Update app.py

app.py

@@ -1886,6 +1886,191 @@ def _normalize_checks_paid_table(grid):
     return new_rows
 
 
+# ── Johnson Bank plain-text section converter ─────────────────────────────
+# Johnson Bank formats Deposits, Withdrawals, Checks, and Daily Account Balance
+# as plain text columns (no HTML tables). This converter detects those sections
+# and emits proper <table> HTML.
+#
+# SAFETY GUARDS — all four must hold before any conversion fires:
+#   1. Line matches exact section-header pattern
+#   2. Next non-blank line matches exact column-header pattern
+#   3. First data row uses MM-DD date format (Johnson Bank's unique date separator)
+#   4. The candidate section contains NO existing <table> tags
+# Guard 3 is the key discriminator: all other supported banks use MM/DD.
+
+_JB_SECTION_HDR = re.compile(
+    r"^(Deposits?|Withdrawals?|Checks?(?:\s+Paid)?|Daily\s+Account\s+Balance)"
+    r"\s*(?:\(cont\.?\))?\s*$",
+    re.IGNORECASE,
+)
+_JB_COL_HDR = re.compile(
+    r"^Date\s+(Description|Number)\s+Amount\s*$"
+    r"|^Date\s+(?:Number\s+)?(?:Balance|Amount)\s*$",
+    re.IGNORECASE,
+)
+# MM-DD date (dash separator) — Johnson Bank's unique format
+_JB_DATE_DD  = re.compile(r"^\d{2}-\d{2}\s+")
+_JB_TXN_ROW  = re.compile(r"^(\d{2}-\d{2})\s+(.+?)\s+(-?\d{1,3}(?:,\d{3})*\.\d{2})\s*$")
+_JB_BAL_ROW  = re.compile(r"^(\d{2}-\d{2})\s+(\d{1,3}(?:,\d{3})*\.\d{2})\s*$")
+
+
+def _jb_rows_to_html(header_cols, rows):
+    th    = "".join("<th>" + h + "</th>" for h in header_cols)
+    parts = ["<table>", "<tr>" + th + "</tr>"]
+    for row in rows:
+        td = "".join("<td>" + str(row.get(h, "")) + "</td>" for h in header_cols)
+        parts.append("<tr>" + td + "</tr>")
+    parts.append("</table>")
+    return "\n".join(parts)
+
+
+def _jb_section_has_table(lines, start, end):
+    """Return True if any line in lines[start:end] contains a <table tag."""
+    return any("<table" in l.lower() for l in lines[start:end])
+
+
+def _find_first_data_row(lines, start):
+    """Return index of first non-blank line after start, or None."""
+    for i in range(start, min(start + 20, len(lines))):
+        if lines[i].strip():
+            return i
+    return None
+
+
+def convert_plaintext_bank_sections(text: str) -> str:
+    """
+    Convert Johnson Bank plain-text transaction sections to HTML tables.
+    All four safety guards must pass before conversion fires.
+    Safe no-op for all other banks.
+    """
+    # Quick exit: if the page already has lots of tables, skip entirely
+    # (handles pages that are already properly formatted)
+    lines = text.splitlines()
+    out   = []
+    i     = 0
+
+    while i < len(lines):
+        line = lines[i].strip()
+
+        # Guard 1: section header
+        if not _JB_SECTION_HDR.match(line):
+            out.append(lines[i])
+            i += 1
+            continue
+
+        section_title = line
+        j = i + 1
+
+        # Skip blank lines to find column header
+        while j < len(lines) and not lines[j].strip():
+            j += 1
+
+        # Guard 2: column header
+        if j >= len(lines) or not _JB_COL_HDR.match(lines[j].strip()):
+            out.append(lines[i])
+            i += 1
+            continue
+
+        col_hdr_line = lines[j].strip()
+        data_start   = j + 1
+
+        # Find first non-blank data line
+        first_data_idx = _find_first_data_row(lines, data_start)
+
+        # Guard 3: first data row must use MM-DD date format
+        if first_data_idx is None or not _JB_DATE_DD.match(lines[first_data_idx].strip()):
+            out.append(lines[i])
+            i += 1
+            continue
+
+        # Find section end (next section header, page separator, or end of text)
+        section_end = len(lines)
+        for k in range(data_start, len(lines)):
+            l = lines[k].strip()
+            if (l != section_title and _JB_SECTION_HDR.match(l)
+                    or l.startswith("---page-separator")
+                    or l.startswith("Page:")):
+                section_end = k
+                break
+
+        # Guard 4: no existing <table> tags in this section
+        if _jb_section_has_table(lines, i, section_end):
+            out.append(lines[i])
+            i += 1
+            continue
+
+        # All guards passed — parse and convert
+        parts_upper = col_hdr_line.upper().split()
+        if len(parts_upper) == 2 and parts_upper[1] == "BALANCE":
+            header_cols = ["Date", "Balance"]
+        elif "NUMBER" in parts_upper:
+            header_cols = ["Date", "Number", "Amount"]
+        else:
+            header_cols = ["Date", "Description", "Amount"]
+
+        out.append(section_title)
+        rows    = []
+        current = None
+        k       = data_start
+
+        while k < section_end:
+            raw      = lines[k]
+            row_line = raw.strip()
+
+            if not row_line:
+                k += 1
+                continue
+
+            if (_JB_SECTION_HDR.match(row_line)
+                    or row_line.startswith("---page-separator")
+                    or row_line.startswith("Page:")):
+                break
+
+            # Balance-only row
+            mb = _JB_BAL_ROW.match(row_line)
+            if mb and header_cols == ["Date", "Balance"]:
+                if current:
+                    rows.append(current)
+                current = {"Date": mb.group(1), "Balance": mb.group(2)}
+                k += 1
+                continue
+
+            # Full transaction row
+            mt = _JB_TXN_ROW.match(row_line)
+            if mt:
+                if current:
+                    rows.append(current)
+                desc_key = "Description" if "Description" in header_cols else "Number"
+                current  = {
+                    "Date":    mt.group(1),
+                    desc_key:  mt.group(2).strip(),
+                    "Amount":  mt.group(3),
+                }
+                k += 1
+                continue
+
+            # Continuation line
+            if current and row_line:
+                desc_key = "Description" if "Description" in header_cols else "Number"
+                if desc_key in current:
+                    current[desc_key] += " " + row_line
+                k += 1
+                continue
+
+            k += 1
+
+        if current:
+            rows.append(current)
+
+        if rows:
+            out.append(_jb_rows_to_html(header_cols, rows))
+
+        i = section_end   # jump past the whole section
+        continue
+
+    return "\n".join(out)
+
+
 def normalize_html_tables(text: str) -> str:
     """
     For every <table>...</table>:
@@ -1978,6 +2163,7 @@ def stabilize_tables_and_text(page_md: str) -> str:
             out_blocks.append(b)
 
     stabilized = "\n\n".join(out_blocks)
+    stabilized = convert_plaintext_bank_sections(stabilized)
     stabilized = normalize_html_tables(stabilized)
     return close_unclosed_html(stabilized)