Kotak mapping adapter.
Kotak's raw file needs two field forms no other broker uses, so this adapter extends _field rather than adding them to the shared transform registry:
{constant: <value>}supplies a fixed value where the raw file has no usable column at all, which is the case for the lot and tick sizes of the NSE commodity and currency underlying segments.{columns: [dstrikeprice, lprecision], transform: dynamic_precision_strike}divides the strike by ten raised to a precision that varies per row: two for equity and MCX commodity contracts, four for currency ones. NSE commodity options are the exception that proves the need for it — their true scale is always four decimal places even though the row reports a precision of two, so they name the fixeddivide_by_10_thousandtransform instead.
Expiry dates come in two flavours on this one broker. NSE contracts count seconds from 1980-01-01, which is what the kotak_expiry_epoch transform exists for; BSE and MCX contracts are plain Unix epochs and use unix_epoch_date.
The classification overrides are the ISIN splits that no equality rule can express: an NSE cash row whose ISIN is fund-prefixed is an exchange traded fund rather than an equity, a BSE row confirmed as a fund by cross-broker agreement is one whatever its group says, and both bond segments require a real non-fund ISIN. classify_extra restores the dual membership that a BSE row genuinely has when it is both a real security and a confirmed exchange traded fund.
Classes:
| Name | Description |
|---|---|
KotakMappingAdapter |
Mapping adapter for Kotak, covering 31 real segments plus the uncategorised buckets. |
Functions:
| Name | Description |
|---|---|
isin_is_fund |
Check whether an ISIN belongs to the fund family, which India marks with an INF prefix. |
isin_valid_bond |
Check whether an ISIN can identify a bond, meaning it is present and is not a fund ISIN. |
Bases: BrokerMappingAdapter
Mapping adapter for Kotak, covering 31 real segments plus the uncategorised buckets.
Example programs
examples/stock_brokers/instruments/mapping/kotak/KotakMappingAdapter/example_1_isin_splits_and_scales.py
"""Classifies Kotak rows whose segment depends on the ISIN, and reads strikes and expiries in Kotak's several scales.
Kotak's groups say little about what a row really is, so `KotakMappingAdapter.classify` looks at the ISIN: an NSE equity-group row with a fund ISIN is an exchange traded fund, a BSE row with one is left uncategorised, bonds need a real non-fund ISIN, and BSE group E rows whose description starts with "INAV" (an indicative value, not a fund) are dropped. A BSE row that other brokers confirm as an exchange traded fund but that also has a company ISIN is both things, and `classify_extra` returns its second segment.
`to_identity` strips Kotak's group suffix, so `NBIFIN-EQ` is stored as NBIFIN. Behind it, Kotak's field reader handles the scales: an NSE expiry counts seconds from 1980-01-01, a BSE one from 1970, and a BSE option strike is divided by ten raised to the row's own precision. Kotak's currency underlying rows have no lot or tick size, so both are stored as None.
`run` normally gathers the BSE exchange traded fund list from the database; this program sets it by hand to one symbol, so no database is needed. `uncategorised_exchange` shows where a dropped row is filed. The dual-membership row is shaped by hand to show the case; its symbol is made up.
Run it from the project root:
python examples/stock_brokers/instruments/mapping/kotak/KotakMappingAdapter/example_1_isin_splits_and_scales.py
"""
from stock_brokers.instruments.mapping.kotak import (
KotakMappingAdapter,
)
class IsinSplitsAndScalesExample:
"""Classifies eight hand-written Kotak rows and prints each one's segments, identity and sizes.
Attributes:
adapter (KotakMappingAdapter): The adapter being shown.
raw_rows (list): The raw rows, as dictionaries of text columns.
"""
def __init__(self):
"""Builds the adapter, sets the list `run` would gather, and writes the rows.
Returns:
None: This method returns nothing.
"""
self.adapter = KotakMappingAdapter()
self.adapter.bse_etf_symbols = {
'DUALETF',
}
self.raw_rows = [
self.raw_row('nse_cm', 'NSE', 'EQ', None, '4749', 'NBIFIN-EQ', 'NBIFIN', 'INE918K01019', 'BAJAJ HOLDINGS'),
self.raw_row('nse_cm', 'NSE', 'EQ', None, '10576', 'NIFTYBEES-EQ', 'NIFTYBEES', 'INF204KB14I2', 'NIPPON INDIA ETF NIFTY 50 BEES'),
self.raw_row('nse_cm', 'NSE', 'GS', None, '15542', '762GS2036-GS', '762GS2036', 'IN0020160019', 'GOI LOAN 7.62% 2036'),
self.raw_row('bse_cm', 'BSE', 'E', None, '540005', 'INAVGOLD', 'INAVGOLD', None, 'INAV OF A GOLD ETF'),
self.raw_row('bse_cm', 'BSE', 'B', None, '543999', 'DUALETF', 'DUALETF', 'INE000X01011', 'A FUND LISTED AS AN EQUITY'),
self.raw_row('nse_fo', 'NFO', None, 'OPTIDX', '40215', 'NIFTY26OCT25000CE', 'NIFTY', None, None),
self.raw_row('bse_fo', 'BFO', None, 'SO', '1165432', 'IRFC26OCT102.5CE', 'IRFC', None, None),
self.raw_row('cde_fo', 'CDS', None, 'UNDCUR', '1', 'USDINR', 'USDINR', None, None),
]
self.raw_rows[5]['lexpirydate'] = '1477578600'
self.raw_rows[5]['dstrikeprice'] = '2500000'
self.raw_rows[5]['poptiontype'] = 'CE'
self.raw_rows[6]['lexpirydate'] = '1793268000'
self.raw_rows[6]['dstrikeprice'] = '10250'
self.raw_rows[6]['lprecision'] = '2'
self.raw_rows[6]['poptiontype'] = 'CE'
def raw_row(self, exchange_segment, exchange, group, instrument_type, token, trading_symbol, symbol_name, isin, description):
"""Builds one raw row in the shape of `kotak.instruments`.
