PIL/Pillow IOError: cannot identify image file u'xxxx.jpg' (Windows only)

Question

I've constructed Python script and it works well on OS X/Linux but I'm having problems in Windows (see title). It's using Pillow module and the error originates in module PIL\Image.py on line 2274.

My code:

# -*- coding: utf-8 -*-

import os
import sys
import urllib2
from PIL import Image, ImageFile
from PyPDF2 import PdfFileReader, PdfFileWriter, PdfFileMerger
from bs4 import BeautifulSoup

ImageFile.LOAD_TRUNCATED_IMAGES = True

def parser():
    try:
        return sys.argv[1].lower()
    except IndexError:
        print 'no argument specified'


the_url = 'http://www.oldgames.sk'
base_url = the_url + '/mags/'

# Add magazines + relative URLs here
magazines = {
        'score': 'score/',
        'level': 'level/',
        'amiga': 'amiga-magazin/',
        'bit': 'bit/',
        'commodore': 'commodore-amater/',
        'CGW': 'cgw/',
        'excalibur': 'excalibur/',
        'hrac': 'hrac-cz/',
        'joystick': 'joystick-sk/',
        'pocitac-aktivne': 'pocitac-aktivne/',
        'pocitacove-hry': 'pocitacove-hry/',
        'riki': 'riki/',
        'zzap64': 'zzap64/'}

issue_links = []
download_list = {}

def parse_args(arg):
    if arg == '--list':
        items = [i for i in magazines.keys()]
        for item in items:
            print item
        sys.exit()
    elif arg in magazines:
        print "Scraping %s magazine..." % arg.capitalize()
        return base_url + magazines[arg]
    else:
        return sys.exit('invalid magazine name')

def extract_links_to_issue(url):
    soup = BeautifulSoup(urllib2.urlopen(url))

    for div in soup.findAll('div','mImage'):
        issue_links.append(the_url + div.a['href'])

    print 'Scraped %d links' % len(issue_links)

def issue_renamer(issue_name):
    char1 = '\\'
    char2 = '/'
    replacement = '-'
    if char1 in issue_name:
        issue_name = issue_name.replace(char1, replacement)
        print 'inv. char (%s): renaming to %s' % (char1, issue_name)
    elif char2 in issue_name:
        issue_name = issue_name.replace(char2, replacement)
        print 'inv. char (%s): renaming to %s' % (char2, issue_name)

    return issue_name

def extract_links_to_images(issue_links):
    for index, link in enumerate(issue_links):
        print 'Scraping issue #%d: %s' % (index + 1, link)
        issue_soup = BeautifulSoup(urllib2.urlopen(link))
        image_list = []
        for image in issue_soup.findAll('div', 'mags_thumb_article'):
            issue_name = issue_renamer(issue_soup.findAll('h1','top')[0].text)
            image_list.append(the_url + image.a['href'])

        download_list[issue_name] = image_list

def clean_up(list_of_files, list_of_pdfs):
    num = len(list_of_files) + len(list_of_pdfs)
    for file in list_of_files:
        os.remove(file)
    for pdf in list_of_pdfs:
        os.remove(pdf)
    print 'Cleaned up %d files' % num

def convert_images(list_of_files, issue):
    list_of_pdfs = []
    for index, file in enumerate(list_of_files):
        im = Image.open(file)
        outfile = file + '.pdf'
        im.save(outfile, 'PDF')
        list_of_pdfs.append(outfile)

        print 'converting ...' + str((index + 1)) + '/' + str(len(list_of_files))

    final_pdf = PdfFileMerger()
    for pdf in list_of_pdfs:
        final_pdf.append(open(pdf, 'rb'))

    issue_name = issue + '.pdf'
    final_pdf.write(open(issue_name, 'wb'))
    final_pdf.close()
    print '--- PDF completed ---'

    clean_up(list_of_files, list_of_pdfs)

def download_images(download_list):
    for issues,image_list in download_list.items():
        print 'Preparing %s ...' % issues
        list_of_files = []
        for image in image_list:
            image_name = os.path.split(image)[1]
            list_of_files.append(image_name)
            f = open(image_name, 'w')
            f.write(urllib2.urlopen(image).read())
            print 'Downloading image: %s' % image
            f.close()
        convert_images(list_of_files, issues)


arg = parser()
extract_links_to_issue(parse_args(arg))
extract_links_to_images(issue_links)
download_images(download_list)

I'd like to fix this, can anyone help me?

Answer 1

You are copying images into a file opened in text mode:

f = open(image_name, 'w')
f.write(urllib2.urlopen(image).read())

On Windows this means that any 0A (newline) bytes are translated to 0D 0A byte sequences (carriage return, newline), as that is the Windows line separator.

Open your files in binary mode:

f = open(image_name, 'wb')
f.write(urllib2.urlopen(image).read())

I'd switch to using the file as a context manager (with the with statement) so you don't have to manually close it, and using shutil.copyfileobj() to stream the data straight to disk (in blocks) rather than read the whole image into memory in one go:

import shutil

# ...
with open(image_name, 'wb') as f:
    shutil.copyfileobj(urllib2.urlopen(image), f)