Markdown table generator: from CSV, JSON or pasted cells

A GitHub Flavored Markdown table has a header row, a delimiter row of hyphens and any number of data rows. Pipes (|) separate the columns, and the pipes at the start and end of a line are optional.

| Column | Alignment | Pipe in cell |
| :--- | :---: | ---: |
| Left | Centered | a \| b |
| Text | More text | 100 |

Colons on the delimiter row set the alignment: :--- aligns left, :---: centers and ---: aligns right. Escape a literal pipe inside a cell as \| so the parser does not split the cell in two.

Writing pipes by hand works for a few lines; for real data, convert what you already have. rows.page reads it inside your browser with DuckDB-WASM, so nothing is uploaded unless you click Share.

1. Open rows.page and click Paste data, paste cells copied from Excel or Google Sheets (they arrive tab-separated) or raw CSV, TSV, JSON or JSONL, and click Open. Or drop a .csv, .txt, .tsv, .tab, .json, .jsonl, .ndjson, .ldjson, .parquet or .pq file onto the page.
2. Trim the table: hide columns in the Columns panel, click a column header to sort, or type a filter like country:NL, price:>20 or -name:* (is empty).
3. For more control, run a DuckDB SQL query that starts with SELECT, WITH or FROM against the table data to pick, rename or aggregate columns.
4. Click Export and choose Markdown.

The export writes the first 1,000 rows of your current view, including its filter, sort, hidden columns or SQL result. Nulls become empty cells, a | inside a value is escaped as \|, line breaks inside a value become spaces, and list columns are written as JSON text. The delimiter row is plain ---, so add colons yourself if you want alignment.

rows.page does not open .xlsx or .xls workbooks: copy the cells and use Paste data, or save the sheet as CSV. You can't edit cells either, so fix typos in the source or rewrite values in a SELECT before exporting.

• Cut the columns: a table with dozens of columns wraps in an editor and scrolls sideways when rendered, so hide what the reader doesn't need or SELECT only the columns to show.
• Right-align numbers: change --- to ---: in the delimiter row so values are easier to compare.
• Keep cells short: nested JSON opens as dot-path columns like user.address.city (up to 8 levels deep, 200 child fields per object), and arrays export as JSON text, so drop or unnest() them in SQL if they make rows too wide.
• Don't bother padding: the pipes don't need to line up for the table to render, which only matters to people reading the raw .md file.

Pipe tables come from GitHub Flavored Markdown, not the original Markdown, so support depends on the parser your platform uses. GitHub renders them in README.md files, issues, pull requests, discussions and gists; check other targets before pasting a large table.

A table with more than a few dozen rows makes an issue or pull request long to scroll. Export a small sample as Markdown and link the full data instead: click Share in rows.page, or run curl -T data.csv https://rows.page, to get a link where readers can filter and sort every row.

Markdown tables are flat grids. There is no syntax for merged cells (colspan or rowspan), footers or per-cell styling, and each row must stay on one line because a newline ends the row.

For a line break inside a cell on GitHub, use <br>. In rows.page, the Markdown export stops at the first 1,000 rows and has no format options, while CSV, Parquet, JSON and JSONL exports include every matching row of a filtered view, or up to 1,000,000 rows of a SQL result.