How does this count words in a PDF without uploading it?

The PDF is parsed entirely in your browser. The tool reads the file as bytes, inflates the FlateDecode content streams and extracts the text-showing operators, then counts words from the resulting text. No data leaves your device.

Why might the word count differ slightly from Adobe or Word?

PDFs store text as positioned glyphs, not flowing paragraphs, so word boundaries are inferred from spacing operators. Different readers reconstruct spaces and line breaks slightly differently, which can shift the count by a small margin on heavily formatted documents.

Does it work on scanned PDFs?

No. Scanned PDFs are images of pages with no embedded text layer, so there are no characters to count. This tool only counts real, selectable text. Run OCR first if you need to count scanned content.

What counts as a word?

A word is any run of non-whitespace characters separated by spaces, tabs or line breaks. The character count includes spaces; a separate count of characters excluding spaces is also shown.

Is there a file size limit?

There is no hard limit, but very large PDFs (hundreds of pages) are parsed in your browser's memory, so extremely big files may be slow on low-memory devices. Most documents process in well under a second.

PDF Word Count — Gera Tools

A PDF word counter tells you how many words, characters and pages a document contains — useful for translators billing per word, students checking essay length, and legal teams sizing disclosure bundles. Unlike most online counters, this tool never uploads your file: the PDF is decoded entirely in your browser, so confidential contracts and unpublished manuscripts stay on your machine.

How it works

A PDF is not a flat text file. Each page’s text lives inside a content stream — a sequence of drawing operators — and those streams are usually compressed with FlateDecode (zlib/deflate). To count words the tool:

Reads the file as raw bytes and locates every stream … endstream object.
Inflates any FlateDecode streams using the browser’s native DecompressionStream.
Scans the decoded operators for text-showing commands — Tj, TJ, ' and " — and pulls the string literals (the parts in parentheses ( )) and hex strings out of them.
Joins the strings into page text, inserting spacing where the layout operators imply a gap, then counts words, characters and sentences.

Words are runs of non-whitespace separated by whitespace. Sentences are approximated by counting terminal punctuation (., !, ?). Because PDFs position glyphs rather than store paragraphs, the count is an accurate estimate rather than a byte-perfect match to the authoring tool.

Tips and notes

If the count comes back as zero, the PDF is almost certainly a scan (image-only) with no text layer — there is nothing to count without OCR.
Encrypted or password-protected PDFs cannot be decoded in the browser; remove the protection first.
The per-page table is handy for spotting where the bulk of a document’s content sits, for example a long appendix.
Everything runs locally, so you can safely use it on sensitive or unpublished material.