v2.2.5

NEWS.md

@@ -6,6 +6,20 @@ width: 128px;
 border-radius: 128px;
 " />
 
+## v2.2.5
+
+- Features
+    - Online tagging combines search and tag by id. A source given an issue id
+      fetches it directly, the rest search, and the results merge (comicbox
+      4.8.0).
+
+- Fixes
+    - Tagging a comic with renaming enabled no longer fails the import and
+      discards the new tags. Comics renamed this way since v2.1.0 need a Force
+      Update Tags to pick up their metadata.
+    - Comics renamed or moved by another program mid-scan no longer leave a
+      permanent failed import.
+
 ## v2.2.4
 
 - Features

codex/librarian/fs/import_task.py

@@ -65,7 +65,18 @@ def _deduplicate(kwargs: dict[str, Any]) -> None:
 
     kwargs["files_modified"] -= kwargs["files_added"]
     kwargs["files_modified"] -= kwargs["files_deleted"]
-    kwargs["files_modified"] -= files_dest_paths
+    # A modified path that takes part in a move refers to content that now
+    # lives at the destination. The watcher emits the source path (an
+    # external tagger writing tags and then renaming produces the modify
+    # before the rename), while the poller emits the destination path on
+    # purpose for a move whose stats also changed. Remap sources and keep
+    # destinations so both reach the read phase alive. A pure rename's
+    # destination costs only a stat() there, because a move preserves the
+    # stored stat (see MOVED_BULK_COMIC_UPDATE_FIELDS).
+    files_moved = kwargs["files_moved"]
+    kwargs["files_modified"] = {
+        files_moved.get(path, path) for path in kwargs["files_modified"]
+    }
 
     kwargs["covers_modified"] -= kwargs["covers_deleted"]
     kwargs["covers_modified"] -= kwargs["covers_added"]

codex/librarian/onlinetag/explicit_id.py

@@ -2,12 +2,18 @@
 Fetch comic metadata by a known online issue id (comicbox explicit-id tagging).
 
 Codex's online tagging normally *searches* a source by filename and ranks
-candidates. When the operator already knows the exact Metron / Comic Vine issue
-id, this module drives comicbox's explicit-id path instead
-(``online.lookup.ids``), which skips the search entirely and calls
-``source.get(issue_id)`` for that one issue. It mirrors the per-file flow of
-``comicbox.online_session.OnlineSession._run_one`` minus the selector — an
-explicit-id fetch never reaches the matcher, so it never prompts.
+candidates. When a comic's issue id is already known, this module drives
+comicbox's explicit-id path instead (``online.lookup.ids``), which skips the
+search entirely and calls ``source.get(issue_id)`` for that one issue. It
+mirrors the per-file flow of ``comicbox.online_session.OnlineSession._run_one``
+minus the selector — an explicit-id fetch never reaches the matcher, so it
+never prompts.
+
+Two callers, both outside a running session: the stored-id prepass (comics
+Codex already holds an identifier for) and prompt resolution (an admin pinned
+the issue by hand). Ids pinned in the tagging request itself go through the
+session — ``OnlineSession(ids=...)`` — so pinned sources and searched sources
+resolve in one lookup and merge.
 """
 
 from __future__ import annotations

codex/librarian/onlinetag/onlinetagd.py

@@ -8,7 +8,6 @@
 from codex.librarian.onlinetag.tasks import (
     BulkOnlineTagTask,
     OnlineTagAbortTask,
-    OnlineTagByIdTask,
     OnlineTagDismissTask,
     OnlineTagPromptResponseTask,
     OnlineTagSkipAllPromptsTask,
@@ -64,8 +63,6 @@ def process_item(self, item) -> None:
         match item:
             case BulkOnlineTagTask():
                 self.session_manager.run_session(item)
-            case OnlineTagByIdTask():
-                self.session_manager.tag_by_id(item)
             case OnlineTagAbortTask():
                 self.session_manager.cancel_session(item.session_id)
             case OnlineTagPromptResponseTask():

codex/librarian/onlinetag/session_manager.py

@@ -58,7 +58,6 @@
 from codex.librarian.onlinetag.tasks import (
     BulkOnlineTagTask,
     OnlineTagAbortTask,
-    OnlineTagByIdTask,
     OnlineTagPromptResponseTask,
     OnlineTagSkipAllPromptsTask,
 )
@@ -130,71 +129,6 @@ def _source_has_credentials(credentials: OnlineCredentials, source: str) -> bool
             return bool(credentials.comicvine_key)
         return False
 
