Fix crash when stylesheet is not available

weasyprint/__init__.py

@@ -29,7 +29,7 @@
 import lxml.etree
 
 from .urls import (default_url_fetcher, wrap_url_fetcher,
-                   path2url, ensure_url, url_is_absolute)
+                   path2url, ensure_url, url_is_absolute, URLError)
 from .logger import LOGGER
 # Some import are at the end of the file (after the CSS class) is defined
 # to work around circular imports.
@@ -77,7 +77,7 @@ def __init__(self, guess=None, filename=None, url=None, file_obj=None,
 
         source_type, source, base_url, protocol_encoding = _select_source(
             guess, filename, url, file_obj, string, tree, base_url,
-            url_fetcher)
+            url_fetcher, required=True)
 
         if source_type == 'tree':
             result = source
@@ -241,7 +241,8 @@ def __init__(self, guess=None, filename=None, url=None, file_obj=None,
 
 def _select_source(guess=None, filename=None, url=None, file_obj=None,
                    string=None, tree=None, base_url=None,
-                   url_fetcher=default_url_fetcher, check_css_mime_type=False):
+                   url_fetcher=default_url_fetcher, check_css_mime_type=False,
+                   required=False):
     """
     Check that only one input is not None, and return it with the
     normalized ``base_url``.
@@ -261,26 +262,34 @@ def _select_source(guess=None, filename=None, url=None, file_obj=None,
             type_ = 'filename'
         return _select_source(
             base_url=base_url, url_fetcher=url_fetcher,
-            check_css_mime_type=check_css_mime_type,
+            check_css_mime_type=check_css_mime_type, required=required,
             **{type_: guess})
     if nones == [True, False, True, True, True, True]:
         if base_url is None:
             base_url = path2url(filename)
         return 'filename', filename, base_url, None
     if nones == [True, True, False, True, True, True]:
-        result = url_fetcher(url)
-        if check_css_mime_type and result['mime_type'] != 'text/css':
-            LOGGER.warn(
-                'Unsupported stylesheet type %s for %s',
-                result['mime_type'], result['redirected_url'])
+        try:
+            result = url_fetcher(url)
+        except URLError as e:
+          if not required:
+            LOGGER.warn('Failed to load resource %s: %s', url, e)
             return 'string', '', base_url, None
-        protocol_encoding = result.get('encoding')
-        if base_url is None:
-            base_url = result.get('redirected_url', url)
-        if 'string' in result:
-            return 'string', result['string'], base_url, protocol_encoding
+          else:
+            raise e
         else:
-            return 'file_obj', result['file_obj'], base_url, protocol_encoding
+          if check_css_mime_type and result['mime_type'] != 'text/css':
+              LOGGER.warn(
+                  'Unsupported stylesheet type %s for %s',
+                  result['mime_type'], result['redirected_url'])
+              return 'string', '', base_url, None
+          protocol_encoding = result.get('encoding')
+          if base_url is None:
+              base_url = result.get('redirected_url', url)
+          if 'string' in result:
+              return 'string', result['string'], base_url, protocol_encoding
+          else:
+              return 'file_obj', result['file_obj'], base_url, protocol_encoding
     if nones == [True, True, True, False, True, True]:
         if base_url is None:
             # filesystem file-like objects have a 'name' attribute.

weasyprint/compat.py

@@ -16,7 +16,7 @@
 import email
 
 
-__all__ = ['Request', 'base64_decode', 'base64_encode', 'basestring',
+__all__ = ['Request', 'URLError', 'base64_decode', 'base64_encode', 'basestring',
            'ints_from_bytes', 'iteritems', 'izip', 'parse_email', 'parse_qs',
            'pathname2url', 'quote', 'unicode', 'unquote', 'unquote_to_bytes',
            'urlencode', 'urljoin', 'urlopen', 'urlopen_contenttype',
@@ -29,6 +29,7 @@
         urljoin, urlsplit, quote, unquote, unquote_to_bytes, parse_qs,
         urlencode, uses_relative as urlparse_uses_relative)
     from urllib.request import urlopen, Request, pathname2url
+    from urllib.error import URLError
     from array import array
     from base64 import (decodebytes as base64_decode,
                         encodebytes as base64_encode)
@@ -60,7 +61,7 @@ def ints_from_bytes(byte_string):
     # Python 2
     from urlparse import (urljoin, urlsplit, parse_qs,
                           uses_relative as urlparse_uses_relative)
-    from urllib2 import urlopen, Request
+    from urllib2 import urlopen, Request, URLError
     from urllib import pathname2url, quote, unquote, urlencode
     from array import array as _array
     from itertools import izip, imap

weasyprint/urls.py

@@ -22,7 +22,7 @@
 from .logger import LOGGER
 from .compat import (
     urljoin, urlsplit, quote, unquote, unquote_to_bytes, urlopen_contenttype,
-    Request, parse_email, pathname2url, unicode, base64_decode)
+    Request, parse_email, pathname2url, unicode, base64_decode, URLError)
 
 
 # Unlinke HTML, CSS and PNG, the SVG MIME type is not always builtin