Spaces:
Sleeping
Sleeping
Update src/streamlit_app.py
Browse files- src/streamlit_app.py +48 -38
src/streamlit_app.py
CHANGED
|
@@ -426,23 +426,31 @@ def show_login_page():
|
|
| 426 |
|
| 427 |
def upload_and_poll_files(api_key: str, uploaded_files: list):
|
| 428 |
"""
|
| 429 |
-
Upload PDFs,
|
|
|
|
| 430 |
"""
|
| 431 |
if not api_key:
|
| 432 |
st.error("Missing API key. Please login first.")
|
| 433 |
return
|
| 434 |
|
| 435 |
-
|
| 436 |
-
progress_bar = st.empty()
|
| 437 |
status_text = st.empty()
|
| 438 |
|
| 439 |
-
#
|
| 440 |
-
|
| 441 |
|
| 442 |
for i, file in enumerate(uploaded_files):
|
| 443 |
status_text.text(f"Uploading {file.name}...")
|
| 444 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 445 |
|
|
|
|
| 446 |
if not result["success"]:
|
| 447 |
st.error(f"β {file.name}: {result['error']}")
|
| 448 |
progress_bar.progress((i + 1) / len(uploaded_files))
|
|
@@ -450,15 +458,15 @@ def upload_and_poll_files(api_key: str, uploaded_files: list):
|
|
| 450 |
|
| 451 |
data = result.get("data") or {}
|
| 452 |
job_id = data.get("document_id") or data.get("job_id")
|
| 453 |
-
status = data.get("status")
|
| 454 |
|
| 455 |
-
|
| 456 |
-
|
| 457 |
poll_text = st.empty()
|
| 458 |
-
poll_progress = st.
|
| 459 |
-
elapsed = 0
|
| 460 |
max_poll_seconds = 600
|
| 461 |
poll_interval = 2
|
|
|
|
| 462 |
|
| 463 |
while elapsed < max_poll_seconds:
|
| 464 |
try:
|
|
@@ -467,47 +475,49 @@ def upload_and_poll_files(api_key: str, uploaded_files: list):
|
|
| 467 |
headers=get_headers(api_key),
|
| 468 |
timeout=10
|
| 469 |
)
|
| 470 |
-
if
|
|
|
|
| 471 |
job = resp.json()
|
| 472 |
-
|
| 473 |
-
|
| 474 |
-
|
| 475 |
-
|
| 476 |
-
|
| 477 |
-
|
| 478 |
-
|
| 479 |
-
|
| 480 |
-
|
| 481 |
-
|
| 482 |
-
break
|
| 483 |
else:
|
| 484 |
-
|
|
|
|
|
|
|
| 485 |
else:
|
| 486 |
-
poll_text.text(f"
|
| 487 |
except Exception as e:
|
| 488 |
poll_text.text(f"Waiting... ({str(e)})")
|
|
|
|
| 489 |
time.sleep(poll_interval)
|
| 490 |
elapsed += poll_interval
|
| 491 |
poll_progress.progress(min(1.0, elapsed / max_poll_seconds))
|
| 492 |
|
| 493 |
-
|
| 494 |
-
|
| 495 |
-
|
| 496 |
elif status in ["success", "completed"]:
|
| 497 |
st.success(f"β
{file.name} uploaded and processed successfully")
|
| 498 |
-
|
|
|
|
|
|
|
| 499 |
|
| 500 |
-
|
|
|
|
|
|
|
|
|
|
| 501 |
|
| 502 |
-
|
| 503 |
-
status_text.empty()
|
| 504 |
-
progress_bar.empty()
|
| 505 |
|
| 506 |
-
|
| 507 |
-
if processed_files and "pdf_uploader" in st.session_state:
|
| 508 |
-
# Streamlit doesn't allow direct modification, so delete and rerun
|
| 509 |
-
del st.session_state["pdf_uploader"]
|
| 510 |
-
st.experimental_rerun()
|
| 511 |
|
| 512 |
# ============================================
|
| 513 |
# MAIN APPLICATION UI
|
|
|
|
| 426 |
|
| 427 |
def upload_and_poll_files(api_key: str, uploaded_files: list):
|
| 428 |
"""
|
| 429 |
+
Upload PDFs to backend, poll for processing, and clean up uploader after completion.
|
| 430 |
+
Works for both local and HF Spaces deployments.
|
| 431 |
"""
|
| 432 |
if not api_key:
|
| 433 |
st.error("Missing API key. Please login first.")
