SRC

All tql scripts must start with one of the data source functions.

There are several SRC functions are included. For example, SQL() produces records by querying machbase-neo database or even external (bridged) database with the given sql statement. FAKE() generates artificial data. CSV() can read csv data, BYTES() reads arbitrary binary data from file system or client’s HTTP request and MQTT payload.

tql_src

SQL()

Syntax: SQL( [bridge(),] sqltext [, params...])

  • bridge() bridge(’name’) If a bridge is given, the SQL query is executed on the bridge
  • sqltext string : SQL SELECT statement to retrieve data from database. Use backtick(`) for multi-line sql string.
  • params : Variadic arguments for the bind arguments for the query.

Example)

  • Query to machbase
SQL (`
    SELECT time, value 
    FROM example 
    WHERE name ='temperature'
    LIMIT 10000
`)
  • Query with variadic arguments
SQL(`SELECT time, value FROM example WHERE name = ? LIMIT ?`,
    param('name') ?? 'temperature',
    param('limit') ?? 10)
  • Query to bridge database
SQL( bridge('sqlite'), `SELECT * FROM EXAMPLE`)
SQL(
    bridge('sqlite'),
    `SELECT time, value FROM example WHERE name = ?`,
    param('name') ?? "temperature")

SQL_SELECT()

Syntax: SQL_SELECT( fields..., from(), between() [, limit()] ) Since v8.0.15

  • fields string : column names, multiple columns are possible.

SQL_SELECT() source function provides same functionality with SQL(), but it simplifies the usage by standardization option functions other than using the raw SQL statement.

This function actually works equivalent to SQL() but it takes query conditions via simplified options instead of a full SQL statement. It assigns time range condition easier way than using WHERE condition in SQL statement.

The example below process data by query SELECT time, value FROM example WHERE NAME = 'temperature' AND time BETWEEN....

SQL_SELECT(
    'time', 'value',
    from('example', 'temperature'),
    between('last-10s', 'last')
)

is equivalent with the SQL() statement below

SQL(`SELECT
        time, value
    FROM
        EXAMPLE
    WHERE
        name = 'TAG1'
    AND time BETWEEN (
        SELECT MAX_TIME-10000000000
        FROM V$EXAMPLE_STAT
        WHERE name = 'temperature')
    AND (
        SELECT MAX_TIME
        FROM V$EXAMPLE_STAT
        WHERE name = 'temperature')
    LIMIT 0, 1000000
`)

from()

Syntax: from( table, tag [, time_column [, name_column] ] )

It provides table name and tag name to SQL_SELECT() function generating SQL internally. It may equivalent to ... FROM <table> WHERE NAME = <tag> ....

  • table string table name
  • tag string tag name
  • time_column string specify “time” column name, if omitted default is 'time'.
  • name_column string specify “name” column name, if omitted default is 'name'. Since v8.0.5

between()

Syntax: between( fromTime, toTime [, period] )

It provides time range condition to SQL_SELECT() function generating SQL internally. It may equivalent to ... WHERE ... TIME BETWEEN <fromTime> AND <toTime>....

  • fromTime string,number time expression with ’now’ and ’last’ as string, Or assign number as unix epoch time in nanosecond
  • toTime string,number time expression
  • period string,number duration expression, Or assign number for the unix epoch time in nanoseconds. Logically only positive period makes sense.

You can specify fromTime and toTime with ’now’ and ’last’ with delta duration expression. For example, 'now-1h30m' specifies the time that 1 hour 30 minutes before from now. 'last-30s' means 30 seconds before the latest(=max) time of the base_time_column.

If period is specified it will generate ‘GROUP BY’ time expression with aggregation SQL functions. In this case the base time column should be included in the fields arguments of the SQL_SELECT().

If it is required to use string expressions for the fromTime, toTime instead of unix epoch nano seconds, use parseTime() to convert string-time-expression to time value. Refer to the document about utility function parseTime()

Example)

between( parseTime("2023-03-01 14:00:00", "DEFAULT", tz("Local")),
         parseTime("2023-03-01 14:05:00", "DEFAULT", tz("Local")))

limit()

Syntax: limit( [offset ,] count )

It will be translated into SELECT... LIMIT offset, count statement.

  • offset number default is 0. if omitted
  • count number

QUERY()

Syntax: QUERY( fields..., from(), between() [, limit()] )

  • fields string : column names, multiple columns are possible.

Deprecated, Use SQL_SELECT() instead.

QUERY() is almost same with SQL_SELECT() except that QUERY produce a query SQL which always implicitly includes the base ’time’ column as the first column, even it is not specified in fields arguments.

The base ’time’ column will be always the first column of the SELECT query.