Args:
exchange_segment (str): The `pexchseg` column, such as nse_cm or bse_fo.
exchange (str): The `pexchange` column.
group (str | None): The `pgroup` column, which holds the series for cash rows.
instrument_type (str | None): The `pinsttype` column, for derivatives.
token (str): The `psymbol` column, Kotak's token.
trading_symbol (str): The `ptrdsymbol` column.
symbol_name (str): The `psymbolname` column.
isin (str | None): The `pisin` column.
description (str | None): The `pdesc` column.
Returns:
dict: The raw row, every column as text as the raw table stores it.
"""
return {
'psymbol': token,
'pgroup': group,
'pexchseg': exchange_segment,
'pinsttype': instrument_type,
'psymbolname': symbol_name,
'ptrdsymbol': trading_symbol,
'poptiontype': None,
'pisin': isin,
'pdesc': description,
'dticksize': '5',
'llotsize': '1',
'lexpirydate': None,
'lprecision': None,
'dstrikeprice': None,
'pexchange': exchange,
}
def run(self):
"""Prints each row's segments, identity and sizes, or where an unmatched row is filed.
Returns:
None: This method returns nothing.
"""
for raw_row in self.raw_rows:
label = f'{raw_row["pexchseg"]} {raw_row["ptrdsymbol"]}'
segment_configuration = self.adapter.classify(raw_row)
if segment_configuration is None:
exchange = self.adapter.uncategorised_exchange(raw_row)
print(f'{label} -> no segment, filed under {exchange}_uncategorised')
continue
extra_segments = []
for extra_configuration in self.adapter.classify_extra(raw_row):
extra_segments.append(extra_configuration['segment'])
identity = self.adapter.to_identity(raw_row, segment_configuration)
broker_fields = self.adapter.to_broker_fields(raw_row, segment_configuration)
print(f'{label} -> {segment_configuration["segment"]}, also {extra_segments}')
print(f' identity {identity}')
print(f' lot size {broker_fields["lot_size"]}, tick size {broker_fields["tick_size"]}')
if __name__ == '__main__':
IsinSplitsAndScalesExample().run()
It prints:
nse_cm NBIFIN-EQ -> nse_equities, also []
identity {'symbol': 'NBIFIN'}
lot size 1.0, tick size 0.05
nse_cm NIFTYBEES-EQ -> nse_exchange_traded_funds, also []
identity {'symbol': 'NIFTYBEES'}
lot size 1.0, tick size 0.05
nse_cm 762GS2036-GS -> nse_fixed_income, also []
identity {'symbol': 'IN0020160019'}
lot size 1.0, tick size 0.05
bse_cm INAVGOLD -> no segment, filed under bse_uncategorised
bse_cm DUALETF -> bse_exchange_traded_funds, also ['bse_equities']
identity {'symbol': 'DUALETF'}
lot size 1.0, tick size 0.05
nse_fo NIFTY26OCT25000CE -> nse_equity_index_options, also []
identity {'underlying_symbol': 'NIFTY', 'expiry_date': datetime.date(2026, 10, 27), 'strike_price': 25000.0, 'option_type': 'CE'}
lot size 1.0, tick size 0.05
bse_fo IRFC26OCT102.5CE -> bse_equity_options, also []
identity {'underlying_symbol': 'IRFC', 'expiry_date': datetime.date(2026, 10, 29), 'strike_price': 102.5, 'option_type': 'CE'}
lot size 1.0, tick size 0.05
cde_fo USDINR -> nse_currencies, also []
identity {'symbol': 'USDINR'}
lot size None, tick size None
examples/stock_brokers/instruments/mapping/kotak/KotakMappingAdapter/example_2_mapping_one_day.py
"""Maps one day of Kotak rows, where one BSE row is written into two segments.
`KotakMappingAdapter.run` first asks the other brokers' tables which BSE symbols are real exchange traded funds, then hands over to the shared mapping run, which reads Kotak's raw rows, classifies them, and writes `unified.instruments` and `unified.broker_mappings` in one transaction. A BSE row that is a confirmed exchange traded fund and also carries a company ISIN is written twice, once as a fund and once as an equity, because `classify_extra` returns the second segment.
The adapter normally talks to PostgreSQL, so this program replaces its `engine` with a stand-in. The stand-in connection answers queries by matching a piece of their text: the raw-row query gets two hand-written Kotak rows, and Flattrade's BSE fund-group query names DUALETF, a made-up symbol shaped to show the dual membership. Every other cross-broker query gets no rows, and every write is recorded instead of run. pandas only reads through something it recognises as an SQLAlchemy connectable, and it recognises one by the `ConnectionEventsTarget` base class, so the stand-in connection inherits from that class.
Notice that the summary counts two raw rows but three segment memberships and three instruments, and that the two DUALETF instruments share one Kotak token.
Run it from the project root:
python examples/stock_brokers/instruments/mapping/kotak/KotakMappingAdapter/example_2_mapping_one_day.py
"""
import collections
import datetime
from sqlalchemy.engine import interfaces
from stock_brokers.instruments.mapping.kotak import (
KotakMappingAdapter,
)
class StandInResult:
"""A stand-in for an SQLAlchemy result holding a fixed set of rows.
Attributes:
columns (list): The column names.
rows (list): The rows, as dictionaries.
"""
def __init__(self, rows):
"""Holds the rows.
Args:
rows (list): The rows, as dictionaries that all share the same keys.
Returns:
None: This method returns nothing.
