[YoutubeDL] Use a progress hook for progress reporting

Instead of every downloader calling two helper functions, let our progress report be an ordinary progress hook like everyone else's.
Closes #4875.

youtube_dl/YoutubeDL.py

@@ -199,18 +199,25 @@ class YoutubeDL(object):
                        postprocessor.
     progress_hooks:    A list of functions that get called on download
                        progress, with a dictionary with the entries
-                       * status: One of "downloading" and "finished".
+                       * status: One of "downloading", "error", or "finished".
                                  Check this first and ignore unknown values.
 
-                       If status is one of "downloading" or "finished", the
+                       If status is one of "downloading", or "finished", the
                        following properties may also be present:
                        * filename: The final filename (always present)
+                       * tmpfilename: The filename we're currently writing to
                        * downloaded_bytes: Bytes on disk
                        * total_bytes: Size of the whole file, None if unknown
-                       * tmpfilename: The filename we're currently writing to
+                       * total_bytes_estimate: Guess of the eventual file size,
+                                               None if unavailable.
+                       * elapsed: The number of seconds since download started.
                        * eta: The estimated time in seconds, None if unknown
                        * speed: The download speed in bytes/second, None if
                                 unknown
+                       * fragment_index: The counter of the currently
+                                         downloaded video fragment.
+                       * fragment_count: The number of fragments (= individual
+                                         files that will be merged)
 
                        Progress hooks are guaranteed to be called at least once
                        (with status "finished") if the download is successful.

youtube_dl/downloader/common.py

@@ -1,4 +1,4 @@
-from __future__ import unicode_literals
+from __future__ import division, unicode_literals
 
 import os
 import re
@@ -54,6 +54,7 @@ def __init__(self, ydl, params):
         self.ydl = ydl
         self._progress_hooks = []
         self.params = params
+        self.add_progress_hook(self.report_progress)
 
     @staticmethod
     def format_seconds(seconds):
@@ -226,42 +227,64 @@ def _report_progress_status(self, msg, is_last_line=False):
             self.to_screen(clear_line + fullmsg, skip_eol=not is_last_line)
         self.to_console_title('youtube-dl ' + msg)
 
-    def report_progress(self, percent, data_len_str, speed, eta):
+    def report_progress(self, s):
-        """Report download progress."""
+        if s['status'] == 'finished':
-        if self.params.get('noprogress', False):
-            return
-        if eta is not None:
-            eta_str = self.format_eta(eta)
-        else:
-            eta_str = 'Unknown ETA'
-        if percent is not None:
-            percent_str = self.format_percent(percent)
-        else:
-            percent_str = 'Unknown %'
-        speed_str = self.format_speed(speed)
-
-        msg = ('%s of %s at %s ETA %s' %
-               (percent_str, data_len_str, speed_str, eta_str))
-        self._report_progress_status(msg)
-
-    def report_progress_live_stream(self, downloaded_data_len, speed, elapsed):
-        if self.params.get('noprogress', False):
-            return
-        downloaded_str = format_bytes(downloaded_data_len)
-        speed_str = self.format_speed(speed)
-        elapsed_str = FileDownloader.format_seconds(elapsed)
-        msg = '%s at %s (%s)' % (downloaded_str, speed_str, elapsed_str)
-        self._report_progress_status(msg)
-
-    def report_finish(self, data_len_str, tot_time):
-        """Report download finished."""
             if self.params.get('noprogress', False):
                 self.to_screen('[download] Download completed')
             else:
+                s['_total_bytes_str'] = format_bytes(s['total_bytes'])
+                if s.get('elapsed') is not None:
+                    s['_elapsed_str'] = self.format_seconds(s['elapsed'])
+                    msg_template = '100%% of %(_total_bytes_str)s in %(_elapsed_str)s'
+                else:
+                    msg_template = '100%% of %(_total_bytes_str)s'
                 self._report_progress_status(
-                ('100%% of %s in %s' %
+                    msg_template % s, is_last_line=True)
-                 (data_len_str, self.format_seconds(tot_time))),
+
-                is_last_line=True)
+        if self.params.get('noprogress'):
+            return
+
+        if s['status'] != 'downloading':
+            return
+
+        if s.get('eta') is not None:
+            s['_eta_str'] = self.format_eta(s['eta'])
+        else:
+            s['_eta_str'] = 'Unknown ETA'
+
+        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 %'
+
+        if s.get('speed') is not None:
+            s['_speed_str'] = self.format_speed(s['speed'])
+        else:
+            s['_speed_str'] = 'Unknown 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'
+        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'
+        else:
+            if s.get('downloaded_bytes') is not None:
+                s['_downloaded_bytes_str'] = format_bytes(s['downloaded_bytes'])
+                if s.get('elapsed'):
+                    s['_elapsed_str'] = self.format_seconds(s['elapsed'])
+                    msg_template = '%(_downloaded_bytes_str)s at %(_speed_str)s (%(_elapsed_str)s)'
+                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'
+
+        self._report_progress_status(msg_template % s)
 
     def report_resuming_byte(self, resume_len):
         """Report attempt to resume at given byte."""

