Skip to content

How to search the text inside PDFs and documents on your computer, offline

"I remember a phrase from the contract, not the file name." Searching by file name cannot answer that; searching by contents can. This guide is about doing it locally — on your own computer, without uploading your documents to a search service.

How ArchiveMind does it

When ArchiveMind scans a folder or drive, it reads the text out of the documents it finds — PDFs, Word and other office documents, plain text, and more — and keeps that text in its catalogue alongside the file's name and location. From then on, a search for a word or phrase finds the documents that contain it, and each result says why it matched: the file name, the text inside, or both. This works for documents on a drive that is no longer plugged in, because the text was read when it was.

Nothing leaves your computer. The catalogue is a file on your disk; the app makes no network connections.

Steps

  1. On the Scan tab, pick the folder or drive that holds the documents and click Start Scan. Text is extracted during the scan.
  2. When the scan is complete, type a word or phrase into Search Explorer. Use a word that would appear in the document, not a description of it ("invoice 2023", "lease agreement", a client's name).
  3. Open Why this matched on a result to see whether it was the name, the contents or both. Use the file-type pills (Documents, PDF) to narrow the list.
  4. Click Reveal in Finder / Show in Explorer or Open File to get to the document itself.

Search results for the word lantern, filtered to PDFs: the same invoice found on an unplugged drive (Offline) and on this computer (Online, with its first page as a thumbnail), each showing why it matched — file name, title, author, PDF text, duplicate

Scanned PDFs (a picture of a page, no text behind it)

A scanned PDF has no text to extract — it is an image of a page. ArchiveMind detects those and labels them, and since version 1.1.6 it can read them with OCR if Tesseract is installed on your computer (free and open source: brew install tesseract on a Mac; the Tesseract installer on Windows). Then tick Enable OCR on the Scan tab before a scan — the checkbox itself says whether Tesseract was found. For scanned documents that are already in your catalogue, tick Force re-extract as well, once. OCR is slow, about half a second per page, which is why it is off by default.

What it cannot do

  • Understand meaning. It finds the words you type (and their variants), not synonyms or paraphrases. "Agreement" does not find "contract".
  • Read protected files. A password-protected PDF or an encrypted archive stays closed; the file is catalogued by name only, and says so.
  • Read every format. Formats without a text extractor are catalogued by name and metadata; the file details say whether text was extracted.
  • Keep up with edits made since the scan. Edited documents are re-read on the next scan of that folder; until then the catalogue holds the text as it was.

See also