Download every liked image on tumblr to a folder. Requires python3, requests, and beautifulsoup.

downtumblrlikes.py

# Either input info here, use command-line arguments, or be asked in-program.
email = ""
password = ""
folder = "" #remember to use / for directories, not \
threads = 8

########################################
##### downtumblrlikes.py v20140101 #####
########################################
###### DO NOT EDIT BELOW THIS LINE #####
########################################

import os, requests, time
from bs4 import BeautifulSoup as bs
from concurrent.futures import ThreadPoolExecutor


def downtumblrlikes(email, password, folder, threads):
    s = requests.Session()
    
    ## get form_key to login properly
    print('Getting form_key...')
    r = s.get('https://www.tumblr.com/login')
    formKey = r.text.split('<input type="hidden" name="form_key" value="')[1]
    formKey = formKey.split('"')[0]


    data = {'user[email]': email, 'user[password]': password, 'user[tos]': 1,
            'context': 'login', 'version': 'STANDARD', 'follow': '',
            'http_referer': 'https://www.tumblr.com/login', 'form_key': formKey}


    ## login
    print('Logging in...')
    r = s.post('https://www.tumblr.com/login', data=data)

    if 'logged_in' not in s.cookies or s.cookies['logged_in'] is not '1':
        print('Error logging in.')
        return
    else:
        print('Successfully logged in.')    

    data = {}
    email = ''
    password = ''

    ## get total pages
    r = s.get('http://www.tumblr.com/likes/page/1')
    totalPages = int(bs(r.text).select('a.likes div')[0]['data-count'])//10+1

    ald = set(os.listdir(folder)) #already downloaded files
    totalBytes = 0
    sTime = time.time()

    def parsePage(page):
        imgs = []

        r = s.get('http://www.tumblr.com/likes/page/%d' % page, headers={
            'X-Requested-With': 'XMLHttpRequest'})
        soup = bs(r.text)


        ## get pics from page (either from script or html)
        posts = soup.find_all("div", class_=lambda x: x in ['is_photo', 'is_photoset'])
        for p in posts:
            script = p('script')
            if not script:
                url = p.select('div.post_media img')[0]['src']
                imgs.append(url)
            else:
                script = script[0].text
                urls = script.split('high_res: \'')
                if len(urls) == 2: #1 url for photo
                    urls = [urls[1].split('\'')[0]]
                else: #many urls for photoset
                    urls = (u.split('"')[0].replace('\\/', '/') for u
                            in script.split('high_res":"')[1:])
                imgs.extend(urls)

        return set(imgs)

    def downloadImage(i):
        filename = i.split('/')[-1]
        if filename in ald:
            return -2 #file already exists
        
        r = s.get(i, stream=True)
        if r.status_code == 200:
            size = 0
            with open(folder+'/'+filename, 'wb') as file:
                for chunk in r.iter_content(1024*4):
                    file.write(chunk)
                size = os.fstat(file.fileno()).st_size
            return size #file sucessfully downloaded
        else:
            return -1
        
    imgs = set()
    downloaded = set()
    existing = set()
    error = set()
         
    print("Grabbing img urls from %d pages:" % totalPages)
    with ThreadPoolExecutor(max_workers=threads) as executor:
        for page, i in zip(range(1,totalPages+1),
                                executor.map(parsePage, range(1,totalPages+1))):
            imgs |= i
            print('\tParsed page %d' % page)


    ## download images
    print('Downloading %d images:' % len(imgs))
    with ThreadPoolExecutor(max_workers=threads) as executor:
        for img, result in zip(imgs, executor.map(downloadImage, imgs)):
            name = img.split('/')[-1]
            if result == -2:
                print('\tAlready exists: %s' % name)
                existing.add(img)
            elif result == -1:
                print('\tError downloading: %s' % name)
                error.add(img)
            else:
                print('\tDownloaded: %s' % name)
                totalBytes += result
                downloaded.add(img)
                    
    eTime = time.time()
                    
    print('Done.')
    print('Downloaded %d new images.' % len(downloaded))
    print('Already had %d images.' % len(existing))
    print('Error dling %d images: \n\t%s' % (len(error), "\n\t".join(error)))
    print('Downloaded %0.2fMB in %0.2fs (%0.2fMb/s)' %
          ((totalBytes/2**20), (eTime-sTime), 8*(totalBytes/2**20)/(eTime-sTime)))

if __name__ == "__main__":
    import argparse, getpass

    parser = argparse.ArgumentParser(description='Download tumblr likes.',
                        prog='downtumblrlikes.py')
    parser.add_argument('-e', '--email', metavar='email')
    parser.add_argument('-p', '--password', metavar='pass')
    parser.add_argument('-f', '--folder', metavar='folder')
    parser.add_argument('-t', '--threads', type=int, metavar='threads')
    parser.add_argument('-v', '--version', action='version', version='%(prog)s 20140101')

    args = parser.parse_args()
    email = args.email if args.email != None else email
    password = args.password if args.password != None else password
    folder = args.folder if args.folder != None else folder
    threads = args.threads if args.threads != None else threads
    
    if len(email) == 0:
        email = input('Email: ')
    if len(password) == 0:
        password = getpass.getpass('Password: ')
    if len(folder) == 0:
        folder = input('Folder: ')
    if threads < 1:
        threads = int(input('Threads: '))
        
    downtumblrlikes(email, password, folder, threads)

Rude! Anyway, I've just made a script for this, but it uses the consumer secret and key to communicate with the tumblr API. However, this seems like it would work for a much more general case. I imagine it was a PITA to get all the scraping and automating login just right.

Nice job. I'll post mine later so you can see how I went about it, if you're interested at all.