"""
self.rows = rows
self.columns = []
if rows:
self.columns = list(rows[0])
def keys(self):
"""The column names, as pandas asks for them.
Returns:
list: The column names.
"""
return self.columns
def fetchall(self):
"""Every row as a plain tuple, as pandas asks for them.
Returns:
list: The rows as tuples in column order.
"""
tuples = []
for row in self.rows:
tuples.append(tuple(row.values()))
return tuples
def all(self):
"""Every row as a named tuple, so a column can be read as an attribute.
Returns:
list: The rows as named tuples.
"""
row_type = collections.namedtuple('StandInRow', self.columns)
named_rows = []
for row in self.rows:
named_rows.append(row_type(**row))
return named_rows
class StandInConnection(interfaces.ConnectionEventsTarget):
"""A stand-in for a PostgreSQL connection that answers queries from prepared rows and records writes.
Attributes:
answers (list): Pairs of (a piece of query text, the rows to answer with).
unanswered_queries (int): How many queries matched no prepared answer and got no rows.
writes (list): Each write statement's first words and its parameters.
"""
def __init__(self, answers):
"""Holds the prepared answers and starts with nothing recorded.
Args:
answers (list): Pairs of (a piece of query text, the rows to answer with).
Returns:
None: This method returns nothing.
"""
self.answers = answers
self.unanswered_queries = 0
self.writes = []
def __enter__(self):
"""Opens the connection for a `with` block.
Returns:
StandInConnection: This connection.
"""
return self
def __exit__(self, exception_type, exception, traceback):
"""Closes the connection at the end of a `with` block.
Args:
exception_type (type | None): The type of any exception raised in the block.
exception (BaseException | None): Any exception raised in the block.
traceback (object | None): The traceback of any exception raised in the block.
Returns:
bool: False, so any exception carries on.
"""
return False
def execute(self, statement, parameters=None):
"""Answers a query from the prepared rows, or records a write.
Args:
statement (sqlalchemy.sql.elements.TextClause): The statement to run.
parameters (dict | list | None): Its bound parameters.
Returns:
StandInResult: The prepared rows whose text matches, otherwise an empty result.
"""
sql = ' '.join(str(statement).split())
if not sql.startswith('SELECT'):
self.writes.append((' '.join(sql.split()[:3]), parameters))
return StandInResult([])
for query_text, rows in self.answers:
if query_text in sql:
return StandInResult(rows)
self.unanswered_queries += 1
return StandInResult([])
class StandInEngine:
"""A stand-in for an SQLAlchemy engine that always hands out the same stand-in connection.
Attributes:
connection (StandInConnection): The connection handed out.
"""
def __init__(self, connection):
"""Holds the connection to hand out.
Args:
connection (StandInConnection): The connection handed out.
Returns:
None: This method returns nothing.
"""
self.connection = connection
def connect(self):
"""Hands out the connection for reading.
Returns:
StandInConnection: The connection.
"""
return self.connection
def begin(self):
"""Hands out the connection for one transaction.
Returns:
StandInConnection: The connection.
"""
return self.connection
class MappingOneDayExample:
"""Maps two Kotak rows for one date and prints what would be written.
Attributes:
mapping_date (datetime.date): The snapshot date being mapped.
connection (StandInConnection): The stand-in connection.
adapter (KotakMappingAdapter): The adapter being shown, using the stand-in engine.
"""
def __init__(self):
"""Builds the adapter and swaps its engine for the stand-in.
Returns:
None: This method returns nothing.
"""
self.mapping_date = datetime.date(2026, 9, 28)
raw_rows = [
self.raw_row('nse_cm', 'EQ', '4749', 'NBIFIN-EQ', 'INE918K01019'),
self.raw_row('bse_cm', 'B', '543999', 'DUALETF', 'INE000X01011'),
]
flattrade_fund_group = [
{
'symbol': 'DUALETF',
},
]
answers = [
(
'SELECT * FROM kotak.instruments',
raw_rows,
),
(
'FROM flattrade.instruments WHERE download_date = :d AND exchange = \'BSE\' AND instrument = \'E\'',
flattrade_fund_group,
),
]
self.connection = StandInConnection(answers)
self.adapter = KotakMappingAdapter()
self.adapter.engine = StandInEngine(self.connection)
def raw_row(self, exchange_segment, group, token, trading_symbol, isin):
"""Builds one cash-market raw row in the shape of `kotak.instruments`.
Args:
exchange_segment (str): The `pexchseg` column, nse_cm or bse_cm.
group (str): The `pgroup` column.
token (str): The `psymbol` column, Kotak's token.
trading_symbol (str): The `ptrdsymbol` column.
isin (str): The `pisin` column.
Returns:
dict: The raw row, every column as text as the raw table stores it.
"""
return {
'psymbol': token,
'pgroup': group,
'pexchseg': exchange_segment,
'pinsttype': None,
'psymbolname': trading_symbol,
'ptrdsymbol': trading_symbol,
'pisin': isin,
'pdesc': trading_symbol,
'dticksize': '1',
'llotsize': '1',
'download_date': self.mapping_date,
}
def run(self):
"""Maps the day and prints the summary and the rows written.
Returns:
None: This method returns nothing.
"""
summary = self.adapter.run(self.mapping_date)
print(f'BSE exchange traded funds from other brokers: {sorted(self.adapter.bse_etf_symbols)}')
print(f'Cross-broker queries that found nothing: {self.connection.unanswered_queries}')
print(f'Summary: raw rows {summary["raw_rows"]}, matched {summary["matched"]}, memberships {summary["memberships"]}, instruments {summary["instruments_upserted"]}')
for statement, parameters in self.connection.writes:
print(statement)
if not isinstance(parameters, list):
continue
for row in parameters:
if 'segment' in row:
print(f' {row["instrument_id"][:8]} {row["segment"]}: {row["symbol"]}')
else:
print(f' {row["instrument_id"][:8]} token {row["broker_token"]}, tick size {row["tick_size"]}')
if __name__ == '__main__':
MappingOneDayExample().run()
It prints:
kotak: 2/2 raw row(s) classified, 0 uncategorised, 3 segment membership(s), 3 distinct instrument(s), 0 error(s).
