HTML Table to CSV

How it works

Paste HTML containing a table element or load a local UTF-8 .html file. Select Inspect HTML tables to parse the source as data using parse5. The parser does not create a browser document, run scripts or request image, frame, stylesheet or link URLs from the input. Table captions or IDs identify the choices; only the selected top-level table is exported. Tables inside script, style, template, noscript, iframe, object, SVG or MathML content are excluded. A selected table containing a nested table is rejected: copy the desired inner table separately instead of silently mixing independent grids.

Rows follow parsed source order across thead, tbody and tfoot groups. Every td or th cell is retained as text; the first row is not inferred as a header or discarded. HTML character references such as & and " are decoded. Inline tag text is joined, ordinary HTML whitespace collapses, br creates a newline, and common block elements create line boundaries. Repeated excess line boundaries are reduced. CSS is not computed, so hidden source text can remain in the export. Links contribute their text, not their URL; images and form-control values are not extracted. Scripts, styles, embedded documents and the other excluded element types contribute no text. This is a defined source-text conversion, not a pixel-perfect copy of a rendered page.

Colspan and rowspan must be explicit positive decimal integers from 1 to 100. Rowspan zero, spans extending beyond their row group, overlapping merged cells and expanded grids beyond the limits stop conversion with an error. Keep text once places a merged cell's value only in its upper-left position. Repeat text fills each covered position with that value. Shorter unmerged rows are padded with blank cells to make rectangular CSV; the result reports the number of these padded blanks and merged cells so this structural change is visible.

Choose comma, semicolon or tab as the delimiter. Fields containing the delimiter, quotes or line breaks are quoted, and embedded quotes are doubled. Downloads are UTF-8 without a BOM, with CRLF record endings and a final record ending. Semicolon and tab options use the same escaping convention as comma CSV. The preview is limited to eight rows and six columns and abbreviates long cell text; the full download contains every supported row and column of the selected table.

Spreadsheet-formula protection is on by default. It prefixes an apostrophe to cell text beginning, after ASCII spacing or line breaks, with =, +, − as the ASCII hyphen, or @. This changes text, including some negative numbers. Disable it when preserving that text is necessary and you trust the receiving application. It is a mitigation rather than a guarantee across every spreadsheet application. Import identifier columns as text to avoid a spreadsheet removing leading zeros or changing dates.

Input is limited to 1 MiB of UTF-8, 50,000 parsed nodes, 100 nesting levels and 50 top-level tables. The selected table supports up to 2,000 rows, 100 expanded columns and 50,000 expanded cells, with an 8 MiB output cap. Invalid UTF-8, NUL characters and unpaired Unicode surrogates are rejected. HTML parsing follows normal HTML repair rules; this is not a strict HTML syntax validator. Review the preview when malformed markup may have changed the grid.

For example, Name/Qty headers and a Tea & coffee cell spanning two data rows export as Name,Qty then Tea & coffee,2 then ,3 using the keep-once policy. Repeat text would put Tea & coffee in both data rows. No value is inferred as a number or date.

Editing the source, loading another file or changing a table or output option clears the old result. Late reads and parser imports cannot overwrite newer edits. The tool processes source and results in browser memory and does not submit their contents. Exported CSV may contain confidential information; review it before sharing.

Common use cases

• Converting a saved report table into reviewable CSV
• Choosing one table from a document containing several grids
• Making merged headings explicit before a spreadsheet import
• Extracting text from table markup without executing embedded resources

Frequently asked questions

Standards and limitations

• WHATWG HTML: tables and cell spans
• parse5: parsing HTML fragments as a tree
• RFC 4180: CSV records and quoting
• OWASP: spreadsheet formula injection