Extract Text
Open the app: Extract text from a PDF → (products.documentize.app)
In one sentence
Extract Text pulls the words that are already inside a PDF out into plain .txt files, in the formatting style you choose.
What makes it unique
- It reads the PDF's existing text; it does not "look at" pictures. That makes it fast and exact, with no recognition mistakes.
- It offers three formatting modes (Pure, Raw, Flatten) that decide how lines and fragments are put together.
- With several PDFs it also produces a combined text file and gives a per-file summary (word counts, previews).
How it differs from its look-alikes:
- OCR Image: input is pictures, output is text read from pixels.
- Searchable: input is a scan, output is a PDF with a hidden text layer.
- Parser: extracts text and images, attachments, bookmarks and more in one ZIP.
When to use it / when not to
- Use it to copy a report into an editor, to feed text to another tool, to count words, or to quickly check what a PDF really contains.
- Do not use it on a scanned PDF: the result will be empty or nearly so. Run Searchable first, then Extract Text. (The site's FAQ mentions an OCR option for scans, but the page shows only the formatting control; use Searchable to be safe, verify on the live app.)
- Do not use it when you need tables or positions; try Parser or Searchable with Excel output.
Before you start
- Check that the PDF has selectable text (try to highlight a word in your viewer).
- Remove the password with Unlock if the file is locked.
- Decide how you will use the text: reading (Pure), maximum fidelity to the file's own order (Raw) or position-based (Flatten).
Step by step
- Open Extract Text and upload one or more PDFs ("Upload a PDF to extract text").
- Under Select Formatting Mode choose Pure, Raw or Flatten.
- Press Extract.
- Download the .txt (or the ZIP with several). Extract again starts a new job.
📷 Screenshot placeholder: the formatting mode dropdown
Options explained
| Option | What it does | Default | Recommended |
|---|---|---|---|
| Pure | Text with a few clean-up routines for readable results | selected | Everyday use |
| Raw | Text without any formatting, exactly as stored in the file | - | When Pure joins or splits lines in a way you do not want |
| Flatten | Puts text fragments in order by their coordinates on the page | - | Forms, columns or layouts where reading order looks wrong |
What you get
- One .txt per PDF, named after the PDF.
- With more than one PDF, an extra combined_extracted_text.txt that lists each file with its mode and word/line statistics, followed by its text. Failed files are listed in it with their error instead of stopping the run.
- The page also receives summary figures (words, characters, success/error counts).
- The original PDF is not modified.
Quality & limits
- Lossless for text: nothing is guessed, but fonts, images, tables and layout are not kept.
- Text hidden in images is not found. Odd fonts or damaged encodings can produce strange characters.
- Each file is handled on its own, so a free-plan time limit can stop a large batch and offer a continuation later (about 12 hours of cool-down by default). Results are kept for about two days.
Works well with
- Use before: Searchable (scans), Unlock, Splitter (just the pages you want).
- Use after: your editor for plain text, Search for a marked-up PDF. For understanding the content use Ask PDF or Abstracts.
- Recipes: Digitise an archive, To spreadsheet.
Common mistakes & fixes
- "The .txt is empty" → the PDF is a scan; use Searchable first.
- "Lines are broken in strange places" → try Raw or Flatten.
- "Columns are mixed up" → try Flatten, or Parser.
- "I got a ZIP, not one file" → several PDFs give one text file each plus the combined file.
Same thing elsewhere
- CLI plugin: Documentize.ExtractText.
- MCP tool: ExtractText (with a text-formatting-mode setting).
- See Beyond the browser.
Try it
Take any text PDF of several pages. Run Extract Text in Pure mode, then again in Raw and in Flatten. Open the three results side by side and note where line breaks and word order differ. Then try a scanned PDF to see an empty result, and fix it with Searchable.