PDF Text Extractor — Extract Text from Any PDF File

PDF Text Extractor is an Apify Actor that extracts text from PDF files. It supports direct URLs and Google Drive / Dropbox / OneDrive share links. For PDFs that have no embedded text (scanned documents, image-based PDFs), an optional OCR fallback powered by Tesseract renders each image-only page and reads the text from the rendered image.

---

🚀 Key Features

• Direct URL support — Fetch and extract text from any publicly accessible PDF URL
• Google Drive — Auto-converts Google Drive share links to direct download URLs
• Dropbox & OneDrive — Converts Dropbox and OneDrive share links automatically
• OCR fallback — Tesseract OCR on pages with no embedded text (scanned / image PDFs)
• Multi-language OCR — 100+ languages supported via Tesseract trained data
• Page limiting — Optionally cap extraction at a specific number of pages
• Page labels in output — When maxPages is set, each page's text is prefixed with [Page N]
• Retry logic — Up to 3 attempts with 1.5× exponential backoff on network failures
• Rotating user agents — Uses randomised browser user-agent headers to avoid blocks
• Pay-per-event pricing — Pay only for successful extractions
• Structured JSON output — Extracted text plus metadata (page count, file size, source type, OCR flag, timestamp)

---

📥 Input Parameters

Parameter	Type	Required	Default	Description
pdfUrl	string	✅ Yes	—	URL of the PDF. Accepts direct URLs, or Google Drive / Dropbox / OneDrive share links.
maxPages	integer	No	0	Maximum pages to extract. 0 = all pages. Range: 0–10 000. When > 0, each page's text is prefixed with [Page N].
ocrFallback	boolean	No	false	Enable OCR for pages with fewer than 50 characters of embedded text. Required for scanned or image-based PDFs. Increases processing time and memory usage.
ocrLanguage	string	No	eng	Tesseract language code. Only used when ocrFallback is true. See supported languages.

Example Input — Standard PDF

{
  "pdfUrl": "https://drive.google.com/file/d/FILE_ID/view?usp=sharing",
  "maxPages": 0
}

Example Input — Scanned / Image PDF with OCR

{
  "pdfUrl": "https://example.com/scanned-document.pdf",
  "ocrFallback": true,
  "ocrLanguage": "eng"
}

---

📤 Output Format

Each run pushes one item to the default dataset.

Successful extraction (text layer)

{
  "originalPdfUrl": "https://drive.google.com/file/d/FILE_ID/view?usp=sharing",
  "processedPdfUrl": "https://drive.google.com/uc?export=download&id=FILE_ID",
  "extractedText": "Full text content extracted from the PDF...",
  "pageCount": 12,
  "extractedPages": 12,
  "fileSizeBytes": 1048576,
  "sourceType": "google-drive",
  "ocrApplied": false,
  "timestamp": "2026-02-18T07:00:00.000Z",
  "success": true
}

Successful extraction (OCR applied to some pages)

{
  "originalPdfUrl": "https://example.com/scanned.pdf",
  "processedPdfUrl": "https://example.com/scanned.pdf",
  "extractedText": "Text recovered by OCR from the scanned pages...",
  "pageCount": 5,
  "extractedPages": 5,
  "fileSizeBytes": 2097152,
  "sourceType": "direct-url",
  "ocrApplied": true,
  "timestamp": "2026-02-18T07:00:00.000Z",
  "success": true
}

Failed extraction

{
  "originalPdfUrl": "https://example.com/missing.pdf",
  "processedPdfUrl": "https://example.com/missing.pdf",
  "extractedText": "",
  "pageCount": 0,
  "extractedPages": 0,
  "fileSizeBytes": 0,
  "sourceType": "direct-url",
  "ocrApplied": false,
  "timestamp": "2026-02-18T07:00:00.000Z",
  "success": false,
  "errorMessage": "HTTP 404: Not Found"
}

Output field reference

Field	Type	Description
originalPdfUrl	string	The input URL exactly as provided
processedPdfUrl	string	The URL actually used to download the file (after cloud-link conversion)
extractedText	string	Full extracted text. When maxPages > 0, each page is prefixed with [Page N]
pageCount	integer	Total number of pages in the PDF
extractedPages	integer	Pages actually extracted (≤ pageCount when maxPages is set)
fileSizeBytes	integer	Downloaded file size in bytes
sourceType	string	One of: direct-url, google-drive, dropbox, onedrive
ocrApplied	boolean	true if OCR was used on at least one page
timestamp	string	ISO 8601 timestamp of when extraction completed
success	boolean	true if extraction succeeded
errorMessage	string	Present only on failure; describes the error

---

🔍 How Text Extraction Works

Step 1 — Text layer extraction (always)

pdf.js-extract parses the PDF and reads the embedded text layer from every page. This is fast and accurate for digitally-created PDFs.

