pdf_text_extraction

PDF Text Extraction

Extract selectable text from a PDF (pdf-parse). Optional useAI=true runs Gemini cleanup on the extracted text (not OCR).

Fields API: GET /v1/tools/pdf_text_extraction/fields · public GET /api/developer/tools/public/pdf_text_extraction/fields

Price: 50 API credits ($0.005) per successful_execution

Version: 1

Accepted types: application/pdf

Max file: 80 MB

Max pages: 250

Timeout: 120000 ms

Batch jobs: supported · max 10 files · job ≤ 200 MB · docs

This tool extracts selectable text via pdf-parse. It does not perform OCR on scanned pages. is_scanned may hint when little text was found.

Multipart fields

Normalized catalog for agents and UIs (same as the sandbox field picker). Send each name as a multipart form key with tool and file.

Name	Type	Default	Notes
useAI	boolean	false	AI-related · When true, run Gemini text cleanup on extracted text (max 450k chars). No extra AI fields — send the PDF file. Billed via ai_pricing (API credits).

Output schema

{
  "text": "string",
  "page_count": "number",
  "is_scanned": "boolean",
  "used_ai": "boolean",
  "ai_text": "string|null"
}

cURL

curl -X POST "$TOOLGRID_API_BASE/v1/execute" \
  -H "Authorization: Bearer tg_api_YOUR_SECRET" \
  -H "Idempotency-Key: YOUR_UNIQUE_KEY" \
  -F "tool=pdf_text_extraction" \
  -F "file=@document.pdf"

Node.js (FormData)

Requires Node.js 18+ (global fetch/FormData/Blob). Set TOOLGRID_API_BASE and TOOLGRID_API_KEY.

import { readFile } from 'node:fs/promises';
import { randomUUID } from 'node:crypto';

const API_BASE = process.env.TOOLGRID_API_BASE || 'http://localhost:5000';
const API_KEY = process.env.TOOLGRID_API_KEY; // tg_api_… — never log this

if (!API_KEY) {
  throw new Error('Set TOOLGRID_API_KEY');
}

const form = new FormData();
form.set('tool', 'pdf_text_extraction');
form.set('file', new Blob([await readFile('./document.pdf')]), 'document.pdf');

const res = await fetch(`${API_BASE}/v1/execute`, {
  method: 'POST',
  headers: {
    Authorization: `Bearer ${API_KEY}`,
    'Idempotency-Key': randomUUID(),
  },
  body: form,
});

const body = await res.json();
if (!res.ok || !body.success) {
  console.error(body.error?.code, body.error?.message, body.request_id);
  process.exit(1);
}

console.log('request_id', body.request_id);
console.log('billing', body.billing);
console.log('pages', body.result.page_count, 'scanned?', body.result.is_scanned);
console.log(body.result.text.slice(0, 500));

Billing rules · Errors

Multipart fields

Normalized catalog for agents and UIs (same as the sandbox field picker). Send each name as a multipart form key with tool and file.

Name	Type	Default	Notes
useAI	boolean	false	AI-related · When true, run Gemini text cleanup on extracted text (max 450k chars). No extra AI fields — send the PDF file. Billed via ai_pricing (API credits).

Node.js (FormData)

Requires Node.js 18+ (global fetch/FormData/Blob). Set TOOLGRID_API_BASE and TOOLGRID_API_KEY.

import { readFile } from 'node:fs/promises';
import { randomUUID } from 'node:crypto';

const API_BASE = process.env.TOOLGRID_API_BASE || 'http://localhost:5000';
const API_KEY = process.env.TOOLGRID_API_KEY; // tg_api_… — never log this

if (!API_KEY) {
  throw new Error('Set TOOLGRID_API_KEY');
}

const form = new FormData();
form.set('tool', 'pdf_text_extraction');
form.set('file', new Blob([await readFile('./document.pdf')]), 'document.pdf');

const res = await fetch(`${API_BASE}/v1/execute`, {
  method: 'POST',
  headers: {
    Authorization: `Bearer ${API_KEY}`,
    'Idempotency-Key': randomUUID(),
  },
  body: form,
});

const body = await res.json();
if (!res.ok || !body.success) {
  console.error(body.error?.code, body.error?.message, body.request_id);
  process.exit(1);
}

console.log('request_id', body.request_id);
console.log('billing', body.billing);
console.log('pages', body.result.page_count, 'scanned?', body.result.is_scanned);
console.log(body.result.text.slice(0, 500));