Update app.py

app.py

@@ -15,8 +15,10 @@ Universal image pipeline (same for every PDF, no keywords / no bank logic):
     thin rules and small glyphs are easier to read before recognition.
 
 Explicitly omitted vs the heavy Space build:
-  - No HTML table rewriting, text-layer row injection, institution-specific
-    splits (UCB / Navy / TD / First Horizon / …), or doc-wide dedupe passes
+  - No text-layer row injection, institution-specific splits (UCB / Navy /
+    TD / First Horizon / …), doc-wide dedupe passes, or HTML table rewriting
+    keyed off column names / dates / amounts (layout fidelity comes from input
+    image quality, not post-hoc string rules).
 
 Configure GLMOCR_API_KEY (environment variable). Optional: glmocr + gradio +
 pymupdf + pillow installed.
@@ -70,24 +72,24 @@ if not GLMOCR_API_KEY:
 
 # Rasterization: higher scale = more pixels per PDF point (helps small type,
 # boxed headers, and narrow columns). Same constant for all uploads.
-RENDER_SCALE = 2.85
+RENDER_SCALE = 3.05
 
 # White margin as a fraction of page width/height after render. Extra right
 # margin helps right-aligned currency columns that hug the page edge.
-PAD_LEFT_FRAC = 0.03
-PAD_RIGHT_FRAC = 0.10
-PAD_TOP_FRAC = 0.015
-PAD_BOTTOM_FRAC = 0.015
+PAD_LEFT_FRAC = 0.035
+PAD_RIGHT_FRAC = 0.11
+PAD_TOP_FRAC = 0.018
+PAD_BOTTOM_FRAC = 0.018
 
 ENABLE_CONTRAST = True
 # Slight contrast lift only; same factor for every file.
-CONTRAST_FACTOR = 1.12
+CONTRAST_FACTOR = 1.16
 
 # Subtle edge enhancement after contrast (helps hairlines and small digits).
 ENABLE_UNSHARP = True
-UNSHARP_RADIUS = 0.85
-UNSHARP_PERCENT = 65
-UNSHARP_THRESHOLD = 2
+UNSHARP_RADIUS = 0.78
+UNSHARP_PERCENT = 72
+UNSHARP_THRESHOLD = 1
 
 DEFAULT_ZONE_FRAC = 0.12
 PDF_HEADER_BAND_FRAC = 0.10