BSE exchange traded funds from other brokers: ['DUALETF']
Cross-broker queries that found nothing: 10
Summary: raw rows 2, matched 2, memberships 3, instruments 3
DELETE FROM unified.broker_mappings
INSERT INTO unified.instruments
1ed7ce71 nse_equities: NBIFIN
7ce5b92e bse_exchange_traded_funds: DUALETF
b7ed1cc6 bse_equities: DUALETF
INSERT INTO unified.broker_mappings
1ed7ce71 token 4749, tick size 0.01
7ce5b92e token 543999, tick size 0.01
b7ed1cc6 token 543999, tick size 0.01
Load and validate the broker's rules file and create the database engine.
Raises:
| Type | Description |
|---|---|
NotImplementedError
|
If the subclass did not set BROKER_NAME. |
ValueError
|
If the rules file fails validation. |
FileNotFoundError
|
If the broker has no rules file. |
Example programs
examples/stock_brokers/instruments/mapping/base/BrokerMappingAdapter/example_1_classifying_rows_by_rules.py
"""Classifies three raw instrument rows using nothing but a broker's rules file.
`BrokerMappingAdapter` is the base class every broker's mapping adapter subclasses. A subclass only has to name its broker in `BROKER_NAME`, and the base class loads the matching rules file from `stock_brokers/instruments/mapping/utilities/rules/`, checks it against the canonical segment vocabulary, and classifies rows by plain equality rules. This program's subclass sets `BROKER_NAME` to Dhan and overrides nothing, so every answer below comes from the base class and Dhan's rules file alone, without the extra filters the real `DhanMappingAdapter` adds.
The rows are written out by hand in the shape of `dhan.instruments`, where every column is text. The adapter builds a database engine when it is constructed, but an engine does not connect until it is used, and this program never asks it to, so nothing touches PostgreSQL.
For each row the program prints the segment the rules chose, any extra segments (the base class never adds any), the identity fields that go into the instrument id, the broker's own fields, and the deterministic instrument id. The third row matches no rule. The base class cannot tell which exchange it belongs to, so it falls into the unprefixed `uncategorised` segment, and the run would store its token as its symbol.
Run it from the project root:
python examples/stock_brokers/instruments/mapping/base/BrokerMappingAdapter/example_1_classifying_rows_by_rules.py
"""
from stock_brokers.instruments.mapping import base
from stock_brokers.instruments.mapping.base import (
BrokerMappingAdapter,
)
class DhanRulesOnlyAdapter(BrokerMappingAdapter):
"""A mapping adapter that uses Dhan's rules file and nothing but the base class's behaviour."""
BROKER_NAME = 'dhan'
class ClassifyingRowsByRulesExample:
"""Runs three hand-written Dhan rows through the base class's rules engine and prints the result.
Attributes:
adapter (DhanRulesOnlyAdapter): The adapter being shown.
raw_rows (list): The raw rows to classify, as dictionaries of text columns.
"""
def __init__(self):
"""Builds the adapter and the three raw rows.
Returns:
None: This method returns nothing.
"""
self.adapter = DhanRulesOnlyAdapter()
self.raw_rows = [
{
'exch_id': 'NSE',
'segment': 'E',
'security_id': '1594',
'isin': 'INE009A01021',
'instrument': 'EQUITY',
'underlying_symbol': 'INFY',
'symbol_name': 'INFOSYS LIMITED',
'display_name': 'Infosys',
'instrument_type': 'ES',
'series': 'EQ',
'lot_size': '1.0',
'sm_expiry_date': None,
'strike_price': None,
'option_type': None,
'tick_size': '10.0',
},
{
'exch_id': 'NSE',
'segment': 'D',
'security_id': '40215',
'isin': None,
'instrument': 'OPTIDX',
'underlying_symbol': 'NIFTY',
'symbol_name': 'NIFTY-Oct2026-25000-CE',
'display_name': 'NIFTY 27 OCT 25000 CALL',
'instrument_type': 'OP',
'series': None,
'lot_size': '75.0',
'sm_expiry_date': '2026-10-27',
'strike_price': '25000.00000',
'option_type': 'CE',
'tick_size': '5.0',
},
{
'exch_id': 'NSE',
'segment': 'X',
'security_id': '990001',
'isin': None,
'instrument': 'SPREAD',
'underlying_symbol': '',
'symbol_name': 'TEST SPREAD',
'display_name': 'Test spread',
'instrument_type': 'SP',
'series': None,
'lot_size': '1.0',
'sm_expiry_date': None,
'strike_price': None,
'option_type': None,
'tick_size': '0',
},
]
def run(self):
"""Prints how the rules file classifies each row.
Returns:
None: This method returns nothing.
