plaggy commited on
Commit
2f42a56
·
1 Parent(s): b97d490

not schedule duplicated calls

Browse files
Files changed (1) hide show
  1. src/main.py +9 -7
src/main.py CHANGED
@@ -25,6 +25,7 @@ HF_TOKEN = os.getenv("HF_TOKEN")
25
  TEI_URL = os.getenv("TEI_URL")
26
 
27
  app = FastAPI()
 
28
 
29
 
30
  @app.get("/")
@@ -40,16 +41,17 @@ async def post_webhook(
40
  if not (
41
  payload.event.action == "update"
42
  and payload.event.scope.startswith("repo.content")
43
- # and payload.repo.name == chunk_config.input_dataset # any input dataset
44
  and payload.repo.type == "dataset"
 
45
  ):
46
  # no-op
47
  logger.info("Update detected, no action taken")
48
  return {"processed": False}
49
 
50
- if payload.repo.name == chunk_config.input_dataset:
51
- task_queue.add_task(chunk_dataset)
52
- task_queue.add_task(embed_dataset)
53
 
54
  return {"processed": True}
55
 
@@ -187,6 +189,6 @@ def embed_dataset():
187
 
188
  # For debugging
189
 
190
- # import uvicorn
191
- # if __name__ == "__main__":
192
- # uvicorn.run(app, host="0.0.0.0", port=7860)
 
25
  TEI_URL = os.getenv("TEI_URL")
26
 
27
  app = FastAPI()
28
+ app.state.last_Sha = None
29
 
30
 
31
  @app.get("/")
 
41
  if not (
42
  payload.event.action == "update"
43
  and payload.event.scope.startswith("repo.content")
44
+ and payload.repo.name == chunk_config.input_dataset
45
  and payload.repo.type == "dataset"
46
+ and (not app.state.last_Sha or app.state.last_Sha != payload.repo.headSha)
47
  ):
48
  # no-op
49
  logger.info("Update detected, no action taken")
50
  return {"processed": False}
51
 
52
+ app.state.last_Sha = payload.repo.headSha
53
+ task_queue.add_task(chunk_dataset)
54
+ task_queue.add_task(embed_dataset)
55
 
56
  return {"processed": True}
57
 
 
189
 
190
  # For debugging
191
 
192
+ import uvicorn
193
+ if __name__ == "__main__":
194
+ uvicorn.run(app, host="0.0.0.0", port=7860)