fix bangumi catalog

catalog/book/tests.py

@@ -409,6 +409,26 @@ def test_scrape(self):
         self.assertEqual(site.resource.item.author[0], "爱潜水的乌贼")
 
 
+class BangumiTestCase(TestCase):
+    databases = "__all__"
+
+    @use_local_response
+    def test_scrape(self):
+        t_url = "https://bgm.tv/subject/342254"
+        site = SiteManager.get_site_by_url(t_url)
+        self.assertEqual(site.ready, False)
+        site.get_resource_ready()
+        self.assertEqual(site.ready, True)
+        self.assertEqual(site.resource.site_name, SiteName.Bangumi)
+        self.assertEqual(site.resource.id_type, IdType.Bangumi)
+        self.assertEqual(site.resource.id_value, "342254")
+        self.assertEqual(site.resource.item.localized_title[0]["text"], "蓦然回首")
+        self.assertEqual(site.resource.item.author[0], "藤本タツキ")
+        self.assertEqual(site.resource.item.isbn, "9784088827827")
+        self.assertEqual(site.resource.item.pages, "144")
+        self.assertEqual(site.resource.item.price, "￥484")
+
+
 class MultiBookSitesTestCase(TestCase):
     databases = "__all__"
 

catalog/sites/bangumi.py

@@ -1,4 +1,5 @@
 import logging
+from collections import OrderedDict
 
 from catalog.book.utils import detect_isbn_asin
 from catalog.common import *
@@ -15,6 +16,7 @@ class Bangumi(AbstractSite):
     URL_PATTERNS = [
         r"https://bgm\.tv/subject/(\d+)",
         r"https://bangumi\.tv/subject/(\d+)",
+        r"https://chii\.in/subject/(\d+)",
     ]
     WIKI_PROPERTY_ID = ""
     DEFAULT_MODEL = None
@@ -70,6 +72,8 @@ def scrape(self):
         authors = None
         site = None
         director = None
+        pages = None
+        price = None
         for i in o.get("infobox", []):
             k = i["key"]
             v = i["value"]
@@ -82,7 +86,7 @@ def scrape(self):
                     )
                 case "imdb_id":
                     imdb_code = v
-                case "isbn":
+                case "isbn" | "ISBN":
                     isbn_type, isbn = detect_isbn_asin(v)
                 case "语言":
                     language = v
@@ -110,6 +114,10 @@ def scrape(self):
                     )
                 case "官方网站" | "website":
                     site = v[0] if isinstance(v, list) else v
+                case "页数":
+                    pages = v
+                case "价格":
+                    price = v
 
         img_url = o["images"].get("large") or o["images"].get("common")
         raw_img = None
@@ -118,10 +126,14 @@ def scrape(self):
             raw_img, ext = BasicImageDownloader.download_image(
                 img_url, None, headers={}
             )
-        titles = set(
+        titles = OrderedDict.fromkeys(
             [title] + (other_title or []) + ([orig_title] if orig_title else [])
         )
-        localized_title = [{"lang": detect_language(t), "text": t} for t in titles]
+        if o.get("name_cn"):
+            titles[o.get("name_cn")] = "zh-cn"
+        localized_title = [
+            {"lang": l or detect_language(t), "text": t} for t, l in titles.items()
+        ]
         localized_desc = (
             [{"lang": detect_language(brief), "text": brief}] if brief else []
         )
@@ -152,6 +164,8 @@ def scrape(self):
             "brief": brief,
             "cover_image_url": img_url,
             "release_date": dt,
+            "pages": pages,
+            "price": price,
         }
         lookup_ids = {}
         if isbn:

test_data/https___api_bgm_tv_v0_subjects_342254

@@ -0,0 +1 @@
+{"date":"2021-09-03","platform":"漫画","images":{"small":"https://lain.bgm.tv/r/200/pic/cover/l/f2/4c/342254_RfvLZ.jpg","grid":"https://lain.bgm.tv/r/100/pic/cover/l/f2/4c/342254_RfvLZ.jpg","large":"https://lain.bgm.tv/pic/cover/l/f2/4c/342254_RfvLZ.jpg","medium":"https://lain.bgm.tv/r/800/pic/cover/l/f2/4c/342254_RfvLZ.jpg","common":"https://lain.bgm.tv/r/400/pic/cover/l/f2/4c/342254_RfvLZ.jpg"},"summary":"学生新聞で４コマ漫画を連載している小学４年生の藤野。クラスメートからは絶賛を受けていたが、ある日、不登校の同級生・京本の４コマを載せたいと先生から告げられるが…!?","name":"ルックバック","name_cn":"蓦然回首","tags":[{"name":"藤本タツキ","count":966,"total_cont":0},{"name":"漫画","count":958,"total_cont":0},{"name":"短篇","count":741,"total_cont":0},{"name":"藤本树","count":470,"total_cont":0},{"name":"2021","count":376,"total_cont":0},{"name":"青春","count":363,"total_cont":0},{"name":"意犹未尽","count":188,"total_cont":0},{"name":"集英社","count":164,"total_cont":0},{"name":"京阿尼","count":132,"total_cont":0},{"name":"短篇漫画","count":98,"total_cont":0},{"name":"少年ジャンプ＋","count":81,"total_cont":0},{"name":"治愈","count":61,"total_cont":0},{"name":"百合","count":50,"total_cont":0},{"name":"日本","count":37,"total_cont":0},{"name":"漫画短篇","count":34,"total_cont":0},{"name":"2021年","count":21,"total_cont":0},{"name":"已完结","count":21,"total_cont":0},{"name":"少年","count":17,"total_cont":0},{"name":"原创","count":17,"total_cont":0},{"name":"已购","count":10,"total_cont":0},{"name":"奇幻","count":8,"total_cont":0},{"name":"日漫","count":8,"total_cont":0},{"name":"全一卷","count":7,"total_cont":0},{"name":"神作","count":6,"total_cont":0},{"name":"日本漫画","count":5,"total_cont":0},{"name":"催泪","count":5,"total_cont":0},{"name":"少年JUMP+","count":5,"total_cont":0},{"name":"JUMP","count":4,"total_cont":0},{"name":"マンガ","count":4,"total_cont":0},{"name":"励志","count":4,"total_cont":0}],"infobox":[{"key":"中文名","value":"蓦然回首"},{"key":"别名","value":[{"v":"頋"},{"v":"Look Back"}]},{"key":"作者","value":"藤本タツキ"},{"key":"出版社","value":"集英社"},{"key":"价格","value":"￥484"},{"key":"连载杂志","value":"少年ジャンプ＋"},{"key":"发售日","value":"2021-09-03"},{"key":"页数","value":"144"},{"key":"话数","value":"1"},{"key":"ISBN","value":"9784088827827"},{"key":"其他","value":"单行本发售"},{"key":"刊载","value":"2021-07-19"}],"rating":{"rank":320,"total":3710,"count":{"1":1,"2":0,"3":3,"4":4,"5":23,"6":135,"7":579,"8":1638,"9":1029,"10":298},"score":8.2},"total_episodes":0,"collection":{"on_hold":17,"dropped":7,"wish":469,"collect":5162,"doing":63},"id":342254,"eps":1,"meta_tags":["日本","漫画","少年","原创","已完结"],"volumes":0,"series":false,"locked":false,"nsfw":false,"type":1}