updating movieposterdb module
This commit is contained in:
parent
4781644eac
commit
acb0e3f0e8
1 changed files with 27 additions and 21 deletions
|
@ -3,28 +3,34 @@
|
|||
|
||||
import re
|
||||
|
||||
from oxlib import cache
|
||||
from oxlib.cache import getUrlUnicode
|
||||
from oxlib import findRe
|
||||
|
||||
def getPosterUrls(imdbId):
|
||||
url = 'http://www.movieposterdb.com/movie/%s' % imdbId
|
||||
posterUrls = []
|
||||
if cache.exists(url):
|
||||
posterUrls = parsePage(url)
|
||||
return posterUrls
|
||||
|
||||
def parsePage(url):
|
||||
posterUrls = []
|
||||
html = cache.getUrlUnicode(url, timeout=86400)
|
||||
groups = re.compile('<a href="(http://www.movieposterdb.com/group/.*?)">', re.DOTALL).findall(html)
|
||||
for group in groups:
|
||||
posterUrls += parsePage(group)
|
||||
posters = re.compile('<a href="(http://www.movieposterdb.com/poster/.*?)">', re.DOTALL).findall(html)
|
||||
for poster in posters:
|
||||
html = cache.getUrlUnicode(poster)
|
||||
posterUrls.append(findRe(html, '"(http://www.movieposterdb.com/posters/.*?\.jpg)"'))
|
||||
return posterUrls
|
||||
def getData(id):
|
||||
data = {
|
||||
"url": getUrl(id)
|
||||
}
|
||||
data["posters"] = getPostersByUrl(data["url"])
|
||||
return data
|
||||
|
||||
def getId(url):
|
||||
return url.split("/")[-2]
|
||||
|
||||
def getPostersByUrl(url):
|
||||
posters = []
|
||||
html = getUrlUnicode(url)
|
||||
results = re.compile('<a href="(http://www.movieposterdb.com/group/.*?)">', re.DOTALL).findall(html)
|
||||
for result in results:
|
||||
posters += getPostersByUrl(result)
|
||||
results = re.compile('<a href="(http://www.movieposterdb.com/poster/.*?)">', re.DOTALL).findall(html)
|
||||
for result in results:
|
||||
html = getUrlUnicode(result)
|
||||
posters.append(findRe(html, '"(http://www.movieposterdb.com/posters/.*?\.jpg)"'))
|
||||
return posters
|
||||
|
||||
def getUrl(id):
|
||||
return "http://www.movieposterdb.com/movie/%s/" % id
|
||||
|
||||
if __name__ == '__main__':
|
||||
print getPosterUrls('0133093')
|
||||
print getPosterUrls('0060304')
|
||||
print getData('0060304')
|
||||
print getData('0133093')
|
||||
|
|
Loading…
Reference in a new issue