fix: wrap MD/TXT content in JSON element when output_format is markdown_json

When output_format='markdown_json', MD and TXT files now return a JSON array
with a single NarrativeText element instead of raw text, ensuring consistent
output format across all file types.

Co-Authored-By: Ryan Waskewich <ryan.waskewich@airbyte.io>

Author: devin-ai-integration[bot]

airbyte_cdk/sources/file_based/file_types/unstructured_parser.py

@@ -233,6 +233,10 @@ def _read_file(
         if filetype in {FileType.MD, FileType.TXT}:
             file_content: bytes = file_handle.read()
             decoded_content: str = optional_decode(file_content)
+            if format.output_format == "markdown_json":
+                return json.dumps(
+                    [{"type": "NarrativeText", "text": decoded_content, "metadata": {}}]
+                )
             return decoded_content
         if format.processing.mode == "local":
             elements = self._read_file_locally_elements(

unit_tests/sources/file_based/file_types/test_unstructured_parser.py

@@ -800,3 +800,40 @@ def test_parse_records_markdown_json_remote(
     # Verify content is valid JSON matching the API response
     content = json.loads(records[0]["content"])
     assert content == json_response
+
+
+@patch("unstructured.partition.pdf.partition_pdf")
+@patch("unstructured.partition.pptx.partition_pptx")
+@patch("unstructured.partition.docx.partition_docx")
+@patch("airbyte_cdk.sources.file_based.file_types.unstructured_parser.detect_filetype")
+def test_parse_records_markdown_json_md_file(
+    mock_detect_filetype,
+    mock_partition_docx,
+    mock_partition_pptx,
+    mock_partition_pdf,
+):
+    """Test that MD/TXT files return a JSON element array when output_format is markdown_json."""
+    stream_reader = MagicMock()
+    mock_open(stream_reader.open_file, read_data=b"# Hello World\n\nSome text content")
+    fake_file = RemoteFile(uri="path/to/file.md", last_modified=datetime.now())
+    logger = MagicMock()
+    config = MagicMock()
+    config.format = UnstructuredFormat(
+        skip_unprocessable_files=False,
+        output_format="markdown_json",
+    )
+    mock_detect_filetype.return_value = FileType.MD
+
+    records = list(
+        UnstructuredParser().parse_records(config, fake_file, stream_reader, logger, MagicMock())
+    )
+    assert len(records) == 1
+    assert records[0]["document_key"] == "path/to/file.md"
+    assert records[0]["_ab_source_file_parse_error"] is None
+    # Verify content is valid JSON with a single NarrativeText element
+    content = json.loads(records[0]["content"])
+    assert isinstance(content, list)
+    assert len(content) == 1
+    assert content[0]["type"] == "NarrativeText"
+    assert content[0]["text"] == "# Hello World\n\nSome text content"
+    assert content[0]["metadata"] == {}