Spaces:
Sleeping
Sleeping
Pointf5ive commited on
Commit ·
2d9935c
1
Parent(s): 7ccec5a
Add debug output to OCR step
Browse files- smoke_signal_tab.py +8 -2
smoke_signal_tab.py
CHANGED
|
@@ -683,12 +683,18 @@ def _profile_status_html() -> str:
|
|
| 683 |
def run_ocr() -> tuple:
|
| 684 |
"""Run Surya OCR on all profiled PDFs."""
|
| 685 |
df = load_manifest_df()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 686 |
if df.empty:
|
| 687 |
-
return _ocr_status_html(), "No sources. Complete Steps 1-2 first."
|
| 688 |
|
| 689 |
eligible = df[df["status"] == "profiled"]
|
| 690 |
if eligible.empty:
|
| 691 |
-
return _ocr_status_html(), "No profiled PDFs. Complete Step 2 first."
|
| 692 |
|
| 693 |
log = []
|
| 694 |
|
|
|
|
| 683 |
def run_ocr() -> tuple:
|
| 684 |
"""Run Surya OCR on all profiled PDFs."""
|
| 685 |
df = load_manifest_df()
|
| 686 |
+
debug = f"[DEBUG] SS_ROOT={SS_ROOT}\nMANIFEST_CSV={MANIFEST_CSV}\nCSV exists={MANIFEST_CSV.exists()}\n"
|
| 687 |
+
if not df.empty:
|
| 688 |
+
debug += f"Manifest rows={len(df)}\nStatuses={df['status'].value_counts().to_dict()}\n"
|
| 689 |
+
else:
|
| 690 |
+
debug += "Manifest is EMPTY\n"
|
| 691 |
+
|
| 692 |
if df.empty:
|
| 693 |
+
return _ocr_status_html(), debug + "No sources. Complete Steps 1-2 first."
|
| 694 |
|
| 695 |
eligible = df[df["status"] == "profiled"]
|
| 696 |
if eligible.empty:
|
| 697 |
+
return _ocr_status_html(), debug + "No profiled PDFs. Complete Step 2 first."
|
| 698 |
|
| 699 |
log = []
|
| 700 |
|