feat(zotero): attach landed AMA files whose folder is not an exact OpenLibrary title

Year-aware containment match (keeps ICD-10-CM apart from ICD-10-PCS and
a 4th edition from a 6th); a local item is minted when no title covers a
landed book, so every file on disk is filed.

dev/scripts/add_ama_coding_to_zotero.py

@@ -51,10 +51,51 @@ def _norm(s: str) -> str:
 
 def _ol_url(key: str) -> str:
     olid = key.rsplit("/", 1)[-1]
+    if key.startswith("/local/"):
+        # Landed file with no OpenLibrary work: a stable local identity.
+        return f"local://ama-coding/{olid}"
     kind = "works" if olid.endswith("W") else "books"
     return f"https://openlibrary.org/{kind}/{olid}"
 
 
+_YEAR_RE = re.compile(r"\b(19|20)\d\d\b")
+
+
+def _year_of(s: str) -> str | None:
+    m = _YEAR_RE.search(s)
+    return m.group(0) if m else None
+
+
+def _folder_matches(folder: str, entries: list[dict]) -> dict | None:
+    """Best title entry for a landed folder: exact normalised title, else a
+    title that contains it (or is contained by it) once spaces are dropped,
+    with agreeing years. Containment keeps ICD-10-CM apart from ICD-10-PCS
+    and a 4th edition apart from a 6th; a similarity ratio does not."""
+    target = _norm(folder)
+    by_norm = {_norm(e["title"]): e for e in entries}
+    if target in by_norm:
+        return by_norm[target]
+    year = _year_of(folder)
+    tight = target.replace(" ", "")
+    best, best_len = None, 0
+    for e in entries:
+        ey = _year_of(e["title"]) or (str(e["year"]) if e.get("year") else None)
+        if year and ey and year != ey:
+            continue
+        cand = _norm(e["title"]).replace(" ", "")
+        if cand in tight or tight in cand:
+            if len(cand) > best_len:
+                best, best_len = e, len(cand)
+    return best
+
+
+def _folder_entry(folder: str) -> dict:
+    """A synthetic entry for a landed book no OpenLibrary title covers."""
+    year = _year_of(folder)
+    return {"key": f"/local/{_norm(folder).replace(' ', '-')}", "title": folder,
+            "year": int(year) if year else None, "isbns": []}
+
+
 def _index_books(books_dir: Path) -> dict[str, list[Path]]:
     """Map normalised folder title -> ebook files LazyLibrarian has landed."""
     index: dict[str, list[Path]] = {}
@@ -113,8 +154,9 @@ def _ensure_item(db: Db, entry: dict, collection_key: str, *, dry_run: bool) ->
         "accessDate": now,
         "language": "en",
         "libraryCatalog": "Open Library",
-        "extra": f"OpenLibrary: {entry['key'].rsplit('/', 1)[-1]}",
     }
+    if not entry["key"].startswith("/local/"):
+        fields["extra"] = f"OpenLibrary: {entry['key'].rsplit('/', 1)[-1]}"
     if entry.get("year"):
         fields["date"] = str(entry["year"])
     if entry.get("isbns"):
@@ -160,6 +202,23 @@ def main() -> None:
                     continue
                 _attach(db, item_id, f, dry_run=args.dry_run)
                 attached += 1
+        # Folders the exact-title pass did not consume: fuzzy-match a title
+        # (same year), else mint a local item so every landed file is filed.
+        exact = {_norm(e["title"]) for e in entries}
+        for folder_norm, files in books.items():
+            if folder_norm in exact:
+                continue
+            folder = files[0].parent.name
+            entry = _folder_matches(folder, entries) or _folder_entry(folder)
+            item_id = _ensure_item(db, entry, collection_key, dry_run=args.dry_run)
+            if item_id is None:
+                continue
+            have = _existing_attachment_names(db, item_id)
+            for f in files:
+                if f.name in have:
+                    continue
+                _attach(db, item_id, f, dry_run=args.dry_run)
+                attached += 1
         if not args.dry_run:
             db.commit()