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

Beta 10.4 更新

Browse Source
This commit is contained in:
wenead99 2019-06-21 18:27:21 +08:00 committed by GitHub
parent 420f995977
commit 45e1d8370c
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
4 changed files with 43 additions and 26 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
AV_Data_Capture.py
View File

@ -71,8 +71,8 @@ if __name__ =='__main__':
if '_' in i: if '_' in i:
os.rename(re.search(r'[^\\/:*?"<>|\r\n]+$', i).group(), rreplace(re.search(r'[^\\/:*?"<>|\r\n]+$', i).group(), '_', '-', 1)) os.rename(re.search(r'[^\\/:*?"<>|\r\n]+$', i).group(), rreplace(re.search(r'[^\\/:*?"<>|\r\n]+$', i).group(), '_', '-', 1))
i = rreplace(re.search(r'[^\\/:*?"<>|\r\n]+$', i).group(), '_', '-', 1) i = rreplace(re.search(r'[^\\/:*?"<>|\r\n]+$', i).group(), '_', '-', 1)
os.system('python core.py' + ' "' + i + '"') #选择从py文件启动 （用于源码py） #os.system('python core.py' + ' "' + i + '"') #选择从py文件启动 （用于源码py）
#os.system('core.exe' + ' "' + i + '"') #选择从exe文件启动（用于EXE版程序） os.system('core.exe' + ' "' + i + '"') #选择从exe文件启动（用于EXE版程序）
print("[*]=====================================") print("[*]=====================================")
print("[!]Cleaning empty folders") print("[!]Cleaning empty folders")

59
fc2fans_club.py
View File

@ -10,6 +10,14 @@ def getTitle(htmlcode): #获取厂商
result2 = str(re.sub('\D{2}2-\d+','',result)).replace(' ','',1) result2 = str(re.sub('\D{2}2-\d+','',result)).replace(' ','',1)
#print(result2) #print(result2)
return result2 return result2
def getActor(htmlcode):
try:
html = etree.fromstring(htmlcode, etree.HTMLParser())
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 def getStudio(htmlcode): #获取厂商
html = etree.fromstring(htmlcode,etree.HTMLParser()) html = etree.fromstring(htmlcode,etree.HTMLParser())
result = str(html.xpath('/html/body/div[2]/div/div[1]/h5[3]/a[1]/text()')).strip(" ['']") result = str(html.xpath('/html/body/div[2]/div/div[1]/h5[3]/a[1]/text()')).strip(" ['']")
@ -19,47 +27,58 @@ def getNum(htmlcode): #获取番号
result = str(html.xpath('/html/body/div[5]/div[1]/div[2]/p[1]/span[2]/text()')).strip(" ['']") result = str(html.xpath('/html/body/div[5]/div[1]/div[2]/p[1]/span[2]/text()')).strip(" ['']")
#print(result) #print(result)
return result return result
def getRelease(number): 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') #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(a,etree.HTMLParser()) 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(" ['']") result = str(html.xpath('//*[@id="container"]/div[1]/div/article/section[1]/div/div[2]/dl/dd[4]/text()')).strip(" ['']")
return result return result
def getCover(htmlcode,number): #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') #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(a, etree.HTMLParser()) 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(" ['']") 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 'http://fc2fans.club' + result2
return 'http:' + result return 'http:' + result
def getOutline(htmlcode,number): #获取番号
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') def getOutline(htmlcode2,number): #获取番号 #
html = etree.fromstring(a, etree.HTMLParser()) #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[4]/p/text()')).replace("\\n",'',10000).strip(" ['']").replace("'",'',10000) result = str(html.xpath('//*[@id="container"]/div[1]/div/article/section[4]/p/text()')).replace("\\n",'',10000).strip(" ['']").replace("'",'',10000)
return result return result
# def getTag(htmlcode,number): #获取番号 def getTag(htmlcode): #获取番号
# 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(htmlcode, etree.HTMLParser())
# html = etree.fromstring(a, etree.HTMLParser()) result = str(html.xpath('/html/body/div[2]/div/div[1]/h5[4]/a/text()'))
# result = str(html.xpath('//*[@id="container"]/div[1]/div/article/section[4]/p/text()')).replace("\\n",'',10000).strip(" ['']").replace("'",'',10000) return result.strip(" ['']").replace("'",'').replace(' ','')
# return result def getYear(release):
try:
result = re.search('\d{4}',release).group()
return result
except:
return ''
def main(number2): def main(number2):
number=number2.replace('PPV','').replace('ppv','') number=number2.replace('PPV','').replace('ppv','')
htmlcode2 = 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')
htmlcode = ADC_function.get_html('http://fc2fans.club/html/FC2-' + number + '.html') htmlcode = ADC_function.get_html('http://fc2fans.club/html/FC2-' + number + '.html')
dic = { dic = {
'title': getTitle(htmlcode), 'title': getTitle(htmlcode),
'studio': getStudio(htmlcode), 'studio': getStudio(htmlcode),
'year': str(re.search('\d{4}',getRelease(number)).group()), 'year': '',#str(re.search('\d{4}',getRelease(number)).group()),
'outline': getOutline(htmlcode,number), 'outline': getOutline(htmlcode,number),
'runtime': '', 'runtime': getYear(getRelease(htmlcode)),
'director': getStudio(htmlcode), 'director': getStudio(htmlcode),
'actor': '', 'actor': getActor(htmlcode),
'release': getRelease(number), 'release': getRelease(number),
'number': 'FC2-'+number, 'number': 'FC2-'+number,
'cover': getCover(htmlcode,number), 'cover': getCover(htmlcode,number,htmlcode2),
'imagecut': 0, 'imagecut': 0,
'tag':" ", 'tag':getTag(htmlcode),
} }
#print(getTitle(htmlcode)) #print(getTitle(htmlcode))
#print(getNum(htmlcode)) #print(getNum(htmlcode))
js = json.dumps(dic, ensure_ascii=False, sort_keys=True, indent=4, separators=(',', ':'),)#.encode('UTF-8') js = json.dumps(dic, ensure_ascii=False, sort_keys=True, indent=4, separators=(',', ':'),)#.encode('UTF-8')
return js return js
#print(main('1104989')) #print(main('1051725'))

2
javbus.py
View File

@ -164,8 +164,6 @@ def main_uncensored(number):
return js return js
#print(main('SIRO-3821'))
# def return1(): # def return1():
# json_data=json.loads(main('ipx-292')) # json_data=json.loads(main('ipx-292'))

2
proxy.ini
View File

@ -3,4 +3,4 @@ proxy=127.0.0.1:1080
[Name_Rule] [Name_Rule]
location_rule='JAV_output/'+actor+'/'+'['+number+']-'+title location_rule='JAV_output/'+actor+'/'+'['+number+']-'+title
naming_rule=number naming_rule=number+'-'+title