tests/test_javdb_cn_names.py
Rama master · réplica de lectura del repositorio original.
"""javdb 资料页取中文名：现名与旧名必须分开，中文写法只认一个。
配错一位就是把另一个人的名字写进真相字段。这里守的是「什么才算证据」。
"""
import importlib.util
import sqlite3
import sys
import tempfile
import unittest
from pathlib import Path
from urllib.parse import quote
from peach import javdb
from peach.entities import normalize_entity_name
REPO = Path(__file__).resolve().parents[1]
#: 资料页顶部的真实结构（2026-09-04 实测）。现名一栏中日并列，紧随的是旧艺名。
PAGE = """
<div class="column section-title">
<h2 class="title is-4">
<span class="actor-section-name">{current}</span>
<br />
<span class="section-meta">{former}</span>
<br>
<span class="section-meta">324 部影片</span>
</h2>
</div>
<a id="button-collect-actor" href="/actors/{actor}/collect"></a>
"""
def page(current: str, former: str = "", actor: str = "RJM8") -> str:
return PAGE.format(current=current, former=former, actor=actor)
def load_script(name: str):
"""按文件路径加载脚本。
要先登记进 `sys.modules` 再执行：`localize_performer_names` 里有 dataclass，
而 `dataclasses` 会去 `sys.modules[cls.__module__]` 取模块字典，没登记就是
`'NoneType' object has no attribute '__dict__'`——报错指向 dataclass，
真正的原因在这一行。
"""
sys.path.insert(0, str(REPO / "src"))
spec = importlib.util.spec_from_file_location(name, REPO / "scripts" / f"{name}.py")
module = importlib.util.module_from_spec(spec)
sys.modules[name] = module
spec.loader.exec_module(module)
return module
def load_module():
return load_script("harvest_javdb_cn_names")
class ParseTests(unittest.TestCase):
def test_the_current_name_and_the_former_names_are_two_different_lists(self):
"""`JULIA` 那页的旧名里有 `京香じゅりあ`。混成一串就分不出这是不是她的中文名。"""
html = page("JULIA", "京香じゅりあ")
self.assertEqual(javdb.current_names(html), ["JULIA"])
self.assertEqual(javdb.former_names(html), ["京香じゅりあ"])
def test_the_film_count_is_not_a_name(self):
self.assertNotIn("324 部影片", javdb.former_names(page("深田詠美, 深田えいみ")))
def test_both_writings_of_the_current_name_come_back_in_order(self):
self.assertEqual(javdb.current_names(page("深田詠美, 深田えいみ")),
["深田詠美", "深田えいみ"])
def test_the_record_kind_suffix_is_not_part_of_the_name(self):
"""无码那条记录的现名挂着 `(無碼)`。留着会让同一个人变成两个人。"""
self.assertEqual(javdb.current_names(page("木下ひまり(無碼)")), ["木下ひまり"])
def test_the_actor_id_is_the_page_identity(self):
self.assertEqual(javdb.actor_id(page("深田詠美, 深田えいみ", actor="Ng03")), "Ng03")
def test_a_search_result_that_names_nobody_we_want_is_not_a_hit(self):
html = ('<div class="box actor-box"><a href="/actors/AAA" title="別人, べつじん">'
'</a></div><div class="box actor-box">'
'<a href="/actors/BBB" title="深田詠美, 深田えいみ"></a></div>')
self.assertEqual(javdb.search_hits(html, {"深田えいみ"}), ["/actors/BBB"])
def test_a_wider_key_can_be_supplied_for_matching(self):
"""账本写 `宫`、站上写 `宮`，字形差别不该让人搜不到自己。"""
html = ('<div class="box actor-box"><a href="/actors/AAA" title="宮本櫻, 宮本さくら">'
'</a></div>')
self.assertEqual(javdb.search_hits(html, {"宫本さくら"}), [])