-    # --- tag by explicit id (no search) --------------------------------
-
-    def tag_by_id(self, task: OnlineTagByIdTask) -> None:
-        """
-        Tag one comic by a known online issue id, skipping search entirely.
-
-        The operator already knows the exact Metron / Comic Vine issue, so we
-        fetch that record directly and hand the tags to the same
-        ``BulkTagWriteTask`` write + re-import path the scan uses. A wrong or
-        unknown id resolves to nothing; that surfaces in the admin Tagging-tab
-        error panel rather than silently re-writing the comic's existing tags.
-        """
-        comic = (
-            Comic.objects.filter(pk=task.comic_pk)
-            .exclude(library__read_only=True)
-            .only("pk", "path")
-            .first()
-        )
-        if not comic:
-            self.log.warning(f"Online tag by id: comic {task.comic_pk} not found.")
-            return
-        path = Path(comic.path)
-        credentials = self._build_credentials()
-        if not credentials or not self._source_has_credentials(
-            credentials, task.source
-        ):
-            self.log.warning(
-                f"Online tag by id: no {task.source} credentials configured."
-            )
-            return
-
-        try:
-            tags = fetch_tags_by_explicit_id(
-                path,
-                task.source,
-                task.issue_id,
-                credentials,
-                extra_ids=task.extra_ids,
-            )
-        except (ComicboxError, OSError) as exc:
-            msg = f"Fetching {task.source} issue {task.issue_id} failed: {exc}"
-            self.log.warning(f"Online tag by id: {msg} ({path})")
-            add_tag_write_error(str(path), msg)
-            self.librarian_queue.put(TAG_WRITE_ERRORS_CHANGED_TASK)
-            return
-        if not tags:
-            msg = f"No {task.source} issue found for id {task.issue_id}."
-            self.log.warning(f"Online tag by id: {msg} ({path})")
-            add_tag_write_error(str(path), msg)
-            self.librarian_queue.put(TAG_WRITE_ERRORS_CHANGED_TASK)
-            return
-
-        write_task = BulkTagWriteTask(
-            comic_pks=frozenset({comic.pk}),
-            per_comic_patches={comic.pk: tags},
-            mode="update",
-            formats=task.formats,
-            delete_original=task.delete_original,
-            rename=task.rename,
-        )
-        self.librarian_queue.put(write_task)
-        self.log.info(
-            f"Online tag by id: applied {task.source}:{task.issue_id} to {path}."
-        )
-
     # --- events --------------------------------------------------------
 
     def _active_state(self) -> SessionState | None:
@@ -527,10 +461,8 @@ def _prefetch_stored_ids(
         Fetches each already-identified comic in one API call, writes the tags
         through the same ``BulkTagWriteTask`` path the scan uses, and drops it
         from ``comic_paths`` so the search session only handles the rest.
-        Skipped on dry runs and for sources without credentials.
+        Skipped for sources without credentials.
         """
-        if task.dry_run:
-            return
         usable_sources = tuple(
             source
             for source in task.sources
@@ -586,6 +518,8 @@ def run_session(self, task: BulkOnlineTagTask) -> None:
         )
         session = OnlineSession(
             sources=task.sources,
+            # Pinned sources fetch their issue id directly; the rest search.
+            ids=dict(task.ids),
             credentials=credentials,
             mode=MatchMode(task.mode),
             defer_prompts=defer_prompts,
@@ -615,7 +549,7 @@ def run_session(self, task: BulkOnlineTagTask) -> None:
                 "delete_original": task.delete_original,
                 "merge_all_sources": task.merge_all_sources,
                 "rename": task.rename,
-                "dry_run": task.dry_run,
+                "ids": dict(task.ids),
             },
         )
         self._active_session_id = task.session_id
@@ -633,7 +567,12 @@ def run_session(self, task: BulkOnlineTagTask) -> None:
             # Fast path: comics codex already has an issue id for are fetched
             # directly by that id (one API call each) and dropped from the set,
             # so the search pass below only handles the unidentified remainder.
-            self._prefetch_stored_ids(state, comic_paths, task, credentials)
+            # Skipped when the request pinned ids: the prepass would drop that
+            # single comic out of the session entirely, but the pinned sources
+            # must be fetched *and* the unpinned ones searched in one lookup so
+            # merge_all_sources can merge both results.
+            if not task.ids:
+                self._prefetch_stored_ids(state, comic_paths, task, credentials)
             # Pass 1: auto-match and write the confident comics. When deferring,
             # ambiguous matches become deferred prompts persisted for later,
             # independent resolution; with "never" prompts they're skipped inline

codex/librarian/onlinetag/tasks.py

@@ -1,6 +1,6 @@
 """Online tagging tasks."""
 
