[docs] How can this validator be used via REST API?

[valentinabojan]

Description
The current documentation doesn't provide any info on how can I create a guard using this validator. More specifically, how could we pass the get_entity_threshold via REST API?

Current documentation
https://hub.guardrailsai.com/validator/guardrails/guardrails_pii

[thisaltennakoon]

You can follow these steps to use this via a REST API

!pip install guardrails-ai
!pip install flask
!guardrails configure
!guardrails hub install hub://guardrails/guardrails_pii

server.py

from flask import Flask, request, jsonify
from guardrails.hub import GuardrailsPII
from guardrails import Guard
from typing import List, Dict, Any, Optional

app = Flask(__name__)

# Setup Guard with PII detection focused on selected entity types
guard = Guard().use(
    GuardrailsPII(entities=[
        "EMAIL_ADDRESS", "PHONE_NUMBER", "PERSON", "DATE_TIME"
    ], on_fail="fix")
)

class ValidationResult:
    """Class to store and format validation results."""
    def __init__(self, passed: bool, summaries: List[Dict[str, Any]], fixed_text: Optional[str] = None):
        self.passed = passed
        self.summaries = summaries
        self.fixed_text = fixed_text

    def to_dict(self) -> Dict[str, Any]:
        return {
            "verdict": self.passed,
            "assessment": self.summaries,
            "anonymized": self.fixed_text
        }

def validate_text(text: str) -> ValidationResult:
    """Validate a single text string for PII and anonymize it."""
    if not text or not isinstance(text, str):
        return ValidationResult(True, [])

    result = guard.validate(text)

    print("Validation result:", result)

    pii_entities = []
    if hasattr(result, 'validation_summaries') and result.validation_summaries:
        for summary in result.validation_summaries:
            for error in summary.error_spans:
                pii_value = text[error.start:error.end]
                pii_entities.append({
                    "piiEntity": error.reason,
                    "piiValue": pii_value
                })

    masked_text = result.validated_output if result.validation_passed else result.fixed_output
    return ValidationResult(result.validation_passed, pii_entities, masked_text)

@app.route('/validate', methods=['POST'])
def validate():
    """Webhook endpoint to validate a single text field for PII."""
    try:
        data = request.json
        if not data or "text" not in data or not data["text"]:
            return jsonify({"error": "Missing or empty 'text' field in request"}), 400

        result = validate_text(data["text"])

        return jsonify({
            "verdict": result.passed,
            "assessment": result.summaries,
            "anonymizedText": result.fixed_text
        })

    except Exception as e:
        return jsonify({"error": str(e)}), 500


if __name__ == '__main__':
    app.run(host='0.0.0.0', port=8000, debug=False)

curl -X POST http://127.0.0.1:8000/validate \
  -H "Content-Type: application/json" \
  -d '{
    "text": "Hi my name is John and my email address is john@email.com"
  }'

[valentinabojan]

Thank you. So without actually exposing it under an endpoint in my server.py, there is no support out of the box for this guardrails in the existing REST API offered by Guardrails AI?

[thisaltennakoon]

I think this might be what you're looking for: https://www.guardrailsai.com/docs/getting_started/guardrails_server