feat(comments): attach combined.md back to bib + fix inline-body lookup

Two changes that, together, complete the loop from comment dir to a
single MD corpus per item in Zotero.

The inline-body lookup in cli.comments was querying items.key for a
CMS comment ID — but item keys are bib hashes (e.g. M5IR9YAE), so
every lookup returned None and every previously-extracted combined.md
was written with an empty inline body. Replace it with a one-shot
URL → comment_id parse over the items table, materialized into a
cache (164k+ rows × LIKE matching can't use any index).

Wire combined.md back to bib so it flows to Zotero via `bib sync`.
walker.walk_and_extract now accepts an `on_combined(comment_id, path)`
callback invoked on the main thread for every dir that ends up with
a combined.md — written or pre-existing. Skipped dirs get the
callback too, so backfill-attaching previously-extracted MDs on a
re-run works without --force. CLI exposes --attach/--no-attach
(default on; auto-off under --no-bib). Idempotent: pre-loads the
set of items already carrying a combined.md attachment.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

src/cli/comments.py

@@ -11,24 +11,32 @@ Phase 1 (this file):
 
 from __future__ import annotations
 
+import logging
 from pathlib import Path
 
 import typer
 
 app = typer.Typer(no_args_is_help=True)
+log = logging.getLogger(__name__)
 
 _DEFAULT_ROOT = Path(".state/comments")
 
 
-def _bib_lookup_factory(use_bib: bool):
-    """Return a callable ``comment_id -> inline body``. Empty if --no-bib.
+def _build_bib_helpers(use_bib: bool, attach: bool):
+    """Return ``(body_lookup, attach_callback)``.
 
-    Opens its own sqlite connection with ``check_same_thread=False`` so
-    walker worker threads can call it directly. Read-only path; SQLite's
-    default thread-safe mode handles the concurrent reads.
+    ``body_lookup(comment_id) -> str`` is called from worker threads and
+    returns the inline comment body from bib.sqlite (or "" if missing or
+    --no-bib). Backed by a comment_id → (item_key, body) cache built
+    once at startup — bib has 164k+ rows and the URL → comment_id parse
+    can't use any index, so we materialize the whole map up front.
+
+    ``attach_callback(comment_id, md_path) -> None`` runs on the main
+    thread and registers combined.md as a bib attachment, skipping items
+    that already have one. Returns None when --no-bib or --no-attach.
     """
     if not use_bib:
-        return lambda _cid: ""
+        return lambda _cid: "", None
 
     import sqlite3
 
@@ -37,13 +45,52 @@ def _bib_lookup_factory(use_bib: bool):
     db = str(path("db.bib"))
     con = sqlite3.connect(db, check_same_thread=False)
 
-    def lookup(comment_id: str) -> str:
-        row = con.execute(
-            "SELECT abstract FROM items WHERE key = ?", (comment_id,)
-        ).fetchone()
-        return (row[0] if row and row[0] else "") or ""
+    cache: dict[str, tuple[str, str]] = {}
+    for row in con.execute(
+        "SELECT key, url, abstract FROM items "
+        "WHERE url LIKE 'https://www.regulations.gov/comment/%'"
+    ):
+        cid = (row[1] or "").rsplit("/", 1)[-1]
+        if cid:
+            cache[cid] = (row[0], row[2] or "")
 
-    return lookup
+    def body_lookup(comment_id: str) -> str:
+        return cache.get(comment_id, ("", ""))[1]
+
+    if not attach:
+        return body_lookup, None
+
+    from bib import connect
+
+    store = connect()
+    existing: set[str] = {
+        row[0]
+        for row in store._con().execute(
+            "SELECT i.key FROM attachments a JOIN items i ON a.item_id = i.id "
+            "WHERE a.filename = 'combined.md'"
+        )
+    }
+
+    def attach_callback(comment_id: str, md_path: Path) -> None:
+        info = cache.get(comment_id)
+        if not info:
+            return  # comment not in bib
+        item_key = info[0]
+        if item_key in existing:
+            return  # already attached on a prior run
+        try:
+            store.attach_file(item_key, md_path, title="combined.md")
+            existing.add(item_key)
+        except Exception as e:  # noqa: BLE001
+            log.warning("attach combined.md failed for %s: %s", item_key, e)
+
+    return body_lookup, attach_callback
+
+
+# Back-compat alias — older code/tests may still import this name.
+def _bib_lookup_factory(use_bib: bool):
+    body, _ = _build_bib_helpers(use_bib, attach=False)
+    return body
 
