Peach: Local-first video metadata scraper and media manager.
fix(entity): 人物页 JavDB 与 MISSAV 入口按现名与有作品的演员页给出
longmeidao12a39525f65d
diff --git a/src/peach/entry_links.py b/src/peach/entry_links.pyindex 879eddfcf..b3a14095c 100644--- a/src/peach/entry_links.py+++ b/src/peach/entry_links.py@@ -22,13 +22,17 @@- 日文艺名可含假名或纯汉字。汉字现名采用同名字形或已有成对译名的来源证据;对应的中文名必须仍是当前规范名,不同姓氏与可信假名艺名的冲突保留给复核。+ r18 的名字映射或目录站、avwikidb 资料把这个纯汉字现名记成她的名字时(`森日向子`),+ 它就是当前日文艺名,采用资料里的写法,不再从别名里改挑旧艺名(#271)。- 别名里的纯平假名读音不作为入口。`r18:performer` 每位都带一条(`しゃくありす`),选择艺名时需要区分它与人物的正式写法(`釈アリス`)。可信艺名来源标明的纯平假名正式艺名照用。- 同样形态的有多条时,优先 `avdb-actor-mapping`、`javdb`、`wiki` 这几类艺名来源。- 一条都挑不出时才用规范名:拼错的地址和没有地址相比,前者更难发现。**同一个站点可以有多枚入口。** 一位女优在 javdb 常有两个演员页,两边挂的作品不同,-所以每个 id 各出一枚,第二枚起在标签后缀一个序号。+所以每个 id 各出一枚,第二枚起在标签后缀一个序号。采集作品时在作品页上署过名的 id+(资料带 `provider_id`)排在前面,按名字回填的排在后面:旧艺名那页常年是空的+(`/actors/MmbDR` 白石アイリ「0 部影片」),排第一会让人以为链接错了(#271)。**两种位置。** minnano-av 是一份资料页,和事务所官网、社媒是同一类东西,所以它排进上面那排链接里。JavDB 与 MISSAV 是「去看片」的入口,另起一行,用两站自己的标识。@@ -216,18 +220,56 @@ def save(root: Path, body: dict) -> dict:return snapshot(root)+def _ref_metadata(ref) -> dict:+ """引用行的资料;没带资料列或坏 JSON 都按空的算。"""+ raw = ref.get("metadata_json")+ if isinstance(raw, dict):+ return raw+ try:+ value = json.loads(raw or "{}")+ except (TypeError, ValueError):+ return {}+ return value if isinstance(value, dict) else {}++def provider_ids(refs) -> dict[str, list[str]]:- """`entity_external_ref` 的行 → 这个站点上的 id 们。只认人物那一档。"""- found: dict[str, list[str]] = {}+ """`entity_external_ref` 的行 → 这个站点上的 id 们。只认人物那一档。++ 作品页上署过名的 id 排前面,同一档内保持原顺序;判据见模块开头。+ """+ found: dict[str, list[tuple[bool, str]]] = {}for ref in refs or ():if str(ref.get("external_kind") or "") != EXTERNAL_KIND:continueexternal_id = str(ref.get("external_id") or "").strip()provider = str(ref.get("provider") or "")ids = found.setdefault(provider, [])- if external_id and external_id not in ids:- ids.append(external_id)- return found+ if external_id and all(value != external_id for _, value in ids):+ ids.append((not _ref_metadata(ref).get("provider_id"), external_id))+ return {provider: [value for _, value in sorted(ids, key=lambda item: item[0])]+ for provider, ids in found.items()}+++#: 记下她现名的外部资料:目录站与 avwikidb 资料里的 `name`;r18 的名字映射另算。+_NAME_RECORDS = JAV_DIRECTORIES | {"avwikidb"}+++def recorded_name(name: str, refs) -> str:+ """外部资料把这个纯汉字现名记成她的名字时,返回资料里的写法;没有就是空串。"""+ if not _han_name(name):+ return ""+ for ref in refs or ():+ provider = str(ref.get("provider") or "")+ if provider == "r18" and ref.get("external_kind") == "performer_name":+ written = ref.get("external_id")+ elif provider in _NAME_RECORDS:+ written = _ref_metadata(ref).get("name")+ else:+ continue+ written = str(written or "").strip()+ if written and _han_name(written) and _glyph_key(written) == _glyph_key(name):+ return written+ return ""def _has_kana(written: str) -> bool:@@ -303,9 +345,12 @@ def _localized_han_name(name: str, aliases, localization) -> str:return jp-def japanese_name(canonical_name: str, aliases, localization=None) -> str:+def japanese_name(canonical_name: str, aliases, localization=None, refs=()) -> str:"""这位的日文艺名。挑不出来时退回规范名,判据见模块开头。"""name = str(canonical_name or "").strip()+ recorded = recorded_name(name, refs)+ if recorded:+ return recordedlocalized = _localized_han_name(name, aliases, localization)current = localized or nameequivalents = _glyph_equivalents(current, aliases)@@ -441,8 +486,13 @@ def build(settings: dict, canonical_name: str, refs, aliases=(), localization=No"""这条人物实体能直达的地址,按 SITES 的先后排好。缺 id 的站点不出现在结果里。"""ids = provider_ids(refs)jav = is_jav_performer(refs)- written = japanese_name(canonical_name, aliases, localization)+ written = japanese_name(canonical_name, aliases, localization, refs)missav_path = directory.path(written, canonical_name) if directory else None+ if missav_path == "":+ # 资料记下的现名在名录里没有页(她已改用别名里的新艺名)时,按别名挑出的那个再找一次。