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

Update 2.0

Browse Source
This commit is contained in:
Yoshiko
2020-01-19 02:16:14 +08:00
committed by GitHub
parent 54db38511d
commit b21d47c55c
7 changed files with 94 additions and 48 deletions
Download Patch File Download Diff File Expand all files Collapse all files

67
fc2fans_club.py
View File

@@ -4,43 +4,50 @@ import json
import ADC_function
def getTitle(htmlcode): #获取厂商
#print(htmlcode)
html = etree.fromstring(htmlcode,etree.HTMLParser())
result = str(html.xpath('//*[@id="container"]/div[1]/div/article/section[1]/h2/text()')).strip(" ['']")
return result
result = str(html.xpath('/html/body/div[2]/div/div[1]/h3/text()')).strip(" ['']")
result2 = str(re.sub('\D{2}2-\d+','',result)).replace(' ','',1)
#print(result2)
return result2
def getActor(htmlcode):
try:
html = etree.fromstring(htmlcode, etree.HTMLParser())
result = str(html.xpath('//*[@id="container"]/div[1]/div/article/section[1]/div/div[2]/dl/dd[5]/a/text()')).strip(" ['']")
result = str(html.xpath('/html/body/div[2]/div/div[1]/h5[5]/a/text()')).strip(" ['']")
return result
except:
return ''
def getStudio(htmlcode): #获取厂商
try:
html = etree.fromstring(htmlcode, etree.HTMLParser())
result = str(html.xpath('//*[@id="container"]/div[1]/div/article/section[1]/div/div[2]/dl/dd[5]/a/text()')).strip(" ['']")
return result
except:
return ''
html = etree.fromstring(htmlcode,etree.HTMLParser())
result = str(html.xpath('/html/body/div[2]/div/div[1]/h5[3]/a[1]/text()')).strip(" ['']")
return result
def getNum(htmlcode): #获取番号
html = etree.fromstring(htmlcode, etree.HTMLParser())
result = str(html.xpath('/html/body/div[5]/div[1]/div[2]/p[1]/span[2]/text()')).strip(" ['']")
#print(result)
return result
def getRelease(htmlcode2): #
#a=ADC_function.get_html('http://adult.contents.fc2.com/article_search.php?id='+str(number).lstrip("FC2-").lstrip("fc2-").lstrip("fc2_").lstrip("fc2-")+'&utm_source=aff_php&utm_medium=source_code&utm_campaign=from_aff_php')
html=etree.fromstring(htmlcode2,etree.HTMLParser())
result = str(html.xpath('//*[@id="container"]/div[1]/div/article/section[1]/div/div[2]/dl/dd[4]/text()')).strip(" ['']")
return result
def getCover(htmlcode2): #获取厂商 #
def getCover(htmlcode,number,htmlcode2): #获取厂商 #
#a = ADC_function.get_html('http://adult.contents.fc2.com/article_search.php?id=' + str(number).lstrip("FC2-").lstrip("fc2-").lstrip("fc2_").lstrip("fc2-") + '&utm_source=aff_php&utm_medium=source_code&utm_campaign=from_aff_php')
html = etree.fromstring(htmlcode2, etree.HTMLParser())
result = str(html.xpath('//*[@id="container"]/div[1]/div/article/section[1]/div/div[1]/a/img/@src')).strip(" ['']")
if result == '':
html = etree.fromstring(htmlcode, etree.HTMLParser())
result2 = str(html.xpath('//*[@id="slider"]/ul[1]/li[1]/img/@src')).strip(" ['']")
return 'https://fc2club.com' + result2
return 'http:' + result
def getOutline(htmlcode2): #获取番号 #
html = etree.fromstring(htmlcode2, etree.HTMLParser())
result = str(html.xpath('//*[@id="container"]/div[1]/div/article/section[4]/p/text()')).strip(" ['']").replace("\\n",'',10000).replace("'",'',10000).replace(', ,','').strip(' ').replace('。,',',')
result = str(html.xpath('/html/body/div[1]/div[2]/div[2]/div[1]/div/article/section[4]/p/text()')).strip(" ['']").replace("\\n",'',10000).replace("'",'',10000).replace(', ,','').strip(' ').replace('。,',',')
return result
def getTag(htmlcode): #获取番号
html = etree.fromstring(htmlcode, etree.HTMLParser())
result = html.xpath('//*[@id="container"]/div[1]/div/article/section[6]/ul/li/a/text()')
return result
result = str(html.xpath('/html/body/div[2]/div/div[1]/h5[4]/a/text()'))
return result.strip(" ['']").replace("'",'').replace(' ','')
def getYear(release):
try:
result = re.search('\d{4}',release).group()
@@ -49,25 +56,29 @@ def getYear(release):
return ''
def main(number):
number=number.replace('PPV','').replace('ppv','').strip('fc2_').strip('fc2-').strip('ppv-').strip('PPV-').strip('FC2_').strip('FC2-').strip('ppv-').strip('PPV-').replace('fc2ppv-','').replace('FC2PPV-','')
htmlcode2 = ADC_function.get_html('http://adult.contents.fc2.com/article_search.php?id='+str(number).lstrip("FC2-").lstrip("fc2-").lstrip("fc2_").lstrip("fc2-")+'')
#htmlcode = ADC_function.get_html('http://fc2fans.club/html/FC2-' + number + '.html')
htmlcode2 = ADC_function.get_html('http://adult.contents.fc2.com/article_search.php?id='+number+'&utm_source=aff_php&utm_medium=source_code&utm_campaign=from_aff_php')
htmlcode = ADC_function.get_html('https://fc2club.com//html/FC2-' + number + '.html')
actor = getActor(htmlcode)
if getActor(htmlcode) == '':
actor = 'FC2系列'
dic = {
'title': getTitle(htmlcode2),
'studio': getStudio(htmlcode2),
'year': getYear(getRelease(htmlcode2)),
'title': getTitle(htmlcode),
'studio': getStudio(htmlcode),
'year': '',#str(re.search('\d{4}',getRelease(number)).group()),
'outline': getOutline(htmlcode2),
'runtime': getYear(getRelease(htmlcode2)),
'director': getStudio(htmlcode2),
'actor': getStudio(htmlcode2),
'release': getRelease(htmlcode2),
'runtime': getYear(getRelease(htmlcode)),
'director': getStudio(htmlcode),
'actor': actor,
'release': getRelease(number),
'number': 'FC2-'+number,
'cover': getCover(htmlcode2),
'cover': getCover(htmlcode,number,htmlcode2),
'imagecut': 0,
'tag': getTag(htmlcode2),
'tag': getTag(htmlcode),
'actor_photo':'',
'website': 'http://adult.contents.fc2.com/article_search.php?id=' + number,
'source': 'fc2fans_club.py',
'website': 'https://fc2club.com//html/FC2-' + number + '.html',
'source':'https://fc2club.com//html/FC2-' + number + '.html',
}
js = json.dumps(dic, ensure_ascii=False, sort_keys=True, indent=4, separators=(',', ':'),)#.encode('UTF-8')
return js
return js
#print(main('1051725'))