Gate fallback OCR by minimum confidence to reduce gibberish queue

smoke_signal_tab.py

@@ -767,6 +767,10 @@ try:
     SS_TESSERACT_PSM = max(1, int(os.environ.get("SS_TESSERACT_PSM", "11")))
 except Exception:
     SS_TESSERACT_PSM = 11
+try:
+    SS_TESSERACT_MIN_CONF_KEEP = max(0.0, min(1.0, float(os.environ.get("SS_TESSERACT_MIN_CONF_KEEP", "0.58"))))
+except Exception:
+    SS_TESSERACT_MIN_CONF_KEEP = 0.58
 
 
 def _load_surya_runtime():
@@ -971,6 +975,11 @@ def _run_tesseract_batch(images):
                 word_count += wc
 
             avg_conf = round(conf_weighted / max(word_count, 1), 4) if regions else 0.0
+
+            # Hard gate for fallback quality: do not keep OCR text below minimum page confidence.
+            if regions and avg_conf < SS_TESSERACT_MIN_CONF_KEEP:
+                return {"regions": [], "confidence": 0.0, "method": "tesseract-lowconf-filtered"}
+
             # Filter obvious OCR noise from illustration texture (common in no-text pages).
             full_text = " ".join(r["text"] for r in regions).strip()
             letters = sum(1 for c in full_text if c.isalpha())
@@ -1049,7 +1058,7 @@ def run_ocr(progress=gr.Progress(track_tqdm=False)) -> tuple:
         log.append(
             log_line(
                 f"ℹ Tesseract fallback config: workers={SS_TESSERACT_WORKERS} timeout={SS_TESSERACT_TIMEOUT_SEC}s "
-                f"psm={SS_TESSERACT_PSM} dpi={SS_RENDER_DPI_FALLBACK}"
+                f"psm={SS_TESSERACT_PSM} dpi={SS_RENDER_DPI_FALLBACK} min_conf_keep={SS_TESSERACT_MIN_CONF_KEEP:.2f}"
             )
         )
 
@@ -1282,7 +1291,7 @@ def run_ocr(progress=gr.Progress(track_tqdm=False)) -> tuple:
                 conf = 0.0
                 method = "skipped-no-surya"
 
-            if method == "tesseract-noise-filtered" and not regions:
+            if method in ("tesseract-noise-filtered", "tesseract-lowconf-filtered") and not regions:
                 conf_class = "auto-accept"
             elif conf >= default_cal["auto_accept"]:
                 conf_class = "auto-accept"