diff --git a/services/app/domain/release_profiles.py b/services/app/domain/release_profiles.py index bd01356b..c7051b32 100644 --- a/services/app/domain/release_profiles.py +++ b/services/app/domain/release_profiles.py @@ -2,6 +2,7 @@ import re import unicodedata +from itertools import permutations from typing import Literal from uuid import UUID @@ -391,6 +392,21 @@ def identifier_values(value): ) +def co_author_credits(names): + """Every author together, in any order, joined by a space, " and " or ", ". + + "&" needs no joiner of its own: normalising turns it into a space. + """ + names = [name for name in names if normalized(name)] + if not 2 <= len(names) <= 3: + return set() + return { + normalized(joiner.join(order)) + for order in permutations(names) + for joiner in (" ", " and ", ", ") + } + + def indexer_title_authors(release, work): """An exact author/title pair can supply the missing structured indexer fields. @@ -426,14 +442,18 @@ def indexer_title_authors(release, work): authors = {} for value in work["authors"]: if author := normalized(value): - authors[author] = value + authors[author] = (value,) # Inverted credits must carry the comma: don't accept arbitrary word # permutations as an author match. Splits also cover compound surnames. words = value.split() for split in range(1, len(words)): given, surname = " ".join(words[:split]), " ".join(words[split:]) if re.search(rf"\b{re.escape(surname)},\s*{re.escape(given)}\b", title, re.I): - authors[normalized(f"{surname} {given}")] = value + authors[normalized(f"{surname} {given}")] = (value,) + # Co-written books often credit every author before the title: + # "Terry Pratchett & Stephen Baxter - The Long Earth". + for credit in co_author_credits(work["authors"]): + authors[credit] = tuple(work["authors"]) matched = set() for expected in expected_titles - {""}: for author in authors: @@ -443,7 +463,7 @@ def indexer_title_authors(release, work): if pair in labelled_titles or re.fullmatch( rf"{re.escape(pair)}(?: (?:19|20)\d{{2}})?(?: retail)?", actual ): - matched.add(authors[author]) + matched.update(authors[author]) if not matched and not re.search(r"\.(?:rar|zip|7z|par2)$", title, re.I): # Indexers often use "Author - Title- Subtitle" when the catalog has # only the short title. Keep the author exact and use the shared subtitle @@ -454,7 +474,7 @@ def indexer_title_authors(release, work): if compatible_title( source_title, work["title"], allow_extra_subtitle=True ) or title_outside_series_prefix(parts[1], work): - matched.add(authors[credit]) + matched.update(authors[credit]) # AudiobookBay names, including Prowlarr's copy of them, are # "Title - Author, Narrator". The catalog author is the first credit. if not matched: @@ -470,7 +490,7 @@ def indexer_title_authors(release, work): normalized(head) in expected_titles or (bare is not None and full_title and normalized(bare) == full_title) ): - matched.add(authors[credit]) + matched.update(authors[credit]) # "Series 5 - Author" omits the catalog subtitle. Accept it only when that # series and position are already on the work and the first credit matches. if not matched: @@ -488,7 +508,7 @@ def indexer_title_authors(release, work): if isinstance(entry, dict) ) ): - matched.add(authors[credit]) + matched.update(authors[credit]) return sorted(matched) @@ -508,7 +528,9 @@ def catalog_language_release(release, work): normalized(parse_title_labels(work["title"]).title), normalized(optional_subtitle_base(work["title"])), } - {""} - authors = {normalized(author) for author in work["authors"]} - {""} + authors = ({normalized(author) for author in work["authors"]} - {""}) | co_author_credits( + work["authors"] + ) for author in work["authors"]: words = author.split() for split in range(1, len(words)): diff --git a/tests/unit/test_release_name_identity.py b/tests/unit/test_release_name_identity.py index 807b4e7b..5dab49de 100644 --- a/tests/unit/test_release_name_identity.py +++ b/tests/unit/test_release_name_identity.py @@ -1,10 +1,16 @@ """Reported indexer layouts must preserve the full catalog-backed identity.""" +from datetime import UTC, datetime from types import SimpleNamespace import pytest -from app.domain.release_profiles import indexer_title_authors +from app.adapters.prowlarr import ProwlarrRelease +from app.domain.release_profiles import ( + catalog_language_release, + co_author_credits, + indexer_title_authors, +) from app.domain.series_identity import series_note_agrees @@ -151,3 +157,58 @@ def test_supported_layout_still_requires_structured_author_agreement_and_catalog assert authors(title, authors=["Other Writer"]) == [] assert authors(title, position=None) == [] assert authors(title) == ["Stephen King"] + + +CO_WRITTEN = {"title": "The Long Earth", "authors": ["Terry Pratchett", "Stephen Baxter"]} + + +def co_written(title): + release = SimpleNamespace(source="prowlarr", title=title, raw_title=title, authors=[]) + return indexer_title_authors(release, CO_WRITTEN) + + +def test_every_co_author_together_is_one_credit(): + credits = co_author_credits(CO_WRITTEN["authors"]) + assert "terry pratchett stephen baxter" in credits + assert "stephen baxter and terry pratchett" in credits + assert co_author_credits(["Terry Pratchett"]) == set() + + +@pytest.mark.parametrize( + "title", + [ + "Terry Pratchett Stephen Baxter - The Long Earth 2012 MP3", + "Terry Pratchett & Stephen Baxter - The Long Earth", + "Stephen Baxter and Terry Pratchett - The Long Earth", + "The Long Earth - Terry Pratchett & Stephen Baxter", + ], +) +def test_release_names_crediting_every_co_author_match(title): + assert co_written(title) == ["Stephen Baxter", "Terry Pratchett"] + + +@pytest.mark.parametrize( + "title", + [ + "Terry Pratchett Neil Gaiman - The Long Earth", + "Terry Pratchett & Stephen Baxter - The Long Mars", + ], +) +def test_another_co_author_or_book_does_not_match(title): + assert co_written(title) == [] + + +def test_language_after_every_co_author_and_title_is_inferred(): + title = "Terry Pratchett & Stephen Baxter - The Long Earth German Hoerbuch" + source = ProwlarrRelease( + source_id="co-author-fixture", + acquisition_supported=True, + title=title, + raw_title=title, + medium="audio", + protocol="nzb", + indexer_name="Fixture", + categories=[3030], + observed_at=datetime.now(UTC), + ) + assert catalog_language_release(source, CO_WRITTEN).language == "de"