BharatStock Python Client

Official Python client for the BharatStock API — reliable Indian stock market data (NSE/BSE): EOD prices, quarterly/annual financials, shareholding patterns, corporate actions, derived per-stock metrics, a screener, bulk/block deals, insider trades, indices, and market-wide FII/DII activity.

Install

pip install bharatstock

Requires Python 3.9+ (the only dependency is httpx).

To work on the client from a checkout of this repo, install it editable:

pip install -e sdk/python

Authentication

Every data endpoint is authenticated with your bsk_live_... key, sent in the X-API-Key header. Get one from the dashboard.

from bharatstock import BharatStock

# Pass the key explicitly...
client = BharatStock(api_key="bsk_live_...")

# ...or set BHARATSTOCK_API_KEY in the environment and omit it:
client = BharatStock()

Quickstart

from bharatstock import BharatStock

client = BharatStock(api_key="bsk_live_...")

# A single stock, with latest price + ~70 derived metrics
stock = client.stocks.get("RELIANCE")
print(stock.company_name, stock.exchange)
print("P/E:", stock.metrics.pe_ratio, "ROE:", stock.metrics.roe)

# Batch quotes for a watchlist (one call, up to 50 symbols)
for q in client.stocks.quotes(["TCS", "INFY", "HDFCBANK"]):
    print(q.symbol, q.close, q.change_pct)

# Search
for hit in client.search("tata"):
    print(hit.symbol, hit.company_name)

# Public data-integrity status (no key required)
print(client.status().status)   # "operational" | "degraded"

Pagination

List endpoints return a Page object: iterate it directly for the rows, or read .total_pages / .has_next to page through manually.

# One page
page = client.stocks.prices("RELIANCE", from_date="2026-01-01", page_size=100)
print(page.total_items, page.total_pages)
for row in page:
    print(row.trade_date, row.close, row.adjusted_close)

# Auto-iterate every stock across all pages (lazy generator)
for s in client.stocks.iter_all(sector="Banking"):
    print(s.symbol)

Date ranges use from_date= / to_date= (sent to the API as from / to), in YYYY-MM-DD form.

Screener

results = client.screener.run(
    filters=["pe_ratio.lt.15", "roe.gt.18", "market_cap.gt.10000"],  # Cr
    sort_by="roe",
    sort_order="desc",
    page_size=25,
)
for r in results:
    print(r.symbol, r.pe_ratio, r.roe)

Filter syntax is metric.operator.value where the operator is one of gt | lt | gte | lte | eq. market_cap values are in Crores.

Rate limits & retries

Plans have a daily request cap. When you exceed it the API returns HTTP 429. The client automatically retries a 429 a few times with exponential backoff (the API does not send a Retry-After header, so the wait is client-side); if it's still capped it raises RateLimitError.

from bharatstock import BharatStock, RateLimitError, NotFoundError

client = BharatStock(api_key="bsk_live_...", max_retries=3)

try:
    stock = client.stocks.get("NONEXISTENT")
except NotFoundError:
    print("no such ticker")
except RateLimitError as e:
    print("slow down:", e.detail)

All errors subclass BharatStockError, so you can catch that one type to handle any API failure. Specific subclasses: AuthenticationError (401), NotFoundError (404), RateLimitError (429), BadRequestError (400/422), APIError (everything else).

Method reference

Every method and its parameters. Types ship with the package (py.typed), so your editor autocompletes each method and every field on the returned objects. Keyword-only params show their default; page/page_size are omitted from the notes below but accepted by every paginated method. Methods that return Page are paginated (iterate directly, or use .total_pages / .has_next); the rest return a single object or a plain list.

client.stocks

client.deals (market-wide, across all stocks)

client.screener

client.indices

client.market

Top-level helpers

Notes

• market_cap units differ by endpoint (this mirrors the current API, so the client reports exactly what the server sends):
  • Rupees: stocks.get, stocks.list / iter_all, search (StockSummary/StockDetail.market_cap), stocks.compare (ComparisonItem), and stocks.ratios (RatioSnapshot).
  • Crores (1 Cr = 10,000,000): the metrics block on stocks.get (StockDetail.metrics.market_cap) and screener.run (ScreenerResult).
  • The screener.run market_cap filter value is also in Crores (e.g. "market_cap.gt.10000" = > 10,000 Cr). So stock.market_cap and stock.metrics.market_cap on the same object are in different units (rupees vs Crores) — divide the rupee value by 1e7 to compare. Convert with crores = rupees / 10_000_000.
• Use the client as a context manager (with BharatStock(...) as c:) to close the underlying HTTP connection pool when you're done.
• Types ship with the package (py.typed), so editors autocomplete every method and response field.

License

MIT