bacow/AV_Data_Capture
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Add anime support #3

Browse Source
This commit is contained in:
yoshiko2
2022-05-21 03:23:52 +08:00
parent 26aeb5bab8
commit 8ed0aa68d7
1 changed files with 16 additions and 12 deletions
Download Patch File Download Diff File Expand all files Collapse all files

28
WebCrawler/dlsite.py
View File

@@ -115,26 +115,30 @@ def main(number):
try: try:
if "RJ" in number or "VJ" in number: if "RJ" in number or "VJ" in number:
number = number.upper() number = number.upper()
htmlcode = get_html('https://www.dlsite.com/maniax/work/=/product_id/' + number + '.html/?locale=zh_CN',cookies={'locale': 'zh-cn'}) htmlcode = get_html('https://www.dlsite.com/maniax/work/=/product_id/' + number + '.html/?locale=zh_CN', cookies={'locale': 'zh-cn'})
html = etree.fromstring(htmlcode, etree.HTMLParser()) html = etree.fromstring(htmlcode, etree.HTMLParser())
else: else:
htmlcode = get_html(f'https://www.dlsite.com/maniax/fsr/=/language/jp/sex_category/male/keyword/{number}/order/trend/work_type_category/movie',cookies={'locale': 'zh-cn'}) htmlcode = get_html(f'https://www.dlsite.com/maniax/fsr/=/language/jp/sex_category/male/keyword/{number}/order/trend/work_type_category/movie', cookies={'locale': 'zh-cn'})
html = etree.HTML(htmlcode) html = etree.HTML(htmlcode)
search_result = html.xpath('//*[@id="search_result_img_box"]/li[1]/dl/dd[2]/div[2]/a/@href') search_result = html.xpath('//*[@id="search_result_img_box"]/li[1]/dl/dd[2]/div[2]/a/@href')
if len(search_result) == 0: if len(search_result) == 0:
number = number.replace("THE ANIMATION", "").replace("he Animation", "").replace("t", "").replace("T", "") number = number.replace("THE ANIMATION", "").replace("he Animation", "").replace("t", "").replace("T","")
htmlcode = get_html( html = etree.HTML(get_html(
f'https://www.dlsite.com/maniax/fsr/=/language/jp/sex_category/male/keyword/{number}/order/trend/work_type_category/movie', f'https://www.dlsite.com/maniax/fsr/=/language/jp/sex_category/male/keyword/{number}/order/trend/work_type_category/movie', cookies={'locale': 'zh-cn'}))
cookies={'locale': 'zh-cn'})
html = etree.HTML(htmlcode)
search_result = html.xpath('//*[@id="search_result_img_box"]/li[1]/dl/dd[2]/div[2]/a/@href') search_result = html.xpath('//*[@id="search_result_img_box"]/li[1]/dl/dd[2]/div[2]/a/@href')
if len(search_result) == 0: if len(search_result) == 0:
number = number.replace('上巻','').replace('下巻','').replace('前編','').replace('後編','') if "～" in number:
htmlcode = get_html( number = number.replace("～","〜")
f'https://www.dlsite.com/maniax/fsr/=/language/jp/sex_category/male/keyword/{number}/order/trend/work_type_category/movie', elif "〜" in number:
cookies={'locale': 'zh-cn'}) number = number.replace("〜","～")
html = etree.HTML(htmlcode) html = etree.HTML(get_html(
f'https://www.dlsite.com/maniax/fsr/=/language/jp/sex_category/male/keyword/{number}/order/trend/work_type_category/movie', cookies={'locale': 'zh-cn'}))
search_result = html.xpath('//*[@id="search_result_img_box"]/li[1]/dl/dd[2]/div[2]/a/@href') search_result = html.xpath('//*[@id="search_result_img_box"]/li[1]/dl/dd[2]/div[2]/a/@href')
if len(search_result) == 0:
number = number.replace('上巻', '').replace('下巻', '').replace('前編', '').replace('後編', '')
html = etree.HTML(get_html(
f'https://www.dlsite.com/maniax/fsr/=/language/jp/sex_category/male/keyword/{number}/order/trend/work_type_category/movie', cookies={'locale': 'zh-cn'}))
search_result = html.xpath('//*[@id="search_result_img_box"]/li[1]/dl/dd[2]/div[2]/a/@href')
a = search_result[0] a = search_result[0]
html = etree.HTML(get_html(a,cookies={'locale': 'zh-cn'})) html = etree.HTML(get_html(a,cookies={'locale': 'zh-cn'}))
number = str(re.findall("\wJ\w+",a)).strip(" [']") number = str(re.findall("\wJ\w+",a)).strip(" [']")