From Support Chaos to ML Gold: Transform Your Tickets into Privacy-Safe Training Data

{
  "ticket_id": "SUP-2024-0892",
  "created_at": "2024-01-15T09:23:00Z",
  "customer": {
    "name": "Milton Waddams",
    "email": "mwaddams@initech.com",
    "company": "Initech",
    "account_id": "ACC-789234"
  },
  "subject": "API Integration Failing - Urgent",
  "description": "Hi support team, I'm Milton Waddams from Initech. Our database connection to server db-prod-01.initech.com is timing out since this morning. This is blocking our entire team. My direct line is 415-555-0123 if you need to call.",
  "priority": "high",
  "tags": ["api", "authentication", "urgent"]
}

import os
from tonic_textual.api import TonicTextual

# Initialize with your API key
tonic = TonicTextual(api_key=os.getenv("TONIC_TEXTUAL_API_KEY"))

# Our sensitive ticket
ticket = {
    "ticket_id": "SUP-2024-0892",
    "customer": {
        "name": "Milton Waddams",
        "email": "jchen@innovatech.com",
        "company": "InnovaTech Solutions"
    },
    "description": "Database timeout errors affecting production..."
}

# Redact the JSON
response = tonic.redact_json(
    ticket,
    generator_config={
        "PERSON": "Tokenize",      # Replace names with tokens
        "EMAIL": "Tokenize",        # Replace emails with tokens
        "COMPANY": "Tokenize",      # Replace company names
        "PHONE": "Tokenize"         # Replace phone numbers
    }
)

print(response.redacted_text)

{
  "ticket_id": "SUP-2024-0892",
  "customer": {
    "name": "[PERSON_TOKEN_1]",
    "email": "[EMAIL_TOKEN_1]",
    "company": "[COMPANY_TOKEN_1]"
  },
  "description": "Customer SSN [SSN_TOKEN_1] exposed..."
}

import pandas as pd

def process_support_csv(file_path: str):
    """Process CSV export from support system"""

    # Read the CSV
    df = pd.read_csv(file_path)

    print(f"Processing {len(df)} tickets...")
    print(f"Columns: {', '.join(df.columns)}")

    # Identify sensitive columns
    sensitive_columns = [
        'customer_name', 'customer_email',
        'company', 'description', 'notes'
    ]

    # Process each sensitive column
    for column in sensitive_columns:
        if column in df.columns:
            print(f"  Redacting {column}...")
            df[column] = df[column].apply(
                lambda x: tonic.redact(str(x)).redacted_text
                if pd.notna(x) else x
            )

    # Add metadata
    df['processed_at'] = pd.Timestamp.now()
    df['processing_version'] = '1.0.0'

    return df

# Process your export
df_clean = process_support_csv('support_tickets_export.csv')

# Save the clean version
df_clean.to_csv('support_tickets_clean.csv', index=False)

print(f"Processed {len(df_clean)} tickets successfully!")

# Process any text - JSON, CSV, plain text, etc.
def process_any_text(text):
    """Redact any text format"""

    response = tonic.redact(text)
    return response.redacted_text

# Test with a complex ticket
complex_ticket = """
Customer: Sarah Johnson (sarah.johnson@weylandcorp.com) from Weyland Corp
Issue: Customer database contains SSN 987-65-4321 and credit card 4916-3385-5678-1234.
Previously worked with your engineer Alexis Kong who can be reached at a.kong@jadetech.com or 212-555-0199.
Our CTO Robert Pereyda (rpereyda@weylandcorp.com) needs this fixed before tomorrow's board meeting.
"""

redacted = process_any_text(complex_ticket)

print("Original Text:")
print(complex_ticket)
print("\n" + "="*50 + "\n")
print("Redacted Text:")
print(redacted)