self.assertEqual(
javdb.search_hits(html, {"宫本さくら"}, lambda name: name.replace("宮", "宫")),
["/actors/AAA"])
def test_two_pages_for_one_name_both_come_back(self):
"""同一个名字在站上真有两位时，取第一个是默默替用户挑了一位。"""
html = ('<div class="box actor-box"><a href="/actors/AAA" title="あおい"></a></div>'
'<div class="box actor-box"><a href="/actors/BBB" title="あおい"></a></div>')
self.assertEqual(javdb.search_hits(html, {"あおい"}), ["/actors/AAA", "/actors/BBB"])
class JudgeTests(unittest.TestCase):
def setUp(self):
self.module = load_module()
self.record = {"entity_id": 1, "name": "深田えいみ", "assets": 6,
"chain": ["深田えいみ", "天海こころ"]}
def judge(self, current, former=""):
return self.module.judge(self.record, page(current, former), "https://javdb.com/x")
def test_a_name_written_both_ways_on_one_page_is_the_evidence(self):
row = self.judge("深田詠美, 深田えいみ")
self.assertEqual(row["verdict"], "ok")
self.assertEqual(row["zh_tw"], "深田詠美")
self.assertEqual(row["jp"], "深田えいみ")
def test_a_ledger_name_found_only_among_former_names_is_left_to_a_human(self):
"""站上已经改了名。用哪个当规范名是人要决定的事，不是这一步该定的。"""
row = self.judge("JULIA", "深田えいみ")
self.assertEqual(row["verdict"], "旧名")
self.assertIn("JULIA", row["evidence"])
def test_a_page_with_only_a_japanese_name_yields_nothing(self):
"""`長谷川るい` 那页只有日文写法。没有就是没有，不转写。"""
self.record.update(name="長谷川るい", chain=["長谷川るい"])
row = self.judge("長谷川るい")
self.assertEqual(row["verdict"], "同形（站上只有日文名）")
self.assertEqual(row["zh_tw"], "")
def test_two_chinese_writings_in_one_field_are_ambiguous(self):
self.record.update(name="田中レモン", chain=["田中レモン", "楓カレン"])
row = self.judge("田中檸檬, 田中レモン, 楓花戀")
self.assertEqual(row["verdict"], "多义")
def test_two_different_stage_names_in_the_current_field_are_not_a_pair(self):
"""`一之瀨亞美莉, 美空あやか` 是两个艺名。姓对不上，中文那个不是这位的名字。"""
self.record.update(name="美空あやか", chain=["美空あやか"])
row = self.judge("一之瀨亞美莉, 美空あやか", "一ノ瀬アメリ")
self.assertEqual(row["verdict"], "不同名（两栏姓氏对不上）")
self.assertEqual(row["zh_cn" if "zh_cn" in row else "zh_tw"], "")
def test_a_shared_single_character_is_not_a_shared_surname(self):
"""`美空` 与 `亞美莉` 都有 `美`。按字取交集会把配错的那一对放过去。"""
self.assertFalse(self.module.same_person("一之瀨亞美莉", "美空あやか"))
self.assertTrue(self.module.same_person("深田詠美", "深田えいみ"))
def test_a_kana_only_japanese_name_has_no_surname_to_compare(self):
"""`あべみかこ` 没有汉字可比。无从判不等于判否。"""
self.assertTrue(self.module.same_person("安部未華子", "あべみかこ"))
def test_a_glyph_difference_is_not_a_name_difference(self):
"""`宫本さくら` 与 `宮本さくら` 是一个人，`宮` 与 `宫` 只是字形。"""
self.record.update(name="宫本さくら", chain=["宫本さくら"])
self.assertEqual(self.judge("宮本櫻, 宮本さくら")["verdict"], "ok")
def test_a_page_matched_only_through_an_alias_is_left_to_a_human(self):
"""账本规范名不在现名栏：站上写的是另一个艺名，换不换规范名是人的决定。"""
self.record.update(name="おっぱい隊長", chain=["おっぱい隊長", "夢見るぅ"])
row = self.judge("夢見露, 夢見るぅ")
self.assertEqual(row["verdict"], "改艺名（现名栏是另一个艺名）")
