QuerySUTRA 0.6.2

AI-powered data analysis for structured and unstructured data. Query PDF, Word, CSV, Excel with natural language.

QuerySUTRA

Natural language to SQL. Upload PDF, CSV, Excel — ask questions in plain English.

pip install QuerySUTRA

Usage

from sutra import SUTRA

sutra = SUTRA(api_key="your-openai-key")
sutra.upload("data.pdf")
result = sutra.ask("How many employees are in each state?")
print(result.data)

Upload Any Format

sutra.upload("report.pdf")       # PDF — AI extracts entities into related tables
sutra.upload("sales.csv")        # CSV
sutra.upload("data.xlsx")        # Excel
sutra.upload("records.json")     # JSON

import pandas as pd
df = pd.DataFrame({"name": ["Alice", "Bob"], "score": [95, 87]})
sutra.upload(df, name="students")

Ask Questions

result = sutra.ask("Show top 5 products by revenue")
print(result.sql)         # Generated SQL
print(result.data)        # pandas DataFrame

# With charts
result = sutra.ask("Sales by region", viz=True)        # Auto-picks best chart
result = sutra.ask("Revenue split", viz="pie")          # Specific chart type
result = sutra.ask("Price over time", viz="line")       # pie | bar | line | scatter

Explore Data

sutra.tables()              # List all tables
sutra.schema()              # Show columns and types
sutra.peek()                # Preview rows
sutra.peek("sales", 10)    # Specific table

Direct SQL

result = sutra.sql("SELECT * FROM products WHERE price > 100")

Export

sutra.save_to_mysql("localhost", "root", "password", "my_db")
sutra.export_db("backup.json", format="json")

How It Works

Upload File → AI extracts entities into related tables (with foreign keys)
            → Stored in SQLite
            → Ask questions in English
            → AI generates SQL with proper JOINs
            → Results as DataFrame + optional chart

Smart JOINs — QuerySUTRA detects foreign key relationships between tables (person_id → people.id) and generates proper JOIN queries automatically. When you ask "Who is the DevOps engineer?", it knows to JOIN work_experience with people to get the name.

Smart Charts — Auto-picks the right visualization. Few categories → pie. Many categories → bar. Two numbers → scatter. Or specify: viz="pie".

Troubleshooting

After installing or upgrading, if your Jupyter notebook still uses the old version:

# Option 1: Restart kernel and re-import
# Kernel → Restart Kernel

# Option 2: Force reload without restarting
import importlib
import sutra
import sutra.sutra
importlib.reload(sutra.sutra)
from sutra.sutra import SUTRA

Check which version is loaded:

import sutra
print(sutra.__version__, sutra.__file__)

Install Extras

pip install QuerySUTRA[mysql]       # MySQL export
pip install QuerySUTRA[embeddings]  # Semantic similarity
pip install QuerySUTRA[all]         # Everything

API

Method	Description
SUTRA(api_key, db)	Initialize. db defaults to "sutra.db"
.upload(file_or_df, name)	Upload from file path or DataFrame
.ask(question, viz)	Natural language query. viz=True or "pie"/"bar"/"line"/"scatter"
.sql(query)	Direct SQL
.tables()	List tables
.schema()	Show structure
.peek(table, n)	Preview rows
.save_to_mysql(host, user, pw, db)	Export to MySQL
.export_db(path, format)	Export SQLite or JSON
.close()	Close connection

QueryResult: result.success · result.sql · result.data · result.viz · result.error

License

MIT — Aditya Batta