Fix modern ICICI template parsing and parser fallback

app/modules/bank_statement_analyzer/parsers/template_engine.py

@@ -77,8 +77,15 @@ def _upper(value) -> str:
     return norm(value).upper().replace("\n", " ")
 
 
-def _is_date(value) -> bool:
+def _normalise_date_cell(value) -> str:
+    """Normalise PDF table line-wrap artefacts without changing the date value."""
     text = norm(value)
+    text = re.sub(r"\s*([/-])\s*", r"\1", text)
+    return text
+
+
+def _is_date(value) -> bool:
+    text = _normalise_date_cell(value)
     return bool(_DATE_CELL_RE.fullmatch(text)) and not pd.isna(parse_flexible_date(text))
 
 
@@ -102,7 +109,7 @@ def _find_header_mapping(row: list) -> dict[str, int]:
 
 
 def _looks_like_data_row(row: list) -> bool:
-    return bool(row) and (_is_date(row[0]) or (len(row) > 1 and _is_date(row[1])))
+    return bool(row) and any(_is_date(row[index]) for index in range(min(4, len(row))))
 
 
 def _infer_structural_template(table: list[list]) -> LayoutTemplate | None:
@@ -110,6 +117,23 @@ def _infer_structural_template(table: list[list]) -> LayoutTemplate | None:
     if sample is None:
         return None
     width = len(sample)
+    # Modern ICICI and similar indexed statements:
+    # serial no, transaction id, transaction date, cheque no, description,
+    # withdrawal, deposit, available balance.
+    if width >= 8 and _is_date(sample[2]):
+        return LayoutTemplate(
+            "indexed_transaction_id_debit_credit_balance",
+            {
+                "transaction_date": 2,
+                "value_date": 2,
+                "reference_no": 1,
+                "narration": 4,
+                "debit": 5,
+                "credit": 6,
+                "balance": 7,
+            },
+            0.92,
+        )
     # SBI Account Summary / SBI YONO family:
     # value date, post date, details, reference, debit, credit, balance.
     if width >= 7 and _is_date(sample[0]) and _is_date(sample[1]):
@@ -354,11 +378,11 @@ class TemplateBasedStatementParser(BaseParser):
                         max_index = max(mapping.values())
                         if len(cells) <= max_index:
                             cells += [None] * (max_index + 1 - len(cells))
-                        txn_text = norm(cells[mapping["transaction_date"]])
+                        txn_text = _normalise_date_cell(cells[mapping["transaction_date"]])
                         if not _is_date(txn_text):
                             continue
                         value_index = mapping.get("value_date", mapping["transaction_date"])
-                        value_text = norm(cells[value_index]) or txn_text
+                        value_text = _normalise_date_cell(cells[value_index]) or txn_text
                         narration = norm(cells[mapping["narration"]])
                         reference = norm(cells[mapping["reference_no"]]) if "reference_no" in mapping else ""
                         debit = amount(cells[mapping["debit"]])