Documentation Portal

The list.sql file will contain a SQL SELECT statement and shall return all the tags with additional metadata as shown below. All your returned types should be mapped onto one of the supported tag types with the type.csv file.

The query should return a result with column names "tagname" (mandatory), "type" (mandatory), "units" (optional), and "description" (optional). All values for the tagnames returned by the query should be unique. Duplicates will lead to synch errors later on when connecting the datasource to TrendMiner in later steps. Stored procedures can be used for retrieving tags as long as they return a result set.

An example of a very simple query is given below. Note that the type has been hardcoded to be "analog". For additional examples, refer to the following article: Examples of SQL queries for ODBC connections.

SELECT
    tagname, description, units, "analog" as type
FROM Table_1

A correct output of the query is shown below:

The values.sql file will contain a SQL SELECT statement and shall return all the values and timestamps for a given tagname and date range. Note that these parameters (tagname, start date, and end date) will be automatically passed on by TrendMiner when in production as binding parameters.

These binding parameters are specified in the query by using question marks "?". If binding parameters (?) are not supported then .NET string formatting can be used, so use '{0}', '{1}', and '{2}' (include single quotes) instead of the question mark.

The query should return a result with column names "ts" (mandatory) and "value" (mandatory). The timestamps in the ts column should be ordered in ascending order (oldest point first, newest point last). Stored procedures can be used for retrieving tag data as long as they return a result set.

Since this is a crucial step, here is a summary of the output requirements:

An example of a simple query is given below. For additional examples, refer to the following article: Examples of SQL queries for ODBC connections.

SELECT
    ts, value
FROM Table_1   
WHERE tagname = ? AND ts IS BETWEEN ? AND ? ORDER BY ts ASC

An example of the correct output is shown below.

The types returned in the list.sql must be mapped to the types supported by TrendMiner, also if the returned type is the same as a TrendMiner supported type. You need to generate your own mapping. It is best practice to explicitly map each possible type returned by list.sql to one supported by TrendMiner. If no mapping/type.csv is provided all tags types will default to analog.

In the type.csv file, create one line per data type, with the output of the list.sql types as the first value, followed by a semicolon (";"), followed by one of the supported datatypes in TrendMiner. The example below ensures the results are returned in the right case, and a DIGITAL result is explicitly returned as a STRING result.

In case the timestamps returned in the values.sql query are not in the format "yyyy-MM-dd HH:mm:ss.fff” (e.g. 2022-12-03 14:28:23.357), then the format can be configured by setting it in the datetime_format.txt file. Simply edit the txt file and paste the format in the file. The format is interpreted following the .NET formatting specification.

Alternatively, the file can contain the string “EPOCH”, in which case dates are converted to the equivalent number of seconds since the Epoch and the converted values are used in the query.

Some examples are specified below.

It is advised to always use timestamps in UTC. This will avoid confusion when troubleshooting, especially when datasources from different timezones are connected since TrendMiner uses UTC by default. If timestamps are stored in the datasource in local time, it is necessary to provide the timezone information.

Specify the time zone name in the timezone.txt file, using "Time zone name" as listed on https://docs.microsoft.com/en-us/previous-versions/windows/embedded/gg154758(v=winembedded.80).