self.assertIn("おっぱい隊長", row["evidence"])
def test_a_latin_stage_name_is_not_a_chinese_name(self):
self.record["name"] = "京香じゅりあ"
self.record["chain"] = ["京香じゅりあ", "JULIA"]
row = self.judge("JULIA, 京香じゅりあ")
self.assertEqual(row["verdict"], "同形（站上只有日文名）")
class SimplifyTests(unittest.TestCase):
def setUp(self):
self.module = load_module()
self.record = {"entity_id": 1, "name": "深田えいみ", "assets": 6,
"chain": ["深田えいみ"]}
def test_the_traditional_writing_is_kept_next_to_the_simplified_one(self):
"""复核件要看得出转了什么：只留简体的话，转错了没人能发现。"""
rows = self.module.localize(
[{"verdict": "ok", "zh_tw": "永瀨未萌", "adopt": "永瀨未萌"},
{"verdict": "旧名", "zh_tw": "", "adopt": "JULIA"}])
self.assertEqual(rows[0]["zh_cn"], "永濑未萌")
self.assertEqual(rows[0]["zh_tw"], "永瀨未萌")
self.assertEqual(rows[1]["zh_cn"], "")
def test_a_japanese_current_name_is_not_half_converted(self):
"""`きみかわ結衣` 转成 `きみかわ结衣` 是半假名半简体，比原样留着更糟。"""
rows = self.module.localize([{"verdict": "旧名", "zh_tw": "", "adopt": "きみかわ結衣"},
{"verdict": "旧名", "zh_tw": "", "adopt": "美月麗花"}])
self.assertEqual(rows[0]["adopt"], "きみかわ結衣")
self.assertEqual(rows[1]["adopt"], "美月丽花")
def test_a_site_renamed_row_carries_the_new_name_as_the_candidate(self):
"""改了名的那些也要有候选，不然复核件里只写着「站上不是这个名字了」。"""
self.record.update(name="京香じゅりあ", chain=["京香じゅりあ", "JULIA"])
row = self.module.judge(self.record, page("JULIA", "京香じゅりあ"), "https://javdb.com/x")
self.assertEqual(row["verdict"], "改艺名（现名栏是另一个艺名）")
self.assertEqual(row["adopt"], "JULIA")
self.assertEqual(row["jp"], "京香じゅりあ")
class TargetTests(unittest.TestCase):
def setUp(self):
self.module = load_module()
self.tmp = tempfile.TemporaryDirectory()
self.con = sqlite3.connect(Path(self.tmp.name).resolve() / "ledger.db")
self.con.executescript(
"CREATE TABLE entity(id INTEGER PRIMARY KEY, kind TEXT, canonical_name TEXT);"
"CREATE TABLE entity_alias(entity_id INTEGER, alias TEXT);"
"CREATE TABLE asset_entity(asset_id INTEGER, entity_id INTEGER, source TEXT);")
self.con.executemany(
"INSERT INTO entity(id,kind,canonical_name) VALUES(?,'performer',?)",
[(1, "深田えいみ"), (2, "三上悠亚"), (3, "MattieDoll"), (4, "河合あすな")])
self.con.executemany(
"INSERT INTO asset_entity(asset_id,entity_id,source) VALUES(?,?,?)",
[(10, 1, "r18:performer"), (11, 2, "r18:performer"),
(12, 3, "legacy:asset"), (13, 4, "javbus:performer")])
self.con.execute("INSERT INTO entity_alias VALUES(1,'天海こころ')")
self.con.commit()
def tearDown(self):
self.con.close()
self.tmp.cleanup()
def test_only_kana_names_backed_by_release_metadata_are_targets(self):
"""已经是中文的不用查；账号型 performer 不是女优，不翻译。"""
found = {row["name"] for row in self.module.targets(self.con)}
self.assertEqual(found, {"深田えいみ", "河合あすな"})
def test_the_aliases_ride_along_as_search_terms(self):
"""按规范名搜不到时，旧名往往搜得到——名字链是入口，不是装饰。"""
chain = next(row["chain"] for row in self.module.targets(self.con)
if row["name"] == "深田えいみ")
self.assertEqual(chain, ["深田えいみ", "天海こころ"])