youtube_dl/downloader/f4m.py

@@ -1,4 +1,4 @@
-from __future__ import unicode_literals
+from __future__ import division, unicode_literals
 
 import base64
 import io
@@ -252,17 +252,6 @@ def real_download(self, filename, info_dict):
         requested_bitrate = info_dict.get('tbr')
         self.to_screen('[download] Downloading f4m manifest')
         manifest = self.ydl.urlopen(man_url).read()
-        self.report_destination(filename)
-        http_dl = HttpQuietDownloader(
-            self.ydl,
-            {
-                'continuedl': True,
-                'quiet': True,
-                'noprogress': True,
-                'ratelimit': self.params.get('ratelimit', None),
-                'test': self.params.get('test', False),
-            }
-        )
 
         doc = etree.fromstring(manifest)
         formats = [(int(f.attrib.get('bitrate', -1)), f)
@@ -298,39 +287,67 @@ def real_download(self, filename, info_dict):
         # For some akamai manifests we'll need to add a query to the fragment url
         akamai_pv = xpath_text(doc, _add_ns('pv-2.0'))
 
+        self.report_destination(filename)
+        http_dl = HttpQuietDownloader(
+            self.ydl,
+            {
+                'continuedl': True,
+                'quiet': True,
+                'noprogress': True,
+                'ratelimit': self.params.get('ratelimit', None),
+                'test': self.params.get('test', False),
+            }
+        )
         tmpfilename = self.temp_name(filename)
         (dest_stream, tmpfilename) = sanitize_open(tmpfilename, 'wb')
+
         write_flv_header(dest_stream)
         write_metadata_tag(dest_stream, metadata)
 
         # This dict stores the download progress, it's updated by the progress
         # hook
         state = {
+            'status': 'downloading',
             'downloaded_bytes': 0,
-            'frag_counter': 0,
+            'frag_index': 0,
+            'frag_count': total_frags,
+            'filename': filename,
+            'tmpfilename': tmpfilename,
         }
         start = time.time()
 
-        def frag_progress_hook(status):
+        def frag_progress_hook(s):
-            frag_total_bytes = status.get('total_bytes', 0)
+            if s['status'] not in ('downloading', 'finished'):
-            estimated_size = (state['downloaded_bytes'] +
+                return
-                              (total_frags - state['frag_counter']) * frag_total_bytes)
+
-            if status['status'] == 'finished':
+            frag_total_bytes = s.get('total_bytes', 0)
+            if s['status'] == 'finished':
                 state['downloaded_bytes'] += frag_total_bytes
-                state['frag_counter'] += 1
+                state['frag_index'] += 1
-                progress = self.calc_percent(state['frag_counter'], total_frags)
+
+            estimated_size = (
+                (state['downloaded_bytes'] + frag_total_bytes)
+                / (state['frag_index'] + 1) * total_frags)
+            time_now = time.time()
+            state['total_bytes_estimate'] = estimated_size
+            state['elapsed'] = time_now - start
+
+            if s['status'] == 'finished':
+                progress = self.calc_percent(state['frag_index'], total_frags)
                 byte_counter = state['downloaded_bytes']
             else:
-                frag_downloaded_bytes = status['downloaded_bytes']
+                frag_downloaded_bytes = s['downloaded_bytes']
                 byte_counter = state['downloaded_bytes'] + frag_downloaded_bytes
                 frag_progress = self.calc_percent(frag_downloaded_bytes,
                                                   frag_total_bytes)
-                progress = self.calc_percent(state['frag_counter'], total_frags)
+                progress = self.calc_percent(state['frag_index'], total_frags)
                 progress += frag_progress / float(total_frags)
 
-            eta = self.calc_eta(start, time.time(), estimated_size, byte_counter)
+                state['eta'] = self.calc_eta(
-            self.report_progress(progress, format_bytes(estimated_size),
+                    start, time_now, estimated_size, state['downloaded_bytes'] + frag_downloaded_bytes)
-                                 status.get('speed'), eta)
+                state['speed'] = s.get('speed')
+            self._hook_progress(state)
+
         http_dl.add_progress_hook(frag_progress_hook)
 
         frags_filenames = []
@@ -354,8 +371,8 @@ def frag_progress_hook(status):
             frags_filenames.append(frag_filename)
 
         dest_stream.close()
-        self.report_finish(format_bytes(state['downloaded_bytes']), time.time() - start)
 
