Merge pull request #490 from lededev/javdb_gdr

fix javdb actors section

WebCrawler/javdb.py

@@ -13,11 +13,24 @@ def getTitle(a):
     html = etree.fromstring(a, etree.HTMLParser())
     result = html.xpath("/html/body/section/div/h2/strong/text()")[0]
     return result
-def getActor(a):  # //*[@id="center_column"]/div[2]/div[1]/div/table/tbody/tr[1]/td/text()
+
-    html = etree.fromstring(a, etree.HTMLParser())  # //table/tr[1]/td[1]/text()
+def getActor(a):
-    result1 = str(html.xpath('//strong[contains(text(),"演員")]/../span/text()')).strip(" ['']")
+    html = etree.fromstring(a, etree.HTMLParser())
-    result2 = str(html.xpath('//strong[contains(text(),"演員")]/../span/a/text()')).strip(" ['']")
+    actors = html.xpath('//span[@class="value"]/a[contains(@href,"/actors/")]/text()')
-    return str(result1 + result2).strip('+').replace(",\\xa0", "").replace("'", "").replace(' ', '').replace(',,', '').replace('N/A', '').lstrip(',').replace(',', ', ')
+    genders = html.xpath('//span[@class="value"]/a[contains(@href,"/actors/")]/../strong/@class')
+    r = []
+    idx = 0
+    actor_gendor = config.Config().actor_gender()
+    if not actor_gendor in ['female','male','both','all']:
+        actor_gendor = 'female'
+    for act in actors:
+        if((actor_gendor == 'all')
+        or (actor_gendor == 'both' and genders[idx] in ['symbol female', 'symbol male'])
+        or (actor_gendor == 'female' and genders[idx] == 'symbol female')
+        or (actor_gendor == 'male' and genders[idx] == 'symbol male')):
+            r.append(act)
+        idx = idx + 1
+    return r
 
 def getaphoto(url):
     html_page = get_html(url)
@@ -276,3 +289,4 @@ if __name__ == "__main__":
     # print(main('AGAV-042'))
     print(main('BANK-022'))
     print(main('FC2-735670'))
+    print(main('MVSD-439'))

config.ini

@@ -7,6 +7,8 @@ failed_move=0
 auto_exit=0
 transalte_to_sc=0
 multi_threading=1
+;actor_gender value: female or male or both or all(含人妖)
+actor_gender=female
 
 [proxy]
 ;proxytype: http or socks5 or socks5h switch: 0 1

config.py

@@ -39,6 +39,9 @@ class Config:
     def success_folder(self) -> str:
         return self.conf.get("common", "success_output_folder")
 
+    def actor_gender(self) -> str:
+        return self.conf.get("common", "actor_gender")
+
     def soft_link(self) -> bool:
         return self.conf.getboolean("common", "soft_link")
     def failed_move(self) -> bool:
@@ -162,6 +165,8 @@ class Config:
         conf.set(sec1, "failed_move", "1")
         conf.set(sec1, "auto_exit", "0")
         conf.set(sec1, "transalte_to_sc", "1")
+        # actor_gender value: female or male or both or all(含人妖)
+        conf.set(sec1, "actor_gender", "female")
 
         sec2 = "proxy"
         conf.add_section(sec2)
@@ -252,3 +257,4 @@ if __name__ == "__main__":
     print(config.get_transalte_key())
     print(config.get_transalte_delay())
     print(config.transalte_values())
+    print(config.actor_gender())