def test_a_chinese_name_is_a_target_when_the_scope_is_all(self):
"""补别名要找的恰恰是已经有中文规范名的人：她们的第二个写法一个都没进过账本。"""
found = {row["name"] for row in self.module.targets(self.con, "all")}
self.assertEqual(found, {"深田えいみ", "三上悠亚", "河合あすな"})
def test_a_named_performer_is_looked_up_whatever_the_scope_says(self):
"""起点常常是「我看着这一位，觉得少个名字」。被范围挡下的正是最该查的一位。"""
for named in ("三上悠亚", "2", "天海こころ"):
found = [row["name"] for row in self.module.targets(self.con, "kana", [named])]
self.assertEqual(len(found), 1, named)
def test_naming_nobody_leaves_the_scope_in_charge(self):
self.assertEqual({row["name"] for row in self.module.targets(self.con, "kana", [])},
{"深田えいみ", "河合あすな"})
def test_a_performer_without_release_metadata_is_out_of_scope_either_way(self):
"""账号型 performer 不是女优，放开范围也不查。"""
for scope in ("kana", "all"):
names = {row["name"] for row in self.module.targets(self.con, scope)}
self.assertNotIn("MattieDoll", names)
class AliasCandidateTests(unittest.TestCase):
"""页面上账本还没有的中文写法。
`桥本有菜` 这种名字在账本里一个字都没有：刮削源给日文与罗马字，译名那条路每人只
给一个中文名，改统称时降为别名的又是日文原规范名。旧艺名的中译正是从这一栏来的。
"""
def setUp(self):
self.module = load_module()
self.record = {"entity_id": 1, "name": "新有菜", "assets": 15,
"chain": ["新有菜", "新ありな"]}
def candidates(self, current, former="", owners=None):
return self.module.alias_candidates(
self.record, page(current, former), "https://javdb.com/actors/RJM8",
owners or {})
def test_a_chinese_writing_in_the_former_name_field_is_a_candidate(self):
rows = self.candidates("新有菜, 新ありな", "橋本有菜, 橋本ありな")
self.assertEqual([(row["alias"], row["origin"], row["verdict"]) for row in rows],
[("桥本有菜", "别名栏", "ok")])
def test_a_nickname_in_that_field_is_offered_like_any_other_writing(self):
"""站上不标这一栏放的是什么，`傻梦` 是爱称。认不认得归复核的人，解析层不猜。"""
self.record.update(entity_id=7746, name="凉森玲梦", chain=["凉森玲梦", "涼森れむ"])
rows = self.candidates("涼森玲夢, 涼森れむ", "傻梦")
self.assertEqual([(row["alias"], row["origin"]) for row in rows], [("傻梦", "别名栏")])
def test_the_writing_on_the_page_is_kept_next_to_the_converted_one(self):
"""复核件要看得出转了什么，也要指得回是哪一页。"""
row = self.candidates("新有菜, 新ありな", "橋本有菜")[0]
self.assertEqual(row["alias_zh_tw"], "橋本有菜")
self.assertEqual(row["actor_id"], "RJM8")
def test_a_writing_the_ledger_already_has_is_not_a_candidate(self):
"""查过、没查出东西，和还没轮到她是两件事。"""
self.assertEqual([row["verdict"] for row in self.candidates("新有菜, 新ありな")],
["无新写法"])
def test_the_same_name_in_traditional_characters_is_not_a_new_writing(self):
self.record["chain"] = ["新有菜", "桥本有菜"]
self.assertEqual(
[row["verdict"] for row in self.candidates("新有菜, 新ありな", "橋本有菜")],
["无新写法"])
def test_a_japanese_or_latin_writing_is_not_a_chinese_alias(self):
"""`JULIA` 是艺名本体，`きみかわ結衣` 掺着假名。两个都不是中文写法。"""
self.assertEqual(
[row["verdict"] for row in self.candidates("JULIA, 京香じゅりあ", "きみかわ結衣")],
["无新写法"])
def test_one_writing_listed_in_both_fields_is_offered_once(self):
rows = self.candidates("新有菜, 橋本有菜", "橋本有菜")
