PDF Text, Tables & Metadata Extractor (from URLs)

Give it a list of PDF links and get back clean, structured data: the full text, the text of every page, tables as rows of cells, every link and email address in the document, and the file's metadata (title, author, creator, creation and modification dates, page count, size). One row per PDF, ready for JSON, CSV, Excel, a database, or an AI pipeline.

No browser, no proxy, no API key. Failed downloads, non-PDF links, oversized files and scanned (image-only) PDFs are returned as rows with an explanation and are not charged.

What you can use it for

• AI and RAG pipelines: feed reports, manuals, filings and papers into embeddings or an LLM, with page numbers kept for citations
• Research and monitoring: pull the text of government reports, annual reports, white papers or academic papers on a schedule
• Finance and data teams: get the number grids out of statistical releases and financial statements as tables
• Lead and contact research: collect the emails and links printed in brochures, price lists and catalogs
• Document audits: check page counts, authors, producers and dates across thousands of files

Why this one

• Page-by-page text with character counts, plus the full text in one field
• Tables two ways: ruled tables (with lines) and whitespace-aligned number grids, the kind used in statistical releases and financial statements
• Page ranges such as 1-3,7 or -2 (the last two pages), so you only pay compute for what you need
• Scanned-file detection: a PDF with no text layer is flagged likelyScanned: true and is free, instead of returning an empty row you pay for
• Password PDFs supported, metadata dates normalized to ISO 8601
• Pay only for PDFs that produced text

Input

Field	What it does	Default
pdfUrls	Direct links to PDF files, one per line	required
pages	Page selection, e.g. 1-3,5 or -2; empty = all	all
includePageText	Adds the pages array (page number, text, characters)	true
extractTables	Detect tables and return them as rows of cells	false
tableStrategy	auto (ruled tables, then number grids), lines, or text	auto
extractLinksAndEmails	Collect hyperlinks and email addresses	true
maxCharsPerPdf	Truncate text per PDF (0 = no limit)	0
maxFileSizeMb	Skip larger files (free error row)	40
password	Password for protected PDFs	none
maxResults	Stop after this many PDFs	1000

Example input:

{
  "pdfUrls": ["https://www.irs.gov/pub/irs-pdf/fw9.pdf", "https://arxiv.org/pdf/1706.03762"],
  "pages": "1-3",
  "extractTables": true
}

Output

One row per PDF:

Field	Meaning
url, finalUrl	The link you gave and where it ended up after redirects
ok, error, warning	Whether the PDF was read; why not; a note such as "looks scanned"
fileName, sizeBytes	File name and size
pageCount, pagesExtracted	Pages in the file and pages read
title, author, subject, keywords, creator, producer	Document metadata
createdAt, modifiedAt	Metadata dates in ISO 8601
encrypted, likelyScanned	Password-protected; no text layer
text, wordCount, charCount	Full text and its size
pages	[{page, text, chars}]
tables	[{page, index, method, rows: [[cell, ...], ...]}]
links, emails	Unique links and email addresses found
processingSeconds	Time spent on this file

Sample (shortened):

{
  "url": "https://www.irs.gov/pub/irs-pdf/fw9.pdf",
  "ok": true,
  "fileName": "fw9.pdf",
  "pageCount": 6,
  "pagesExtracted": 3,
  "title": "Form W-9 (Rev. March 2024)",
  "createdAt": "2024-03-06T08:18:13",
  "likelyScanned": false,
  "wordCount": 3595,
  "text": "W-9\nRequest for Taxpayer\nIdentification Number and Certification ...",
  "pages": [{"page": 1, "text": "W-9 ...", "chars": 5659}],
  "tables": [{"page": 1, "index": 0, "method": "lines", "rows": [["...", "..."]]}],
  "links": [],
  "emails": []
}

How much does it cost?

Pay per result: $2.50 per 1,000 PDFs extracted on the Free plan (a small one-time start fee per run), with automatic discounts on paid Apify plans. Failed downloads, non-PDF links, oversized files and scanned PDFs without text are free.

Worked examples:

• 50 annual reports for a research project: 50 × $0.0025 = $0.13
• 1,000 papers for a RAG index: $2.50
• 10,000 manuals with 300 that turn out to be scanned images: 9,700 × $0.0025 = $24.25 (the 300 scanned files are free)

Platform compute is included in the price.

FAQ

Can it read scanned PDFs? It detects them (likelyScanned: true, with a warning) and does not charge for them. It does not run OCR, so an image-only page returns little or no text.

Why is a table split into odd cells? Tables without ruling lines are rebuilt from the position of words on the page. Number columns come out reliably; long labels can be split across two cells. Use tableStrategy: lines for ruled tables only, or text to force the whitespace method on every page.

Some links return "HTTP 403". A few sites refuse automated downloads. Those rows are free. Download the file yourself and host it anywhere with a direct link, or use a link from a site that allows downloads.

Is it legal? It downloads the files you link to, the same way a browser does, and reads them. You are responsible for having the right to use the documents you process (public reports and documents you own are the usual case).

How big can a PDF be? Up to maxFileSizeMb (40 MB by default, 200 MB maximum). For very long documents, use pages to read only the part you need.

Does it keep my files? No. Files are processed in memory during the run and only the extracted data is written to your dataset.

Integrations

Run it from the Apify API, a schedule, or any integration Apify supports: Make, Zapier, n8n, Google Sheets, webhooks, and AI agents through the Apify MCP server. The dataset downloads as JSON, CSV, Excel, XML or HTML.

Changelog

• 0.1 (October 2026): first release: text, page text, tables (ruled and number grids), links, emails, metadata, page ranges, password PDFs, scanned-file detection.