table-extract

Extract structured tables from HTML or CSV sources into normalized rows and columns.

Why install it

Research workflows frequently encounter useful data trapped in tables. This tool gives agents a portable table extraction primitive without requiring a full scraping stack.

Inputs

• source_type: html, csv, or auto
• path: optional local file path for the source
• html_text: optional raw HTML content
• csv_text: optional raw CSV content
• table_index: zero-based table index for HTML extraction
• header_row: whether the first row should be treated as the header

Outputs

• tables: extracted tables with columns and rows
• detected_count: number of tables detected in the source
• warnings: extraction warnings
• metadata: summary metadata about the extraction

Local development

python -m unittest discover -s tests -p 'test_*.py'

Example invocation

python -u table_extract/__main__.py < input.json

With input.json containing:

{
  "html_text": "<table><tr><th>Name</th></tr><tr><td>Ada</td></tr></table>",
  "source_type": "html"
}