Update app.py

Header and footer returns. Remove the logs

app.py

@@ -18,20 +18,12 @@ except Exception:
 import logging
 import os
 import re
-import sys
 import tempfile
 import yaml
 import gradio as gr
 
 import glmocr
 
-# Log to stderr so Hugging Face / Docker show it in logs. Set GLMOCR_LOG=DEBUG for more detail.
-_log_level = os.environ.get("GLMOCR_LOG", "INFO").upper()
-logging.basicConfig(
-    level=getattr(logging, _log_level, logging.INFO),
-    format="[%(levelname)s] %(name)s: %(message)s",
-    stream=sys.stderr,
-)
 log = logging.getLogger("glmocr_app")
 
 GLMOCR_BASE = os.path.dirname(glmocr.__file__)
@@ -118,7 +110,6 @@ except Exception:
 
 def get_header_footer_zones(regions, norm_height=1000):
     if not regions:
-        log.debug("get_header_footer_zones: no regions")
         return None, None
     y_tops, y_bottoms = [], []
     for r in regions:
@@ -127,24 +118,8 @@ def get_header_footer_zones(regions, norm_height=1000):
             y_tops.append(bbox[1])
             y_bottoms.append(bbox[3])
     if not y_tops:
-        log.debug("get_header_footer_zones: no bboxes in regions")
         return None, None
-    he_frac = min(y_tops) / norm_height
-    fs_frac = max(y_bottoms) / norm_height
-    log.debug("get_header_footer_zones: regions=%s -> header_end_frac=%.3f footer_start_frac=%.3f", len(regions), he_frac, fs_frac)
-    return he_frac, fs_frac
-
-
-def _log_text(prefix, text, max_display=400):
-    """Log extracted text for comparison with OCR; truncate long content."""
-    if not text or not text.strip():
-        log.info("%s: (empty)", prefix)
-        return
-    s = text.strip()
-    if len(s) <= max_display:
-        log.info("%s: %d chars | %s", prefix, len(s), s)
-    else:
-        log.info("%s: %d chars | %s ...", prefix, len(s), s[:max_display].rstrip())
+    return min(y_tops) / norm_height, max(y_bottoms) / norm_height
 
 
 def extract_zone_text_pdf(pdf_path, page_num, y_start_frac, y_end_frac):
@@ -157,26 +132,11 @@ def extract_zone_text_pdf(pdf_path, page_num, y_start_frac, y_end_frac):
         rect = fitz.Rect(0, h * y_start_frac, w, h * y_end_frac)
         text = page.get_text(clip=rect).strip()
         doc.close()
-        _log_text("pymupdf clip page=%s y=%.2f-%.2f" % (page_num, y_start_frac, y_end_frac), text)
         return text
-    except Exception as e:
-        log.debug("extract_zone_text_pdf failed: %s", e)
+    except Exception:
         return ""
 
 
-def log_pymupdf_full_page(pdf_path, page_num):
-    """Log full-page PyMuPDF text for comparison (length + prefix)."""
-    try:
-        import pymupdf as fitz
-        doc = fitz.open(pdf_path)
-        page = doc[page_num]
-        text = page.get_text().strip()
-        doc.close()
-        _log_text("pymupdf full page=%s" % page_num, text, max_display=500)
-    except Exception as e:
-        log.debug("log_pymupdf_full_page failed: %s", e)
-
-
 def extract_pdf_text_in_band(pdf_path, page_num, y_start_frac, y_end_frac):
     """Extract all text whose word bbox falls in the given vertical band (by position).
     Use a wider band (e.g. top 15% / bottom 15%) to capture header/footer that clip might miss."""
@@ -195,11 +155,8 @@ def extract_pdf_text_in_band(pdf_path, page_num, y_start_frac, y_end_frac):
                 y0, y1 = float(w[1]), float(w[3])
                 if y0 < y_hi and y1 > y_lo:
                     parts.append(w[4])
-        text = " ".join(parts).strip()
-        _log_text("pymupdf band page=%s y=%.2f-%.2f" % (page_num, y_start_frac, y_end_frac), text)
-        return text
-    except Exception as e:
-        log.debug("extract_pdf_text_in_band failed: %s", e)
+        return " ".join(parts).strip()
+    except Exception:
         return ""
 
