tubearchivist/tubearchivist-frontend
1
0
Fork 0
mirror of https://github.com/tubearchivist/tubearchivist-frontend.git synced 2024-11-22 11:50:14 +00:00
Code Issues Projects Releases Wiki Activity

fix autocaption extraction, flatten words, #180

Browse Source
This commit is contained in:
simon 2022-03-10 19:58:13 +07:00
parent 6a6c8fa5d8
commit 1ce832b846
No known key found for this signature in database
GPG Key ID: 2C15AA5E89985DD4
1 changed files with 25 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

28
tubearchivist/home/src/index/video.py
View File

@ -114,12 +114,13 @@ class YoutubeSubtitle:
for subtitle in relevant_subtitles: for subtitle in relevant_subtitles:
dest_path = os.path.join(videos_base, subtitle["media_url"]) dest_path = os.path.join(videos_base, subtitle["media_url"])
source = subtitle["source"] source = subtitle["source"]
lang = subtitle.get("lang")
response = requests.get(subtitle["url"]) response = requests.get(subtitle["url"])
if not response.ok: if not response.ok:
print(f"{self.video.youtube_id}: failed to download subtitle") print(f"{self.video.youtube_id}: failed to download subtitle")
continue continue
parser = SubtitleParser(response.text, subtitle.get("lang")) parser = SubtitleParser(response.text, lang, source)
parser.process() parser.process()
subtitle_str = parser.get_subtitle_str() subtitle_str = parser.get_subtitle_str()
self._write_subtitle_file(dest_path, subtitle_str) self._write_subtitle_file(dest_path, subtitle_str)
@ -144,15 +145,20 @@ class YoutubeSubtitle:
class SubtitleParser: class SubtitleParser:
"""parse subtitle str from youtube""" """parse subtitle str from youtube"""
def __init__(self, subtitle_str, lang): def __init__(self, subtitle_str, lang, source):
self.subtitle_raw = json.loads(subtitle_str) self.subtitle_raw = json.loads(subtitle_str)
self.lang = lang self.lang = lang
self.source = source
self.all_cues = False self.all_cues = False
def process(self): def process(self):
"""extract relevant que data""" """extract relevant que data"""
all_events = self.subtitle_raw.get("events")
if self.source == "auto":
all_events = self._flat_auto_caption(all_events)
self.all_cues = [] self.all_cues = []
for idx, event in enumerate(self.subtitle_raw.get("events")): for idx, event in enumerate(all_events):
cue = { cue = {
"start": self.ms_conv(event["tStartMs"]), "start": self.ms_conv(event["tStartMs"]),
"end": self.ms_conv(event["tStartMs"] + event["dDurationMs"]), "end": self.ms_conv(event["tStartMs"] + event["dDurationMs"]),
@ -161,6 +167,22 @@ class SubtitleParser:
} }
self.all_cues.append(cue) self.all_cues.append(cue)
@staticmethod
def _flat_auto_caption(all_events):
"""flatten autocaption segments"""
flatten = []
for event in all_events:
if "segs" not in event.keys():
continue
text = "".join([i.get("utf8") for i in event.get("segs")])
if not text.strip():
continue
event.update({"segs": [{"utf8": text}]})
flatten.append(event)
return flatten
@staticmethod @staticmethod
def ms_conv(ms): def ms_conv(ms):
"""convert ms to timestamp""" """convert ms to timestamp"""