fix: support Indonesian currency format and fix NoneType error in pdfoxide parser

pdfparser/pdfoxide_parser.py

@@ -69,7 +69,7 @@ def parse_pdf_pdfoxide(path: str) -> Dict[str, Any]:
         # Extract transactions from all pages
         all_text = ""
         for page_num in range(page_count):
-            page_text = doc.extract_text(page_num)  # type: ignore[attr-defined] or ""
+            page_text = doc.extract_text(page_num) or ""  # type: ignore[attr-defined]
             all_text += page_text + "\n"
         transactions = extract_transactions(all_text)
 

pdfparser/utils.py

@@ -71,7 +71,8 @@
 _WHITESPACE_PATTERN: Pattern = re.compile(r"\s+")
 _NUMERIC_LINE_PATTERN: Pattern = re.compile(r"^[\d,.]+\s*$")
 _NUMERIC_ONLY_PATTERN: Pattern = re.compile(r"^[\d,.]*$")
-_AMOUNT_PATTERN: Pattern = re.compile(r"^[\d,]+\.\d{2}$")
+# Support both US (1,234.56) and Indonesian (1.234,56) formats
+_AMOUNT_PATTERN: Pattern = re.compile(r"^(?:[\d,]+\.\d{2}|[\d.]+\,\d{2})$")
 _USER_ID_PATTERN: Pattern = re.compile(r"^\d{6,8}$")
 
 # Summary section label patterns (compiled for extract_summary_totals)

tests/test_utils.py

@@ -110,6 +110,45 @@ def test_extract_transactions_keys_are_strings(self, transaction_text):
             for key in item.keys():
                 assert isinstance(key, str), f"Key '{key}' is not a string"
 
+    def test_extract_transactions_indonesian_format(self):
+        """Verify extraction of Indonesian number format (comma decimal)."""
+        text = """
+01/01/23 12:00:00
+TRANSFER IN
+1.000.000,00
+0,00
+10.000.000,00
+"""
+        transactions = extract_transactions(text)
+
+        assert len(transactions) == 1
+        txn = transactions[0]
+
+        # Verify parsing
+        assert txn["debit"] == "1.000.000,00"
+        assert txn["credit"] == "0,00"
+        assert txn["balance"] == "10.000.000,00"
+        assert txn["user"] == ""
+
+    def test_extract_transactions_us_format(self):
+        """Verify extraction of US number format (dot decimal)."""
+        text = """
+01/01/23 12:00:00
+TRANSFER IN
+1,000,000.00
+0.00
+10,000,000.00
+"""
+        transactions = extract_transactions(text)
+
+        assert len(transactions) == 1
+        txn = transactions[0]
+
+        assert txn["debit"] == "1,000,000.00"
+        assert txn["credit"] == "0.00"
+        assert txn["balance"] == "10,000,000.00"
+        assert txn["user"] == ""
+
 
 class TestTransactionDatePattern:
     """Tests for transaction date regex pattern using hypothesis."""