self.assertEqual([(row["alias"], row["origin"]) for row in rows],
[("桥本有菜", "现名栏")])
def test_a_writing_that_belongs_to_someone_else_is_flagged_not_offered(self):
"""要么两条该合并、要么真有两位重名，都得人判。"""
owners = {self.module.key("橋本有菜"): (99, "桥本有菜")}
row = self.candidates("新有菜, 新ありな", "橋本有菜", owners)[0]
self.assertEqual(row["verdict"], "占用（另一条实体已有这个名字）")
self.assertIn("99", row["evidence"])
def box(path: str, title: str) -> str:
return f'<div class="box actor-box"><a href="{path}" title="{title}"></a></div>'
class FakeSite:
"""按地址发页的假站。没登记的地址抛错：取页失败要落进 CSV，不是静默的空页。"""
def __init__(self, pages: dict):
self.pages = dict(pages)
self.asked: list[str] = []
def get(self, url: str) -> str:
self.asked.append(url)
if url not in self.pages:
raise LookupError(f"没有这一页 {url}")
return self.pages[url]
class HarvestTests(unittest.TestCase):
"""一趟走完两份产物。分两个脚本就要按 5 秒一页把同一批资料页再走一遍。"""
def setUp(self):
self.module = load_module()
self.tmp = tempfile.TemporaryDirectory()
self.con = sqlite3.connect(Path(self.tmp.name).resolve() / "ledger.db")
self.con.executescript(
"CREATE TABLE entity(id INTEGER PRIMARY KEY, kind TEXT, canonical_name TEXT);"
"CREATE TABLE entity_alias(entity_id INTEGER, alias TEXT);"
"CREATE TABLE asset_entity(asset_id INTEGER, entity_id INTEGER, source TEXT);")
self.con.executemany(
"INSERT INTO entity(id,kind,canonical_name) VALUES(?,'performer',?)",
[(1, "深田えいみ"), (2, "三上悠亚")])
self.con.executemany(
"INSERT INTO asset_entity(asset_id,entity_id,source) VALUES(?,?,'r18:performer')",
[(10, 1), (11, 2)])
self.con.execute("INSERT INTO entity_alias VALUES(1,'天海こころ')")
self.con.commit()
def tearDown(self):
self.con.close()
self.tmp.cleanup()
@staticmethod
def search(name: str) -> str:
return javdb.SEARCH.format(quote(name))
def test_one_page_feeds_both_sheets(self):
site = FakeSite({
self.search("深田えいみ"): box("/actors/BBB", "深田詠美, 深田えいみ"),
"https://javdb.com/actors/BBB": page("深田詠美, 深田えいみ", "天海こころ")})
names, aliases = self.module.harvest(self.con, site, 0)
self.assertEqual([row["verdict"] for row in names], ["ok"])
self.assertEqual([(row["alias"], row["origin"]) for row in aliases],
[("深田咏美", "现名栏")])
def test_two_pages_for_one_name_yield_no_alias_candidate(self):
"""别名进的是身份。配错了人，比缺一个写法更难查回来。"""
site = FakeSite({
self.search("深田えいみ"): box("/actors/BBB", "深田詠美, 深田えいみ"),
"https://javdb.com/actors/BBB": page("深田詠美, 深田えいみ"),
self.search("三上悠亚"): (box("/actors/AAA", "三上悠亞, 三上悠亜")
+ box("/actors/CCC", "三上悠亞, 三上悠亜"))})
names, aliases = self.module.harvest(self.con, site, 0, "all")
crowded = [row for row in aliases if row["current_name"] == "三上悠亚"]
self.assertEqual([row["verdict"] for row in crowded],
["多页（同名两位，分不清是哪一页）"])
self.assertNotIn("https://javdb.com/actors/AAA", site.asked)
def test_the_name_sheet_keeps_to_the_performers_still_named_in_kana(self):
"""`--scope all` 放开的只是别名那一份。中文规范名不需要再定一次规范名。"""
site = FakeSite({
