otp.ODBC#

class ODBC(dsn=None, connection_string=None, authentication_type='credentials', user=None, password=None, sql=None, start_expr=None, end_expr=None, tz=None, allow_unordered_ticks=False, numeric_scale=8, preserve_unicode_fields=None, symbology='', apply_symbol_name_history=False, start=utils.adaptive, end=utils.adaptive, db=utils.adaptive, tick_type=utils.adaptive, symbols=utils.adaptive_to_default, presort=False, **kwargs)#

Bases: onetick.py.core.source.Source

Read from ODBC-compatible database and propagate the resulting data as a time series of ticks.

A time series is generated for every symbol of the query and the symbol name can be passed down to SQL query.

If an attribute with the name TIMESTAMP is present in the database schema, it is assumed to be of either SQL DATE or SQL TIMESTAMP type, in which case output tick timestamps will carry values of that field and the field itself will not be propagated. If such a field is absent, output tick timestamps will be equal to query end time.

ODBC source supported only on OneTick versions starting from release 1.24 (or from development build 20231108-0)

Parameters

dsn (str) – Target database’s source name registered in ODBC configuration.
connection_string (str) –
ODBC connection string. The format depends on the database you are trying to connect. Example for SQLite database:
```
DRIVER={SQLite3};Database=/path/to/the/database.db
```
This parameter is mutually exclusive with parameters dsn, user, password and authentication_type.
authentication_type (str ('system' or 'credentials')) –
- system: in this case a trusted connection to the database will be used. If authentication is enabled in the server configuration file (see the OneTick Installation and Administration Guide for details), the client will be authenticated and the client’s credentials will be used for connection; otherwise, server credentials will be used.
- credentials: username and password specified with parameter connection_string or with parameters user and password will be used to connect to the database.
user (str, optional) –
The connection user name.

Ignored if authentication_type is system.
password (str, optional) –
The connection password.

Ignored if authentication_type is system.
sql (str) –
A query in SQL language, which may optionally contain parameter placeholders.

There are 3 types of placeholders: <_SYMBOL_NAME>, <_START_TIME>, and <_END_TIME>.

<_SYMBOL_NAME> will be replaced by the pure symbol name part of the query.

<_START_TIME> and <_END_TIME> will be replaced by values of parameters start_expr and end_expr or, if they are not set, with values taken from query start and end times.

These timestamps will be inserted in the format YYYY-MM-DD HH:MM:SS.sss TIMEZONE.
start_expr (str or Operation) – A constant string expression used to replace the <_START_TIME> placeholder in an sql query. Tick-dependent columns can’t be used in this expression, only meta fields such as TIMESTAMP, _START_TIME, and _END_TIME are allowed.
end_expr (str or Operation) – A constant string expression used to replace the <_END_TIME> placeholder in an sql query. Tick-dependent columns can’t be used in this expression, only meta fields such as TIMESTAMP, _START_TIME, and _END_TIME are allowed.
tz (str) – The timezone used to interpret TIMESTAMP and other datetime columns in the database. By default tz is used.
allow_unordered_ticks (bool) –
If set to False, this class will raise an exception when unordered ticks will be encountered.

If set to True, processing ticks unordered by timestamp will be allowed (exception may still be raised by other EPs of the graph that require ticks to be ordered).
numeric_scale (int) – Number of digits after the decimal point for SQL_NUMERIC and SQL_DECIMAL data types. Precision (maximum number of digits) is always set to 34.
preserve_unicode_fields (list of str) –
By default ODBC queries all data as ANSI, ODBC Driver Manager then tries to convert Unicode characters to ANSI.

This parameter is a list of fields, which have Unicode types in the data source and will be propagated without conversion.

UNICODE_CHAR_TYPE field property will be set to UCS-2 for these fields.
symbology (str) – If specified, <_SYMBOL_NAME> placeholder in sql query will be replaced with the mapping of the symbol name of the query for provided symbology.
apply_symbol_name_history (bool) –
If set to False, ODBC will not take into account the symbol name history when substituting the SQL query with actual symbol names.

If set to True, ODBC will resolve symbol name changes according to the reference data by substituting the SQL query accordingly, provided the symbol date is specified in otp.run when running the query and the SQL query contains all three placeholders <_SYMBOL_NAME>, <_START_TIME>, <_END_TIME>.

Note that ODBC data source needs to support “UNION ALL” syntax for this functionality to work.
start – Custom start time of the query. By default the start time used by otp.run will be inherited.
end – Custom end time of the query. By default the start time used by otp.run will be inherited.
db (str) – Custom database name for the node of the graph. By default the database used by otp.run will be inherited.
tick_type (str) – Custom tick type for the node of the graph. By default “ANY” tick type will be set.
symbols (str or list of str) –
Custom symbol name for the node of the graph or list of symbols. If list of symbols is specified, ticks from different symbols will be merged into one source.

Separate ODBC connection will be created for each processed symbol. To avoid this and have a single connection per thread, presort parameter can be specified.

By default the symbol name used by otp.run will be inherited.
presort (bool) – Adds PRESORT EP before merging bound symbols specified in symbols. That make ODBC use single connection for all symbols.
kwargs –
Set the schema of the python Source object of this class.

Schema can’t be taken automatically from the database, so it should be set manually for python-level type checking to work.

Note

To be able to use this class, you need to have a ODBC driver manager available on your machine (comes with Windows OS; unixodbc package may need to be installed on UNIX OS).

Also, the following entry needs to be added to the main configuration file:

LOAD_ODBC_UDF=true

Examples

Connect with database’s dsn, manually set schema, get all data from TEST_TABLE:

>>> data = otp.ODBC(dsn='testdb_dsn', sql='select * from TEST_TABLE',
...                 A=str, B=int, C=float, D=otp.nsectime)  
>>> otp.run(data)  
        Time   A     B        C                       D
0 2003-12-04  A1  1975  8.12345 2022-01-01 12:13:14.111
1 2003-12-04  A2  1971  7.98765 2022-01-02 22:23:24.222

Connect using connection_string parameter:

>>> data = otp.ODBC(connection_string='DRIVER={SQLite3};Database=/path/to/the/database',
...                 sql='select * from TEST_TABLE')  
>>> otp.run(data)  
        Time   A     B        C                       D
0 2003-12-04  A1  1975  8.12345 2022-01-01 12:13:14.111
1 2003-12-04  A2  1971  7.98765 2022-01-02 22:23:24.222

Substitute start time placeholder with start_expr parameter:

>>> data = otp.ODBC(
...     dsn='testdb_dsn',
...     sql='select * from TEST_TIMESTAMP where TIMESTAMP >= "<_START_TIME>"',
...     start_expr=(otp.meta_fields['START_TIME'] + otp.Day(1)).dt.strftime('%Y-%m-%d %H:%M:%S.%q')
... )  
>>> otp.run(data, start=otp.dt(2022, 1, 1), end=otp.dt(2022, 1, 3))  
                     Time  A
0 2022-01-02 22:23:24.222  2

Use parameter allow_unordered_ticks if needed:

>>> data = otp.ODBC(dsn='testdb_dsn',
...                 sql='select * from TEST_UNORDERED',
...                 allow_unordered_ticks=True)  
>>> otp.run(data, start=otp.dt(2022, 1, 1), end=otp.dt(2022, 1, 3))  
                     Time  A
0 2022-01-02 22:23:24.222  2
1 2022-01-01 12:13:14.111  1