Merge ffd4b5970d9796526dd17f505f9adbb3fec4d829 into 3eb8d22ddb8982ca4fb56bb7a8d6517538bf14c6

* thx seproDev, yt-dlp/yt-dlp/pull/12694/commits/ae69e3c
* also add documenting comment in `InfoExtractor`

youtube_dl/downloader/common.py

@@ -57,6 +57,7 @@ class FileDownloader(object):
 
     _TEST_FILE_SIZE = 10241
     params = None
+    progress_update = time.time()
 
     def __init__(self, ydl, params):
         """Create a FileDownloader object with the given options."""
@@ -120,8 +121,8 @@ class FileDownloader(object):
     @staticmethod
     def format_speed(speed):
         if speed is None:
-            return '%10s' % '---b/s'
-        return '%10s' % ('%s/s' % format_bytes(speed))
+            return '%12s' % '--- B/s'
+        return '%12s' % ('%s/s' % format_bytes(speed))
 
     @staticmethod
     def format_retries(retries):
@@ -276,32 +277,33 @@ class FileDownloader(object):
         if s['status'] != 'downloading':
             return
 
+        # Show non-final progress updates no more often than once per second
+        tstamp = time.time()
+        if tstamp - self.progress_update < 1:
+            return
+        self.progress_update = tstamp
+
         if s.get('eta') is not None:
             s['_eta_str'] = self.format_eta(s['eta'])
         else:
-            s['_eta_str'] = 'Unknown ETA'
+            s['_eta_str'] = 'unknown'
 
         if s.get('total_bytes') and s.get('downloaded_bytes') is not None:
             s['_percent_str'] = self.format_percent(100 * s['downloaded_bytes'] / s['total_bytes'])
         elif s.get('total_bytes_estimate') and s.get('downloaded_bytes') is not None:
             s['_percent_str'] = self.format_percent(100 * s['downloaded_bytes'] / s['total_bytes_estimate'])
         else:
-            if s.get('downloaded_bytes') == 0:
-                s['_percent_str'] = self.format_percent(0)
-            else:
-                s['_percent_str'] = 'Unknown %'
+            s['_percent_str'] = self.format_percent(0 if s.get('downloaded_bytes') == 0 else None)
 
-        if s.get('speed') is not None:
-            s['_speed_str'] = self.format_speed(s['speed'])
-        else:
-            s['_speed_str'] = 'Unknown speed'
+        # Missing 'speed' is already handled properly in self.format_speed()
+        s['_speed_str'] = self.format_speed(s.get('speed'))
 
         if s.get('total_bytes') is not None:
             s['_total_bytes_str'] = format_bytes(s['total_bytes'])
-            msg_template = '%(_percent_str)s of %(_total_bytes_str)s at %(_speed_str)s ETA %(_eta_str)s'
+            msg_template = '%(_percent_str)s of %(_total_bytes_str)s at %(_speed_str)s, ETA %(_eta_str)s'
         elif s.get('total_bytes_estimate') is not None:
             s['_total_bytes_estimate_str'] = format_bytes(s['total_bytes_estimate'])
-            msg_template = '%(_percent_str)s of ~%(_total_bytes_estimate_str)s at %(_speed_str)s ETA %(_eta_str)s'
+            msg_template = '%(_percent_str)s of ~%(_total_bytes_estimate_str)s at %(_speed_str)s, ETA %(_eta_str)s'
         else:
             if s.get('downloaded_bytes') is not None:
                 s['_downloaded_bytes_str'] = format_bytes(s['downloaded_bytes'])
@@ -311,7 +313,7 @@ class FileDownloader(object):
                 else:
                     msg_template = '%(_downloaded_bytes_str)s at %(_speed_str)s'
             else:
-                msg_template = '%(_percent_str)s % at %(_speed_str)s ETA %(_eta_str)s'
+                msg_template = '%(_percent_str)s % at %(_speed_str)s, ETA %(_eta_str)s'
 
         self._report_progress_status(msg_template % s)
 

youtube_dl/extractor/bokecc.py

@@ -32,7 +32,7 @@ class BokeCCBaseIE(InfoExtractor):
 
 
 class BokeCCIE(BokeCCBaseIE):
-    _IE_DESC = 'CC视频'
+    IE_DESC = 'CC视频'
     _VALID_URL = r'https?://union\.bokecc\.com/playvideo\.bo\?(?P<query>.*)'
 
     _TESTS = [{

youtube_dl/extractor/cloudy.py

@@ -9,7 +9,7 @@ from ..utils import (
 
 
 class CloudyIE(InfoExtractor):
-    _IE_DESC = 'cloudy.ec'
+    IE_DESC = 'cloudy.ec'
     _VALID_URL = r'https?://(?:www\.)?cloudy\.ec/(?:v/|embed\.php\?.*?\bid=)(?P<id>[A-Za-z0-9]+)'
     _TESTS = [{
         'url': 'https://www.cloudy.ec/v/af511e2527aac',

youtube_dl/extractor/common.py

@@ -422,6 +422,8 @@ class InfoExtractor(object):
     _GEO_COUNTRIES = None
     _GEO_IP_BLOCKS = None
     _WORKING = True
+    # supply this in public subclasses: used in supported sites list, etc
+    # IE_DESC = 'short description of IE'
 
     def __init__(self, downloader=None):
         """Constructor. Receives an optional downloader."""

