Fix #226: avwiki: 修复演员名的字符串被当做列表处理的问题

Yuukiy · Feb 25, 2024 · fe86a74 · fe86a74
1 parent aae4612
commit fe86a74
Show file tree

Hide file tree

Showing 4 changed files with 42 additions and 5 deletions.
diff --git a/unittest/data/259LUXU-593 (avwiki).json b/unittest/data/259LUXU-593 (avwiki).json
@@ -13,7 +13,9 @@
   "ori_title": null,
   "magnet": null,
   "serial": "ラグジュTV",
-  "actress": "白咲りの",
+  "actress": [
+    "白咲りの"
+  ],
   "actress_pics": null,
   "director": null,
   "duration": null,

diff --git a/unittest/data/300MIUM-1001 (avwiki).json b/unittest/data/300MIUM-1001 (avwiki).json
@@ -13,7 +13,9 @@
   "ori_title": null,
   "magnet": null,
   "serial": "レンタル彼女",
-  "actress": "沙優七羽",
+  "actress": [
+    "沙優七羽"
+  ],
   "actress_pics": null,
   "director": null,
   "duration": null,

diff --git a/unittest/data/SSIS-698 (avwiki).json b/unittest/data/SSIS-698 (avwiki).json
@@ -0,0 +1,30 @@
+{
+  "dvdid": "SSIS-698",
+  "cid": null,
+  "url": "https://av-wiki.net/SSIS-698",
+  "plot": null,
+  "cover": "https://pics.dmm.co.jp/digital/video/ssis00698/ssis00698pl.jpg",
+  "big_cover": null,
+  "genre": null,
+  "genre_id": null,
+  "genre_norm": null,
+  "score": null,
+  "title": "三上悠亜と新ありなと相沢みなみ",
+  "ori_title": null,
+  "magnet": null,
+  "serial": null,
+  "actress": [
+    "三上悠亜",
+    "新ありな",
+    "相沢みなみ"
+  ],
+  "actress_pics": null,
+  "director": null,
+  "duration": null,
+  "producer": "エスワン ナンバーワンスタイル",
+  "publisher": null,
+  "uncensored": false,
+  "publish_date": "2023-05-05",
+  "preview_pics": null,
+  "preview_video": null
+}
diff --git a/web/avwiki.py b/web/avwiki.py
@@ -46,12 +46,15 @@ def parse_data(movie: MovieInfo):
     dt_txt_ls, dd_tags = info.xpath("dt/text()"), info.xpath("dd")
     data = {}
     for dt_txt, dd in zip(dt_txt_ls, dd_tags):
+        dt_txt = dt_txt.strip()
         a_tag = dd.xpath('a')
         if len(a_tag) == 0:
-            dd_txt = dd.text
+            dd_txt = dd.text.strip()
         else:
-            dd_txt = a_tag[0].text
-        data[dt_txt.strip()] = dd_txt.strip()
+            dd_txt = [i.text.strip() for i in a_tag]
+        if isinstance(dd_txt, list) and dt_txt != 'AV女優名':    # 只有女优名以列表的数据格式保留
+            dd_txt = dd_txt[0]
+        data[dt_txt] = dd_txt
 
     ATTR_MAP = {'メーカー': 'producer', 'AV女優名': 'actress', 'メーカー品番': 'dvdid', 'シリーズ': 'serial', '配信開始日': 'publish_date'}
     for key, attr in ATTR_MAP.items():