KDB-X Database Service API (0.1.0)

Target table name.

Path to input data on the server filesystem.

Accepted forms:

• Single file
• Directory
• Glob pattern

Glob semantics:

• * matches within a single path segment (does not cross /).
• ** matches across subdirectories, e.g. quotes/**/quote*.csv.

The service ingests all matched files.

Example values:

• quote.csv
• quotes/
• quote_2026-01-*.csv
• quotes/**/quote*.csv

Input data format. If omitted, the service may auto-detect it from the file(s).

Optional ingest mode. If omitted, the service uses merge behavior. Mode behavior is applied per partition.

Delimiter character for delimited text. Ignored for parquet and qbinary.

Decimal separator character for delimited text. Ignored for parquet and qbinary.

If true and the table does not exist, create it prior to ingest.

If the file has headers, this is optional.

If provided, it overrides header names by position.

If the file has no headers and either postparse is used or the file has unused columns, this is required (names by position).

If createTable=true and the table doesn’t exist, these headers are used as the new table’s column names.

Optional list of column names to ingest (and their order).

Only these columns are kept.

Applied after both headers are resolved & postparse is applied.

Optional type string in CSV parse format (uppercase), e.g. F=float, J=long, P=timestamp, S=symbol, *=string. See https://code.kx.com/q/ref/tok/ for full details.

A dictionary of data transforms, made up of column names as keys and q expression strings as values.

Existing columns can be modified, and new ones created.

Columns from the file are available for use in the expression via a data dictionary variable, and the file name is available via a path string variable.

For example: {'timestamp': 'data[`date]+data[`time]'}.

Runs after loading and parsing (including typecasting) of data in the file is done.

Note that columns used as inputs in the transform are by default not included after use, therefore must be re-added if they are required.

For example, to preserve the date column in the above, the postparse dictionary should be: {'timestamp': 'data[`date]+data[`time]', 'date': 'data[`date]'}

Path to a kdb-native data location readable by the service, such as a partitioned HDB directory, a parent directory containing one or more splayed tables, or a parent directory containing basic/q-binary tables. Path is resolved relative to the configured import root.

Optional ingest mode. If omitted, the service uses merge behavior. Mode behavior is applied per partition.

Target table name.

Optional ingest mode. If omitted, the service uses merge behavior. Mode behavior is applied per partition.

Table name

Inclusive start of timestamp range

Exclusive end of timestamp range

Timezones should be supplied in tz database format, for example America/New_York or Europe/London. inputTZ is used to interpret startTS and endTS.

Timezones should be supplied in tz database format, for example America/New_York or Europe/London. outputTZ is used for timestamp conversion in query output.

Columns that should be converted to 'outputTZ'

The filter parameter takes a list of filter triples in the following form:

• [operator, column, values]

It is also possible to nest filters with the logical operators and, not, and or, for example:

• ["and", ["=", "size", 100], ["<", "price", 500]]

Supported filter operators are =, <>, <, >, <=, >=, in, like, within.

A list of column names to group by

agg controls what the query returns. Use it to select columns, rename columns, or compute aggregated values.

1. A simple list of column names specifies what columns to include in the result:

["sym", "bid", "ask"]

2. A list of tuples specifies what columns to include in the result, and renames them:

[["Symbol", "sym"], ["BidPrice", "bid"], ["AskPrice", "ask"]]

3. A list of triples runs the given aggregation function on the specified column and renames it:

[
  ["o", "first", "bid"],
  ["h", "max", "bid"],
  ["l", "min", "bid"],
  ["c", "last", "bid"]
]

Available aggregation functions are:

all, any, avg, cor, count, cov, dev, distinct, first, last, max, min, prd, scov, sdev, sum, svar, var, wavg, wsum.

Note: distinct always returns a list.

fill can be used to fill in null values:

• forward will replace nulls with the previous non-null value.
• zero will replace nulls with the 0 equivalent value of the column type.

Temporality controls how the query interprets the time range between startTS and endTS.

Supported values are:

• snapshot Returns one continuous block of data from startTS to endTS. This is the default if temporality is not specified.

• slice Returns data between the time of startTS and the time of endTS for each day in the date range. For example, if startTS is 2021.01.01D10:00:00 and endTS is 2021.01.03D13:00:00, data is returned from 10:00:00 to 13:00:00 on each of these dates:

  • 2021.01.01
  • 2021.01.02
  • 2021.01.03

Columns to be sorted, post query execution

Freeform q query