QUERY(
    'value',
    from('example', 'temperature'),
    between('last-10s', 'last')
)

is equivalent with the SQL() statement below

SQL(`SELECT
        time, value
    FROM
        EXAMPLE
    WHERE
        name = 'TAG1'
    AND time BETWEEN (
        SELECT MAX_TIME-10000000000
        FROM V$EXAMPLE_STAT
        WHERE name = 'temperature')
    AND (
        SELECT MAX_TIME
        FROM V$EXAMPLE_STAT
        WHERE name = 'temperature')
    LIMIT 0, 1000000
`)

CSV()

Syntax: CSV( file(file_path_string) | payload() [, charset()] [,field()...[, header()]] )

Load CSV and yield key-value records, the key is generated in sequence number and the fields of CSV become the value of the record. The string parameter of ‘file’ should be absolute path to the CSV. If payload() is used, it will reads CSV from HTTP POST request body stream. It is useful to make an API that writes data into database when remote client sends data by HTTP POST.

  • file() | payload() input stream
  • field(idx, type, name) specifying fields
  • header(bool) specifies if the first line of input stream is a header
  • charset(string) specify charset if the CSV data is in non UTF-8. Since v8.0.8 .

Example)

// Read CSV from HTTP request body.
// ex)
// barn,1677646800,0.03135
// dew_point,1677646800,24.4
// dishwasher,1677646800,3.33e-05
CSV(payload(), 
    field(0, stringType(), 'name'),
    field(1, datetimeType('s'), 'time'),
    field(2, doubleType(), 'value'),
    header(false)
)
APPEND(table('example'))

Combination of CSV() and APPEND() as above example, it is simple, useful. Be aware that it is 5 times slower than command line import command, but still faster than INSERT() function when writing more than several thousands records per a HTTP request.

Use ?? operator to make it works with or without HTTP POST request.

CSV(payload() ?? file('/absolute/path/to/data.csv'),
    field(0, doubleType(), 'freq'),
    field(1, doubleType(), 'ampl')
)
CHART_LINE()

file()

Syntax: file(path)

Open the given file and returns input stream for the content. the path should be the absolute path to the file.

  • path string path to the file to open, or http url where to get resource from.

If path starts with “http://” or “https://”, it retrieves the content of the specified http url Since v8.0.7 . Otherwise it is looking for the path on the file system.

The code below shows how to call a remote HTTP api with file() that it actually invokes machbase-neo itself for the demonstration purpose, the SQL query which safely url escaped by escapeParam().

CSV( file(`http://127.0.0.1:5654/db/query?`+
        `format=csv&`+
        `q=`+escapeParam(`select * from example limit 10`)
))
CSV() // or JSON()

payload()

Syntax: payload()

Returns the input stream of the request content if the tql script has been invoked from HTTP POST or MQTT PUBLISH.

field()

Syntax: field(idx, typefunc, name)

Specify field-types of the input CSV data.

  • idx number 0-based index of the field.
  • typefunc specify the type of the field. (see below)
  • name string specify the name of the field.
type functiontype
stringType()string
doubleType()number
datetimeType()datetime

The stringType() and doubleType() take no arguments, datetimeType() function takes one or two parameters for proper conversion of date-time data.

If the input data of the field specifies time in unix epoch time, specify the one of the time units ns, us, ms and s.

  • datetimeType('s')
  • datetimeType('ms')
  • datetimeType('us')
  • datetimeType('ns')

The input field represents time in human readable format, it is requires to specifying how to parse them including time zone.

  • datetimeType('DEFAULT', 'Local')
CSV(payload() ??
`name,2006-01-02 15:04:05.999,10`,
field(1, datetimeType('DEFAULT', 'Local'), 'time'))
CSV()
  • datetimeType('RFC3339', 'EST')
CSV(payload() ??
`name,2006-01-02T15:04:05.999Z,10`,
field(1, datetimeType('RFC3339', 'EST'), 'time'))
CSV()

If the timezone is omitted, it assumes ‘UTC’ by default.

  • datetimeType('RC822')

The first argument of the datetimeType() directs how to parse the input data that uses the same syntax with timeformat() function. Please refer to the description of the timeformat() function for the timeformat spec.

charset()

Syntax: charset(name) Since v8.0.8

  • name string character set name

Supported character set.