self.search("深田えいみ"): box("/actors/BBB", "深田詠美, 深田えいみ"),
"https://javdb.com/actors/BBB": page("深田詠美, 深田えいみ"),
self.search("三上悠亚"): box("/actors/AAA", "三上悠亞, 三上悠亜"),
"https://javdb.com/actors/AAA": page("三上悠亞, 三上悠亜")})
names, aliases = self.module.harvest(self.con, site, 0, "all")
self.assertEqual({row["current_name"] for row in names}, {"深田えいみ"})
self.assertEqual({row["current_name"] for row in aliases},
{"深田えいみ", "三上悠亚"})
def test_a_performer_the_site_does_not_have_is_recorded_in_both_sheets(self):
site = FakeSite({self.search("深田えいみ"): "", self.search("天海こころ"): ""})
names, aliases = self.module.harvest(self.con, site, 1)
self.assertEqual([row["verdict"] for row in names], ["未取得"])
self.assertEqual([row["verdict"] for row in aliases], ["未取得"])
def test_a_failed_fetch_lands_in_the_sheets_instead_of_stopping_the_batch(self):
site = FakeSite({})
names, aliases = self.module.harvest(self.con, site, 1)
self.assertEqual([row["verdict"] for row in names], ["取页失败"])
self.assertIn("LookupError", str(aliases[0]["evidence"]))
class ApplyAliasTests(unittest.TestCase):
"""候选写进 `entity_alias` 的那一步。四种不写，每种都要留下为什么。"""
def setUp(self):
self.module = load_script("apply_alias_candidates")
self.tmp = tempfile.TemporaryDirectory()
self.con = sqlite3.connect(Path(self.tmp.name).resolve() / "ledger.db")
self.con.executescript(
"CREATE TABLE entity(id INTEGER PRIMARY KEY, kind TEXT,"
" canonical_name TEXT, normalized_name TEXT);"
"CREATE TABLE entity_alias(entity_id INTEGER, alias TEXT,"
" normalized_alias TEXT, source TEXT, confidence REAL);")
for entity_id, name in ((1, "桥本有菜"), (2, "三上悠亚")):
self.con.execute(
"INSERT INTO entity VALUES(?,'performer',?,?)",
(entity_id, name, normalize_entity_name(name)))
self.con.commit()
def tearDown(self):
self.con.close()
self.tmp.cleanup()
def candidate(self, **over):
row = {"entity_id": 1, "current_name": "桥本有菜", "alias": "新有菜",
"origin": "别名栏", "verdict": "ok", "url": "https://javdb.com/actors/RJM8"}
row.update(over)
return row
def plan(self, **over):
return self.module.plan(self.con, [self.candidate(**over)])[0]
def aliases_of(self, entity_id: int):
return [(row[0], row[1]) for row in self.con.execute(
"SELECT alias,source FROM entity_alias WHERE entity_id=?", (entity_id,))]
def test_a_new_writing_is_written_with_the_page_as_its_source(self):
rows = self.module.plan(self.con, [self.candidate()])
self.assertEqual(rows[0]["action"], "写入")
self.assertEqual(self.module.apply_rows(self.con, rows, "javdb-20260912"), 1)
self.assertEqual(self.aliases_of(1),
[("新有菜", f"{javdb.ALIAS_SOURCE}@javdb-20260912")])
def test_the_batch_is_named_in_the_source_so_it_can_be_taken_back_whole(self):
"""解析判错时，认得出批次才能按 `source` 把那一趟整批撤回。"""
self.module.apply_rows(self.con, self.module.plan(self.con, [self.candidate()]),
"javdb-20260912")
self.module.apply_rows(self.con, self.module.plan(
