A deliberately small table dialect
The parser expects single-line pipe-table cells, at least two columns, and a separator row with at least three hyphens per cell. Alignment colons are accepted. Outer border pipes are optional. Fenced code blocks are excluded, so a documentation example is not accidentally exported as a real table.
Escaped pipes and pipes within balanced backtick code spans belong to the cell. Multiple-backtick delimiters are supported. This is an intentional convenience dialect: some Markdown implementations require escaping pipes even within code. Do not use this extractor as a claim of complete GFM or CommonMark conformance.
| Expression | Note |
| --- | --- |
| `A|B` | draft \| final |
# Two cells: `A|B` and draft | finalRefuse ambiguous repairs
A body row with the wrong number of cells raises an error. The tool does not guess which value was omitted or silently discard a trailing column. It does not support merged cells, HTML tables, multiline cells or tables nested in blockquotes/lists. Repair such source data explicitly, or use another format.
Preserve values, then choose a safety policy
Default CSV preserves inline markup and numeric strings, apart from surrounding cell whitespace and Markdown pipe escaping. No locale conversion occurs. --text removes only simple code/link wrappers; it does not sanitize HTML or render a full document.
If a cell begins with a formula-like prefix, CSV quoting alone may not prevent spreadsheet evaluation. --spreadsheet-safe adds an apostrophe to values whose first non-whitespace character is =, +, - or @. That includes negative numeric strings: -20 becomes '-20. Use this option knowingly and keep a raw copy for reconciliation. It is a precaution, not a guarantee for every spreadsheet application.
The exporter cannot establish whether a number was extracted correctly. Compare important cells with the source before any later calculation.
If a cell is wrong because of an earlier PDF extraction, repair or repeat that upstream step before exporting again. The related KolmoPDF PDF-to-Markdown converter is a separate cloud option that uploads the PDF. This local table parser cannot read the PDF or certify that the extracted figure matches its source.
← Field notes