-from dataclasses import dataclass
+from dataclasses import dataclass, field
 from typing import Any
 
 from comicbox.formats.base.online import SOURCE_NAMES
@@ -22,12 +22,15 @@ class BulkOnlineTagTask(OnlineTagTask):
     mode: str = "auto"
     prompts_mode: str = "ask"
     delete_original: bool = False
-    dry_run: bool = False
     # Query every source per comic and merge (comicbox first_wins=False)
     # instead of stopping at the first match.
     merge_all_sources: bool = False
     # Rename each written archive to the comicbox filename scheme.
     rename: bool = False
+    # Pinned issue ids by source for a single-comic session. A pinned source
+    # fetches that issue directly; the rest of ``sources`` search as usual, so
+    # one session mixes tag-by-id and search and merges both results.
+    ids: dict[str, int] = field(default_factory=dict)
 
 
 @dataclass
@@ -37,22 +40,6 @@ class OnlineTagAbortTask(OnlineTagTask):
     session_id: str = ""
 
 
-@dataclass
-class OnlineTagByIdTask(OnlineTagTask):
-    """Tag one comic by a known online issue id (skips search)."""
-
-    comic_pk: int
-    source: str
-    issue_id: int
-    formats: tuple[str, ...] = ("COMIC_INFO",)
-    delete_original: bool = False
-    rename: bool = False
-    # Additional (source, issue_id) pairs to fetch by explicit id and merge
-    # onto the primary fetch — the Tag-by-id "merge all sources" path. Empty
-    # means a single-source fetch.
-    extra_ids: tuple[tuple[str, int], ...] = ()
-
-
 @dataclass
 class OnlineTagPromptResponseTask(OnlineTagTask):
     """Admin response to a deferred online tagging prompt (keyed by fingerprint)."""

codex/librarian/scribe/importer/const.py

@@ -482,7 +482,16 @@ def get_through_model(field: ManyToManyField) -> type[BaseModel]:
 #########
 # MOVED #
 #########
-MOVED_BULK_COMIC_UPDATE_FIELDS = ("path", "parent_folder", "stat", "updated_at")
+# ``stat`` is deliberately absent: the stored stat means "the state of the
+# file the last time its tags were imported", not "the last time codex saw
+# the file". A pure rename changes neither inode, mtime, nor size, so the
+# preserved stat still matches disk exactly and inode move-detection keeps
+# working. When content changed too (an external tagger writes tags and
+# then renames), the preserved stat is the only evidence the read phase's
+# filesystem prefilter has that the archive needs re-reading. update_comics
+# writes a fresh stat once the new tags actually import
+# (ALWAYS_UPDATE_COMIC_FIELDS).
+MOVED_BULK_COMIC_UPDATE_FIELDS = ("path", "parent_folder", "updated_at")
 CUSTOM_COVER_UPDATE_FIELDS = ("path", "stat", "updated_at", "sort_name", "collection")
 
 ###########

codex/librarian/scribe/importer/failed/create.py

@@ -72,8 +72,12 @@ def _bulk_create_failed_imports(
                         library=self.library, path=path, parent_folder=None
                     )
                     fi.set_reason(exc)
-                    create_objs.append(fi)
+                    # presave() stats the path, so it must succeed before
+                    # the row is queued. A file that vanished mid-import
+                    # (renamed out from under us) raises here and gets no
+                    # failed import row.
                     fi.presave()
+                    create_objs.append(fi)
                 except OSError:
                     self.log.warning(
                         f"Error preparing failed import create for {path}: {exc}"

codex/librarian/scribe/importer/read/extract.py

@@ -194,7 +194,9 @@ def _extract_post_process_comic(
         result, exc = value
         path_str = str(path)
         if exc:
-            self.metadata[FIS].update({path: exc})
+            # str keys, not Path: the failed import phase matches these
+            # against path strings queried from the db.
+            self.metadata[FIS].update({path_str: exc})
             self.metadata[SKIPPED].add(path_str)
             status.increment_complete()
             self.status_controller.update(status)