self.con, [self.candidate(alias="岩谷志季")]), "javdb-20261001")
gone = self.con.execute("DELETE FROM entity_alias WHERE source=?",
(self.module.source_for("javdb-20261001"),)).rowcount
self.assertEqual(gone, 1)
self.assertEqual([alias for alias, _ in self.aliases_of(1)], ["新有菜"])
def test_a_writing_the_entity_already_has_is_left_alone(self):
self.con.execute("INSERT INTO entity_alias VALUES(1,'新有菜',?,'javdb-actor-page',1.0)",
(normalize_entity_name("新有菜"),))
self.assertEqual(self.plan()["action"], "已有")
def test_the_canonical_name_is_not_added_to_its_own_alias_list(self):
self.assertEqual(self.plan(alias="桥本有菜")["action"], "已有")
def test_a_writing_that_belongs_to_another_performer_is_refused(self):
"""静默写下去只会给后面的合并判定多送一个假信号。"""
self.con.execute("INSERT INTO entity_alias VALUES(2,'新有菜',?,'javdb-actor-page',1.0)",
(normalize_entity_name("新有菜"),))
row = self.plan()
self.assertEqual(row["action"], "占用")
self.assertIn("三上悠亚", row["detail"])
def test_a_row_whose_entity_was_renamed_after_the_harvest_is_not_written(self):
"""CSV 是某一刻的快照。统称变过说明这个人后来被动过，该重抓一遍再看。"""
row = self.plan(current_name="新ありな")
self.assertEqual(row["action"], "已变")
self.assertIn("桥本有菜", row["detail"])
def test_a_row_pointing_at_nobody_is_not_written(self):
self.assertEqual(self.plan(entity_id=404)["action"], "查无此人")
def test_a_writing_from_another_site_records_that_site_as_its_source(self):
rows = self.module.plan(self.con, [self.candidate()])
self.module.apply_rows(self.con, rows, "namewikis-20260924", base="av_name")
self.assertEqual(self.aliases_of(1), [("新有菜", "av_name@namewikis-20260924")])
def test_a_writing_judged_wrong_is_taken_off_whatever_wrote_it(self):
"""javdb 与 Wiki 的別名栏都混有共演男优名，人看过判「删除」就摘掉，来源留在复核 CSV。"""
self.con.execute("INSERT INTO entity_alias VALUES(1,'島修也',?,'javdb-actor-page@x',1.0)",
(normalize_entity_name("島修也"),))
rows = self.module.plan(self.con, [self.candidate(alias="島修也", verdict="删除")])
self.assertEqual(rows[0]["action"], "删除")
self.assertIn("javdb-actor-page@x", rows[0]["detail"])
self.assertEqual(self.module.apply_rows(self.con, rows, "r"), 0)
self.assertEqual(self.module.remove_rows(self.con, rows), 1)
self.assertEqual(self.aliases_of(1), [])
def test_removing_a_writing_the_entity_does_not_have_changes_nothing(self):
row = self.plan(alias="島修也", verdict="删除")
self.assertEqual(row["action"], "名下没有")
self.assertEqual(self.module.remove_rows(self.con, [row]), 0)
def test_only_the_verdicts_named_on_the_command_line_are_taken(self):
rows = [self.candidate(), self.candidate(verdict="占用（另一条实体已有这个名字）"),
self.candidate(alias="")]
self.assertEqual(len(self.module.accepted(rows, ["ok"])), 1)
self.assertEqual(len(self.module.accepted(rows, ["ok", "占用（另一条实体已有这个名字）"])), 2)
class MappingCsvTests(unittest.TestCase):
def setUp(self):
self.module = load_script("localize_performer_names")
self.tmp = tempfile.TemporaryDirectory()