 
 @app.command()
@@ -64,6 +111,12 @@ def extract(
         "--bib/--no-bib",
         help="Look up inline body from bib.sqlite items.abstract.",
     ),
+    attach: bool = typer.Option(
+        True,
+        "--attach/--no-attach",
+        help="Attach combined.md back to its bib item (so it flows to "
+        "Zotero via `bib sync`). Implies --bib; no-op when --no-bib.",
+    ),
 ) -> None:
     """Walk comment dirs and write combined.md (PDF/DOCX → text)."""
     from rex.comments.walker import walk_and_extract
@@ -72,7 +125,7 @@ def extract(
         typer.echo(f"root not found: {root}", err=True)
         raise typer.Exit(1)
 
-    lookup = _bib_lookup_factory(use_bib)
+    lookup, attach_cb = _build_bib_helpers(use_bib, attach=use_bib and attach)
     stats = walk_and_extract(
         root,
         inline_body_lookup=lookup,
@@ -80,6 +133,7 @@ def extract(
         limit=limit or None,
         workers=workers,
         force=force,
+        on_combined=attach_cb,
     )
     for k, v in stats.items():
         typer.echo(f"  {k}: {v}")

src/rex/comments/walker.py

@@ -27,6 +27,7 @@ def walk_and_extract(
     limit: int | None = None,
     workers: int = 8,
     force: bool = False,
+    on_combined: Callable[[str, Path], None] | None = None,
 ) -> dict[str, int]:
     """Process every comment dir under *root*.
 
@@ -35,22 +36,35 @@ def walk_and_extract(
     ``comment_id`` and returns the inline body text from bib.sqlite (or
     "" if missing).
 
+    *on_combined*, if given, is called on the main thread with
+    ``(comment_id, combined_md_path)`` for every dir that ends up with a
+    combined.md — both newly-written ones and pre-existing skips. Lets
+    callers attach the markdown to bib without crossing thread boundaries.
+
     Returns counts: ``{"written": N, "skipped": N, "failed": N}``.
     """
-    candidate_dirs, skipped = _collect_dirs(
+    candidate_dirs, skipped_dirs = _collect_dirs(
         root, docket=docket, limit=limit, force=force
     )
-    stats = {"written": 0, "skipped": skipped, "failed": 0}
-
-    if not candidate_dirs:
-        return stats
+    stats = {"written": 0, "skipped": len(skipped_dirs), "failed": 0}
 
+    if candidate_dirs:
         with ThreadPoolExecutor(max_workers=workers) as pool:
             futures = {
-            pool.submit(_one, d, inline_body_lookup, force): d for d in candidate_dirs
+                pool.submit(_one, d, inline_body_lookup, force): d
+                for d in candidate_dirs
             }
             for fut in as_completed(futures):
-            stats[fut.result()] += 1
+                status = fut.result()
+                stats[status] += 1
+                if on_combined and status == "written":
+                    cdir = futures[fut]
+                    on_combined(cdir.name, cdir / _COMBINED)
+
+    if on_combined:
+        for cdir in skipped_dirs:
+            on_combined(cdir.name, cdir / _COMBINED)
+
     return stats
 
 
@@ -60,10 +74,16 @@ def _collect_dirs(
     docket: str | None,
     limit: int | None,
     force: bool,
-) -> tuple[list[Path], int]:
-    """Return (dirs to process, count of dirs skipped due to existing combined.md)."""
+) -> tuple[list[Path], list[Path]]:
+    """Return ``(dirs_to_process, dirs_skipped)``.
+
+    ``dirs_skipped`` are dirs that already have combined.md and would not
+    be re-processed. We return them as a list (not a count) so callers
+    can still drive per-dir post-processing — e.g. attaching the existing
+    combined.md to bib on a re-run.
+    """
     todo: list[Path] = []
-    skipped = 0
+    skipped: list[Path] = []
     for docket_dir in sorted(root.iterdir()):
         if not docket_dir.is_dir():
             continue
@@ -73,7 +93,7 @@ def _collect_dirs(
             if not cdir.is_dir():
                 continue
             if (cdir / _COMBINED).is_file() and not force:
-                skipped += 1
+                skipped.append(cdir)
                 continue
             todo.append(cdir)
             if limit and len(todo) >= limit:

tests/cli/test_comments.py

@@ -57,3 +57,60 @@ def test_stats_after_extract(tmp_path: Path):
     result = runner.invoke(app, ["stats", "--root", str(tmp_path)])
     assert result.exit_code == 0, result.output
     assert "1" in result.output  # at least one comment counted
+
+
+def test_extract_attaches_combined_md_to_bib(tmp_path: Path, monkeypatch):
+    """End-to-end: extract writes combined.md AND attaches it to the bib item."""
+    from bib import connect
+    from bib.item import Source
+
+    bib_db = tmp_path / "bib.sqlite"
+    monkeypatch.setattr("conf.path", lambda _: bib_db)
+
+    cdir = _seed(tmp_path)
+    comment_id = cdir.name  # CMS-2024-0001-0001
+
+    store = connect(str(bib_db))
+    item = Source(
+        title=f"Comment {comment_id}",
+        url=f"https://www.regulations.gov/comment/{comment_id}",
+    )
+    item_key = store.upsert(item)
+    store.close()
+
+    result = runner.invoke(
+        app,
+        [
+            "extract",
+            "--root",
+            str(tmp_path),
+            "--workers",
+            "1",
+        ],
+    )
+    assert result.exit_code == 0, result.output
+    assert (cdir / "combined.md").is_file()
+
+    store = connect(str(bib_db))
+    rows = list(
+        store._con().execute(
+            "SELECT a.filename FROM attachments a JOIN items i ON a.item_id=i.id "
+            "WHERE i.key=? AND a.filename='combined.md'",
+            (item_key,),
+        )
+    )
+    store.close()
+    assert len(rows) == 1, "combined.md should be attached exactly once"
+
+    # Re-running must not duplicate the attachment (idempotence)
+    runner.invoke(app, ["extract", "--root", str(tmp_path), "--workers", "1"])
+    store = connect(str(bib_db))
+    rows = list(
+        store._con().execute(
+            "SELECT COUNT(*) FROM attachments a JOIN items i ON a.item_id=i.id "
+            "WHERE i.key=? AND a.filename='combined.md'",
+            (item_key,),
+        )
+    )
+    store.close()
+    assert rows[0][0] == 1, "second run created a duplicate attachment"

tests/rex/comments/test_walker.py

@@ -75,3 +75,37 @@ def test_walk_and_extract_respects_limit(tmp_path: Path):
         tmp_path, inline_body_lookup=_stub_inline_body, limit=1, workers=1
     )
     assert stats["written"] == 1
+
+
+def test_on_combined_called_for_written_dirs(tmp_path: Path):
+    a, b = _setup_two_comments(tmp_path)
+    seen: list[tuple[str, Path]] = []
+
+    walk_and_extract(
+        tmp_path,
+        inline_body_lookup=_stub_inline_body,
+        workers=2,
+        on_combined=lambda cid, p: seen.append((cid, p)),
+    )
+
+    assert sorted(cid for cid, _ in seen) == sorted([a.name, b.name])
+    assert all(p.name == "combined.md" for _, p in seen)
+
+
+def test_on_combined_called_for_skipped_dirs(tmp_path: Path):
+    """Pre-existing combined.md still gets the callback — so previously
+    extracted MDs can be backfill-attached on a later run."""
+    a, b = _setup_two_comments(tmp_path)
+    (a / "combined.md").write_text("---\n---\nstale\n")
+
+    seen: list[str] = []
+    stats = walk_and_extract(
+        tmp_path,
+        inline_body_lookup=_stub_inline_body,
+        workers=1,
+        on_combined=lambda cid, _p: seen.append(cid),
+    )
+
+    assert stats["written"] == 1
+    assert stats["skipped"] == 1
+    assert sorted(seen) == sorted([a.name, b.name])