UTF-8, ISO-2022-JP, EUC-KR, SJIS , CP932, SHIFT_JIS, EUC-JP, UTF-16, UTF-16BE, UTF-16LE, CP437, CP850, CP852, CP855, CP858, CP860, CP862, CP863, CP865, CP866, LATIN-1, ISO-8859-1, ISO-8859-2, ISO-8859-3, ISO-8859-4, ISO-8859-5, ISO-8859-6, ISO-8859-7, ISO-8859-8, ISO-8859-10, ISO-8859-13, ISO-8859-14, ISO-8859-15, ISO-8859-16, KOI8R, KOI8U, MACINTOSH, MACINTOSHCYRILLIC, WINDOWS1250, WINDOWS1251, WINDOWS1252, WINDOWS1253, WINDOWS1254, WINDOWS1255, WINDOWS1256, WINDOWS1257, WINDOWS1258, WINDOWS874, XUSERDEFINED, HZ-GB2312

BYTES(), STRING()

Syntax: BYTES( src [, separator(char), trimspace(boolean) ] )

Syntax: STRING( src [, separator(char), trimspace(boolean) ] )

  • src data source, it can be one of payload(), file() and a constant text string.
  • separator(char) optional setseparator("\n") to read a line by line, or omit it to read whole string in a time.
  • trimspace(boolean) optional trim spaces, default is false

Split the input content by separator and yield records that separated sub content as value and the key is increment number of records.

Example)

  • STRING('A,B,C', separator(",")) yields 3 records ["A"], ["B"] and ["C"].
  • STRING('A,B,C') yields 1 record ["A,B,C"].

The both of BYTES() and STRING() works exactly same, except the value types that yields, as the function’s name implies, BYTES() yields value in ‘array of byte’ and STRING() yields string value.

STRING(payload() ?? `12345
                    23456
                    78901`, separator("\n"))

The example code above generates 3 records ["12345"], ["␣␣␣␣␣␣␣␣␣␣23456"], ["␣␣␣␣␣␣␣␣␣␣78901"].

STRING(payload() ?? `12345
                    23456
                    78901`, separator("\n"), trimspace(true))

The example code above generates 3 records ["12345"], ["23456"], ["78901"].

STRING( file(`http://example.com/data/words.txt`), separator("\n") )

It retrieves the content from the http address. file() supports http url Since v8.0.7 .

ARGS()

Syntax: ARGS() Since v8.0.7

ARGS generates a record that values are passed by parent TQL flow as arguments. It is purposed to be used as a SRC of a sub flow within a WHEN...do() statement.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
FAKE( json({
    [ 1, "hello" ],
    [ 2, "world" ]
}))
WHEN( value(0) == 2, do( value(0), strToUpper(value(1)), {
    ARGS()
    WHEN( true, doLog("OUTPUT:", value(0), value(1)) )
    DISCARD()
}))
CSV()

The code print log message on the output console.

OUTPUT: 2 WORLD

FAKE()

Syntax: FAKE( generator )

  • generator one of the oscillator(), meshgrid(), linspace(), arrange(), csv(), json()

Producing “fake” data by given generator.

oscillator()

Syntax: oscillator( freq() [, freq()...], range() )

Generating wave data by given frequency and time range. If provide multiple freq() arguments, it composites waves.

freq()

Syntax: freq( frequency, amplitude [, bias, phase])

It produce sine wave by time amplitude * SIN( 2*Pi * frequency * time + phase) + bias.

  • frequency number frequency in Hertz (Hz).
  • amplitude number
  • bias number
  • phase number in radian

range()

Syntax: range( fromTime, duration, period )

It specifies time range from fromTime to fromTime+duration.

  • fromTime string|number ’now’ and ’last’ is available for string type, Or assign number as unix epoch time in nanosecond
  • duration string|number duration expression, Or assign number in nanoseconds. ex) '-1d2h30m', '1s100ms'
  • period string|number duration expression, Or assign number in nanoseconds. Logically only positive period makes sense.

arrange()

Syntax: arrange(start, stop, step) Since v8.0.12

  • start number
  • stop number
  • step number
1
2
3
4
FAKE(
   arrange(1, 2, 0.5)
)
CSV()
1
1.5
2

linspace()

Syntax: linspace(start, stop, num)

It generates 1 dimension linear space.

meshgrid()

Syntax: meshgrid(xseries, yseries)

It generates meshed values - [xseries, yseries].

csv()

Syntax: csv(content) Since v8.0.7

  • content string csv content

It generates records from the given csv content.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
FAKE(
    csv( strTrimSpace(`
        A,1,true
        B,2,false
        C,3,true
    `))
)
MAPVALUE(0, strTrimSpace(value(0)))
MAPVALUE(1, parseFloat(value(1))*10)
MAPVALUE(2, parseBool(value(2)))
CSV()
A,10,true
B,20,false
C,30,true

json()

Syntax: json({...}) Since v8.0.7

It generates records from the given multiple json array.

1
2
3
4
5
6
7
8
9
FAKE(
    json({
        ["A", 1, true],
        ["B", 2, false],
        ["C", 3, true]
    })
)
MAPVALUE(1, value(1)*10)
CSV()
A,10,true
B,20,false
C,30,true
Last updated on