Update app.py

app.py

@@ -1102,10 +1102,16 @@ def _patch_ocr_with_textlayer(page_md: str, pdf_path: str, page_num: int) -> str
             key = (_norm(r["date"]), _norm(r["desc"]), _norm(r["amount"]))
             tl_freq[key] = tl_freq.get(key, 0) + 1
 
-        # Share with _dedupe_duplicate_rows_in_da3_table so it can use
-        # the confirmed count as the cap (Fix-20 override).
+        # Share with _dedupe_duplicate_rows_in_da3_table (Fix-20 override).
+        # Keys MUST match _transaction_row_dedupe_key (date/desc/amount, no spaces in
+        # date/desc; amount stripped of $/,/space). The injection map tl_freq uses
+        # _norm() with spaces — different tuple shape — so we keep a parallel count.
         global _tl_freq_context
-        _tl_freq_context = dict(tl_freq)
+        tl_dedupe = {}
+        for r in tl_rows:
+            dk = _transaction_row_dedupe_key([r["date"], r["desc"], r["amount"]])
+            tl_dedupe[dk] = tl_dedupe.get(dk, 0) + 1
+        _tl_freq_context = tl_dedupe
 
         table_pattern = re.compile(r"<table[^>]*>.*?</table>", re.DOTALL | re.IGNORECASE)
         result = page_md
@@ -1280,10 +1286,8 @@ def _patch_ocr_with_textlayer(page_md: str, pdf_path: str, page_num: int) -> str
 
     except Exception as e:
         log.warning("_patch_ocr_with_textlayer failed (page %d): %s", page_num, e)
-        return page_md
-    finally:
-        # Always clear so stale frequencies do not affect subsequent pages.
         _tl_freq_context.clear()
+        return page_md
 
 # --------------------------
 # Fix 5: Mid-table separator row reconstruction
@@ -2731,11 +2735,10 @@ _DA3_MONEY_TOKEN = re.compile(
 _MSFT_PIN_G_REF = re.compile(r"MICROSOFT#(G\d{6,16})", re.IGNORECASE)
 
 # Module-level text-layer frequency context (Fix-20 override).
-# Set by _patch_ocr_with_textlayer before post-patch normalize passes;
-# cleared in a finally clause so it never bleeds across pages.
-# Keys are (date_norm, desc_norm, amount_norm) tuples matching tl_freq.
-# When a key is present, _dedupe_duplicate_rows_in_da3_table uses its value
-# as the cap instead of the default 2, preserving all text-layer-confirmed rows.
+# Set by _patch_ocr_with_textlayer; consumed by normalize_html_tables in run_ocr's
+# post-patch pass. Cleared at the start of stabilize_tables_and_text (new page)
+# and on patch failure so keys never bleed across pages.
+# Keys match _transaction_row_dedupe_key (not the _norm()-based tl_freq map).
 _tl_freq_context: dict = {}
 
 _DA3_LEADING_DATE = re.compile(r"^\d{1,2}/\d{1,2}(?:/\d{2,4})?$")
@@ -3714,6 +3717,10 @@ def stabilize_tables_and_text(page_md: str) -> str:
     if not page_md:
         return page_md
 
+    # Fresh page: drop any text-layer dedupe caps so the first normalize pass
+    # (before _patch_ocr_with_textlayer) does not see stale keys from a prior page.
+    _tl_freq_context.clear()
+
     page_md = normalize_money_glyphs(page_md)
 
     blocks = re.split(r"\n\s*\n", page_md.strip())
@@ -3829,6 +3836,8 @@ def run_ocr(uploaded_file):
 
         all_pages = []
         for page_num, page_result in enumerate(results):
+            # Header normalize runs before body patch; do not reuse prior page caps.
+            _tl_freq_context.clear()
             page_md, regions = get_page_md_and_regions(page_result)
             img_h = page_heights[page_num] if page_num < len(page_heights) else 1000
             header_end_frac, footer_start_frac = get_header_footer_zones(regions, img_h)