extract

extract

The extract function extracts structured data from unstructured text.

Import

from pyai import extract

Basic Usage

# Extract with schema
data = extract(text, schema={"name": str, "email": str})

# Extract common entities
entities = extract.entities("John works at Microsoft")

Parameters

Parameter	Type	Default	Description
text	str	required	Source text
schema	dict	None	Expected data structure
type	str	None	Preset: "contact", "date", "product"

Examples

Extract Contact Info

from pyai import extract

text = """
Contact John Smith at john@example.com
Phone: 555-123-4567
"""

contact = extract(text, schema={
    "name": str,
    "email": str,
    "phone": str
})

print(contact)
# {"name": "John Smith", "email": "john@example.com", "phone": "555-123-4567"}

Extract Entities

text = "Apple announced iPhone 15 on September 12, 2023"

entities = extract.entities(text)
# {
#   "organizations": ["Apple"],
#   "products": ["iPhone 15"],
#   "dates": ["September 12, 2023"]
# }

Extract from Documents

# Extract invoice data
invoice_data = extract(
    invoice_text,
    schema={
        "invoice_number": str,
        "date": str,
        "total": float,
        "items": list
    }
)

Custom Extraction

# Extract product info
product = extract(
    description,
    schema={
        "name": str,
        "price": float,
        "features": list,
        "specifications": dict
    }
)

Async Usage

import asyncio
from pyai import extract

async def main():
    data = await extract.async_(
        email_text,
        schema={"sender": str, "subject": str}
    )
    print(data)

asyncio.run(main())

Preset Types

Type	Extracted Fields
contact	name, email, phone, address
date	dates, times, durations
product	name, price, description, features
receipt	items, totals, date, vendor
invoice	number, date, items, total, tax

See Also

• ask - Question answering
• summarize - Summarization
• analyze - Data analysis