diff options
| -rw-r--r-- | youtube/comments.py | 69 | ||||
| -rw-r--r-- | youtube/proto.py | 85 | ||||
| -rw-r--r-- | youtube/templates/comments.html | 20 | ||||
| -rw-r--r-- | youtube/yt_data_extract/everything_else.py | 6 |
4 files changed, 129 insertions, 51 deletions
diff --git a/youtube/comments.py b/youtube/comments.py index 8ab2b2c..66b5353 100644 --- a/youtube/comments.py +++ b/youtube/comments.py @@ -33,8 +33,8 @@ def make_comment_ctoken(video_id, sort=0, offset=0, lc='', secret_key=''): video_id = proto.as_bytes(video_id) secret_key = proto.as_bytes(secret_key) - page_info = proto.string(4, video_id) + proto.uint(6, sort) + page_info = proto.string(4,video_id) + proto.uint(6, sort) offset_information = proto.nested(4, page_info) + proto.uint(5, offset) if secret_key: offset_information = proto.string(1, secret_key) + offset_information @@ -47,15 +47,6 @@ def make_comment_ctoken(video_id, sort=0, offset=0, lc='', secret_key=''): return base64.urlsafe_b64encode(result).decode('ascii') -def comment_replies_ctoken(video_id, comment_id, max_results=500): - - params = proto.string(2, comment_id) + proto.uint(9, max_results) - params = proto.nested(3, params) - - result = proto.nested(2, proto.string(2, video_id)) + proto.uint(3, 6) + proto.nested(6, params) - return base64.urlsafe_b64encode(result).decode('ascii') - - mobile_headers = { 'User-Agent': 'Mozilla/5.0 (iPhone; CPU iPhone OS 10_3_1 like Mac OS X) AppleWebKit/603.1.30 (KHTML, like Gecko) Version/10.0 Mobile/14E304 Safari/602.1', 'Accept': '*/*', @@ -66,10 +57,11 @@ mobile_headers = { def request_comments(ctoken, replies=False): - if replies: # let's make it use different urls for no reason despite all the data being encoded - base_url = "https://m.youtube.com/watch_comment?action_get_comment_replies=1&ctoken=" + base_url = 'https://m.youtube.com/watch_comment?' + if replies: + base_url += 'action_get_comment_replies=1&ctoken=' else: - base_url = "https://m.youtube.com/watch_comment?action_get_comments=1&ctoken=" + base_url += 'action_get_comments=1&ctoken=' url = base_url + ctoken.replace("=", "%3D") + "&pbj=1" content = util.fetch_url( @@ -99,17 +91,24 @@ def post_process_comments_info(comments_info): comment['permalink'] = concat_or_none( util.URL_ORIGIN, '/watch?v=', - comments_info['video_id'], '&lc=', comment['id']) + comments_info['video_id'], + '&lc=', comment['id'] + ) reply_count = comment['reply_count'] - - if reply_count == 0: - comment['replies_url'] = None - else: - comment['replies_url'] = concat_or_none( - util.URL_ORIGIN, - '/comments?parent_id=', comment['id'], - '&video_id=', comments_info['video_id']) + comment['replies_url'] = None + if comment['reply_ctoken']: + # change max_replies field to 250 in ctoken + ctoken = comment['reply_ctoken'] + ctoken, err = proto.set_protobuf_value( + ctoken, + 'base64p', 6, 3, 9, value=250) + if err: + print('Error setting ctoken value:') + print(err) + comment['replies_url'] = None + comment['replies_url'] = concat_or_none(util.URL_ORIGIN, + '/comments?replies=1&ctoken=' + ctoken) if reply_count == 0: comment['view_replies_text'] = 'Reply' @@ -118,6 +117,7 @@ def post_process_comments_info(comments_info): else: comment['view_replies_text'] = str(reply_count) + ' replies' + if comment['like_count'] == 1: comment['likes_text'] = '1 like' else: @@ -125,10 +125,12 @@ def post_process_comments_info(comments_info): comments_info['include_avatars'] = settings.enable_comment_avatars if comments_info['ctoken']: + replies_param = '&replies=1' if comments_info['is_replies'] else '' comments_info['more_comments_url'] = concat_or_none( util.URL_ORIGIN, '/comments?ctoken=', - comments_info['ctoken'] + comments_info['ctoken'], + replies_param ) comments_info['page_number'] = page_number = str(int(comments_info['offset']/20) + 1) @@ -137,14 +139,11 @@ def post_process_comments_info(comments_info): comments_info['sort_text'] = 'top' if comments_info['sort'] == 0 else 'newest' comments_info['video_url'] = concat_or_none( - util.URL_ORIGIN, - '/watch?v=', - comments_info['video_id'] - ) - + util.URL_ORIGIN, '/watch?v=', comments_info['video_id']) comments_info['video_thumbnail'] = concat_or_none( settings.img_prefix, 'https://i.ytimg.com/vi/', - comments_info['video_id'], '/mqdefault.jpg') + comments_info['video_id'], '/mqdefault.jpg' + ) def video_comments(video_id, sort=0, offset=0, lc='', secret_key=''): @@ -198,17 +197,9 @@ def video_comments(video_id, sort=0, offset=0, lc='', secret_key=''): @yt_app.route('/comments') def get_comments_page(): ctoken = request.args.get('ctoken', '') - replies = False - if not ctoken: - video_id = request.args['video_id'] - parent_id = request.args['parent_id'] - - ctoken = comment_replies_ctoken(video_id, parent_id) - replies = True - - comments_info = yt_data_extract.extract_comments_info( - request_comments(ctoken, replies)) + replies = request.args.get('replies', '0') == '1' + comments_info = yt_data_extract.extract_comments_info(request_comments(ctoken, replies)) post_process_comments_info(comments_info) if not replies: diff --git a/youtube/proto.py b/youtube/proto.py index ff59eac..933ac92 100644 --- a/youtube/proto.py +++ b/youtube/proto.py @@ -1,6 +1,7 @@ from math import ceil import base64 import io +import traceback def byte(n): @@ -92,7 +93,6 @@ def read_group(data, end_sequence): data.seek(index + len(end_sequence)) return data.original[start:index] - def read_protobuf(data): data_original = data data = io.BytesIO(data) @@ -122,12 +122,89 @@ def read_protobuf(data): yield (wire_type, field_number, value) -def parse(data): - return {field_number: value for _, field_number, value in read_protobuf(data)} +def parse(data, include_wire_type=False): + '''Returns a dict mapping field numbers to values + + data is the protobuf structure, which must not be b64-encoded''' + if include_wire_type: + return {field_number: [wire_type, value] + for wire_type, field_number, value in read_protobuf(data)} + return {field_number: value + for _, field_number, value in read_protobuf(data)} + + +base64_enc_funcs = { + 'base64': base64.urlsafe_b64encode, + 'base64s': unpadded_b64encode, + 'base64p': percent_b64encode, +} + + +def _make_protobuf(data): + # must be dict mapping field_number to [wire_type, value] + if isinstance(data, dict): + new_data = [] + for field_num, (wire_type, value) in sorted(data.items()): + new_data.append((wire_type, field_num, value)) + data = new_data + if isinstance(data, str): + return data.encode('utf-8') + elif len(data) == 2 and data[0] in base64_enc_funcs: + return base64_enc_funcs[data[0]](make_proto(data[1])) + elif isinstance(data, list): + result = b'' + for field in data: + if field[0] == 0: + result += uint(field[1], field[2]) + elif field[0] == 2: + result += string(field[1], _make_protobuf(field[2])) + else: + raise NotImplementedError('Wire type ' + str(field[0]) + + ' not implemented') + return result + return data + + +def make_protobuf(data): + return _make_protobuf(data).decode('ascii') + + +def _set_protobuf_value(data, *path, value): + if not path: + return value + op = path[0] + if op in base64_enc_funcs: + inner_data = b64_to_bytes(data) + return base64_enc_funcs[op]( + _set_protobuf_value(inner_data, *path[1:], value=value) + ) + pb_dict = parse(data, include_wire_type=True) + pb_dict[op][1] = _set_protobuf_value( + pb_dict[op][1], *path[1:], value=value + ) + return _make_protobuf(pb_dict) + + +def set_protobuf_value(data, *path, value): + '''Set a field's value in a raw protobuf structure + + path is a list of field numbers and/or base64 encoding directives + + The directives are + base64: normal base64 encoding with equal signs padding + base64s ("stripped"): no padding + base64p: %3D instead of = for padding + + return new_protobuf, err''' + try: + new_protobuf = _set_protobuf_value(data, *path, value=value) + return new_protobuf.decode('ascii'), None + except Exception: + return None, traceback.format_exc() def b64_to_bytes(data): if isinstance(data, bytes): data = data.decode('ascii') data = data.replace("%3D", "=") - return base64.urlsafe_b64decode(data + "="*((4 - len(data)%4)%4)) + return base64.urlsafe_b64decode(data + "="*((4 - len(data) % 4) % 4)) diff --git a/youtube/templates/comments.html b/youtube/templates/comments.html index ceb31b8..808f98a 100644 --- a/youtube/templates/comments.html +++ b/youtube/templates/comments.html @@ -23,14 +23,18 @@ <span class="comment-likes">{{ comment['likes_text'] if comment['like_count'] else ''}}</span> <div class="button-row"> - {% if settings.use_comments_js and comment['reply_count'] %} - <details class="replies" data-src="{{ comment['replies_url'] }}"> - <summary>{{ comment['view_replies_text'] }}</summary> - <a href="{{ comment['replies_url'] }}" class="replies-open-new-tab" target="_blank">Open in new tab</a> - <div class="comment_page">loading..</div> - </details> - {% elif comment['reply_count'] %} - <a href="{{ comment['replies_url'] }}" class="replies">{{ comment['view_replies_text'] }}</a> + {% if comment['reply_count'] %} + {% if settings.use_comments_js and comment['replies_url'] %} + <details class="replies" src="{{ comment['replies_url'] }}"> + <summary>{{ comment['view_replies_text'] }}</summary> + <a href="{{ comment['replies_url'] }}" class="replies-open-new-tab" target="_blank">Open in new tab</a> + <div class="comment_page">loading...</div> + </details> + {% elif comment['replies_url'] %} + <a href="{{ comment['replies_url'] }}" class="replies">{{ comment['view_replies_text'] }}</a> + {% else %} + <a class="replies">{{ comment['view_replies_text'] }} (error constructing url)</a> + {% endif %} {% endif %} </div> </div> diff --git a/youtube/yt_data_extract/everything_else.py b/youtube/yt_data_extract/everything_else.py index ae8715f..197cf88 100644 --- a/youtube/yt_data_extract/everything_else.py +++ b/youtube/yt_data_extract/everything_else.py @@ -251,13 +251,19 @@ def extract_comments_info(polymer_json): info['video_title'] = extract_str(comment_thread.get('commentTargetTitle')) if 'replies' not in comment_thread: comment_info['reply_count'] = 0 + comment_info['reply_ctoken'] = None else: comment_info['reply_count'] = extract_int(deep_get(comment_thread, 'replies', 'commentRepliesRenderer', 'moreText' ), default=1) # With 1 reply, the text reads "View reply" + comment_info['reply_ctoken'] = deep_get(comment_thread, + 'replies', 'commentRepliesRenderer', 'continuations', 0, + 'nextContinuationData', 'continuation' + ) comment_renderer = deep_get(comment_thread, 'comment', 'commentRenderer', default={}) elif 'commentRenderer' in comment: # replies comment_info['reply_count'] = 0 # replyCount, below, not present for replies even if the reply has further replies to it + comment_info['reply_ctoken'] = None conservative_update(info, 'is_replies', True) comment_renderer = comment['commentRenderer'] else: |