 
@@ -213,7 +170,6 @@ def ocr_zone(image_path, y_start_frac, y_end_frac):
         y0 = max(0, int(h * y_start_frac))
         y1 = min(h, int(h * y_end_frac))
         if y1 <= y0:
-            log.info("[%s] ocr_zone: skip (zero height band y0=%s y1=%s)", zone_name, y0, y1)
             return ""
         crop = img.crop((0, y0, w, y1))
         cw, ch = crop.size
@@ -230,7 +186,6 @@ def ocr_zone(image_path, y_start_frac, y_end_frac):
                 canvas.paste(crop, (0, need_h - ch))  # crop at bottom
             crop = canvas
             cw, ch = crop.size
-            log.info("[%s] ocr_zone: padded to %dx%d for API", zone_name, cw, ch)
         fd, path = tempfile.mkstemp(suffix=".jpg")
         os.close(fd)
         try:
@@ -241,10 +196,7 @@ def ocr_zone(image_path, y_start_frac, y_end_frac):
                 out = [out]
             if out and getattr(out[0], "markdown_result", None):
                 text = (out[0].markdown_result or "").strip()
-                log.info("[%s] ocr_zone: crop %dx%d -> %d chars", zone_name, cw, ch, len(text))
-                _log_text("ocr_zone %s result" % zone_name, text)
                 return text
-            log.info("[%s] ocr_zone: crop %dx%d -> empty result from API", zone_name, cw, ch)
         finally:
             try:
                 os.unlink(path)
@@ -281,7 +233,6 @@ def run_ocr(uploaded_file):
 
         path = uploaded_file.name if hasattr(uploaded_file, "name") else str(uploaded_file)
         is_pdf = path.lower().endswith(".pdf")
-        log.info("run_ocr: path=%s is_pdf=%s", path, is_pdf)
         parser = get_parser()
 
         if is_pdf:
@@ -293,7 +244,6 @@ def run_ocr(uploaded_file):
                 pix.save(img_path)
                 page_images.append(img_path)
             doc.close()
-            log.info("run_ocr: PDF has %d pages, running MaaS on page images", len(page_images))
             results = parser.parse(page_images)
         else:
             page_images = [path]
@@ -304,8 +254,6 @@ def run_ocr(uploaded_file):
 
         all_pages = []
         for page_num, page_result in enumerate(results):
-            if is_pdf:
-                log_pymupdf_full_page(path, page_num)
             page_md, regions = get_page_md_and_regions(page_result)
             header_end_frac, footer_start_frac = get_header_footer_zones(regions, 1000)
 
@@ -317,11 +265,6 @@ def run_ocr(uploaded_file):
             if fs >= 1.0:
                 fs = 1.0 - DEFAULT_ZONE_FRAC  # ensure we always OCR a bottom band
 
-            log.info(
-                "run_ocr: page %d regions=%s he=%.3f fs=%.3f body_md_len=%d",
-                page_num, len(regions), he, fs, len(page_md or ""),
-            )
-
             parts = []
 
             # Always run header band (top 8–12% of page)
@@ -332,15 +275,10 @@ def run_ocr(uploaded_file):
                     hdr = extract_zone_text_pdf(path, page_num, 0, he)
                     if not (hdr and hdr.strip()):
                         hdr = extract_pdf_text_in_band(path, page_num, 0, PDF_HEADER_BAND_FRAC)
-                    if hdr and hdr.strip():
-                        log.info("run_ocr: page %d header from PDF: %d chars", page_num, len(hdr.strip()))
                 if not (hdr and hdr.strip()):
                     hdr = ocr_zone(img_path, 0, he)
                 if hdr and hdr.strip():
                     parts.append(hdr.strip())
-                    _log_text("run_ocr page %d final header (used in output)" % page_num, hdr.strip())
-                elif not (hdr and hdr.strip()) and he > 0:
-                    log.info("run_ocr: page %d header empty (PDF + ocr_zone)", page_num)
 
             if page_md:
                 parts.append(page_md)
@@ -353,20 +291,14 @@ def run_ocr(uploaded_file):
                     ftr = extract_zone_text_pdf(path, page_num, fs, 1.0)
                     if not (ftr and ftr.strip()):
                         ftr = extract_pdf_text_in_band(path, page_num, PDF_FOOTER_BAND_FRAC, 1.0)
-                    if ftr and ftr.strip():
-                        log.info("run_ocr: page %d footer from PDF: %d chars", page_num, len(ftr.strip()))
                 if not (ftr and ftr.strip()):
                     ftr = ocr_zone(img_path, fs, 1.0)
                 if ftr and ftr.strip():
                     parts.append(ftr.strip())
-                    _log_text("run_ocr page %d final footer (used in output)" % page_num, ftr.strip())
-                elif not (ftr and ftr.strip()) and fs < 1.0:
-                    log.info("run_ocr: page %d footer empty (PDF + ocr_zone)", page_num)
 
             if parts:
                 all_pages.append("\n\n".join(parts))
 
-        log.info("run_ocr: done, %d pages in output", len(all_pages))
         return "\n\n---\n\n".join(all_pages) if all_pages else "(No content)"
     except Exception as e:
         import traceback