PDF Text Extractor

PDF Text Extractor pulls the text out of public PDF files. Paste PDF URLs and get back, for each file, the plain text in reading order, clean Markdown with headings, lists and tables, every detected table as rows of cells, the web links inside the PDF, its bookmarks outline, the page count and the document metadata. Choose one row per PDF or one row per page. Pages with no text layer (scans) are flagged, so you know which files still need OCR.

It costs $2 per 1,000 PDFs, and PDFs that fail are free.

What does PDF Text Extractor return?

• Plain text (text): the words of the PDF in reading order, with line breaks as in the file and a form feed between pages. Two-column pages are read column by column.
• Markdown (markdown): headings (#, ##, ###) from the font sizes, bullet and numbered lists, tables as Markdown tables, and words split by a hyphen at a line end joined back. Running headers, footers and page numbers are left out, so the Markdown is ready for an LLM, a RAG pipeline or a search index.
• Tables (tables): each table with pageNumber, rowCount, columnCount and rows, an array of cell texts per row. Easy to load into a spreadsheet.
• Metadata: title, subject, keywords, creator, producer, creationDate, modificationDate, pdfVersion, language, isTagged, hasForm, pageCount and fileSizeBytes.
• Structure: links (web links in the PDF), outline (bookmarks with their level) and pages (size, rotation, word count and table count per page).
• Scan check: pdfType is text, scanned or mixed, and pagesWithoutText lists the pages that are images only.
• Status for every file: status is ok, or says why nothing was extracted (no_text_layer, password_protected, invalid_pdf, not_a_pdf, too_large, not_found, robots_disallowed, blocked, timeout and others), with a plain-language error.

How to extract text from a PDF, step by step

1. Open PDF Text Extractor and go to the Input tab.
2. In PDF URLs, paste one PDF address per line, for example https://example.org/annual-report-2025.pdf.
3. Pick Row per: Document gives one row with the whole text of each PDF; Page gives one row per page, which helps when an answer must cite a page or when you split text into chunks.
4. Keep Plain text, Markdown and Tables on, or turn off what you do not need.
5. Optional: turn on Save .txt and .md files to get a download link for each PDF's text.
6. Click Start. The example PDF takes a few seconds.
7. Open the Output tab. PDFs shows the status, pages, words and text of each file, Pages shows the page rows, and Metadata shows the document details. Download as JSON, CSV or Excel, or read the results through the Apify API.

How much does PDF text extraction cost?

You pay per PDF, and only for PDFs that gave text:

• pdf-extracted: $0.002 per PDF ($2 per 1,000 PDFs). This covers a PDF of up to 100 pages.
• extra-100-pages: $0.002 for each further 100 pages (or part of 100) of a long PDF.

Files with no text are not charged: scanned-only PDFs, password-protected files, broken links, pages that are not PDFs, files over the size limit and addresses that robots.txt does not allow all return a row with the reason and cost nothing.

Worked example: 1,000 PDFs of 20 pages, plus one 250-page report, cost 1,001 x $0.002 + 2 x $0.002 = $2.006. If 50 of the 1,000 are scans with no text layer, those 50 are free and the run costs $1.906.

Set a maximum cost on the run and PDF Text Extractor stops starting new PDFs when the next one would go over it.

Input example

{
    "pdfUrls": [
        "https://raw.githubusercontent.com/py-pdf/sample-files/main/026-latex-multicolumn/multicolumn.pdf"
    ],
    "rowPer": "document",
    "includeText": true,
    "includeMarkdown": true,
    "extractTables": true
}

Output example

{
    "rowType": "document",
    "url": "https://raw.githubusercontent.com/py-pdf/sample-files/main/026-latex-multicolumn/multicolumn.pdf",
    "fileName": "multicolumn.pdf",
    "status": "ok",
    "producer": "pdfTeX-1.40.21",
    "creationDate": "2024-01-03T08:38:26.000Z",
    "pageCount": 3,
    "pagesExtracted": 3,
    "pdfType": "text",
    "wordCount": 1075,
    "tableCount": 1,
    "text": "Two-Column Document with Lorem Ipsum\n\n...",
    "markdown": "# Two-Column Document with Lorem Ipsum\n\n...",
    "tables": [
        {
            "pageNumber": 3,
            "rowCount": 6,
            "columnCount": 5,
            "rows": [
                ["Country", "Population (millions)", "Area (km2)", "Capital", "Official Language"],
                ["Austria", "8.9", "83,879", "Vienna", "German"]
            ]
        }
    ],
    "charged": true
}

Use cases for a PDF to text converter

• Feed reports, papers, manuals and filings to an LLM or a vector database as Markdown.
• Pull tables from statistical releases, price lists and annual reports into a spreadsheet.
• Index a document library for search, with page numbers for citations.
• Sort a pile of PDFs into text PDFs and scans before paying for OCR.
• Read PDF metadata (dates, producer, language, page count) in bulk.

Bulk lists

Use a long list of PDF addresses from an Apify Dataset or from a file. Pick the Dataset under Dataset with the list and write the name of its field under Field in the dataset. Leave the field name empty to use the first text field. Or paste the web address of a .txt, .csv or .json file under List file URL. A .csv file uses its first column. A header row is skipped. The file can be up to 10 MB. Entries from the Dataset and the file are added to the ones typed under PDF URLs, and duplicates are removed. The limit under Maximum PDFs still applies. The price per PDF does not change.

Related Actors

• Document to Markdown Converter turns Word, PowerPoint, Excel, EPUB and other formats into Markdown, and finds document links on a web page.
• For scanned PDFs, use an OCR Actor on the pages listed in pagesWithoutText.

FAQ

Is it legal to extract text from PDFs? PDF Text Extractor reads only files at public addresses you give it, with no login. It checks each site's robots.txt, identifies itself honestly with a contact address, and stops contacting a site that answers 401, 403 or 429. It does not return author names from the metadata. You are responsible for having the right to process the documents you submit.

Does it do OCR? No. It reads the text layer that the PDF already has. Pages that are only images are listed in pagesWithoutText, and a PDF with no text at all returns no_text_layer and is free.

How big can a PDF be? Up to 200 MB and 5,000 pages per file. The defaults are 50 MB and 500 pages; a PDF with more pages returns the first pages and truncated: true. Give the run more memory (2 GB or more) for files over 50 MB. Very long text (over 1,000,000 characters) is saved as a file and linked in textFileUrl and markdownFileUrl.

Why is a table missing or split? Tables are found from the position of the text. Tables drawn as images, or with merged cells across many rows, may come back as plain lines. The text is still in text and markdown.

Something went wrong? Look at status and error in the row. If a PDF opens in your browser but fails here, open an issue on the Issues tab with the URL.