feat(file-based): Add Calamine-first with Openpyxl fallback for Excel parser

Implements a fallback mechanism for Excel file parsing to handle edge cases
where Calamine fails (e.g., invalid date values like year 20225). The parser
now tries Calamine first for performance, then falls back to Openpyxl if
Calamine encounters an error.

Changes:
- Modified open_and_parse_file() to implement try-catch with fallback logic
- Added logger parameter to log when fallback is triggered
- Added openpyxl as optional dependency in pyproject.toml
- Added openpyxl to file-based extras list

This resolves crashes in Google Drive source when processing large numbers
of Excel files with malformed data, allowing syncs to complete successfully
instead of failing entirely.

Fixes: airbytehq/oncall#10097
Co-Authored-By: unknown <>

Author: devin-ai-integration[bot]

airbyte_cdk/sources/file_based/file_types/excel_parser.py

@@ -64,7 +64,7 @@ async def infer_schema(
         fields: Dict[str, str] = {}
 
         with stream_reader.open_file(file, self.file_read_mode, self.ENCODING, logger) as fp:
-            df = self.open_and_parse_file(fp)
+            df = self.open_and_parse_file(fp, logger, file.uri)
             for column, df_type in df.dtypes.items():
                 # Choose the broadest data type if the column's data type differs in dataframes
                 prev_frame_column_type = fields.get(column)  # type: ignore [call-overload]
@@ -92,7 +92,7 @@ def parse_records(
         discovered_schema: Optional[Mapping[str, SchemaType]] = None,
     ) -> Iterable[Dict[str, Any]]:
         """
-        Parses records from an Excel file based on the provided configuration.
+        Parses records from an Excel file with fallback error handling.
 
         Args:
             config (FileBasedStreamConfig): Configuration for the file-based stream.
@@ -111,7 +111,7 @@ def parse_records(
         try:
             # Open and parse the file using the stream reader
             with stream_reader.open_file(file, self.file_read_mode, self.ENCODING, logger) as fp:
-                df = self.open_and_parse_file(fp)
+                df = self.open_and_parse_file(fp, logger, file.uri)
                 # Yield records as dictionaries
                 # DataFrame.to_dict() method returns datetime values in pandas.Timestamp values, which are not serializable by orjson
                 # DataFrame.to_json() returns string with datetime values serialized to iso8601 with microseconds to align with pydantic behavior
@@ -181,14 +181,30 @@ def validate_format(excel_format: BaseModel, logger: logging.Logger) -> None:
             raise ConfigValidationError(FileBasedSourceError.CONFIG_VALIDATION_ERROR)
 
     @staticmethod
-    def open_and_parse_file(fp: Union[IOBase, str, Path]) -> pd.DataFrame:
+    def open_and_parse_file(
+        fp: Union[IOBase, str, Path],
+        logger: Optional[logging.Logger] = None,
+        file_uri: Optional[str] = None,
+    ) -> pd.DataFrame:
         """
-        Opens and parses the Excel file.
-
-        Args:
-            fp: File pointer to the Excel file.
+        Opens and parses the Excel file with Calamine-first and Openpyxl fallback.
 
         Returns:
             pd.DataFrame: Parsed data from the Excel file.
         """
-        return pd.ExcelFile(fp, engine="calamine").parse()  # type: ignore [arg-type, call-overload, no-any-return]
+        try:
+            return pd.ExcelFile(fp, engine="calamine").parse()  # type: ignore [arg-type, call-overload, no-any-return]
+        except Exception as calamine_exc:
+            if logger:
+                logger.warning(
+                    "Calamine parsing failed for %s, falling back to openpyxl: %s",
+                    file_uri or "file",
+                    str(calamine_exc),
+                )
+
+            try:
+                fp.seek(0)  # type: ignore [union-attr]
+            except (AttributeError, OSError):
+                pass
+
+            return pd.ExcelFile(fp, engine="openpyxl").parse()  # type: ignore [arg-type, call-overload, no-any-return]

pyproject.toml

@@ -73,6 +73,7 @@ pdf2image = { version = "1.16.3", optional = true }
 pyarrow = { version = "^19.0.0", optional = true }
 pytesseract = { version = "0.3.10", optional = true }  # Used indirectly by unstructured library
 python-calamine = { version = "0.2.3", optional = true }  # TODO: Remove if unused
+openpyxl = { version = "^3.1.0", optional = true }
 python-snappy = { version = "0.7.3", optional = true }  # TODO: remove if unused
 tiktoken = { version = "0.8.0", optional = true }
 nltk = { version = "3.9.1", optional = true }
@@ -120,7 +121,7 @@ deptry = "^0.23.0"
 dagger-io = "0.19.0"
 
 [tool.poetry.extras]
-file-based = ["avro", "fastavro", "pyarrow", "unstructured", "pdf2image", "pdfminer.six", "unstructured.pytesseract", "pytesseract", "markdown", "python-calamine", "python-snappy"]
+file-based = ["avro", "fastavro", "pyarrow", "unstructured", "pdf2image", "pdfminer.six", "unstructured.pytesseract", "pytesseract", "markdown", "python-calamine", "openpyxl", "python-snappy"]
 vector-db-based = ["langchain_community", "langchain_core", "langchain_text_splitters", "openai", "cohere", "tiktoken"]
 sql = ["sqlalchemy"]
 dev = ["pytest"]