"""
equities = self.adapter.segment_config('nse_equities')
print(f'Rules file: {self.adapter.config["raw_table"]}, {len(self.adapter.config["segments"])} segments')
print(f'nse_equities is a {equities["shape"]} on {equities["exchange"]} with {len(equities["rules"])} rules')
print(f'Unconfigured segment: {self.adapter.segment_config("nse_bonds")}')
for raw_row in self.raw_rows:
print()
print(f'Row {raw_row["security_id"]} ({raw_row["symbol_name"]})')
segment_configuration = self.adapter.classify(raw_row)
if segment_configuration is None:
exchange = self.adapter.uncategorised_exchange(raw_row)
print(f' No rule matched, and the base class finds no exchange: {exchange}')
segment_configuration = self.adapter.segment_config('uncategorised')
extra_segments = self.adapter.classify_extra(raw_row)
identity = self.adapter.to_identity(raw_row, segment_configuration)
broker_fields = self.adapter.to_broker_fields(raw_row, segment_configuration)
identifier = base.instrument_id(
segment_configuration['exchange'],
segment_configuration['segment'],
segment_configuration['shape'],
identity,
)
print(f' Segment: {segment_configuration["segment"]} ({segment_configuration["shape"]})')
print(f' Extra segments: {extra_segments}')
print(f' Identity: {identity}')
print(f' Broker fields: {broker_fields}')
print(f' Instrument id: {identifier}')
if __name__ == '__main__':
ClassifyingRowsByRulesExample().run()
It prints:
Rules file: dhan.instruments, 32 segments
nse_equities is a security on nse with 2 rules
Unconfigured segment: None
Row 1594 (INFOSYS LIMITED)
Segment: nse_equities (security)
Extra segments: []
Identity: {'symbol': 'INFY'}
Broker fields: {'broker_token': '1594', 'broker_symbol': 'Infosys', 'order_symbol': None, 'lot_size': 1.0, 'tick_size': 0.1}
Instrument id: f310e1b0-e082-5b16-808a-25f741abb9de
Row 40215 (NIFTY-Oct2026-25000-CE)
Segment: nse_equity_index_options (option)
Extra segments: []
Identity: {'underlying_symbol': 'NIFTY', 'expiry_date': datetime.date(2026, 10, 27), 'strike_price': 25000.0, 'option_type': 'CE'}
Broker fields: {'broker_token': '40215', 'broker_symbol': 'NIFTY-Oct2026-25000-CE', 'order_symbol': None, 'lot_size': 75.0, 'tick_size': 0.05}
Instrument id: fe85c704-ce49-5190-b068-d4d3646ed162
Row 990001 (TEST SPREAD)
No rule matched, and the base class finds no exchange: None
Segment: uncategorised (security)
Extra segments: []
Identity: {'symbol': ''}
Broker fields: {'broker_token': '990001', 'broker_symbol': 'Test spread', 'order_symbol': None, 'lot_size': 1.0, 'tick_size': None}
Instrument id: 55a8952d-9071-5670-a4fe-a5384d3876bd
examples/stock_brokers/instruments/mapping/base/BrokerMappingAdapter/example_2_mapping_one_day.py
"""Maps one day's raw rows into the two unified tables, against a stand-in database.
`run` is what the daily mapping job calls for each broker. It reads the broker's raw rows for one date with `read_raw_rows`, classifies each row, and then, inside one transaction, deletes that broker's mappings for the date and upserts one row per instrument into `unified.instruments` and one per broker mapping into `unified.broker_mappings`. It returns a summary and prints a one-line report.
The adapter normally reads and writes PostgreSQL, so this program replaces its `engine` with a small stand-in. The stand-in connection answers the raw-row query with four hand-written rows in the shape of `dhan.instruments`, and records every other statement it is sent instead of running it. pandas only reads through something it recognises as an SQLAlchemy connectable, and it recognises one by the `ConnectionEventsTarget` base class, so the stand-in connection inherits from that class and nothing else.
The subclass uses Dhan's rules file and overrides nothing. The fourth row has a strike price that is not a number, so building its identity fails; `run` records that as an error and carries on with the other rows rather than stopping. The unmatched row is stored under its token as its symbol, and its tick size of 5.0 is kept as written, because Dhan's rules divide the tick size by 100 only in the real segments and not in the uncategorised one. The last lines show that a subclass which forgets to set `BROKER_NAME` cannot be built.
Run it from the project root:
python examples/stock_brokers/instruments/mapping/base/BrokerMappingAdapter/example_2_mapping_one_day.py
"""
import datetime
from sqlalchemy.engine import interfaces
from stock_brokers.instruments.mapping.base import (
BrokerMappingAdapter,
)
class DhanRulesOnlyAdapter(BrokerMappingAdapter):
"""A mapping adapter that uses Dhan's rules file and nothing but the base class's behaviour."""
BROKER_NAME = 'dhan'
class UnnamedAdapter(BrokerMappingAdapter):
"""A mapping adapter that forgets to name its broker."""
class StandInResult:
"""A stand-in for an SQLAlchemy result holding a fixed set of rows.
Attributes:
columns (list): The column names.
rows (list): The rows, as tuples in column order.
"""
def __init__(self, columns, rows):
"""Holds the columns and rows.
Args:
columns (list): The column names.
rows (list): The rows, as tuples in column order.
Returns:
None: This method returns nothing.
"""
self.columns = columns
self.rows = rows
def keys(self):
"""The column names, as pandas asks for them.
Returns:
list: The column names.
"""
return self.columns
def fetchall(self):
"""Every row, as pandas asks for them.
Returns:
list: The rows.
"""
return self.rows
def all(self):
"""Every row, as the adapter's own queries ask for them.
Returns:
list: The rows.
"""
return self.rows
class StandInConnection(interfaces.ConnectionEventsTarget):
"""A stand-in for a PostgreSQL connection that serves one raw table and records every other statement.
Attributes:
raw_table (str): The raw table whose rows this connection serves.
raw_rows (list): The raw rows, as dictionaries of column values.
statements (list): Each recorded statement's first line and its parameters.
"""
def __init__(self, raw_table, raw_rows):
"""Holds the raw rows and starts with no statements recorded.
Args:
raw_table (str): The raw table whose rows this connection serves.
raw_rows (list): The raw rows, as dictionaries of column values.
