diff --git a/backend/app/utils/sidecar.py b/backend/app/utils/sidecar.py index ae35b28..8df78b4 100644 --- a/backend/app/utils/sidecar.py +++ b/backend/app/utils/sidecar.py @@ -4,6 +4,7 @@ No per-platform branching: a small common key set with fallbacks; the full JSON is kept in raw so anything unmapped is recoverable later. """ +import re from dataclasses import dataclass from datetime import UTC, datetime from pathlib import Path @@ -21,10 +22,28 @@ class SidecarData: raw: dict +# gallery-dl prefixes media filenames with `NN_` for in-post ordering +# (`01_HOLLOW-ICHIGO.png`, `02_HOLOW ICHIGO.zip`) but writes the sidecar +# under the attachment's stem WITHOUT that ordering prefix +# (`HOLLOW-ICHIGO.json`). Strip the prefix when looking for sidecars. +# Confirmed against real Patreon downloads 2026-05-26 — without this +# strip, every gallery-dl post-level sidecar was invisible to FC since +# FC-3 shipped (24 deep-refresh calls produced 0 Posts in operator's DB). +_NUMBERING_PREFIX = re.compile(r"^\d+_(.+)$") + + def find_sidecar(media: Path) -> Path | None: + # Attachment-level sidecars (image.jpg.json, image.json). for cand in (media.with_suffix(".json"), Path(str(media) + ".json")): if cand.is_file(): return cand + # gallery-dl post-numbered convention: strip the `NN_` prefix from + # the stem and look for that.json in the same directory. + m = _NUMBERING_PREFIX.match(media.stem) + if m: + cand = media.parent / f"{m.group(1)}.json" + if cand.is_file(): + return cand return None diff --git a/tests/test_sidecar_util.py b/tests/test_sidecar_util.py index 0fd3e40..3abff31 100644 --- a/tests/test_sidecar_util.py +++ b/tests/test_sidecar_util.py @@ -23,6 +23,51 @@ def test_find_sidecar_none(tmp_path): assert find_sidecar(media) is None +def test_find_sidecar_gallerydl_numbered_prefix(tmp_path): + """gallery-dl prefixes media filenames with NN_ for in-post ordering + (e.g. `01_HOLLOW-ICHIGO.png`) but writes the post-level sidecar under + the attachment stem WITHOUT the prefix (`HOLLOW-ICHIGO.json`). + Confirmed against real Patreon downloads 2026-05-26 — operator's deep + scan produced 24 refresh calls but 0 Posts because the unprefixed + sidecar was invisible to find_sidecar.""" + media = tmp_path / "01_HOLLOW-ICHIGO.png" + media.write_bytes(b"x") + sc = tmp_path / "HOLLOW-ICHIGO.json" + sc.write_text("{}") + assert find_sidecar(media) == sc + + +def test_find_sidecar_multidigit_prefix(tmp_path): + """Numbering prefix can be wider than 2 digits (`001_...`); the strip + handles any \\d+_ form.""" + media = tmp_path / "001_mirko-sketch.png" + media.write_bytes(b"x") + sc = tmp_path / "mirko-sketch.json" + sc.write_text("{}") + assert find_sidecar(media) == sc + + +def test_find_sidecar_prefers_attachment_level_over_post_level(tmp_path): + """If BOTH a per-attachment sidecar and a post-level sidecar exist, + the attachment-level one wins (it's more specific).""" + media = tmp_path / "01_image.png" + media.write_bytes(b"x") + per_attachment = tmp_path / "01_image.json" + per_attachment.write_text('{"specific": true}') + post_level = tmp_path / "image.json" + post_level.write_text('{"specific": false}') + assert find_sidecar(media) == per_attachment + + +def test_find_sidecar_no_underscore_not_treated_as_prefix(tmp_path): + """`01.png` (just digits, no underscore-separated stem) shouldn't + match. The regex requires NN_.""" + media = tmp_path / "01.png" + media.write_bytes(b"x") + (tmp_path / ".json").write_text("{}") # would be matched only if buggy + assert find_sidecar(media) is None + + def test_parse_empty_dict_all_none(): sd = parse_sidecar({}) assert isinstance(sd, SidecarData)