X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fnytimes.py;h=681683e86f54e796f1c954de2c0cb374016fe303;hb=07d7689f2eb4698cf98ee837b56489b7ff1924df;hp=03f0a4de6dccd8485091b802422bbfaf5f7402df;hpb=ef249a2cd7a7cfbd92a030cb72e238ba4ad52604;p=youtube-dl

diff --git a/youtube_dl/extractor/nytimes.py b/youtube_dl/extractor/nytimes.py
index 03f0a4de6..681683e86 100644
--- a/youtube_dl/extractor/nytimes.py
+++ b/youtube_dl/extractor/nytimes.py
@@ -8,30 +8,8 @@ from ..utils import (
 )
 
 
-class NYTimesIE(InfoExtractor):
-    _VALID_URL = r'https?://(?:(?:www\.)?nytimes\.com/video/(?:[^/]+/)+?|graphics8\.nytimes\.com/bcvideo/\d+(?:\.\d+)?/iframe/embed\.html\?videoId=)(?P<id>\d+)'
-
-    _TESTS = [{
-        'url': 'http://www.nytimes.com/video/opinion/100000002847155/verbatim-what-is-a-photocopier.html?playlistId=100000001150263',
-        'md5': '18a525a510f942ada2720db5f31644c0',
-        'info_dict': {
-            'id': '100000002847155',
-            'ext': 'mov',
-            'title': 'Verbatim: What Is a Photocopier?',
-            'description': 'md5:93603dada88ddbda9395632fdc5da260',
-            'timestamp': 1398631707,
-            'upload_date': '20140427',
-            'uploader': 'Brett Weiner',
-            'duration': 419,
-        }
-    }, {
-        'url': 'http://www.nytimes.com/video/travel/100000003550828/36-hours-in-dubai.html',
-        'only_matching': True,
-    }]
-
-    def _real_extract(self, url):
-        video_id = self._match_id(url)
-
+class NYTimesBaseIE(InfoExtractor):
+    def _extract_video_from_id(self, video_id):
         video_data = self._download_json(
             'http://www.nytimes.com/svc/video/api/v2/video/%s' % video_id,
             video_id, 'Downloading video JSON')
@@ -40,8 +18,9 @@ class NYTimesIE(InfoExtractor):
         description = video_data.get('summary')
         duration = float_or_none(video_data.get('duration'), 1000)
 
-        uploader = video_data['byline']
-        timestamp = parse_iso8601(video_data['publication_date'][:-8])
+        uploader = video_data.get('byline')
+        publication_date = video_data.get('publication_date')
+        timestamp = parse_iso8601(publication_date[:-8]) if publication_date else None
 
         def get_file_size(file_size):
             if isinstance(file_size, int):
@@ -59,7 +38,7 @@ class NYTimesIE(InfoExtractor):
                 'width': int_or_none(video.get('width')),
                 'height': int_or_none(video.get('height')),
                 'filesize': get_file_size(video.get('fileSize')),
-            } for video in video_data['renditions']
+            } for video in video_data['renditions'] if video.get('url')
         ]
         self._sort_formats(formats)
 
@@ -68,7 +47,7 @@ class NYTimesIE(InfoExtractor):
                 'url': 'http://www.nytimes.com/%s' % image['url'],
                 'width': int_or_none(image.get('width')),
                 'height': int_or_none(image.get('height')),
-            } for image in video_data['images']
+            } for image in video_data.get('images', []) if image.get('url')
         ]
 
         return {
@@ -81,3 +60,59 @@ class NYTimesIE(InfoExtractor):
             'formats': formats,
             'thumbnails': thumbnails,
         }
+
+
+class NYTimesIE(NYTimesBaseIE):
+    _VALID_URL = r'https?://(?:(?:www\.)?nytimes\.com/video/(?:[^/]+/)+?|graphics8\.nytimes\.com/bcvideo/\d+(?:\.\d+)?/iframe/embed\.html\?videoId=)(?P<id>\d+)'
+
+    _TESTS = [{
+        'url': 'http://www.nytimes.com/video/opinion/100000002847155/verbatim-what-is-a-photocopier.html?playlistId=100000001150263',
+        'md5': '18a525a510f942ada2720db5f31644c0',
+        'info_dict': {
+            'id': '100000002847155',
+            'ext': 'mov',
+            'title': 'Verbatim: What Is a Photocopier?',
+            'description': 'md5:93603dada88ddbda9395632fdc5da260',
+            'timestamp': 1398631707,
+            'upload_date': '20140427',
+            'uploader': 'Brett Weiner',
+            'duration': 419,
+        }
+    }, {
+        'url': 'http://www.nytimes.com/video/travel/100000003550828/36-hours-in-dubai.html',
+        'only_matching': True,
+    }]
+
+    def _real_extract(self, url):
+        video_id = self._match_id(url)
+
+        return self._extract_video_from_id(video_id)
+
+
+class NYTimesArticleIE(NYTimesBaseIE):
+    _VALID_URL = r'https?://(?:www\.)?nytimes\.com/(.(?<!video))*?/(?:[^/]+/)*(?P<id>[^.]+)(?:\.html)?'
+    _TESTS = [{
+        'url': 'http://www.nytimes.com/2015/04/14/business/owner-of-gravity-payments-a-credit-card-processor-is-setting-a-new-minimum-wage-70000-a-year.html?_r=0',
+        'md5': 'e2076d58b4da18e6a001d53fd56db3c9',
+        'info_dict': {
+            'id': '100000003628438',
+            'ext': 'mov',
+            'title': 'New Minimum Wage: $70,000 a Year',
+            'description': 'Dan Price, C.E.O. of Gravity Payments, surprised his 120-person staff by announcing that he planned over the next three years to raise the salary of every employee to $70,000 a year.',
+            'timestamp': 1429033037,
+            'upload_date': '20150414',
+            'uploader': 'Matthew Williams',
+        }
+    }, {
+        'url': 'http://www.nytimes.com/news/minute/2014/03/17/times-minute-whats-next-in-crimea/?_php=true&_type=blogs&_php=true&_type=blogs&_r=1',
+        'only_matching': True,
+    }]
+
+    def _real_extract(self, url):
+        video_id = self._match_id(url)
+
+        webpage = self._download_webpage(url, video_id)
+
+        video_id = self._html_search_regex(r'data-videoid="(\d+)"', webpage, 'video id')
+
+        return self._extract_video_from_id(video_id)