Returns:
None: This method returns nothing.
"""
self.raw_table = raw_table
self.raw_rows = raw_rows
self.statements = []
def __enter__(self):
"""Opens the connection for a `with` block.
Returns:
StandInConnection: This connection.
"""
return self
def __exit__(self, exception_type, exception, traceback):
"""Closes the connection at the end of a `with` block.
Args:
exception_type (type | None): The type of any exception raised in the block.
exception (BaseException | None): Any exception raised in the block.
traceback (object | None): The traceback of any exception raised in the block.
Returns:
bool: False, so any exception carries on.
"""
return False
def execute(self, statement, parameters=None):
"""Answers the raw-row query from the held rows and records anything else.
Args:
statement (sqlalchemy.sql.elements.TextClause): The statement to run.
parameters (dict | list | None): Its bound parameters.
Returns:
StandInResult: The raw rows for the raw-row query, otherwise an empty result.
"""
sql = ' '.join(str(statement).split())
if sql.startswith(f'SELECT * FROM {self.raw_table} '):
columns = list(self.raw_rows[0])
rows = []
for raw_row in self.raw_rows:
values = []
for column in columns:
values.append(raw_row[column])
rows.append(tuple(values))
return StandInResult(columns, rows)
self.statements.append((sql, parameters))
return StandInResult([], [])
class StandInEngine:
"""A stand-in for an SQLAlchemy engine that always hands out the same stand-in connection.
Attributes:
connection (StandInConnection): The connection handed out.
"""
def __init__(self, connection):
"""Holds the connection to hand out.
Args:
connection (StandInConnection): The connection handed out.
Returns:
None: This method returns nothing.
"""
self.connection = connection
def connect(self):
"""Hands out the connection for reading.
Returns:
StandInConnection: The connection.
"""
return self.connection
def begin(self):
"""Hands out the connection for one transaction.
Returns:
StandInConnection: The connection.
"""
return self.connection
class MappingOneDayExample:
"""Maps four raw rows for one date and prints what would be written.
Attributes:
mapping_date (datetime.date): The snapshot date being mapped.
connection (StandInConnection): The stand-in connection holding the raw rows.
adapter (DhanRulesOnlyAdapter): The adapter being shown, using the stand-in engine.
"""
def __init__(self):
"""Builds the adapter and swaps its engine for the stand-in.
Returns:
None: This method returns nothing.
"""
self.mapping_date = datetime.date(2026, 9, 28)
raw_rows = [
self.raw_row('1594', 'NSE', 'E', 'EQUITY', 'ES', 'INFY', 'INFOSYS LIMITED', None, None, None),
self.raw_row('40215', 'NSE', 'D', 'OPTIDX', 'OP', 'NIFTY', 'NIFTY-Oct2026-25000-CE', '2026-10-27', '25000.00000', 'CE'),
self.raw_row('990001', 'NSE', 'X', 'SPREAD', 'SP', '', 'TEST SPREAD', None, None, None),
self.raw_row('40216', 'NSE', 'D', 'OPTIDX', 'OP', 'NIFTY', 'NIFTY-Oct2026-XX-PE', '2026-10-27', 'not a number', 'PE'),
]
self.connection = StandInConnection('dhan.instruments', raw_rows)
self.adapter = DhanRulesOnlyAdapter()
self.adapter.engine = StandInEngine(self.connection)
def raw_row(self, security_id, exchange, segment, instrument, instrument_type, underlying_symbol, symbol_name, expiry, strike_price, option_type):
"""Builds one raw row in the shape of `dhan.instruments`.
Args:
security_id (str): Dhan's token for the row.
exchange (str): The `exch_id` column.
segment (str): Dhan's one-letter segment code.
instrument (str): The `instrument` column.
instrument_type (str): The `instrument_type` column.
underlying_symbol (str): The `underlying_symbol` column.
symbol_name (str): The `symbol_name` column.
expiry (str | None): The `sm_expiry_date` column.
strike_price (str | None): The `strike_price` column.
option_type (str | None): The `option_type` column.
Returns:
dict: The raw row, every column as text as the raw table stores it.
"""
return {
'exch_id': exchange,
'segment': segment,
'security_id': security_id,
'isin': None,
'instrument': instrument,
'underlying_security_id': None,
'underlying_symbol': underlying_symbol,
'symbol_name': symbol_name,
'display_name': symbol_name,
'instrument_type': instrument_type,
'series': None,
'lot_size': '1.0',
'sm_expiry_date': expiry,
'strike_price': strike_price,
'option_type': option_type,
'tick_size': '5.0',
'download_date': self.mapping_date,
}
def short_row(self, row):
"""Shortens one written row to its filled-in columns, leaving out the dates the example already names.
Args:
row (dict): One row the adapter asked to insert.
Returns:
dict: The row's columns that hold a value, with the instrument id cut to its first eight characters.
"""
shortened = {}
for column, value in row.items():
if value is None or isinstance(value, datetime.date):
continue
if column == 'instrument_id':
value = value[:8]
shortened[column] = value
return shortened
def run(self):
"""Reads the raw rows, maps them, and prints the summary and the statements sent.
Returns:
None: This method returns nothing.
