tej

Python SDK for tej-api: open, free, end-of-day market data for NSE and BSE. No signup, no keys, no catch.

Zero runtime dependencies. Sync and async clients with identical surface. Returns plain list[dict] rows so you can feed them straight to polars, pandas, or your own code without an extra conversion step.

Install

pip install tejhq

That is the entire install. No httpx, no aiohttp, no pydantic. Just stdlib. Distribution is tejhq; import as tej.

Quick start (sync)

from tej import Client

c = Client()

# Last 90 days of RELIANCE (default range)
rows = c.ohlcv("RELIANCE", "nse")
print(rows[-1])
# {'date': '2026-06-05', 'open': 1442.0, 'high': 1455.7, ...}

# Full history since 2010
rows = c.ohlcv("RELIANCE", "nse", "2010-01-04", "2026-06-05")
print(len(rows))  # ~4040

# Full-market snapshot for one trading date
snap = c.snapshot("nse", "2025-05-28")
print(len(snap))  # ~2700 symbols

# Corporate actions (dividends, splits, bonuses)
acts = c.actions("RELIANCE")
print(acts[0])
# {'exchange': 'NSE', 'type': 'dividend', 'ex_date': '2024-10-28', 'cash_amount': 10.0, ...}

Quick start (async)

import asyncio
from tej import AsyncClient

async def main():
    async with AsyncClient() as c:
        # Fan out 10 symbols concurrently
        symbols = ["RELIANCE", "TCS", "INFY", "HDFCBANK", "ICICIBANK",
                   "SBIN", "HINDUNILVR", "ITC", "LT", "AXISBANK"]
        results = await asyncio.gather(*[c.ohlcv(s, "nse") for s in symbols])
        for sym, rows in zip(symbols, results):
            print(sym, len(rows))

asyncio.run(main())

Use with polars / pandas

tej returns list[dict] so you bring your own dataframe library.

import polars as pl
from tej import Client

rows = Client().ohlcv("RELIANCE", "nse", "2010-01-04", "2026-06-05")
df = pl.DataFrame(rows)
df = df.with_columns(pl.col("date").str.to_date())
df.head()

import pandas as pd
from tej import Client

rows = Client().ohlcv("RELIANCE", "nse", "2025-01-01", "2025-06-01")
df = pd.DataFrame(rows)
df["date"] = pd.to_datetime(df["date"])
df.set_index("date", inplace=True)
df["close"].plot()

Optional extras install the dataframe library alongside the SDK:

pip install "tejhq[polars]"
pip install "tejhq[pandas]"

Free key endpoints

Get a key at tejhq.dev/keys, no card, no password.

from tej import Client

c = Client(api_key="tej_live_...")

# Back-adjusted prices, continuous through splits, bonuses and dividends
adj = c.adjusted("RELIANCE", "nse", "2024-10-01", "2024-11-30")
print(adj[-1]["adj_close"], adj[-1]["adj_factor_cumulative"])

# Symbol history by ISIN or symbol
c.symbols("nse", isin="INE040A01034")
c.symbols("nse", symbol="HDFCBANK")

# Account, keys, today's usage
c.me()

Pro endpoints

c = Client(api_key="tej_live_...")   # a key on the Pro tier

# Derived metrics per day: returns, 52w high/low, average volume and turnover
m = c.metrics("RELIANCE", "nse", "2026-01-01")

# Point-in-time liquidity universe, survivorship-bias-free
members = c.universe("liquid500", "nse", as_of="2019-03-15")
symbols = [x["symbol"] for x in members]

# Up to 50 symbols in one request, dict keyed by symbol
bars = c.batch(["RELIANCE", "TCS", "INFY"], "nse", "2026-01-01")

# Free text to symbol
c.resolve("tata motors", exchange="nse", limit=3)
c.resolve("zomato")   # former ticker resolves to the current one

API reference

Need the response envelope (with meta)? Use c.ohlcv_envelope(...), which returns the raw {"data": [...], "meta": {...}} dict.

AsyncClient exposes the same surface with await.

Configuration

from tej import Client

c = Client(
    base_url="https://api.tejhq.dev",   # override for self-hosted or staging
    api_key=None,                        # tej_live_... from https://tejhq.dev/keys; keyless works for ohlcv, snapshot, actions
    timeout=30.0,                        # seconds
    max_retries=3,                       # exponential backoff on 5xx/429/network
    user_agent_suffix="my-app/1.0",      # for attribution
    default_headers={"X-My-Header": "hi"},
)

Errors

All errors inherit from tej.TejError. The specific subclass tells you what happened:

from tej import Client, ProRequiredError

try:
    Client().ohlcv("RELIANCE", "nse")
except ProRequiredError as e:
    print(e.error_code)  # 'pro_required'
    print(e.request_id)  # tej-api request id from response headers

Why a separate package

tej-api is a Go service. tej-bazaar is the Python ingestion pipeline. This repo is the thin client that anyone can pip install without dragging in either, and that releases on its own cadence as the API evolves.

Coverage

• NSE bhavcopy + corp actions: 2010-01-04 to today, ~4,047 trading days, ~7M rows
• BSE bhavcopy + corp actions: 2024-07-08 to today (the SEBI CMTS cutover), ~470 days, ~1M rows
• Cron refresh: weekdays 20:00 IST

Keyless endpoints are served at the Cloudflare edge from pre-rendered JSON, so most requests return in well under a second and repeat requests are cache hits. Keyless access is rate limited to 100 requests per 10 seconds per IP at the edge and 120 per minute at origin; the SDK retries 429s with backoff. Keyed requests bypass the edge cache and carry X-RateLimit-Limit-Day and X-RateLimit-Remaining-Day headers: 1,000 a day and 300 a minute on a free key.

License

MIT. Use it, fork it, ship products on top of it.

Links

• API docs: tejhq.dev/docs
• OpenAPI spec: api.tejhq.dev/openapi.yaml
• Bulk parquet downloads: data.tejhq.dev and huggingface.co/datasets/tejhq/indian-markets
• Issues: github.com/tejhq/tej-sdk-py/issues