+        elapsed = time.time() - start
         self.try_rename(tmpfilename, filename)
         for frag_file in frags_filenames:
             os.remove(frag_file)
@@ -366,6 +383,7 @@ def frag_progress_hook(status):
             'total_bytes': fsize,
             'filename': filename,
             'status': 'finished',
+            'elapsed': elapsed,
         })
 
         return True

youtube_dl/downloader/http.py

@@ -200,16 +200,16 @@ def real_download(self, filename, info_dict):
             else:
                 percent = self.calc_percent(byte_counter, data_len)
                 eta = self.calc_eta(start, time.time(), data_len - resume_len, byte_counter - resume_len)
-            self.report_progress(percent, data_len_str, speed, eta)
 
             self._hook_progress({
+                'status': 'downloading',
                 'downloaded_bytes': byte_counter,
                 'total_bytes': data_len,
                 'tmpfilename': tmpfilename,
                 'filename': filename,
-                'status': 'downloading',
                 'eta': eta,
                 'speed': speed,
+                'elapsed': now - start,
             })
 
             if is_test and byte_counter == data_len:
@@ -221,7 +221,13 @@ def real_download(self, filename, info_dict):
             return False
         if tmpfilename != '-':
             stream.close()
-        self.report_finish(data_len_str, (time.time() - start))
+
+        self._hook_progress({
+            'downloaded_bytes': byte_counter,
+            'total_bytes': data_len,
+            'tmpfilename': tmpfilename,
+            'status': 'error',
+        })
         if data_len is not None and byte_counter != data_len:
             raise ContentTooShortError(byte_counter, int(data_len))
         self.try_rename(tmpfilename, filename)
@@ -235,6 +241,7 @@ def real_download(self, filename, info_dict):
             'total_bytes': byte_counter,
             'filename': filename,
             'status': 'finished',
+            'elapsed': time.time() - start,
         })
 
         return True

youtube_dl/downloader/rtmp.py

@@ -51,23 +51,23 @@ def run_rtmpdump(args):
                     if not resume_percent:
                         resume_percent = percent
                         resume_downloaded_data_len = downloaded_data_len
-                    eta = self.calc_eta(start, time.time(), 100 - resume_percent, percent - resume_percent)
+                    time_now = time.time()
-                    speed = self.calc_speed(start, time.time(), downloaded_data_len - resume_downloaded_data_len)
+                    eta = self.calc_eta(start, time_now, 100 - resume_percent, percent - resume_percent)
+                    speed = self.calc_speed(start, time_now, downloaded_data_len - resume_downloaded_data_len)
                     data_len = None
                     if percent > 0:
                         data_len = int(downloaded_data_len * 100 / percent)
-                    data_len_str = '~' + format_bytes(data_len)
-                    self.report_progress(percent, data_len_str, speed, eta)
-                    cursor_in_new_line = False
                     self._hook_progress({
+                        'status': 'downloading',
                         'downloaded_bytes': downloaded_data_len,
-                        'total_bytes': data_len,
+                        'total_bytes_estimate': data_len,
                         'tmpfilename': tmpfilename,
                         'filename': filename,
-                        'status': 'downloading',
                         'eta': eta,
+                        'elapsed': time_now - start,
                         'speed': speed,
                     })
+                    cursor_in_new_line = False
                 else:
                     # no percent for live streams
                     mobj = re.search(r'([0-9]+\.[0-9]{3}) kB / [0-9]+\.[0-9]{2} sec', line)
@@ -75,15 +75,15 @@ def run_rtmpdump(args):
                         downloaded_data_len = int(float(mobj.group(1)) * 1024)
                         time_now = time.time()
                         speed = self.calc_speed(start, time_now, downloaded_data_len)
-                        self.report_progress_live_stream(downloaded_data_len, speed, time_now - start)
-                        cursor_in_new_line = False
                         self._hook_progress({
                             'downloaded_bytes': downloaded_data_len,
                             'tmpfilename': tmpfilename,
                             'filename': filename,
                             'status': 'downloading',
+                            'elapsed': time_now - start,
                             'speed': speed,
                         })
+                        cursor_in_new_line = False
                     elif self.params.get('verbose', False):
                         if not cursor_in_new_line:
                             self.to_screen('')