"""
raw = self.adapter.read_raw_rows(self.connection, self.mapping_date)
print(f'read_raw_rows returned {len(raw)} rows with {len(raw.columns)} columns')
print()
summary = self.adapter.run(self.mapping_date)
print()
for key in summary:
print(f'{key}: {summary[key]}')
print()
print('Statements sent inside the write transaction:')
for sql, parameters in self.connection.statements:
print(f' {sql[:60]}...')
if isinstance(parameters, list):
for row in parameters:
print(f' {self.short_row(row)}')
else:
print(f' {parameters}')
print()
try:
UnnamedAdapter()
except NotImplementedError as error:
print(f'NotImplementedError: {error}')
if __name__ == '__main__':
MappingOneDayExample().run()
It prints:
read_raw_rows returned 4 rows with 17 columns
dhan: 3/4 raw row(s) classified, 1 uncategorised, 3 segment membership(s), 3 distinct instrument(s), 1 error(s).
broker: dhan
mapping_date: 2026-09-28
raw_rows: 4
matched: 3
uncategorised: 1
memberships: 3
instruments_upserted: 3
errors: [('nse_equity_index_options', "could not convert string to float: 'not a number'")]
Statements sent inside the write transaction:
DELETE FROM unified.broker_mappings WHERE broker = :b AND ma...
{'b': 'dhan', 'd': datetime.date(2026, 9, 28)}
INSERT INTO unified.instruments AS master (instrument_id, ex...
{'instrument_id': '8fdcd38e', 'exchange': 'unknown', 'segment': 'uncategorised', 'shape': 'security', 'symbol': '990001'}
{'instrument_id': 'f310e1b0', 'exchange': 'nse', 'segment': 'nse_equities', 'shape': 'security', 'symbol': 'INFY'}
{'instrument_id': 'fe85c704', 'exchange': 'nse', 'segment': 'nse_equity_index_options', 'shape': 'option', 'underlying_symbol': 'NIFTY', 'strike_price': 25000.0, 'option_type': 'CE'}
INSERT INTO unified.broker_mappings (instrument_id, broker, ...
{'instrument_id': '8fdcd38e', 'broker': 'dhan', 'broker_token': '990001', 'broker_symbol': 'TEST SPREAD', 'lot_size': 1.0, 'tick_size': 5.0, 'attributes': '{"display_name": "TEST SPREAD", "instrument_type": "SP"}'}
{'instrument_id': 'f310e1b0', 'broker': 'dhan', 'broker_token': '1594', 'broker_symbol': 'INFOSYS LIMITED', 'lot_size': 1.0, 'tick_size': 0.05, 'attributes': '{"display_name": "INFOSYS LIMITED", "instrument_type": "ES"}'}
{'instrument_id': 'fe85c704', 'broker': 'dhan', 'broker_token': '40215', 'broker_symbol': 'NIFTY-Oct2026-25000-CE', 'lot_size': 1.0, 'tick_size': 0.05, 'attributes': '{"display_name": "NIFTY-Oct2026-25000-CE", "instrument_type": "OP"}'}
NotImplementedError: UnnamedAdapter must set BROKER_NAME
Methods:
| Name | Description |
|---|---|
run |
Map Kotak's raw rows for one date, precomputing the BSE exchange traded fund allowlist first. |
classify |
Classify a raw row, applying the ISIN splits and the cross-broker fund reroutes on top of the rules. |
classify_extra |
Return the second segment a BSE row also belongs to when it is both a real security and a confirmed exchange traded fund. |
to_identity |
Build the unified identity fields, stripping Kotak's trading symbol suffix and applying the BSE index alias. |
uncategorised_exchange |
Determine the canonical exchange for an unmatched Kotak row from its exchange segment column. |
segment_config |
Look up one segment configuration by its segment value. |
to_broker_fields |
Build the per-broker mapping fields for a raw row under one segment. |
read_raw_rows |
Read the broker's raw rows for one date. |
Source code in stock_brokers/instruments/mapping/base.py
Map Kotak's raw rows for one date, precomputing the BSE exchange traded fund allowlist first.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
mapping_date
|
date
|
The raw snapshot date to map. |
required |
Returns:
| Type | Description |
|---|---|
dict
|
The run summary, as described in BrokerMappingAdapter.run. |
Example usage
Source code in stock_brokers/instruments/mapping/kotak.py
Read one raw column value, adding Kotak's two broker-local field forms to the shared ones.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
raw_row
|
dict
|
One raw row from kotak.instruments. |
required |
specification
|
str | dict
|
A bare column name, a {column, transform} pair, a {constant} value, or a {columns, transform} pair. |
required |
Returns:
| Type | Description |
|---|---|
Any
|
The resolved value. |
Raises:
| Type | Description |
|---|---|
ValueError
|
If a multi-column specification names a transform this adapter does not implement. |
Source code in stock_brokers/instruments/mapping/kotak.py
Classify a raw row, applying the ISIN splits and the cross-broker fund reroutes on top of the rules.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
raw_row
|
dict
|
One raw row from kotak.instruments. |
required |
Returns:
| Type | Description |
|---|---|
dict | None
|
The matched segment configuration, or None when no segment matches. |
Example usage
Source code in stock_brokers/instruments/mapping/kotak.py
Return the second segment a BSE row also belongs to when it is both a real security and a confirmed exchange traded fund.
A group B or F row with a real non-fund ISIN whose symbol the other brokers independently confirm as an exchange traded fund is genuinely both things. classify picks the fund classification as primary, being the more specific signal; this restores the equity or bond membership rather than losing it.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
raw_row
|
dict
|
One raw row from kotak.instruments. |
required |
Returns:
| Type | Description |
|---|---|
list[dict]
|
The additional segment configuration, or an empty list. |
Example usage
Source code in stock_brokers/instruments/mapping/kotak.py
Build the unified identity fields, stripping Kotak's trading symbol suffix and applying the BSE index alias.