|
| 434 |
return
|
| 435 |
|
| 436 |
+
progress_bar = st.progress(0)
|
|
|
|
| 437 |
status_text = st.empty()
|
| 438 |
|
| 439 |
+
# Make a copy of uploaded_files so we can remove processed ones
|
| 440 |
+
remaining_files = uploaded_files.copy()
|
| 441 |
|
| 442 |
for i, file in enumerate(uploaded_files):
|
| 443 |
status_text.text(f"Uploading {file.name}...")
|
| 444 |
+
|
| 445 |
+
# Upload
|
| 446 |
+
try:
|
| 447 |
+
result = upload_pdf(api_key, file)
|
| 448 |
+
except Exception as e:
|
| 449 |
+
st.error(f"β {file.name} upload failed: {e}")
|
| 450 |
+
progress_bar.progress((i + 1) / len(uploaded_files))
|
| 451 |
+
continue
|
| 452 |
|
| 453 |
+
# If upload request itself failed
|
| 454 |
if not result["success"]:
|
| 455 |
st.error(f"β {file.name}: {result['error']}")
|
| 456 |
progress_bar.progress((i + 1) / len(uploaded_files))
|
|
|
|
| 458 |
|
| 459 |
data = result.get("data") or {}
|
| 460 |
job_id = data.get("document_id") or data.get("job_id")
|
| 461 |
+
status = data.get("status", "processing")
|
| 462 |
|
| 463 |
+
# Poll backend for processing status
|
| 464 |
+
if job_id and status == "processing":
|
| 465 |
poll_text = st.empty()
|
| 466 |
+
poll_progress = st.progress(0)
|
|
|
|
| 467 |
max_poll_seconds = 600
|
| 468 |
poll_interval = 2
|
| 469 |
+
elapsed = 0
|
| 470 |
|
| 471 |
while elapsed < max_poll_seconds:
|
| 472 |
try:
|
|
|
|
| 475 |
headers=get_headers(api_key),
|
| 476 |
timeout=10
|
| 477 |
)
|
| 478 |
+
# Always attempt JSON parse, fallback if HF wraps response
|
| 479 |
+
try:
|
| 480 |
job = resp.json()
|
| 481 |
+
except Exception:
|
| 482 |
+
job = {}
|
| 483 |
+
|
| 484 |
+
final_status = job.get("status", "processing")
|
| 485 |
+
display_name = job.get("username") or job.get("user_id") or "You"
|
| 486 |
+
|
| 487 |
+
if final_status not in ["processing"]:
|
| 488 |
+
if final_status in ["success", "completed"]:
|
| 489 |
+
res = job.get("result", {})
|
| 490 |
+
st.success(f"β
{file.name} processed for {display_name}: {res.get('filename', file.name)}")
|
|
|
|
| 491 |
else:
|
| 492 |
+
error_msg = job.get('error', 'unknown error')
|
| 493 |
+
st.error(f"β {file.name} failed for {display_name}: {error_msg}")
|
| 494 |
+
break
|
| 495 |
else:
|
| 496 |
+
poll_text.text(f"Processing {file.name} for {display_name}... elapsed {elapsed}s")
|
| 497 |
except Exception as e:
|
| 498 |
poll_text.text(f"Waiting... ({str(e)})")
|
| 499 |
+
|
| 500 |
time.sleep(poll_interval)
|
| 501 |
elapsed += poll_interval
|
| 502 |
poll_progress.progress(min(1.0, elapsed / max_poll_seconds))
|
| 503 |
|
| 504 |
+
if elapsed >= max_poll_seconds:
|
| 505 |
+
st.error(f"β {file.name}: Processing timed out after {max_poll_seconds}s")
|
| 506 |
+
|
| 507 |
elif status in ["success", "completed"]:
|
| 508 |
st.success(f"β
{file.name} uploaded and processed successfully")
|
| 509 |
+
else:
|
| 510 |
+
# Unknown status β assume processing and continue
|
| 511 |
+
st.info(f"βΉοΈ {file.name}: Upload in progress")
|
| 512 |
|
| 513 |
+
# Remove the file from remaining_files to clear uploader
|
| 514 |
+
remaining_files.remove(file)
|
| 515 |
+
# Force re-render of file uploader with remaining files
|
| 516 |
+
st.session_state["pdf_uploader"] = remaining_files
|
| 517 |
|
| 518 |
+
progress_bar.progress((i + 1) / len(uploaded_files))
|
|
|
|
|
|
|
| 519 |
|
| 520 |
+
status_text.text("β
Upload complete!")
|
|
|
|
|
|
|
|
|
|
|
|
|
| 521 |
|
| 522 |
# ============================================
|
| 523 |
# MAIN APPLICATION UI
|