smshine/mediaplayer
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

add chinese subtitles

Browse Source
This commit is contained in:
Shuming Liu
2026-10-03 12:14:17 +08:00
parent 138d446d73
commit 6bd64ee4a9
22 changed files with 922 additions and 77 deletions
Download Patch File Download Diff File Expand all files Collapse all files

126
sentence_api/repository.py
View File

@@ -56,6 +56,7 @@ class VideoRepository:
end_ms INTEGER NOT NULL,
text TEXT,
language TEXT,
zh_text TEXT,
reference_speech_duration_ms INTEGER,
PRIMARY KEY (video_hash, sentence_index),
FOREIGN KEY (video_hash) REFERENCES videos(video_hash) ON DELETE CASCADE
@@ -193,6 +194,14 @@ class VideoRepository:
if "remote_url" not in video_columns:
connection.execute("ALTER TABLE videos ADD COLUMN remote_url TEXT")
with self._connect() as connection:
sentence_columns = {
row["name"]
for row in connection.execute("PRAGMA table_info(sentences)")
}
if "zh_text" not in sentence_columns:
connection.execute("ALTER TABLE sentences ADD COLUMN zh_text TEXT")
def upsert_upload(
self,
*,
@@ -431,17 +440,7 @@ class VideoRepository:
video_hash=video["video_hash"],
duration_ms=video["duration_ms"],
algorithm_version=video["algorithm_version"] or "unknown",
sentences=[
SentenceBoundary(
index=row["sentence_index"],
start_ms=row["start_ms"],
end_ms=row["end_ms"],
text=row["text"],
language=row["language"],
reference_speech_duration_ms=row["reference_speech_duration_ms"],
)
for row in rows
],
sentences=[_sentence_from_row(row) for row in rows],
)
def update_sentence_text(
@@ -462,6 +461,10 @@ class VideoRepository:
)
if cursor.rowcount == 0:
return None
connection.execute(
"UPDATE sentences SET zh_text = NULL WHERE video_hash = ?",
(normalized_hash,),
)
connection.execute(
"UPDATE videos SET updated_at = ? WHERE video_hash = ?",
(utc_now(), normalized_hash),
@@ -470,14 +473,7 @@ class VideoRepository:
"SELECT * FROM sentences WHERE video_hash = ? AND sentence_index = ?",
(normalized_hash, sentence_index),
).fetchone()
return SentenceBoundary(
index=row["sentence_index"],
start_ms=row["start_ms"],
end_ms=row["end_ms"],
text=row["text"],
language=row["language"],
reference_speech_duration_ms=row["reference_speech_duration_ms"],
)
return _sentence_from_row(row)
def get_sentence(
self,
@@ -495,14 +491,7 @@ class VideoRepository:
).fetchone()
if row is None:
return None
return SentenceBoundary(
index=row["sentence_index"],
start_ms=row["start_ms"],
end_ms=row["end_ms"],
text=row["text"],
language=row["language"],
reference_speech_duration_ms=row["reference_speech_duration_ms"],
)
return _sentence_from_row(row)
def replace_sentence(
self,
@@ -614,6 +603,10 @@ class VideoRepository:
replacement.reference_speech_duration_ms,
),
)
connection.execute(
"UPDATE sentences SET zh_text = NULL WHERE video_hash = ?",
(normalized_hash,),
)
connection.execute(
"UPDATE videos SET updated_at = ? WHERE video_hash = ?",
(utc_now(), normalized_hash),
@@ -627,17 +620,7 @@ class VideoRepository:
(normalized_hash, sentence_index, sentence_index + len(rows)),
).fetchall()
return [
SentenceBoundary(
index=row["sentence_index"],
start_ms=row["start_ms"],
end_ms=row["end_ms"],
text=row["text"],
language=row["language"],
reference_speech_duration_ms=row["reference_speech_duration_ms"],
)
for row in result_rows
]
return [_sentence_from_row(row) for row in result_rows]
def merge_with_next_sentence(
@@ -727,6 +710,10 @@ class VideoRepository:
(old_index - 1, normalized_hash, old_index),
)
connection.execute(
"UPDATE sentences SET zh_text = NULL WHERE video_hash = ?",
(normalized_hash,),
)
connection.execute(
"UPDATE videos SET updated_at = ? WHERE video_hash = ?",
(utc_now(), normalized_hash),
@@ -739,14 +726,7 @@ class VideoRepository:
(normalized_hash, sentence_index),
).fetchone()
return SentenceBoundary(
index=row["sentence_index"],
start_ms=row["start_ms"],
end_ms=row["end_ms"],
text=row["text"],
language=row["language"],
reference_speech_duration_ms=row["reference_speech_duration_ms"],
)
return _sentence_from_row(row)
def adjust_sentence_boundary(
self,
@@ -851,17 +831,36 @@ class VideoRepository:
(normalized_hash, *updated_indexes),
).fetchall()
return [
SentenceBoundary(
index=row["sentence_index"],
start_ms=row["start_ms"],
end_ms=row["end_ms"],
text=row["text"],
language=row["language"],
reference_speech_duration_ms=row["reference_speech_duration_ms"],
return [_sentence_from_row(row) for row in result_rows]
def clear_sentence_translations(self, video_hash: str) -> None:
with self._connect() as connection:
connection.execute(
"UPDATE sentences SET zh_text = NULL WHERE video_hash = ?",
(normalize_video_hash(video_hash),),
)
for row in result_rows
]
def save_sentence_translations(
self,
video_hash: str,
translations: Dict[int, str],
) -> Optional[SentenceBoundaryDocument]:
normalized_hash = normalize_video_hash(video_hash)
with self._connect() as connection:
for index, zh_text in translations.items():
connection.execute(
"""
UPDATE sentences
SET zh_text = ?
WHERE video_hash = ? AND sentence_index = ?
""",
(zh_text.strip(), normalized_hash, index),
)
connection.execute(
"UPDATE videos SET updated_at = ? WHERE video_hash = ?",
(utc_now(), normalized_hash),
)
return self.get_document(normalized_hash)
def create_dub_share(
self,
@@ -1278,6 +1277,21 @@ class VideoRepository:
return video
def _sentence_from_row(row: sqlite3.Row) -> SentenceBoundary:
keys = set(row.keys())
zh_text = row["zh_text"] if "zh_text" in keys else None
return SentenceBoundary(
index=row["sentence_index"],
start_ms=row["start_ms"],
end_ms=row["end_ms"],
text=row["text"],
language=row["language"],
zh_text=zh_text,
reference_speech_duration_ms=row["reference_speech_duration_ms"],
)
def _join_adjacent_sentence_text(
left: str,
right: str,