import os, sys, re, urllib, stringreg = re.compile(r"&#[0-9]{1,3};")def fix(s): codes = reg.findall(s) for i in codes: s = s.replace(i,'%c'%(int(i.replace('&#','').replace(';','')))).replace('<em>','').replace('</em>','') #.replace('<br>','') return(s)def replaceOne(args): return (args * 1) + (args == 0)class IMDb: def __init__(self, url, Info=False, Details=False, Gallery=False): if Info == True: fromInfo = urllib.urlopen(url).read() self.fromInfo = fix(fromInfo).replace(' / ','').replace('</a> <i>',' ').replace('</i>','</a>') if Details == True: self.fromDetails = urllib.urlopen(url + 'plotsummary').read() if Gallery == True: self.fromGallery = urllib.urlopen(url + 'photogallery').read()def Galleries(fromGallery, page='0'): in_gallery = re.compile('"(photogallery.*?|/gallery/.*?)"><img src="(.*?jpg)"') in_pgLeft = re.compile('There are.*?>(.*?) more</a>') Gallery = [] GalInfo = [] imgLeft = [] pgLeft = [] for a in in_gallery.findall(fromGallery): Gallery.append(a[0]) imgLeft.append(urllib.urlopen(url + a[0]).read()) for a in imgLeft: pgLeft.append(in_pgLeft.findall(a)) for a in pgLeft: GalInfo.append(int(str(a)[2:-2]) / 12) for cnt, item in enumerate(GalInfo): GalInfo[cnt] = (item * 1) + (item == 0) #for cnt in range(0,25,12): #Links = Gallery[0][:Gallery[0].rfind('-')] + str(int(Gallery[0][Gallery[0].rfind('-') + 1:]) + 12) Links = [] for b in GalInfo: for cnt in range(0,int(b*12+1),12): for a in Gallery: Links.append(url + a[:-1] + str(cnt)) print Links #print thumbs[1] #for a in pgleft: # pgLeft = int(a) / 12 #if '0' in pgLeft: # pgLeft.replace('0','1') #print pgLeft #for a in Gallery: # thumbs = urllib.urlopen(url + a).read() #gallery = in_gallery.findall(thumbs) # find thumbs for 1st Gallery page of each Gallery #for a in gallery: # Thumbs.append(a[1].replace('th-','')) # changes Thumb images to full-size Images #pgleft = in_pgLeft.findall(thumbs) # parse how many thumbs are left #for a in pgleft: # pgLeft = int(a) / 12 # how many pages are in each gallery #return Gallery, Thumbsnum = '0'url = 'http://www.imdb.com/title/tt0315824/'imdb = IMDb(url, Gallery=True)gal = Galleries(imdb.fromGallery, page=num)#print gal#print thumb