Add CSV to Markdown table conversion - fixes #1144

[erinshek]

Description

This PR adds support for converting CSV files to Markdown tables. The converter handles various CSV formats and edge cases while maintaining proper table structure.

Features

• Convert CSV files to properly formatted Markdown tables
• Support for text/csv and application/csv MIME types
• UTF-8 encoding with error replacement for invalid characters
• Preserve table structure with headers and data rows
• Handle edge cases:
  • Empty cells
  • Rows with mismatched column counts
  • Special characters in content

Example

Input CSV:

Name,Age,City
John,30,New York
Alice,25,London

Output Markdown:

| Name  | Age | City     |
|-------|-----|----------|
| John  | 30  | New York |
| Alice | 25  | London   |

Usage

Command Line

markitdown input.csv > output.md

Python Library

from markitdown import MarkItDown

converter = MarkItDown()
result = converter.convert("input.csv")
print(result.markdown)

Changes

• Add new CsvConverter class
• Register converter in MarkItDown class
• Add CSV converter to available converters
• Fix Azure Document Intelligence dependency handling

Testing

• Tested with various CSV formats
• Verified UTF-8 encoding handling
• Checked edge cases with empty cells and mismatched columns
• Validated Markdown table formatting

Related Issues

Closes #1144

[erinshek]

@microsoft-github-policy-service agree

[erinshek]

Hi, @afourney

Could you please take a look at the changes I’ve made? I tried to address the issue mentioned in #1144. It should now be possible to convert a CSV file to Markdown.

If you notice anything that could be improved, feel free to let me know—I'm happy to make adjustments.

[afourney]

This is pretty solid.

Note that #1171 also adds CSV support, but requires pandas (same as xlsx and xls currently), so I think I prefer basing csv support on this one here (#1176) for basic CSV support.

Only thing missing is a test vector added here (also include a test file in tests/test_files)
https://github.com/microsoft/markitdown/blob/main/packages/markitdown/tests/_test_vectors.py

Perhaps see how it is done in #1171.

Also, I would like to credit you both on the release notes.

EDIT: I went ahead and fixed these while also fixing the pre-commit format. No further action needed

[erinshek]

Thank you