Step 2 — OCR fallback (optional, ocrFallback: true)

Any page that returns fewer than 50 characters of embedded text is treated as an image-only page. The Actor:

1. Renders that page to a PNG image at 300 DPI using pdftoppm (from poppler-utils)
2. Runs Tesseract OCR (tesseract.js) on the image
3. Replaces the page's text with the OCR result if it yields more characters

This hybrid approach means both regular and scanned PDFs are handled in a single run. Pages that already have good embedded text are never sent to OCR, keeping processing fast.

OCR accuracy notes:

• Works best on clean, high-contrast scans
• Handwritten text accuracy varies by quality
• Complex multi-column layouts may have word-ordering issues
• Non-Latin scripts require the matching ocrLanguage code

---

🌐 Supported OCR Language Codes

Pass any Tesseract language code as ocrLanguage:

Language	Code
English	eng
French	fra
German	deu
Spanish	spa
Portuguese	por
Italian	ita
Chinese (Simplified)	chi_sim
Chinese (Traditional)	chi_tra
Japanese	jpn
Korean	kor
Arabic	ara
Hindi	hin
Russian	rus

---

☁️ Supported Link Types

Source	Example
Direct URL	https://example.com/document.pdf
Google Drive	https://drive.google.com/file/d/FILE_ID/view
Google Drive (open)	https://drive.google.com/open?id=FILE_ID
Dropbox	https://www.dropbox.com/s/HASH/filename.pdf
OneDrive (1drv.ms)	https://1drv.ms/b/s!SHARE_ID

Cloud share links are automatically converted to direct download URLs before fetching.

---

💰 Pricing

Pay-per-event pricing — charged only on successful extractions.

Event	Price	Trigger
pdf-processed	$0.005	Per successfully processed PDF
page-extracted	$0.0005	Per page extracted (only when extractedPages > 1)

Cost examples

PDF	Cost
1-page PDF	$0.0050
3-page PDF	$0.0065 ($0.005 + 3 × $0.0005)
10-page PDF	$0.0100 ($0.005 + 10 × $0.0005)

Failed extractions are not charged. Spending limits can be controlled via ACTOR_MAX_TOTAL_CHARGE_USD.

---

⚙️ Technical Details

Property	Value
Runtime	Node.js 20
Actor version	1.4
Memory	256 MB min — 512 MB max
Download timeout	120 seconds (fixed)
Max pages input	0–10 000 (0 = all pages)
PDF library	pdf.js-extract ^0.2.1
OCR engine	tesseract.js ^5.1.0 (LSTM mode)
PDF-to-image	pdftoppm from poppler-utils at 300 DPI
OCR threshold	Pages with < 50 embedded chars
HTTP client	node-fetch ^2.7.0
Retry attempts	3 (1.5× exponential backoff, 1.5 s base delay)
Output format	JSON dataset

---

🔧 Use Cases

• Document processing — Invoices, contracts, reports, forms (including scanned paper copies)
• Research — Extract text from academic papers, white papers, and archival PDFs
• Data pipelines — Feed PDF content into downstream NLP or search systems
• Content management — Index PDF archives for full-text search
• Automation — Process PDFs at scale via the Apify API or Zapier/Make integrations
• Historical documents — OCR old scanned records and books

---

🌐 Integration

Platform	Details
Apify API	Full REST API access
Apify SDK (Python / Node.js)	Official SDKs supported
Zapier	Connect with 5 000+ apps
Make (Integromat)	Visual workflow automation
Webhooks	Real-time completion notifications

---

🔒 Security & Privacy

• Processing runs inside Apify's secure cloud infrastructure
• Data is not persisted beyond the Actor run's dataset retention period
• All transfers use HTTPS
• Spending limits enforced via Apify's pay-per-event system

---

🛠️ Local Development

# Install dependencies
npm install

# Run Actor locally (reads from input.json)
apify run --input-file input.json

# Validate input/output schemas
apify validate-schema

# Deploy to Apify platform
apify push

# Run on the platform
apify call

Local run storage is written to ./storage/ (git-ignored).

Note for local OCR testing: pdftoppm must be installed locally. The Apify platform uses Alpine Linux and installs it via apk add poppler-utils automatically. For local dev: brew install poppler (macOS), apk add poppler-utils (Alpine), apt-get install poppler-utils (Debian/Ubuntu), or choco install poppler (Windows).

---

📖 Quick Start

1. Open the Actor on Apify Store
2. Click Try for free
3. Enter a PDF URL in the pdfUrl field
4. For scanned PDFs, toggle Enable OCR Fallback to true
5. Click Start and wait for the run to finish
6. Download extracted text from the Dataset tab in JSON, CSV, or XLSX format

---

Ready to extract text from any PDF? Start using PDF Text Extractor →