", "reappeared", "drifted")
# created_at already say that on the row; history is for what CHANGED.
# A `moved` event carries the path the shape left; the row says where it went.
SHAPE_EVENTS = ("classified", "vanished", "reappeared", "drifted", "moved")
class CodeShapeEvent(Base):
@@ -321,7 +322,7 @@ class CodeShapeEvent(Base):
"We used #N here from <date>, #M replaced it at commit C, reason R" is a
question the ledger row alone cannot answer once it has moved on; this
table keeps each judgment (status, snippet, who, why, at which commit)
and each presence change (vanished / reappeared / drifted) as it
and each presence change (vanished / reappeared / drifted / moved) as it
happened. Denormalised path/symbol/kind so a directory's history reads
without joining; `snippet_id` is deliberately FK-free — history outlives
the snippet it names, which is the point.
+79
View File
@@ -170,6 +170,85 @@ async def sync_repo_shapes(
await session.commit()
def _under(path: str, prefix: str) -> bool:
return path == prefix or path.startswith(prefix + "/")
async def move_shapes(
user_id: int, project_id: int, from_path: str, to_path: str,
) -> dict:
"""Carry the ledger rows under ``from_path`` (a file or a directory) to
the same place under ``to_path``, in every repo of the project.
Rows are keyed by path, so a directory moved in the tree would otherwise
vanish every judged row at the next sync and start the same shapes again,
unclassified, under the new path. A moved row is the same row — its id,
judgment, uses edges and history ride along — and a `moved` event
records the path it left. A row already at the target (the sync ran
first, or a provisional judgment landed there) is replaced when nobody
judged it; a judged one stands, and the old row is left where it is and
reported in ``kept``. Bodies are not compared here: the next sync does
that as always, and asks for a recheck where a judged body changed.
"""
from sqlalchemy import or_
from grimoire.services import access
old = (from_path or "").strip().strip("/")
new = (to_path or "").strip().strip("/")
if not old or not new:
raise ValueError("from_path and to_path are both required")
if _under(old, new) or _under(new, old):
raise ValueError(f"`{old}` and `{new}` overlap; move to a path outside the source")
if not await access.can_write_project(user_id, project_id):
raise ValueError(f"project {project_id} not found or no write access")
now = datetime.now(timezone.utc)
async with async_session() as session:
def beneath(prefix: str):
return select(CodeShape).where(
CodeShape.project_id == project_id,
or_(CodeShape.path == prefix, CodeShape.path.like(prefix + "/%")),
)
# LIKE reads `_` as a wildcard; the prefix test is the real filter.
sources = [
r for r in (await session.execute(beneath(old))).scalars().all()
if _under(r.path, old)
]
if not sources:
raise ValueError(f"project {project_id} has no shapes under `{old}`")
targets = {
(r.repo_key, r.path, r.symbol, r.kind): r
for r in (await session.execute(beneath(new))).scalars().all()
}
moved, replaced, kept = 0, 0, []
for row in sources:
dest = new + row.path[len(old):]
there = targets.get((row.repo_key, dest, row.symbol, row.kind))
if there is not None:
if there.status not in _MECHANICAL_TODO:
kept.append({"path": row.path, "symbol": row.symbol, "kind": row.kind})
continue
# The target is the live sighting when the sync got there
# first; the moved row takes over its presence.
row.vanished_at = there.vanished_at
row.last_seen_commit = there.last_seen_commit or row.last_seen_commit
await session.delete(there)
await session.flush()
replaced += 1
session.add(CodeShapeEvent(
shape_id=row.id, project_id=project_id,
path=row.path, symbol=row.symbol, kind=row.kind,
event="moved", status=row.status, snippet_id=row.snippet_id,
classified_by=row.classified_by, reason=row.reason,
commit=row.last_seen_commit or "", at=now,
))
row.path = dest
moved += 1
await session.commit()
return {"moved": moved, "replaced": replaced, "kept": kept}
def _event(row: CodeShape, event: str, at: datetime, *, commit: str = "") -> CodeShapeEvent:
"""A history row for a state change on ``row`` — status/snippet/by/reason
are the row's CURRENT values, which for `classified` is the judgment