Extract PDF pages containing a word
Extracting PDF pages by keyword searches the document's selectable text for a word or phrase, case-insensitively, and saves a new PDF containing only the pages that match — or, in inverted mode, only the pages that do not. This page runs that filter directly in your browser: type a term, see which pages hit, and download the filtered file. The PDF is never uploaded.
How it works
A PDF page does not store sentences; it stores positioned text runs — short fragments placed at coordinates by drawing operators. To search, the tool first extracts every page's text runs with a PDF renderer's text engine, concatenates them in reading order with spaces, and lowercases the result. Your search term is lowercased the same way, so matching is case-insensitive by construction. A page counts as a match if the term appears anywhere in that assembled string, whether in body text, a header, or a table cell.
The matched page numbers then drive a structural copy: each matching page object is copied into a new document together with the fonts, images, and graphics streams it references. This is the same lossless operation as extracting pages by number — nothing is rasterized, so text stays selectable and file size stays proportional to the pages kept. With the invert option enabled, the copy list is simply complemented: pages without the term are kept instead, which turns the tool into a keyword-based page remover.
The one hard requirement is a text layer. Scanned pages are pictures of text, and a picture contains nothing the matcher can read — the search will legitimately report zero hits on a pure scan. Run the file through the searchable PDF (OCR) tool first to add an invisible text layer, then filter. The same page-level matcher powers two sibling tools: highlight text in PDF marks every occurrence instead of extracting, and redact by search blacks the occurrences out permanently.
Illustrative example: pulling one vendor out of a 90-page invoice batch
Illustrative scenario, not a recorded benchmark. File sizes, timings and results depend on the input and device. Try the reproducible samples.
An accountant has a 90-page, 6.8 MB PDF containing a quarter's worth of scanned-then-OCRed invoices from many vendors, and needs only the ones from "Northwind Traders". She types northwind into the search box. The matcher reports hits on 11 pages: 7, 8, 19, 33, 34, 35, 51, 60, 72, 73, and 88.
Choosing "keep matching pages" produces an 11-page PDF of about 0.9 MB — the invoices, in their original order and quality, ready to forward. Had she instead needed everything except that vendor, the invert option would have produced the complementary 79-page file. The whole pass over 90 pages took roughly six seconds, dominated by text extraction.
Frequently asked questions
Why does my search find nothing in a scanned PDF?
Scanned pages are images and contain no selectable text to search. Run the file through an OCR tool such as our searchable PDF tool first to add a text layer, then extract by keyword.
Is the keyword search case-sensitive?
No. Both the page text and your search term are lowercased before comparison, so "Invoice", "INVOICE", and "invoice" all match the same pages.
Can I remove matching pages instead of keeping them?
Yes. An invert option flips the selection so the output contains only the pages that do not mention the term. This is useful for stripping boilerplate pages such as fax cover sheets.
Can I search for a phrase with spaces in it?
Yes, multi-word phrases are matched as a whole against the page's assembled text. A phrase split across a line break may not match if the extracted text lacks a space at the break, so shorter distinctive terms are more reliable.
Does extraction change the quality of the kept pages?
No. Matching pages are copied structurally into the new file with their fonts, images, and vector content intact, exactly as a manual page extraction would. Nothing is re-rendered or recompressed.