self.path = Path(self.tmp.name).resolve() / "mapping.csv"
def tearDown(self):
self.tmp.cleanup()
def write(self, *rows):
from peach.review_csv import write_rows
write_rows(self.path, load_module().FIELDS, list(rows), fill_missing=True)
def test_only_the_rows_judged_ok_become_mappings(self):
self.write({"current_name": "深田えいみ", "adopt": "深田咏美", "verdict": "ok",
"actor_id": "A1"},
{"current_name": "長谷川るい", "adopt": "", "verdict": "同形（站上只有日文名）"})
mappings = self.module.read_mapping_csv(self.path)
self.assertEqual([m.jp for m in mappings], ["深田えいみ"])
def test_a_renamed_row_is_only_taken_when_it_is_named_on_the_command_line(self):
"""站上改了名要不要跟着改是人的决定，不能因为查到了就默认采用。"""
self.write({"current_name": "京香じゅりあ", "adopt": "JULIA", "verdict": "改艺名"})
self.assertEqual(self.module.read_mapping_csv(self.path), [])
taken = self.module.read_mapping_csv(self.path, ["ok", "改艺名"])
self.assertEqual([(m.jp, m.zh_cn) for m in taken], [("京香じゅりあ", "JULIA")])
def test_one_performer_matched_on_several_pages_yields_one_mapping(self):
"""`アンナ` 命中了三页。三条映射进去，落库时只会变成一次「没有唯一映射」。"""
self.write({"current_name": "アンナ", "adopt": "Anna", "verdict": "改艺名"},
{"current_name": "アンナ", "adopt": "安娜", "verdict": "ok"},
{"current_name": "アンナ", "adopt": "ANNA", "verdict": "改艺名"})
taken = self.module.read_mapping_csv(self.path, ["ok", "改艺名"])
self.assertEqual([(m.jp, m.zh_cn) for m in taken], [("アンナ", "安娜")])
def test_a_latin_current_name_is_the_stage_name_not_a_romanization(self):
"""XML 那份偶尔把罗马字填进 `zh_cn`；javdb 的现名栏写 `JULIA` 时那就是艺名。"""
self.write({"current_name": "京香じゅりあ", "adopt": "JULIA", "verdict": "改艺名"})
taken = self.module.read_mapping_csv(self.path, ["改艺名"])
self.assertTrue(taken[0].latin_is_a_stage_name)
def test_a_row_with_nothing_to_adopt_is_not_a_mapping(self):
"""判定收了但这一行没给出可用的名字，落库时就是把规范名改成空。"""
self.write({"current_name": "長谷川るい", "adopt": "", "verdict": "ok"})
self.assertEqual(self.module.read_mapping_csv(self.path), [])
def test_the_page_id_becomes_the_provenance_key(self):
"""回溯「这个中文名是谁写的」时，要指得回具体那一页。"""
self.write({"current_name": "深田えいみ", "adopt": "深田咏美", "verdict": "ok",
"actor_id": "A1"})
self.assertEqual(self.module.read_mapping_csv(self.path)[0].key, "javdb:A1")
def test_the_index_points_back_into_the_returned_list(self):
"""`collect` 拿 `index` 当下标回查。跳过的行也算数就会指到别人身上。"""
self.write({"current_name": "跳过", "adopt": "", "verdict": "同形（站上只有日文名）"},
{"current_name": "深田えいみ", "adopt": "深田咏美", "verdict": "ok",
"actor_id": "A1"})
mappings = self.module.read_mapping_csv(self.path)
self.assertEqual(mappings[mappings[0].index].jp, "深田えいみ")
def test_a_japanese_glyph_left_by_the_traditional_conversion_is_normalized(self):
"""opencc 转的是繁简，不管日本字形：`滝` 中文写 `泷`。字形不需要外部证据。"""
self.write({"current_name": "滝田あゆ", "adopt": "滝田亚由", "verdict": "ok",
"actor_id": "A2"})
self.assertEqual(self.module.read_mapping_csv(self.path)[0].zh_cn, "泷田亚由")
if __name__ == "__main__":
unittest.main()