Skip to main content
Version: 0.3.x

sqrl compile

The sqrl compile command is used to facilitate testing of generated SQL queries from SQL templates and parameter selections.

All the details on command line arguments can be found using sqrl compile --help or sqrl compile -h. The result is as follows:

usage: sqrl compile [-h] (-d DATASET | -D) [-t TEST_SET | -T] [-s SELECT] [-r]

optional arguments:
  -h, --help            Show this help message and exit
  -d DATASET, --dataset DATASET
                        Select dataset to use for dataset traits. Is required, unless using --all-datasets
  -D, --all-datasets    Compile models for all datasets. Only required if --dataset is not specified
  -t TEST_SET, --test-set TEST_SET
                        The selection test set to use. If not specified, default selections are used, unless using --all-test-sets
  -T, --all-test-sets   Compile models for all selection test sets
  -s SELECT, --select SELECT
                        Select single model to compile. If not specified, all models for the dataset are compiled. Ignored if using --all-datasets    
  -r, --runquery        Runs all target models, and produce the results as csv files

Unless using the -h option, only one of --dataset and --all-datasets options (or -d and -D for short) can be specified, and both cannot be specified at the same time. Also, --test-set and --all-test-sets options (or -t and -T for short) cannot both be specified at the same time.

For each combination of dataset and test set being compiled for, the following files are produced:

  • SQL files for the compiled SQL models
  • A JSON file for the dictionary of placeholder names and values
  • A PNG file for the lineage of the models
  • If running with --runquery or -r option, CSV files for the result of the models

All files are written to some target/compile/<dataset>/<test_set>/ folder (after deleting all pre-existing files in that folder). The folders created depends on the command options specified:

  • If --dataset is specified, then the specified dataset is the name of the <dataset>/ folder.
  • If --all-dataset is specified, then a <dataset>/ folder is created for every dataset.
  • If neither --test-set or --all-test-sets is specified, then the <test_set>/ folder is simply the default test set for the dataset (typically called default/ unless specified otherwise).
  • If --test-set is specified, then the specified test set is the name of the <test_set>/ folder
  • If --all-test-sets is specified, then a <test_set>/ folder is created for all test sets applicable to the dataset.

Also, if --select is specified (in addition to --dataset), the compiled query of the selected model is printed in terminal. The <dataset>/ folder will only contain the compiled query of the selected model as well (without the upstream models). An exception to this is when --runquery is specified, since upstream model results must be created first in order to create the selected model result.

For more information on using selection test sets, see the "selection_test_sets" section of the Squirrels Project File page.