+ fallback = japanese_name(canonical_name, aliases, localization)+ if fallback != written and directory.path(fallback, canonical_name):+ written, missav_path = fallback, directory.path(fallback, canonical_name)out: list[dict] = []for site in SITES:row = settings.get(site.key) or {}diff --git a/src/peach/web_entity.py b/src/peach/web_entity.pyindex ffcfb90f9..ebd7d8931 100644--- a/src/peach/web_entity.py+++ b/src/peach/web_entity.py@@ -159,10 +159,13 @@ def _performer_entries(contract: WebContract, c, d: dict, alias_rows) -> dict:"""metadata = d.get("metadata") or {}localization = metadata.get("name_localization") if isinstance(metadata, dict) else None+ # 选艺名与排入口要看引用资料(记下的现名、作品页署名),资料不随页面下发。+ refs = [dict(r) for r in c.execute(+ "SELECT provider,external_kind,external_id,metadata_json FROM entity_external_ref "+ "WHERE entity_id=? ORDER BY provider,external_kind,external_id", (d["id"],))]out = {"entry_links": entry_links.entry_links(- contract.entry_links_root, d["canonical_name"], d["external_refs"], alias_rows,- localization)}- if (entry_links.javdb_actor_pages(d["canonical_name"], d["external_refs"])+ contract.entry_links_root, d["canonical_name"], refs, alias_rows, localization)}+ if (entry_links.javdb_actor_pages(d["canonical_name"], refs)or feeds.dmm_actress_ids(c, d["id"])):out["feed"] = {"following": feeds.entity_following(c, d["id"])}return outdiff --git a/tests/test_entry_links.py b/tests/test_entry_links.pyindex 5c27a891e..b6d8913ba 100644--- a/tests/test_entry_links.py+++ b/tests/test_entry_links.py@@ -47,6 +47,14 @@ class EntryLinkAddressTests(unittest.TestCase):["https://javdb.com/actors/d45k9","https://javdb.com/actors/ZX5z7"])+ def test_the_page_credited_on_her_works_comes_before_a_name_backfilled_one(self):+ """#271:旧艺名那页(白石アイリ)常年是空的,排第一会让人以为链接错了。"""+ backfilled = ref("javdb", "MmbDR") | {"metadata_json": json.dumps({"source": "backfill_performer_entry_ids"})}+ credited = ref("javdb", "bkxd") | {"metadata_json": json.dumps({"source": "javdb", "provider_id": "NNPJ-466"})}+ rows = entry_links.build(entry_links.defaults(), "森日向子", [backfilled, credited])+ self.assertEqual([row["url"] for row in rows if row["site"] == "javdb"],+ ["https://javdb.com/actors/bkxd", "https://javdb.com/actors/MmbDR"])+def test_a_site_without_an_id_is_absent_instead_of_a_search_address(self):rows = entry_links.build(entry_links.defaults(), "深田えいみ",[ref("r18dev", "1041521")])@@ -337,6 +345,33 @@ class MissavDirectoryTests(unittest.TestCase):[alias(jp)], directory=directory)self.assertEqual(next(row["url"] for row in rows if row["site"] == "missav"), address)+ def test_a_han_current_name_recorded_by_directories_beats_a_former_name_alias(self):+ """#271:r18 与 avwikidb 都把「森日向子」记成她的名字,可信来源的旧艺名不能顶替。"""+ aliases = [alias("白石アイリ", "av_name@namewikis-20260924"), alias("朝比奈えみり", "auto:performer-alias@1")]+ recorded = [ref("javdb", "bkxd"), ref("r18", "森日向子", "performer_name"),+ ref("avwikidb", "1") | {"metadata_json": json.dumps({"name": "森日向子", "kana": "もり ひなこ"})}]+ for refs in (recorded[:2], [recorded[0], recorded[2]]):+ with self.subTest(refs=[row["provider"] for row in refs]):+ self.assertEqual(entry_links.japanese_name("森日向子", aliases, refs=refs), "森日向子")+ self.assertEqual(entry_links.japanese_name("森日向子", aliases, refs=recorded[:1]), "白石アイリ")+ # 记下的是日文字形时用资料里的写法,中文规范名只用来对上她。+ self.assertEqual(entry_links.japanese_name("三上悠亚", [alias("ヒュージョン亀頭", "wiki")],+ refs=[ref("r18", "三上悠亜", "performer_name")]), "三上悠亜")+ address = "https://missav.ws/dm127/cn/actresses/" + quote("森日向子")+ directory = self.directory([{"name": "森日向子", "url": address},+ {"name": "白石アイリ", "url": "https://missav.ws/dm31/cn/actresses/X"}], complete=True)+ rows = entry_links.build(entry_links.defaults(), "森日向子", recorded, aliases, directory=directory)+ self.assertEqual(next(row["url"] for row in rows if row["site"] == "missav"), address)++ def test_a_recorded_name_without_a_directory_page_keeps_the_alias_page(self):+ """资料记下的是她改名前的写法、名录里只有新艺名的页时,照旧链到别名挑出的那页。"""+ address = "https://missav.ws/dm62/cn/actresses/" + quote("とみの伊織")+ directory = self.directory([{"name": "とみの伊織", "url": address}], complete=True)+ rows = entry_links.build(entry_links.defaults(), "富野伊织",+ [ref("javdb", "a"), ref("r18", "富野伊織", "performer_name")],+ [alias("とみの伊織", "wiki")], directory=directory)+ self.assertEqual(next(row["url"] for row in rows if row["site"] == "missav"), address)+def test_ambiguous_directory_names_have_no_missav_entry(self):directory = self.directory([{"name": "同名艺人", "url": "https://missav.ws/dm1/cn/actresses/A"},