Spaces:
Sleeping
Sleeping
Commit
·
529a32e
1
Parent(s):
1fa646f
Fix bug early enrich again
Browse files- app/api/transcribe.py +21 -18
- app/jobs/transcribe_job.py +37 -10
app/api/transcribe.py
CHANGED
|
@@ -123,25 +123,28 @@ async def _run_sync_pipeline(tmp_wav: str, note_id: str):
|
|
| 123 |
|
| 124 |
|
| 125 |
async def _create_placeholder_note(note_id: str, duration: float):
|
| 126 |
-
|
| 127 |
-
|
| 128 |
-
|
| 129 |
-
|
| 130 |
-
|
| 131 |
-
|
| 132 |
-
|
| 133 |
-
|
| 134 |
-
|
| 135 |
-
|
| 136 |
-
|
| 137 |
-
|
| 138 |
-
|
| 139 |
-
|
| 140 |
-
|
| 141 |
}
|
| 142 |
-
|
| 143 |
-
|
| 144 |
-
|
|
|
|
|
|
|
|
|
|
| 145 |
|
| 146 |
|
| 147 |
# ============================================================
|
|
|
|
| 123 |
|
| 124 |
|
| 125 |
async def _create_placeholder_note(note_id: str, duration: float):
|
| 126 |
+
client = NoteServiceClient()
|
| 127 |
+
res = await client.create_audio_note(
|
| 128 |
+
{
|
| 129 |
+
"note_id": note_id,
|
| 130 |
+
"type": "audio",
|
| 131 |
+
"status": "processing",
|
| 132 |
+
"raw_text": "",
|
| 133 |
+
"metadata": {
|
| 134 |
+
"audio": {
|
| 135 |
+
"duration": duration,
|
| 136 |
+
"chunks": [],
|
| 137 |
+
"asr_model": "PhoWhisper-base",
|
| 138 |
+
}
|
| 139 |
+
},
|
| 140 |
+
# ❌ KHÔNG generate ở đây
|
| 141 |
}
|
| 142 |
+
)
|
| 143 |
+
|
| 144 |
+
if res is None:
|
| 145 |
+
# 🔥 FAIL FAST
|
| 146 |
+
raise RuntimeError("Failed to create placeholder note")
|
| 147 |
+
|
| 148 |
|
| 149 |
|
| 150 |
# ============================================================
|
app/jobs/transcribe_job.py
CHANGED
|
@@ -2,11 +2,24 @@ import asyncio
|
|
| 2 |
import tempfile
|
| 3 |
import os
|
| 4 |
import requests
|
|
|
|
|
|
|
| 5 |
|
| 6 |
from app.core.asr_engine import load_model, transcribe_file, transcribe_file_chunks
|
| 7 |
from app.services.note_client import NoteServiceClient
|
| 8 |
from app.core.audio_utils import get_audio_info
|
| 9 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10 |
|
| 11 |
def download_audio(audio_url: str) -> str:
|
| 12 |
r = requests.get(audio_url, timeout=30)
|
|
@@ -62,18 +75,32 @@ def transcribe_job(audio_url: str, note_id: str, user_id: str | None = None):
|
|
| 62 |
"generate": ["normalize", "keywords", "summary", "mindmap"],
|
| 63 |
}
|
| 64 |
|
| 65 |
-
|
| 66 |
-
|
| 67 |
-
|
| 68 |
-
|
| 69 |
-
{
|
| 70 |
-
"status": note_status,
|
| 71 |
-
"raw_text": text,
|
| 72 |
-
"metadata": payload["metadata"],
|
| 73 |
-
},
|
| 74 |
-
)
|
| 75 |
)
|
| 76 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 77 |
|
| 78 |
finally:
|
| 79 |
if wav_path and os.path.exists(wav_path):
|
|
|
|
| 2 |
import tempfile
|
| 3 |
import os
|
| 4 |
import requests
|
| 5 |
+
import httpx
|
| 6 |
+
import time
|
| 7 |
|
| 8 |
from app.core.asr_engine import load_model, transcribe_file, transcribe_file_chunks
|
| 9 |
from app.services.note_client import NoteServiceClient
|
| 10 |
from app.core.audio_utils import get_audio_info
|
| 11 |
|
| 12 |
+
def run_async(coro):
|
| 13 |
+
try:
|
| 14 |
+
loop = asyncio.get_running_loop()
|
| 15 |
+
except RuntimeError:
|
| 16 |
+
loop = None
|
| 17 |
+
|
| 18 |
+
if loop and loop.is_running():
|
| 19 |
+
return asyncio.create_task(coro)
|
| 20 |
+
else:
|
| 21 |
+
return asyncio.run(coro)
|
| 22 |
+
|
| 23 |
|
| 24 |
def download_audio(audio_url: str) -> str:
|
| 25 |
r = requests.get(audio_url, timeout=30)
|
|
|
|
| 75 |
"generate": ["normalize", "keywords", "summary", "mindmap"],
|
| 76 |
}
|
| 77 |
|
| 78 |
+
generate_tasks = (
|
| 79 |
+
["normalize", "keywords", "summary", "mindmap"]
|
| 80 |
+
if note_status == "transcribed"
|
| 81 |
+
else []
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 82 |
)
|
| 83 |
|
| 84 |
+
time.sleep(1.0)
|
| 85 |
+
|
| 86 |
+
client = NoteServiceClient()
|
| 87 |
+
try:
|
| 88 |
+
run_async(
|
| 89 |
+
client.update_note(
|
| 90 |
+
note_id,
|
| 91 |
+
{
|
| 92 |
+
"status": note_status,
|
| 93 |
+
"raw_text": text,
|
| 94 |
+
"metadata": payload["metadata"],
|
| 95 |
+
**({"generate": generate_tasks} if generate_tasks else {}),
|
| 96 |
+
},
|
| 97 |
+
)
|
| 98 |
+
)
|
| 99 |
+
except httpx.HTTPStatusError as e:
|
| 100 |
+
if e.response.status_code == 404:
|
| 101 |
+
run_async(client.create_audio_note(payload))
|
| 102 |
+
else:
|
| 103 |
+
raise
|
| 104 |
|
| 105 |
finally:
|
| 106 |
if wav_path and os.path.exists(wav_path):
|