fix(cdsl): recover split demat account summaries and broker names

casparser/parsers/cdsl.py

@@ -585,11 +585,20 @@ def parse_cdsl(
 
 
 def _is_summary_demat_row(block: Block) -> bool:
-    if len(block.cells) != 4:
+    """Recognise both physical layouts of the page-2 demat summary row.
+
+    PDFium can either join the broker and DP/client details into one
+    cell or expose them as two adjacent cells.  Both layouts represent
+    the same logical row::
+
+        Type | "<BROKER>\nDP Id:... Client Id:..." | folios | value
+        Type | <BROKER> | "DP Id:... Client Id:..." | folios | value
+    """
+    if len(block.cells) not in (4, 5):
         return False
     if not DEMAT_TYPE_RE.match(block.cells[0].text.strip()):
         return False
-    return bool(SUMMARY_DPC_RE.search(block.cells[1].text))
+    return any(SUMMARY_DPC_RE.search(cell.text) for cell in block.cells[1:])
 
 
 def _is_summary_mf_folios_row(block: Block) -> bool:
@@ -601,15 +610,29 @@ def _is_summary_mf_folios_row(block: Block) -> bool:
 def _account_from_summary_row(
     block: Block, owners: List[DematOwner]
 ) -> Tuple[DematAccount, Tuple[str, str, str]]:
+    """Build an account from either supported page-2 summary layout."""
     type_word = DEMAT_TYPE_RE.match(block.cells[0].text.strip()).group(1).upper()
-    broker_dp = block.cells[1].text
-    lines = [ln.strip() for ln in broker_dp.split("\n") if ln.strip()]
-    broker = lines[0] if lines else ""
-    dpc = SUMMARY_DPC_RE.search(broker_dp)
+    dp_cell_idx = next(
+        (
+            index
+            for index in range(1, len(block.cells))
+            if SUMMARY_DPC_RE.search(block.cells[index].text)
+        ),
+        1,
+    )
+    dp_cell_text = block.cells[dp_cell_idx].text
+    dpc = SUMMARY_DPC_RE.search(dp_cell_text)
     dp_id = dpc.group(1) if dpc else ""
     client_id = dpc.group(2) if dpc else ""
-    folios = int(_to_decimal(block.cells[2].text))
-    balance = _to_decimal(block.cells[3].text)
+    broker_fragments = []
+    for cell in block.cells[1 : dp_cell_idx + 1]:
+        for line in cell.text.split("\n"):
+            fragment = SUMMARY_DPC_RE.sub("", line).strip()
+            if fragment:
+                broker_fragments.append(fragment)
+    broker = " ".join(broker_fragments)
+    folios = int(_to_decimal(block.cells[-2].text))
+    balance = _to_decimal(block.cells[-1].text)
     ac = DematAccount(
         name=broker,
         type=_full_type(type_word),

tests/test_demat_units.py

@@ -872,6 +872,13 @@ def test_summary_demat_row_4_cell(self):
         assert ac.folios == 12
         assert ac.balance == Decimal("10400929.50")
 
+        assert cdsl_p._is_summary_demat_row(block)
+        cdsl_ac, cdsl_key = cdsl_p._account_from_summary_row(block, owners=[])
+        assert cdsl_key == ("NSDL", "IN123456", "99998888")
+        assert cdsl_ac.name == "ACME BROKER LIMITED"
+        assert cdsl_ac.folios == 12
+        assert cdsl_ac.balance == Decimal("10400929.50")
+
     def test_summary_demat_row_5_cell(self):
         """5-cell variant: broker name and DP/Client line as separate
         cells (observed on CDSL rows in some NSDL CAS layouts)."""
@@ -890,6 +897,28 @@ def test_summary_demat_row_5_cell(self):
         assert ac.folios == 25
         assert ac.balance == Decimal("9734823.11")
 
+        assert cdsl_p._is_summary_demat_row(block)
+        cdsl_ac, cdsl_key = cdsl_p._account_from_summary_row(block, owners=[])
+        assert cdsl_key == ("CDSL", "11112222", "33334444")
+        assert cdsl_ac.name == "BETA BROKER LIMITED"
+        assert cdsl_ac.folios == 25
+        assert cdsl_ac.balance == Decimal("9734823.11")
+
+    def test_summary_demat_row_reconstructs_broker_split_across_cells(self):
+        block = _block(
+            _cell("CDSL Demat Account"),
+            _cell("EXAMPLE BROKER PRIVATE"),
+            _cell("LIMITED\nDP ID:11112222 Client ID:33334444"),
+            _cell("25"),
+            _cell("97,34,823.11"),
+            page=2,
+        )
+
+        ac, key = cdsl_p._account_from_summary_row(block, owners=[])
+
+        assert key == ("CDSL", "11112222", "33334444")
+        assert ac.name == "EXAMPLE BROKER PRIVATE LIMITED"
+
     def test_summary_demat_row_rejects_wrong_cell_count(self):
         # 3 cells: too short.
         block = _block(
@@ -899,6 +928,7 @@ def test_summary_demat_row_rejects_wrong_cell_count(self):
             page=2,
         )
         assert not nsdl_p._is_summary_demat_row(block)
+        assert not cdsl_p._is_summary_demat_row(block)
 
     def test_parse_bond_summary_row(self):
         """NSDL-flavour summary bonds row — discriminates frequency