Kotak appends the group to its NSE cash trading symbols, so "NBIFIN-EQ" has to lose the "-EQ" before it can converge with the other brokers' plain symbol.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
raw_row
|
dict
|
One raw row from kotak.instruments. |
required |
segment_configuration
|
dict
|
The segment configuration whose identity mapping applies. |
required |
Returns:
| Type | Description |
|---|---|
dict
|
Identity fields, as described in BrokerMappingAdapter.to_identity. |
Raises:
| Type | Description |
|---|---|
ValueError
|
If an expiry value is numeric with no explicit transform. |
Example usage
From example_1_isin_splits_and_scales.py
identity = self.adapter.to_identity(raw_row, segment_configuration)
Source code in stock_brokers/instruments/mapping/kotak.py
Determine the canonical exchange for an unmatched Kotak row from its exchange segment column.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
raw_row
|
dict
|
One raw row from kotak.instruments. |
required |
Returns:
| Type | Description |
|---|---|
str | None
|
"nse", "bse", or "mcx", or None for any other value. |
Example usage
Source code in stock_brokers/instruments/mapping/kotak.py
Check the loaded rules file against the canonical vocabulary.
Every configured segment must be a real canonical segment for its exchange with the matching shape, and the segments must appear in the canonical asset-class order. The file must end with the unprefixed uncategorised entry on exchange unknown, which is the guaranteed fallback for rows whose exchange cannot be determined; unknown is allowed for that one entry only, since it is the value stored in the exchange column of the mapped table for those rows.
Raises:
| Type | Description |
|---|---|
ValueError
|
On any invalid exchange, segment, shape, ordering, or missing or misplaced uncategorised entry. |
Source code in stock_brokers/instruments/mapping/base.py
Look up one segment configuration by its segment value.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
segment
|
str
|
The exchange-prefixed segment value, for example "nse_equities". |
required |
Returns:
| Type | Description |
|---|---|
dict | None
|
The matching segment configuration, or None when the rules file does not configure that segment. |
Example usage
From example_1_classifying_rows_by_rules.py
segment_configuration = self.adapter.segment_config('uncategorised')
From example_1_classifying_rows_by_rules.py
print(f'Unconfigured segment: {self.adapter.segment_config("nse_bonds")}')
Source code in stock_brokers/instruments/mapping/base.py
Check one equality rule against a raw row.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
raw_row
|
dict
|
One raw row from the broker's instrument table. |
required |
match
|
dict
|
Mapping of raw column name to expected value, or to a list of acceptable values. |
required |
Returns:
| Type | Description |
|---|---|
bool
|
True when every column in the rule matches. |
Source code in stock_brokers/instruments/mapping/base.py
Build the per-broker mapping fields for a raw row under one segment.
A tick size of zero or below is stored as None, because no instrument moves in steps of nothing, and brokers send 0 or -1 where they have no tick for a row.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
raw_row
|
dict
|
One raw row from the broker's instrument table. |
required |
segment_configuration
|
dict
|
The segment configuration whose broker field mapping applies. |
required |
Returns:
| Type | Description |
|---|---|
dict
|
Keys "broker_token" (str), "broker_symbol", "order_symbol", "lot_size" (float | None), and "tick_size" (float | None). |
Example usage
From example_1_classifying_rows_by_rules.py
broker_fields = self.adapter.to_broker_fields(raw_row, segment_configuration)
Source code in stock_brokers/instruments/mapping/base.py
Pick the uncategorised segment configuration a row falls back into.
The per-exchange bucket is used when uncategorised_exchange resolves an exchange and the rules file configures that bucket; otherwise the unprefixed uncategorised entry is used.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
raw_row
|
dict
|
One raw row from the broker's instrument table. |
required |
Returns:
| Type | Description |
|---|---|
dict
|
The uncategorised segment configuration. |
Raises:
| Type | Description |
|---|---|
ValueError
|
If no uncategorised configuration exists, which validation should have made impossible. |
Source code in stock_brokers/instruments/mapping/base.py
Read the broker's raw rows for one date.
Adapters override this where the fetch itself is part of the mapping: to impose a row order the classification depends on, or to bring in rows the plain query would miss.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
connection
|
Connection
|
An open SQLAlchemy connection. |
required |
mapping_date
|
date
|
The raw snapshot date to read. |
required |
Returns:
| Type | Description |
|---|---|
DataFrame
|
The raw rows for that date. |
Example usage
From example_2_mapping_one_day.py
raw = self.adapter.read_raw_rows(self.connection, self.mapping_date)
Source code in stock_brokers/instruments/mapping/base.py
Write one run's rows into the mapped tables inside a single transaction.
Both row lists are sorted by instrument id before the insert so concurrent writers acquire locks in the same order, which is the standard fix for the multi-writer deadlock this design was observed hitting when adapters ran in parallel.
The seen dates widen rather than overwrite, so they do not depend on the order dates happen to be mapped in. Assigning the run's date to the last seen date directly looks equivalent while a backfill walks forward, and is wrong the moment any older date is mapped afterwards: it drags the last seen date backwards, and leaves the first seen date frozen at whichever date was written first rather than the earliest the instrument was actually seen.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
instrument_rows
|
dict
|
Mapping of instrument id to the unified.instruments row. |
required |
broker_rows
|
dict
|
Mapping of instrument id to the unified.broker_mappings row. |
required |
mapping_date
|
date
|
The mapping date the run covers. |
required |
Source code in stock_brokers/instruments/mapping/base.py
Check whether an ISIN belongs to the fund family, which India marks with an INF prefix.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
isin
|
str | None
|
The raw ISIN value from a Kotak row. |
required |
Returns:
| Type | Description |
|---|---|
bool
|
True when the ISIN is present and starts with "INF". |
Source code in stock_brokers/instruments/mapping/kotak.py
Check whether an ISIN can identify a bond, meaning it is present and is not a fund ISIN.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
isin
|
str | None
|
The raw ISIN value from a Kotak row. |
required |
Returns:
| Type | Description |
|---|---|
bool
|
True when the ISIN is usable as a fixed income identity. |