[youtube] Adds #1312 Download annotations

Adds #1321 Download annotations from youtube
Annotations are downloaded and written to a .annotations.xml file using the https://www.youtube.com/annotations_invideo?features=1&legacy=1&video_id=$VIDEOID API.
Added unit test for annotations.

test/test_write_annotations.py

@@ -0,0 +1,82 @@
+#!/usr/bin/env python
+# coding: utf-8
+
+import xml.etree.ElementTree
+import os
+import sys
+import unittest
+
+# Allow direct execution
+sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+
+import youtube_dl.YoutubeDL
+import youtube_dl.extractor
+from youtube_dl.utils import *
+from .helper import try_rm
+
+PARAMETERS_FILE = os.path.join(os.path.dirname(os.path.abspath(__file__)), "parameters.json")
+
+# General configuration (from __init__, not very elegant...)
+jar = compat_cookiejar.CookieJar()
+cookie_processor = compat_urllib_request.HTTPCookieProcessor(jar)
+proxy_handler = compat_urllib_request.ProxyHandler()
+opener = compat_urllib_request.build_opener(proxy_handler, cookie_processor, YoutubeDLHandler())
+compat_urllib_request.install_opener(opener)
+
+class YoutubeDL(youtube_dl.YoutubeDL):
+    def __init__(self, *args, **kwargs):
+        super(YoutubeDL, self).__init__(*args, **kwargs)
+        self.to_stderr = self.to_screen
+
+with io.open(PARAMETERS_FILE, encoding='utf-8') as pf:
+    params = json.load(pf)
+params['writeannotations'] = True
+params['skip_download'] = True
+params['writeinfojson'] = False
+params['format'] = 'flv'
+
+TEST_ID = 'gr51aVj-mLg'
+ANNOTATIONS_FILE = TEST_ID + '.flv.annotations.xml'
+EXPECTED_ANNOTATIONS = ['Speech bubble', 'Note', 'Title', 'Spotlight', 'Label']
+
+class TestAnnotations(unittest.TestCase):
+    def setUp(self):
+        # Clear old files
+        self.tearDown()
+
+
+    def test_info_json(self):
+        expected = list(EXPECTED_ANNOTATIONS) #Two annotations could have the same text.
+        ie = youtube_dl.extractor.YoutubeIE()
+        ydl = YoutubeDL(params)
+        ydl.add_info_extractor(ie)
+        ydl.download([TEST_ID])
+        self.assertTrue(os.path.exists(ANNOTATIONS_FILE))
+        annoxml = None
+        with io.open(ANNOTATIONS_FILE, 'r', encoding='utf-8') as annof:
+                annoxml = xml.etree.ElementTree.parse(annof)
+        self.assertTrue(annoxml is not None, 'Failed to parse annotations XML')
+        root = annoxml.getroot()
+        self.assertEqual(root.tag, 'document')
+        annotationsTag = root.find('annotations')
+        self.assertEqual(annotationsTag.tag, 'annotations')
+        annotations = annotationsTag.findall('annotation')
+
+        #Not all the annotations have TEXT children and the annotations are returned unsorted.
+        for a in annotations:
+                self.assertEqual(a.tag, 'annotation')
+                if a.get('type') == 'text':
+                        textTag = a.find('TEXT')
+                        text = textTag.text
+                        self.assertTrue(text in expected) #assertIn only added in python 2.7
+                        #remove the first occurance, there could be more than one annotation with the same text
+                        expected.remove(text)
+        #We should have seen (and removed) all the expected annotation texts.
+        self.assertEqual(len(expected), 0, 'Not all expected annotations were found.')
+        
+
+    def tearDown(self):
+        try_rm(ANNOTATIONS_FILE)
+
+if __name__ == '__main__':
+    unittest.main()