youtube_dl/extractor/itv.py

@@ -35,15 +35,6 @@ from ..utils import (
 
 class ITVBaseIE(InfoExtractor):
 
-    def _search_nextjs_data(self, webpage, video_id, **kw):
-        transform_source = kw.pop('transform_source', None)
-        fatal = kw.pop('fatal', True)
-        return self._parse_json(
-            self._search_regex(
-                r'''<script\b[^>]+\bid=('|")__NEXT_DATA__\1[^>]*>(?P<js>[^<]+)</script>''',
-                webpage, 'next.js data', group='js', fatal=fatal, **kw),
-            video_id, transform_source=transform_source, fatal=fatal)
-
     def __handle_request_webpage_error(self, err, video_id=None, errnote=None, fatal=True):
         if errnote is False:
             return False
@@ -109,7 +100,9 @@ class ITVBaseIE(InfoExtractor):
 
 class ITVIE(ITVBaseIE):
     _VALID_URL = r'https?://(?:www\.)?itv\.com/(?:(?P<w>watch)|hub)/[^/]+/(?(w)[\w-]+/)(?P<id>\w+)'
-    _IE_DESC = 'ITVX'
+    IE_DESC = 'ITVX'
+    _WORKING = False
+
     _TESTS = [{
         'note': 'Hub URLs redirect to ITVX',
         'url': 'https://www.itv.com/hub/liar/2a4547a0012',
@@ -270,7 +263,7 @@ class ITVIE(ITVBaseIE):
                 'ext': determine_ext(href, 'vtt'),
             })
 
-        next_data = self._search_nextjs_data(webpage, video_id, fatal=False, default='{}')
+        next_data = self._search_nextjs_data(webpage, video_id, fatal=False, default={})
         video_data.update(traverse_obj(next_data, ('props', 'pageProps', ('title', 'episode')), expected_type=dict)[0] or {})
         title = traverse_obj(video_data, 'headerTitle', 'episodeTitle')
         info = self._og_extract(webpage, require_title=not title)
@@ -323,7 +316,7 @@ class ITVIE(ITVBaseIE):
 
 class ITVBTCCIE(ITVBaseIE):
     _VALID_URL = r'https?://(?:www\.)?itv\.com/(?!(?:watch|hub)/)(?:[^/]+/)+(?P<id>[^/?#&]+)'
-    _IE_DESC = 'ITV articles: News, British Touring Car Championship'
+    IE_DESC = 'ITV articles: News, British Touring Car Championship'
     _TESTS = [{
         'note': 'British Touring Car Championship',
         'url': 'https://www.itv.com/btcc/articles/btcc-2018-all-the-action-from-brands-hatch',

youtube_dl/extractor/senateisvp.py

@@ -47,7 +47,7 @@ class SenateISVPIE(InfoExtractor):
         ['vetaff', '76462', 'http://vetaff-f.akamaihd.net'],
         ['arch', '', 'http://ussenate-f.akamaihd.net/']
     ]
-    _IE_NAME = 'senate.gov'
+    IE_NAME = 'senate.gov'
     _VALID_URL = r'https?://(?:www\.)?senate\.gov/isvp/?\?(?P<qs>.+)'
     _TESTS = [{
         'url': 'http://www.senate.gov/isvp/?comm=judiciary&type=live&stt=&filename=judiciary031715&auto_play=false&wmode=transparent&poster=http%3A%2F%2Fwww.judiciary.senate.gov%2Fthemes%2Fjudiciary%2Fimages%2Fvideo-poster-flash-fit.png',

youtube_dl/jsinterp.py

@@ -686,6 +686,8 @@ class JSInterpreter(object):
             raise self.Exception('Cannot get index {idx!r:.100}'.format(**locals()), expr=repr(obj), cause=e)
 
     def _dump(self, obj, namespace):
+        if obj is JS_Undefined:
+            return 'undefined'
         try:
             return json.dumps(obj)
         except TypeError:

youtube_dl/utils.py

@@ -3586,10 +3586,12 @@ def format_bytes(bytes):
     if bytes == 0.0:
         exponent = 0
     else:
-        exponent = int(math.log(bytes, 1024.0))
+        # Display user-friendly values, e.g. "1001.45 KiB" is much less
+        # user-friendly than "1.00 MiB"
+        exponent = int(math.log(bytes, 1000.0))
     suffix = ['B', 'KiB', 'MiB', 'GiB', 'TiB', 'PiB', 'EiB', 'ZiB', 'YiB'][exponent]
     converted = float(bytes) / float(1024 ** exponent)
-    return '%.2f%s' % (converted, suffix)
+    return '%.2f %s' % (converted, suffix)
 
 
 def lookup_unit_table(unit_table, s):