jav321: 清洗网站数据里URL中的连续//

unittest/data/DCV-137 (jav321).json

@@ -36,43 +36,43 @@
   "uncensored": null,
   "publish_date": "2019-05-17",
   "preview_pics": [
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_0_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_1_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_2_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_3_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_4_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_5_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_6_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_7_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_8_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_9_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_10_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_11_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_12_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_13_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_14_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_15_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_16_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_17_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_18_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_19_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_20_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_21_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_22_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_23_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_24_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_25_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_26_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_27_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_28_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_29_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_30_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_31_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_32_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_33_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_34_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_35_277dcv-137.jpg",
-    "https://www.jav321.com//images/documentv/277dcv/137/cap_e_36_277dcv-137.jpg"
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_0_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_1_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_2_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_3_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_4_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_5_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_6_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_7_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_8_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_9_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_10_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_11_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_12_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_13_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_14_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_15_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_16_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_17_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_18_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_19_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_20_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_21_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_22_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_23_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_24_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_25_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_26_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_27_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_28_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_29_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_30_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_31_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_32_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_33_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_34_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_35_277dcv-137.jpg",
+    "https://www.jav321.com/images/documentv/277dcv/137/cap_e_36_277dcv-137.jpg"
   ],
   "preview_video": "https://sample.mgstage.com/sample/documentv/277dcv/137/277DCV-137_sample.mp4"
 }

unittest/data/IPX-177 (jav321).json

@@ -3,7 +3,7 @@
   "cid": "ipx00177",
   "url": "https://en.jav321.com/video/ipx00177",
   "plot": "ミニスカートとニーハイから覗くきれいな素肌、ニーソの太ももへの食い込み…全てを兼ね備えた「絶対領域」！ピチピチの肌感、ぷにぷにハミもも思わず触りたくなること間違いなし！さらに絶対領域を眺めながら足コキ、股コキ、尻コキ！完全着衣で堪能！！ニーハイを履いた小悪魔に罵られながらイク！！「マヂきもいんだけど！触り方が！！」「完璧すぎる、、絶対領域はぁはぁ」",
-  "cover": "http://pics.dmm.co.jp//digital/video/ipx00177/ipx00177pl.jpg",
+  "cover": "http://pics.dmm.co.jp/digital/video/ipx00177/ipx00177pl.jpg",
   "big_cover": null,
   "genre": [
     "デジモ",

web/jav321.py

@@ -71,6 +71,8 @@ def parse_data(movie: MovieInfo):
     if len(preview_pics) == 0:
         # 尝试搜索另一种布局下的封面，需要使用onerror过滤掉明明没有封面时网站往里面塞的默认URL
         preview_pics = html.xpath("//div/div/div[@class='col-md-3']/img[@onerror and @class='img-responsive']/@src")
+    # 有的图片链接里有多个//，网站质量堪忧……
+    preview_pics = [i[:8] + i[8:].replace('//', '/') for i in preview_pics]
     # 磁力和ed2k链接是依赖js脚本加载的，无法通过静态网页来解析
 
     movie.url = page_url