Use a wrapper around xml.etree.ElementTree.fromstring in python 2.x (#7178)

Attributes aren't unicode objects, so they couldn't be directly used in info_dict fields (for example '--write-description' doesn't work with bytes).

test/test_compat.py

@@ -13,8 +13,10 @@
 from youtube_dl.utils import get_filesystem_encoding
 from youtube_dl.compat import (
     compat_getenv,
+    compat_etree_fromstring,
     compat_expanduser,
     compat_shlex_split,
+    compat_str,
     compat_urllib_parse_unquote,
     compat_urllib_parse_unquote_plus,
 )
@@ -71,5 +73,10 @@ def test_compat_urllib_parse_unquote_plus(self):
     def test_compat_shlex_split(self):
         self.assertEqual(compat_shlex_split('-option "one two"'), ['-option', 'one two'])
 
+    def test_compat_etree_fromstring(self):
+        xml = '<el foo="bar"></el>'
+        doc = compat_etree_fromstring(xml.encode('utf-8'))
+        self.assertTrue(isinstance(doc.attrib['foo'], compat_str))
+
 if __name__ == '__main__':
     unittest.main()

test/test_utils.py

@@ -68,6 +68,9 @@
     cli_valueless_option,
     cli_bool_option,
 )
+from youtube_dl.compat import (
+    compat_etree_fromstring,
+)
 
 
 class TestUtil(unittest.TestCase):
@@ -242,7 +245,7 @@ def test_find_xpath_attr(self):
             <node x="b" y="d" />
             <node x="" />
         </root>'''
-        doc = xml.etree.ElementTree.fromstring(testxml)
+        doc = compat_etree_fromstring(testxml)
 
         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
@@ -263,7 +266,7 @@ def test_xpath_with_ns(self):
                 <url>http://server.com/download.mp3</url>
             </media:song>
         </root>'''
-        doc = xml.etree.ElementTree.fromstring(testxml)
+        doc = compat_etree_fromstring(testxml)
         find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
         self.assertTrue(find('media:song') is not None)
         self.assertEqual(find('media:song/media:author').text, 'The Author')
@@ -285,7 +288,7 @@ def test_xpath_text(self):
                 <p>Foo</p>
             </div>
         </root>'''
-        doc = xml.etree.ElementTree.fromstring(testxml)
+        doc = compat_etree_fromstring(testxml)
         self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
         self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
         self.assertTrue(xpath_text(doc, 'div/bar') is None)
@@ -297,7 +300,7 @@ def test_xpath_attr(self):
                 <p x="a">Foo</p>
             </div>
         </root>'''
-        doc = xml.etree.ElementTree.fromstring(testxml)
+        doc = compat_etree_fromstring(testxml)
         self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
         self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
         self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)

youtube_dl/compat.py

@@ -14,6 +14,7 @@
 import subprocess
 import sys
 import itertools
+import xml.etree.ElementTree
 
 
 try:
@@ -212,6 +213,29 @@ def data_open(self, req):
 except ImportError:  # Python 2.6
     from xml.parsers.expat import ExpatError as compat_xml_parse_error
 
+if sys.version_info[0] >= 3:
+    compat_etree_fromstring = xml.etree.ElementTree.fromstring
+else:
+    # on python 2.x the the attributes of a node are str objects instead of
+    # unicode
+    etree = xml.etree.ElementTree
+
+    # on 2.6 XML doesn't have a parser argument, function copied from CPython
+    # 2.7 source
+    def _XML(text, parser=None):
+        if not parser:
+            parser = etree.XMLParser(target=etree.TreeBuilder())
+        parser.feed(text)
+        return parser.close()
+
+    def _element_factory(*args, **kwargs):
+        el = etree.Element(*args, **kwargs)
+        for k, v in el.items():
+            el.set(k, v.decode('utf-8'))
+        return el
+
+    def compat_etree_fromstring(text):
+        return _XML(text, parser=etree.XMLParser(target=etree.TreeBuilder(element_factory=_element_factory)))
 
 try:
     from urllib.parse import parse_qs as compat_parse_qs
@@ -507,6 +531,7 @@ def compat_itertools_count(start=0, step=1):
     'compat_chr',
     'compat_cookiejar',
     'compat_cookies',
+    'compat_etree_fromstring',
     'compat_expanduser',
     'compat_get_terminal_size',
     'compat_getenv',

youtube_dl/downloader/f4m.py

@@ -5,10 +5,10 @@
 import itertools
 import os
 import time
-import xml.etree.ElementTree as etree
 
 from .fragment import FragmentFD
 from ..compat import (
+    compat_etree_fromstring,
     compat_urlparse,
     compat_urllib_error,
     compat_urllib_parse_urlparse,
@@ -290,7 +290,7 @@ def real_download(self, filename, info_dict):
         man_url = urlh.geturl()
         manifest = urlh.read()
 
