What gets extracted — and what doesn't

A PDF can contain text in two very different ways. The first is actual text data — characters stored with their positions, fonts, and encoding. That's what this tool extracts. The second is text that appears in images — a scanned page, a screenshot embedded in the file, a photograph of a page. That kind of text is invisible to extraction tools because it's just pixels, not characters.

Formatting doesn't carry over. Headings, bold text, italics, columns, tables — none of that survives extraction. What you get is a flat stream of plain text, page by page. Headers and footers usually get mixed in too. The output is raw material, not a formatted document.

The scanned PDF problem

This is the most common reason text extraction "doesn't work." If you open the PDF, highlight text with your cursor, and can copy it — extraction will work. If you try to highlight text and nothing gets selected, the document is scanned and the content is stored as images. No extraction tool can pull text from that without OCR (optical character recognition), which is a separate process.

What you can do with extracted text

• Search or grep through the content without opening the PDF in a viewer
• Feed it into a language model, translation service, or summarisation tool
• Import it into a spreadsheet or database after light cleanup
• Index document contents for full-text search in a custom application