Extract PDF pages

Page extraction makes a new PDF that contains only the pages you name. The source stays untouched. Unlike copying text into a new document, extraction copies complete PDF pages, including their visible layout, embedded graphics, and page dimensions. The output follows ascending page order because the selection is normalized and sorted before pages are copied. Entering `8,2,4` therefore produces pages 2, 4, and 8, not pages 8, 2, and 4.

Repeated numbers are included once. If you enter `2,2,1-3`, the result contains source pages 1, 2, and 3. A reversed range such as `7-5` is accepted and treated as pages 5 through 7, again in ascending order. Those rules make the selector forgiving, but they also mean it is not a page-reordering control. Use a dedicated organizer if the result must place page 8 before page 2 or duplicate a page intentionally.

The downloaded file is one PDF, not a ZIP and not one file per chosen page. This makes extraction a good fit for a chapter, evidence subset, appendix, signature page, or client-facing excerpt that should travel as a single document.

Page numbers are one-based, matching the numbers people normally see in a PDF viewer. Separate individual pages or ranges with commas. `1,3-5,9` selects page 1, pages 3 through 5, and page 9. Spaces around commas are fine. Letters, decimal numbers, open ranges such as `4-`, and labels such as `iv` are rejected. Every requested number must fall between 1 and the document's page count.

Printed page labels can differ from PDF positions. A report may show roman numeral `iii` on the page that the viewer counts as PDF page 3, then restart the printed body at page 1. The selector uses the physical PDF position reported by the browser, not the number printed in the footer. Check the viewer's page counter and a distinctive heading before entering a long range. If the table of contents says Chapter 2 starts on page 14 but the viewer shows position 18, use 18.

For a long request, build the expression from verified blocks. Suppose a 120-page audit needs the cover, executive findings, and two exhibits. Confirm each block in the reader, then enter something like `1,6-14,83-87,112`. Read the expression back against your notes before processing. One missing comma can turn two pages into a broad range.

Choose one readable PDF under 100 MB. After the browser displays its page count, open the source beside the tool and identify the first and last page of each needed segment. Enter the selection, run Extract PDF pages, and download the new file. The operation copies selected pages in their original dimensions, so a landscape exhibit remains landscape and a small receipt page does not become letter size.

Name the excerpt for its subject, not only for the extraction action. `lease-signature-pages.pdf` is easier to recognize than `document-extract-pdf-pages.pdf`. If the excerpt belongs to a case or project, include the approved reference number but avoid personal data in the filename unless policy allows it. Filenames can appear in email subjects, download histories, shared-drive logs, and backup indexes.

Keep a short mapping when the excerpt will be reviewed formally. For example: output page 1 came from source page 2; output pages 2 through 5 came from source pages 17 through 20. The tool does not stamp original page numbers onto the copy. A mapping prevents reviewers from confusing an excerpt's new position with its location in the source.

Open the downloaded PDF and compare its page count with the count implied by your selection. The expression `1,4-6,10` should yield five pages. Inspect the first and last chosen pages, then check each boundary where two nonadjacent source sections meet. A section may start with a heading on the previous page, or a table may continue onto a page you did not request. Extraction can follow your numbers exactly and still create an incomplete excerpt if the selection itself lacked context.

Search for names, section titles, or exhibit labels that must be present. Then search for confidential terms that must not appear. Text search is only one check because words inside scans may not be searchable. Visually inspect headers, footers, sticky-note annotations, margin comments, and page backgrounds. If the goal is disclosure, have a second person compare the excerpt against the approved page list.

Links and bookmarks deserve attention. A copied page can contain a link to a destination page that is no longer in the excerpt. Bookmarks may disappear or point somewhere unexpected after a new document is assembled. Test navigation that readers will use, especially a table of contents copied without all of its target chapters.

This tool does not crop a region from a page, pull embedded images, export tables, or turn selected text into another format. It copies whole pages. If a page contains one paragraph you may share and a client identifier you may not share, extracting that page does not solve the disclosure problem. Use a redaction tool that removes the underlying content, then verify the redacted output before extracting or distributing it.

Encrypted or malformed PDFs are rejected during the initial read. The tool does not ask for a password and does not bypass document restrictions. Some advanced PDF content can behave differently when pages are copied into a fresh document, including interactive forms, embedded files, multimedia, scripts, layers, digital signatures, and links to pages outside the selection. Signed PDFs need special caution because copying pages into a new file does not preserve the source document's cryptographic signature as a valid signature on the excerpt.

Very large or graphically dense documents can strain browser memory even below 100 MB. If processing fails, close other heavy tabs or use a desktop device. Splitting the source with an approved desktop application may help, but do not upload sensitive material to an unapproved service as a workaround.

The compatible PDF is handled locally in the browser and is not uploaded to a PWRKIT processing server. The resulting download still becomes a new copy on the device. Browser downloads, synchronized folders, automatic backups, endpoint monitoring, and shared accounts can expose that copy. Decide where it should be saved before working with health records, contracts, personnel files, financial statements, or discovery material.

Extraction reduces the number of pages, not necessarily the amount of hidden information. Selected pages may carry document metadata, invisible text, annotations, form values, attachments, or links that reveal names and locations. The absence of an unselected page does not prove every reference to it disappeared. Use sanitization and redaction software for controlled disclosure, and check document properties as well as visible pages.

An excerpt can also change meaning. A clause may depend on a definition several pages earlier, a chart may omit its methodology, or a signature page may not identify the agreement it signs. When context has legal or safety consequences, include enough surrounding material and label the copy as an excerpt. Preserve the full source and the approved selection record.

A shorter PDF is not automatically more accessible. Check the output with a screen reader or PDF accessibility checker when assistive technology matters. Page copying may not repair missing tags, incorrect reading order, unlabeled links, or image-only scans. The tool does not run OCR or create alternative text. If the selected page begins halfway through a tagged structure, the excerpt may also lack the heading context that made the source understandable.

Provide a descriptive filename and a clear document title in the publishing system. If you distribute a chapter excerpt, say which source pages it contains in nearby HTML or accessible text. Do not rely on color, page position, or the phrase 'see above' to explain why those pages were chosen. Readers who navigate by headings need a meaningful entry point, not just a stack of copied pages.

Interactive forms require keyboard testing after extraction. Tab through fields, inspect their accessible names, and submit only a test copy. A field can look present while its calculation, validation, or destination action depends on content that was left behind.

Choose extraction when a defined subset should become one new PDF. Remove PDF Pages starts from the opposite question: which pages should be excluded while everything else stays? For a 200-page manual where only pages 40 through 52 are needed, extraction is easier to audit. For that same manual with two accidental blank pages, removal is clearer because the intended output is almost the whole source.

Split PDF creates a ZIP containing one PDF for every source page. It does not use the selection expression and it does not group ranges into chapter files. Use it when each page must stand alone, such as a batch of one-page invoices. Merge PDF moves in the other direction by joining several PDFs. Compression addresses file size, while rotation changes page orientation. None of those operations redact content.

For a multi-step job, choose an order that leaves evidence. Redact first in suitable software, verify the redaction, extract the approved pages, then check the excerpt. Compress only after its content is final. Each saved stage should have a distinct filename so a reviewer can identify the source, sanitized copy, excerpt, and delivery version without guessing.

If the excerpt will replace a prior one, compare both page mappings before publication. A changed source revision can shift every later position even when the requested section title stays the same.