PDF to Markdown converter

A PDF describes where marks are drawn on a page. It does not normally contain the same document outline that a Markdown file needs. Frisqoo rebuilds a useful outline from the text layer instead of treating every line as an unrelated string.

• Paragraphs joined from nearby lines in the same reading flow
• Headings inferred from larger or heavier text
• Bullet lists and numbered lists found at the start of a line
• Fenced code blocks when the PDF identifies a monospaced font
• Simple tables with repeated, aligned columns
• Exact page design, fonts, colours, backgrounds, and decorative shapes
• Pictures, chart artwork, handwritten notes, and words stored only inside images
• Guaranteed reading order for every magazine, brochure, or nested layout

Structured is the default. It adds Markdown syntax when the page gives enough evidence to identify a heading, list, code block, or table. It also joins nearby prose lines into paragraphs and repairs words split with a hyphen at a line break when the next line starts with a lowercase letter.

Plain keeps the detected reading order but does not add inferred structure. It is useful for unusual reports, forms, and heavily designed files where you would rather format the result yourself. Switching the mode updates the result in the workspace without reading the PDF again.

Reports often repeat a short document name, date, page number, or confidentiality label on every page. With Remove repeating headers and footers enabled, Frisqoo looks only near the outer page edges and removes short matching lines that occur across at least three pages. This conservative rule avoids treating a one-off title as page furniture.

Turn the option off when a repeated line carries information you need. You can also enable Add page comments. The comments mark boundaries without appearing as visible prose in most Markdown viewers.

Ordinary single-column documents follow top-to-bottom reading order. When a page contains two consistent text columns, the converter reads the left column before moving to the right. Full-width titles above the columns stay first. This avoids alternating between two unrelated sentences on the same horizontal line.

Simple tables are identified from repeated column starts across at least three rows. The first row becomes the Markdown header. A two-column block must also look like structured values before it becomes a table, which reduces the chance of turning normal prose columns into rows and cells. Merged cells, nested tables, and complex financial layouts can still need manual correction.

The Markdown box is editable because conversion is an informed reconstruction, not a source-format guarantee. Check heading levels, list boundaries, table headers, hyphenated words, and the order of sidebars or callouts before using the file in documentation, a knowledge base, or an automated workflow.

Pay particular attention to documents that mix portrait and landscape pages, place captions far from their pictures, use several small columns, or draw tables without regular alignment. Mathematical notation, footnote links, page references, and text positioned inside diagrams may need manual markup. The original PDF stays visible beside the result so you can compare the two without reopening the file elsewhere.

If a conversion will feed a search index, retrieval system, or language model, review a representative sample before processing a large batch. Confirm that section names, lists, and tables mean the same thing after conversion. Markdown makes the structure easier to process, but it cannot restore meaning that the PDF never encoded.

The tool accepts files chosen from your device. It does not accept a PDF URL, fetch linked resources, follow document links, open attachments, or run document actions. PDF.js reads the supplied bytes with dynamic evaluation disabled, and the Markdown preview is an editable text box rather than rendered document HTML.

Links written as text remain text in this release. Embedded files, JavaScript actions, videos, and interactive widgets are not copied into the Markdown output. This reduces the active-content surface and keeps the result predictable. It does not prove that every PDF is harmless, so the parser and its dependencies still need routine security updates and regression tests.

PDF.js loads only when you start the tool. Text is then read one page at a time, and finished page objects are cleaned up instead of keeping every rendered page in memory. Changing the output options reuses the extracted page data, so switching between structured and plain output does not parse the PDF again.

Small text PDFs usually finish quickly. Image-heavy files, complex fonts, long tables, and large page counts take more work, even when the images themselves are not exported. The 50 MB, 300 page, and 20 file limits prevent unbounded jobs, but available memory and speed still vary by browser and device. Keep the tab open until a batch finishes.

An image-only scan contains pixels rather than characters. If the file has no usable text layer, Frisqoo stops and offers OCR PDF as the next step. It does not invent Markdown from an empty extraction.

Dynamic XFA forms are also refused. Their visible fields are assembled by Adobe software and may not exist as normal page text. A damaged PDF that PDF.js cannot parse returns an error rather than a partial download. If the file is repairable, try Repair PDF before converting it.

A single PDF opens in the editor with its pages on the left and editable Markdown on the right. A file named report.pdf downloads as report.md. For a batch, each successful PDF keeps its base filename, and Download all as ZIP saves the set as frisqoo_pdf-to-markdown.zip.

Files are processed in sequence rather than rendering the whole batch at once. Up to 20 PDFs can be selected, with a 50 MB and 300 page limit for each file. Those limits make failures predictable and reduce memory pressure, but actual speed still depends on the document and the available device memory.