]> jfr.im git - yt-dlp.git/blobdiff - yt_dlp/downloader/fragment.py
Improved progress reporting (See desc) (#1125)
[yt-dlp.git] / yt_dlp / downloader / fragment.py
index cd1e2350dee1541d6fcc20199eabb6893dee013d..6a490131b112ed69e2af35300f8c0fab097c0cce 100644 (file)
@@ -3,6 +3,7 @@
 import os
 import time
 import json
+from math import ceil
 
 try:
     import concurrent.futures
@@ -120,6 +121,7 @@ def _download_fragment(self, ctx, frag_url, info_dict, headers=None, request_dat
             'url': frag_url,
             'http_headers': headers or info_dict.get('http_headers'),
             'request_data': request_data,
+            'ctx_id': ctx.get('ctx_id'),
         }
         success = ctx['dl'].download(fragment_filename, fragment_info_dict)
         if not success:
@@ -219,6 +221,7 @@ def _prepare_frag_download(self, ctx):
     def _start_frag_download(self, ctx, info_dict):
         resume_len = ctx['complete_frags_downloaded_bytes']
         total_frags = ctx['total_frags']
+        ctx_id = ctx.get('ctx_id')
         # This dict stores the download progress, it's updated by the progress
         # hook
         state = {
@@ -242,6 +245,12 @@ def frag_progress_hook(s):
             if s['status'] not in ('downloading', 'finished'):
                 return
 
+            if ctx_id is not None and s.get('ctx_id') != ctx_id:
+                return
+
+            state['max_progress'] = ctx.get('max_progress')
+            state['progress_idx'] = ctx.get('progress_idx')
+
             time_now = time.time()
             state['elapsed'] = time_now - start
             frag_total_bytes = s.get('total_bytes') or 0
@@ -301,6 +310,9 @@ def _finish_frag_download(self, ctx, info_dict):
             'filename': ctx['filename'],
             'status': 'finished',
             'elapsed': elapsed,
+            'ctx_id': ctx.get('ctx_id'),
+            'max_progress': ctx.get('max_progress'),
+            'progress_idx': ctx.get('progress_idx'),
         }, info_dict)
 
     def _prepare_external_frag_download(self, ctx):
@@ -324,7 +336,66 @@ def _prepare_external_frag_download(self, ctx):
             'fragment_index': 0,
         })
 
-    def download_and_append_fragments(self, ctx, fragments, info_dict, *, pack_func=None, finish_func=None):
+    def decrypter(self, info_dict):
+        _key_cache = {}
+
+        def _get_key(url):
+            if url not in _key_cache:
+                _key_cache[url] = self.ydl.urlopen(self._prepare_url(info_dict, url)).read()
+            return _key_cache[url]
+
+        def decrypt_fragment(fragment, frag_content):
+            decrypt_info = fragment.get('decrypt_info')
+            if not decrypt_info or decrypt_info['METHOD'] != 'AES-128':
+                return frag_content
+            iv = decrypt_info.get('IV') or compat_struct_pack('>8xq', fragment['media_sequence'])
+            decrypt_info['KEY'] = decrypt_info.get('KEY') or _get_key(info_dict.get('_decryption_key_url') or decrypt_info['URI'])
+            # Don't decrypt the content in tests since the data is explicitly truncated and it's not to a valid block
+            # size (see https://github.com/ytdl-org/youtube-dl/pull/27660). Tests only care that the correct data downloaded,
+            # not what it decrypts to.
+            if self.params.get('test', False):
+                return frag_content
+            decrypted_data = aes_cbc_decrypt_bytes(frag_content, decrypt_info['KEY'], iv)
+            return decrypted_data[:-decrypted_data[-1]]
+
+        return decrypt_fragment
+
+    def download_and_append_fragments_multiple(self, *args, pack_func=None, finish_func=None):
+        '''
+        @params (ctx1, fragments1, info_dict1), (ctx2, fragments2, info_dict2), ...
+                all args must be either tuple or list
+        '''
+        max_progress = len(args)
+        if max_progress == 1:
+            return self.download_and_append_fragments(*args[0], pack_func=pack_func, finish_func=finish_func)
+        max_workers = self.params.get('concurrent_fragment_downloads', max_progress)
+        self._prepare_multiline_status(max_progress)
+
+        def thread_func(idx, ctx, fragments, info_dict, tpe):
+            ctx['max_progress'] = max_progress
+            ctx['progress_idx'] = idx
+            return self.download_and_append_fragments(ctx, fragments, info_dict, pack_func=pack_func, finish_func=finish_func, tpe=tpe)
+
+        class FTPE(concurrent.futures.ThreadPoolExecutor):
+            # has to stop this or it's going to wait on the worker thread itself
+            def __exit__(self, exc_type, exc_val, exc_tb):
+                pass
+
+        spins = []
+        for idx, (ctx, fragments, info_dict) in enumerate(args):
+            tpe = FTPE(ceil(max_workers / max_progress))
+            job = tpe.submit(thread_func, idx, ctx, fragments, info_dict, tpe)
+            spins.append((tpe, job))
+
+        result = True
+        for tpe, job in spins:
+            try:
+                result = result and job.result()
+            finally:
+                tpe.shutdown(wait=True)
+        return result
+
+    def download_and_append_fragments(self, ctx, fragments, info_dict, *, pack_func=None, finish_func=None, tpe=None):
         fragment_retries = self.params.get('fragment_retries', 0)
         is_fatal = (lambda idx: idx == 0) if self.params.get('skip_unavailable_fragments', True) else (lambda _: True)
         if not pack_func:
@@ -332,7 +403,7 @@ def download_and_append_fragments(self, ctx, fragments, info_dict, *, pack_func=
 