-        doc = etree.fromstring(manifest)
+        doc = compat_etree_fromstring(manifest)
         formats = [(int(f.attrib.get('bitrate', -1)), f)
                    for f in self._get_unencrypted_media(doc)]
         if requested_bitrate is None:

youtube_dl/extractor/bbc.py

@@ -2,7 +2,6 @@
 from __future__ import unicode_literals
 
 import re
-import xml.etree.ElementTree
 
 from .common import InfoExtractor
 from ..utils import (
@@ -14,7 +13,10 @@
     remove_end,
     unescapeHTML,
 )
-from ..compat import compat_HTTPError
+from ..compat import (
+    compat_etree_fromstring,
+    compat_HTTPError,
+)
 
 
 class BBCCoUkIE(InfoExtractor):
@@ -344,7 +346,7 @@ def _download_media_selector_url(self, url, programme_id=None):
                 url, programme_id, 'Downloading media selection XML')
         except ExtractorError as ee:
             if isinstance(ee.cause, compat_HTTPError) and ee.cause.code == 403:
-                media_selection = xml.etree.ElementTree.fromstring(ee.cause.read().decode('utf-8'))
+                media_selection = compat_etree_fromstring(ee.cause.read().decode('utf-8'))
             else:
                 raise
         return self._process_media_selector(media_selection, programme_id)

youtube_dl/extractor/bilibili.py

@@ -4,9 +4,11 @@
 import re
 import itertools
 import json
-import xml.etree.ElementTree as ET
 
 from .common import InfoExtractor
+from ..compat import (
+    compat_etree_fromstring,
+)
 from ..utils import (
     int_or_none,
     unified_strdate,
@@ -88,7 +90,7 @@ def _real_extract(self, url):
         except ValueError:
             pass
 
-        lq_doc = ET.fromstring(lq_page)
+        lq_doc = compat_etree_fromstring(lq_page)
         lq_durls = lq_doc.findall('./durl')
 
         hq_doc = self._download_xml(

youtube_dl/extractor/brightcove.py

@@ -3,10 +3,10 @@
 
 import re
 import json
-import xml.etree.ElementTree
 
 from .common import InfoExtractor
 from ..compat import (
+    compat_etree_fromstring,
     compat_parse_qs,
     compat_str,
     compat_urllib_parse,
@@ -119,7 +119,7 @@ def _build_brighcove_url(cls, object_str):
         object_str = fix_xml_ampersands(object_str)
 
         try:
-            object_doc = xml.etree.ElementTree.fromstring(object_str.encode('utf-8'))
+            object_doc = compat_etree_fromstring(object_str.encode('utf-8'))
         except compat_xml_parse_error:
             return
 

youtube_dl/extractor/common.py

@@ -10,7 +10,6 @@
 import socket
 import sys
 import time
-import xml.etree.ElementTree
 
 from ..compat import (
     compat_cookiejar,
@@ -23,6 +22,7 @@
     compat_urllib_request,
     compat_urlparse,
     compat_str,
+    compat_etree_fromstring,
 )
 from ..utils import (
     NO_DEFAULT,
@@ -461,7 +461,7 @@ def _download_xml(self, url_or_request, video_id,
             return xml_string
         if transform_source:
             xml_string = transform_source(xml_string)
-        return xml.etree.ElementTree.fromstring(xml_string.encode('utf-8'))
+        return compat_etree_fromstring(xml_string.encode('utf-8'))
 
     def _download_json(self, url_or_request, video_id,
                        note='Downloading JSON metadata',

youtube_dl/extractor/crunchyroll.py

@@ -5,12 +5,12 @@
 import json
 import base64
 import zlib
-import xml.etree.ElementTree
 
 from hashlib import sha1
 from math import pow, sqrt, floor
 from .common import InfoExtractor
 from ..compat import (
+    compat_etree_fromstring,
     compat_urllib_parse,
     compat_urllib_parse_unquote,
     compat_urllib_request,
@@ -234,7 +234,7 @@ def ass_bool(strvalue):
         return output
 
     def _extract_subtitles(self, subtitle):
-        sub_root = xml.etree.ElementTree.fromstring(subtitle)
+        sub_root = compat_etree_fromstring(subtitle)
         return [{
             'ext': 'srt',
             'data': self._convert_subtitles_to_srt(sub_root),

youtube_dl/extractor/vevo.py

@@ -1,10 +1,10 @@
 from __future__ import unicode_literals
 
 import re
-import xml.etree.ElementTree
 
 from .common import InfoExtractor
 from ..compat import (
+    compat_etree_fromstring,
     compat_urllib_request,
 )
 from ..utils import (
@@ -97,7 +97,7 @@ def _formats_from_json(self, video_info):
         if last_version['version'] == -1:
             raise ExtractorError('Unable to extract last version of the video')
 
-        renditions = xml.etree.ElementTree.fromstring(last_version['data'])
+        renditions = compat_etree_fromstring(last_version['data'])
         formats = []
         # Already sorted from worst to best quality
         for rend in renditions.findall('rendition'):
@@ -114,7 +114,7 @@ def _formats_from_json(self, video_info):
 
     def _formats_from_smil(self, smil_xml):
         formats = []
-        smil_doc = xml.etree.ElementTree.fromstring(smil_xml.encode('utf-8'))
+        smil_doc = compat_etree_fromstring(smil_xml.encode('utf-8'))
         els = smil_doc.findall('.//{http://www.w3.org/2001/SMIL20/Language}video')
         for el in els:
             src = el.attrib['src']

youtube_dl/utils.py

@@ -36,6 +36,7 @@
 from .compat import (
     compat_basestring,
     compat_chr,
+    compat_etree_fromstring,
     compat_html_entities,
     compat_http_client,
     compat_kwargs,
@@ -1974,7 +1975,7 @@ def parse_node(node):
 
         return out
 
-    dfxp = xml.etree.ElementTree.fromstring(dfxp_data.encode('utf-8'))
+    dfxp = compat_etree_fromstring(dfxp_data.encode('utf-8'))
     out = []
     paras = dfxp.findall(_x('.//ttml:p')) or dfxp.findall(_x('.//ttaf1:p')) or dfxp.findall('.//p')