How to use DocumentScraperGraph with OCR for local files?

[ksudhir-bentleighgroup]

pdf_scraper.scrape() --> Scrape Doesn't exist,

pdf_scraper.run() ---> Returns N/A (my file exist in path)

Example followed : https://github.com/ScrapeGraphAI/Scrapegraph-ai/tree/main/examples/document_scraper_graph

[dosubot]

Hi @ksudhir-bentleighgroup! I'm Dosu and I’m helping the Scrapegraph-ai team.

The correct method to run DocumentScraperGraph is .run()—there is no .scrape() method. If .run() returns "N/A", it's often due to how the source is provided or a config issue.

For OCR on local files (like PDFs or images), make sure:

• Your config includes "ocr": True (or "use_ocr": True in some versions) and the required LLM settings.
• The source parameter should be the path to your local file (not the file content).
• Tesseract OCR must be installed and available in your system path.

Example:

from scrapegraphai.graphs import DocumentScraperGraph

graph_config = {
    "llm": {
        "api_key": "YOUR_API_KEY",
        "model": "openai/gpt-3.5-turbo"
    },
    "ocr": True,  # or "use_ocr": True depending on your version
    "verbose": True,
    "headless": False
}

pdf_path = "path/to/your/file.pdf"

pdf_scraper = DocumentScraperGraph(
    prompt="Summarize the document.",
    source=pdf_path,
    config=graph_config
)

result = pdf_scraper.run()
print(result)

If you still get "N/A", check:

• The file path is correct and accessible.
• Tesseract is installed and working (tesseract --version in your terminal).
• Your file is a supported type (PDF, image).

If your version expects file content instead of a path, read the file as text or bytes and pass that as source. But most recent versions accept the file path directly for local files with OCR enabled.

There is no .scrape() method—.run() is the only supported execution method for this graph.
See related issue and maintainer guidance and code reference for more details.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}