         def download_fragment(fragment, ctx):
             frag_index = ctx['fragment_index'] = fragment['frag_index']
-            headers = info_dict.get('http_headers', {})
+            headers = info_dict.get('http_headers', {}).copy()
             byte_range = fragment.get('byte_range')
             if byte_range:
                 headers['Range'] = 'bytes=%d-%d' % (byte_range['start'], byte_range['end'] - 1)
@@ -369,26 +440,6 @@ def download_fragment(fragment, ctx):
                 return False, frag_index
             return frag_content, frag_index
 
-        _key_cache = {}
-
-        def _get_key(url):
-            if url not in _key_cache:
-                _key_cache[url] = self.ydl.urlopen(self._prepare_url(info_dict, url)).read()
-            return _key_cache[url]
-
-        def decrypt_fragment(fragment, frag_content):
-            decrypt_info = fragment.get('decrypt_info')
-            if not decrypt_info or decrypt_info['METHOD'] != 'AES-128':
-                return frag_content
-            iv = decrypt_info.get('IV') or compat_struct_pack('>8xq', fragment['media_sequence'])
-            decrypt_info['KEY'] = decrypt_info.get('KEY') or _get_key(info_dict.get('_decryption_key_url') or decrypt_info['URI'])
-            # Don't decrypt the content in tests since the data is explicitly truncated and it's not to a valid block
-            # size (see https://github.com/ytdl-org/youtube-dl/pull/27660). Tests only care that the correct data downloaded,
-            # not what it decrypts to.
-            if self.params.get('test', False):
-                return frag_content
-            return aes_cbc_decrypt_bytes(frag_content, decrypt_info['KEY'], iv)
-
         def append_fragment(frag_content, frag_index, ctx):
             if not frag_content:
                 if not is_fatal(frag_index - 1):
@@ -402,6 +453,8 @@ def append_fragment(frag_content, frag_index, ctx):
             self._append_fragment(ctx, pack_func(frag_content, frag_index))
             return True
 
+        decrypt_fragment = self.decrypter(info_dict)
+
         max_workers = self.params.get('concurrent_fragment_downloads', 1)
         if can_threaded_download and max_workers > 1:
 
@@ -411,7 +464,7 @@ def _download_fragment(fragment):
                 return fragment, frag_content, frag_index, ctx_copy.get('fragment_filename_sanitized')
 
             self.report_warning('The download speed shown is only of one thread. This is a known issue and patches are welcome')
-            with concurrent.futures.ThreadPoolExecutor(max_workers) as pool:
+            with tpe or concurrent.futures.ThreadPoolExecutor(max_workers) as pool:
                 for fragment, frag_content, frag_index, frag_filename in pool.map(_download_fragment, fragments):
                     ctx['fragment_filename_sanitized'] = frag_filename
                     ctx['fragment_index'] = frag_index