Edit book: When downloading external resources, also convert data URLs into files. Fixes #1774945 [[Enhancement] Image DataURI](https://bugs.launchpad.net/calibre/+bug/1774945)

src/calibre/ebooks/oeb/polish/download.py

@@ -2,20 +2,27 @@
 # vim:fileencoding=utf-8
 # License: GPLv3 Copyright: 2016, Kovid Goyal <kovid at kovidgoyal.net>
 
-from __future__ import (unicode_literals, division, absolute_import,
+from __future__ import absolute_import, division, print_function, unicode_literals
-                        print_function)
+
-import shutil, os, posixpath, cgi, mimetypes
+import cgi
+import mimetypes
+import os
+import posixpath
+import re
+import shutil
+from base64 import standard_b64decode
 from collections import defaultdict
 from contextlib import closing
-from urlparse import urlparse
-from multiprocessing.dummy import Pool
 from functools import partial
+from io import BytesIO
+from multiprocessing.dummy import Pool
 from tempfile import NamedTemporaryFile
 from urllib2 import urlopen
+from urlparse import urlparse
 
 from calibre import as_unicode, sanitize_file_name2
+from calibre.ebooks.oeb.base import OEB_DOCS, OEB_STYLES, barename, iterlinks
 from calibre.ebooks.oeb.polish.utils import guess_type
-from calibre.ebooks.oeb.base import OEB_DOCS, iterlinks, barename, OEB_STYLES
 from calibre.ptempfile import TemporaryDirectory
 from calibre.web import get_download_filename_from_response
 
@@ -25,7 +32,7 @@ def is_external(url):
         purl = urlparse(url)
     except Exception:
         return False
-    return purl.scheme in ('http', 'https', 'file', 'ftp')
+    return purl.scheme in ('http', 'https', 'file', 'ftp', 'data')
 
 
 def iterhtmllinks(container, name):
@@ -104,6 +111,24 @@ def download_one(tdir, timeout, progress_report, url):
                 src = lopen(purl.path, 'rb')
                 filename = os.path.basename(src)
                 sz = (src.seek(0, os.SEEK_END), src.tell(), src.seek(0))[1]
+            elif purl.scheme == 'data':
+                prefix, payload = purl.path.split(',', 1)
+                parts = prefix.split(';')
+                if parts and parts[-1].lower() == 'base64':
+                    payload = re.sub(r'\s+', '', payload)
+                    payload = standard_b64decode(payload)
+                else:
+                    payload = payload.encode('utf-8')
+                src = BytesIO(payload)
+                sz = len(payload)
+                ext = 'unknown'
+                for x in parts:
+                    if '=' not in x and '/' in x:
+                        exts = mimetypes.guess_all_extensions(x)
+                        if exts:
+                            ext = exts[0]
+                            break
+                filename = 'data-uri.' + ext
             else:
                 src = urlopen(url, timeout=timeout)
                 filename = get_filename(purl, src)

src/calibre/gui2/tweak_book/download.py

@@ -42,14 +42,20 @@ class ChooseResources(QWidget):
 
     @property
     def resources(self):
-        return {i.text():self.original_resources[i.text()] for i in self if i.checkState() == Qt.Checked}
+        return {i.data(Qt.UserRole):self.original_resources[i.data(Qt.UserRole)] for i in self if i.checkState() == Qt.Checked}
 
     @resources.setter
     def resources(self, resources):
         self.items.clear()
         self.original_resources = resources
+        dc = 0
         for url in resources:
-            i = QListWidgetItem(url, self.items)
+            text = url
+            if text.startswith('data:'):
+                dc += 1
+                text = _('Data URL ({})').format(dc)
+            i = QListWidgetItem(text, self.items)
+            i.setData(Qt.UserRole, url)
             i.setCheckState(Qt.Checked)
             i.setFlags(Qt.ItemIsUserCheckable | Qt.ItemIsEnabled)