The Complete Production SOP: How to Deploy and Run Tesseract OCR Searchable Sandwich PDF in Autonomous AI Agents

---
name: "tesseract-ocr-searchable-pdf"
description: "Convert scanned paper documents and images into searchable sandwich PDF files with OCR."
metadata:
  author: "Tesseract OCR"
  version: "1.0.0"
  runtime_compatibility:
    - "codex"
    - "claude-code"
    - "cursor"
    - "windsurf"
triggers:
  - "convert scanned pdf to searchable"
  - "run tesseract ocr on image"
  - "deskew and ocr document"
permissions:
  fs_read: true
  fs_write: "workspace"
  shell_exec: false
  net_access: false
---

# Tesseract OCR Searchable Sandwich PDF Rules
1. Enforce ocrmypdf with --deskew, --clean, and --optimize flags.
2. Require resolution check: raise error if DPI < 300 unless --force-ocr is explicitly mandated.
3. Keep original raster stream intact; embed searchable text layer via invisible text font matrix.
4. Output files must validate against PDF/A standards via pdfminer or pdftotext verification.

import shutil
import sys
import ocrmypdf

required_bins = ["tesseract", "gs", "unpaper"]
missing = [b for b in required_bins if shutil.which(b) is None]

if missing:
    print(f"[ERROR] Missing required system binaries: {missing}", file=sys.stderr)
    sys.exit(1)

print("[OK] All OCR dependencies and binary paths verified.")
sys.exit(0)

import subprocess
from pathlib import Path

def generate_searchable_pdf(input_path: Path, output_path: Path, languages: str = "eng") -> None:
    cmd = [
        "ocrmypdf",
        "--deskew",
        "--clean",
        "--rotate-pages",
        "--output-type", "pdfa",
        "-l", languages,
        str(input_path),
        str(output_path)
    ]
    result = subprocess.run(cmd, capture_output=True, text=True)
    if result.returncode != 0:
        raise RuntimeError(f"OCR processing failed: {result.stderr}")
    print(f"Successfully generated sandwich PDF: {output_path}")

import pdfplumber
import pytest
from pathlib import Path

def test_sandwich_pdf_contains_text():
    output_file = Path("output_searchable.pdf")
    assert output_file.exists(), "Output file was not created."
    
    with pdfplumber.open(output_file) as pdf:
        total_text = ""
        for idx, page in enumerate(pdf.pages):
            page_text = page.extract_text()
            assert page_text is not None and len(page_text.strip()) > 0, f"Page {idx} has no text layer"
            total_text += page_text
            
    assert "INVOICE" in total_text.upper() or len(total_text) > 100