Development¶

This module contains the classes

refactor, a module that assists in refactoring Modelica classes,
Tester that runs the unit tests of the Buildings library,
Validator that validates the html code of the info section of the .mo files, and
IBPSA that synchronizes Modelica libraries with the IBPSA library.
ErrorDictionary that contains information about possible error strings.
Comparator that compares the simulation performance across simulation tools or git branches.

Refactoring code¶

This module provides functions to

create Modelica packages and autopopulate for example the package.mo and package.order files
move Modelica classes include any associated .mos scripts, reference results and images, and
rewrite the package.order file.

buildingspy.development.refactor.create_modelica_package(directory)¶

Create in directory a Modelica package.

If directory/package.mo exists, this method returns and does nothing. Otherwise, it creates the directory and populates it with a package.mo and package.order file.

Parameters: directory – The directory in which the package is created.

buildingspy.development.refactor.move_class(source, target)¶

Move the class from the source to the target name.

Both arguments need to be Modelica class names such as Buildings.Fluid.Sensors.TemperatureTwoPort, or a directory with a top-level package.mo file, such as Buildings/Fluid, that contains a file Buildings/Fluid/package.mo.

Parameters

source – Class name of the source.
target – Class name of the target.

Usage: Type

>>> import buildingspy.development.refactor as r
>>> old = "Buildings.Fluid.Movers.FlowControlled_dp"
>>> new = "Buildings.Fluid.Movers.Flow_dp"
>>> r.move_class(old, new) 

buildingspy.development.refactor.write_package_order(directory='.', recursive=False)¶

Write the package.order file in the directory directory.

Any existing package.order file will be overwritten.

Parameters

directory – The name of the directory in which the package.order file will be written.
recursive – Set to True to recursively include all sub directories.

Usage: To rewrite package.order in the current directory, type

>>> import buildingspy.development.refactor as r
>>> r.write_package_order(".") 

Regression tests¶

class buildingspy.development.regressiontest.Tester(check_html=True, tool='dymola', cleanup=True, comp_tool='funnel', tol=0.001, skip_verification=False, color=False, rewriteConfigurationFile=False)¶

Class that runs regression tests of the Modelica library.

Initiate with the following optional arguments:

Parameters

check_html – Boolean (default True). Specify whether to load tidylib and perform validation of html documentation.
tool – string {'dymola', 'openmodelica', 'optimica'}. Default is 'dymola', specifies the tool to use for running the regression test with run().
cleanup – Boolean (default True). Specify whether to delete temporary directories.
tol – float or dict (default=1E-3). Comparison tolerance If a float is provided, it is assigned to the absolute tolerance along x axis and to the absolute and relative tolerance along y axis. (If comp_tool='legacy', only the absolute tolerance in y is used.) If a dict is provided, keys must conform with pyfunnel.compareAndReport arguments.
skip_verification – Boolean (default False). If True, unit test results are not verified against reference points.
color – Boolean (default False). If True, command line output will be in color (ignored on Windows).
rewriteConfigurationFile – Boolean (default False). If True, rewrite conf.yml file. (Currently only supported for OpenModelica.)

This class can be used to run all regression tests.

Regression testing using Dymola

For Dymola, this module searches the directory CURRENT_DIRECTORY/Resources/Scripts/Dymola for all *.mos files that contain the string simulate, where CURRENT_DIRECTORY is the name of the directory in which the Python script is started, as returned by the function getLibraryName(). All these files will be executed as part of the regression tests. Any variables or parameters that are plotted by these *.mos files will be compared to previous results that are stored in CURRENT_DIRECTORY/Resources/ReferenceResults/Dymola. If no reference results exist, then they will be created. Otherwise, the accuracy of the new results is compared to the reference results. If they differ by more than a prescibed tolerance, a plot such as the one below is shown.

_images/unitTestPlot.png — Plot that compares the new results (solid line) of the regression test with the old results (dotted line). The blue line indicates the time where the largest error occurs.¶

In this plot, the vertical line indicates the time where the biggest error occurs. The user is then asked to accept or reject the new results.

For Dymola, the regression tests also store and compare the following statistics for the initialization problem and the time domain simulation:

The number and the size of the linear system of equations,

the number and the size of the nonlinear system of equations, and

the number of the numerical Jacobians.

To run the regression tests, type

>>> import os
>>> import buildingspy.development.regressiontest as r
>>> rt = r.Tester(tool="dymola")
>>> myMoLib = os.path.join("buildingspy", "tests", "MyModelicaLibrary")
>>> rt.setLibraryRoot(myMoLib)
>>> rt.run()                    
MyModelicaLibrary.Examples.NoSolution: Excluded from simulation. Model excluded from simulation as it has no solution.
Number of models   : 11
          blocks   : 2
          functions: 0
Using ... of ... processors to run unit tests for dymola.
Generated 7 regression tests.

Comparison files output by funnel are stored in the directory 'funnel_comp' of size ... MB.
Run 'python -c "import buildingspy.development.regressiontest as t; t.Tester(tool=\"dymola\").report()"'
to access a summary of the comparison results.

Script that runs unit tests had 0 warnings and 0 errors.

See 'simulator-dymola.log' for details.
Unit tests completed successfully.

Execution time = ...

To run regression tests only for a single package, call setSinglePackage() prior to run().

Regression testing using OpenModelica or OPTIMICA

For OpenModelica and OPTIMICA, the selection of test cases is done the same way as for Dymola. However, the solver tolerance is obtained from the .mo file by reading the annotation Tolerance=”value”.

For OpenModelica and OPTIMICA, a JSON file stored as Resources/Scripts/BuildingsPy/conf.yml (or for backward compatibility, in conf.json) can be used to further configure tests. The file has the syntax below, where openmodelica or optimica specifies the tool.

- model_name: Buildings.Fluid.Examples.FlowSystem.Simplified2
  optimica:
    ncp: 500,
    rtol: 1E-6,
    solver: "CVode",
    simulate: True,
    translate: True,
    time_out: 600
    comment: 'Add some comment and ideally a link to a github issue'

For OpenModelica, replace optimica with openmodelica. For the detailed specification of allowed fields, see buildingspy/templates/regressiontest_conf.py.

Any entries are optional, and the entries shown above are the default values, except for the relative tolerance rtol which is read from the .mo file. However, with rtol, this value can be overwritten. Note that this syntax is still experimental and may be changed.

areResultsEqual(tOld, yOld, tNew, yNew, varNam, data_idx)¶

Return True if the data series are equal within a tolerance.

Parameters

tOld – List of old time values.
yOld – Old simulation results.
tNew – Time stamps of new results.
yNew – New simulation results.
varNam – Variable name, used for reporting.
filNam – File name, used for reporting.
model_name – Model name, used for reporting.

Returns

A list with False if the results are not equal, and the time of the maximum error, and an error message or None. In case of errors, the time of the maximum error may by None.

are_statistics_equal(s1, s2)¶: Compare the simulation statistics s1 and s2 and return True if they are equal, or False otherwise.

batchMode(batchMode)¶

Set the batch mode flag.

Parameters: batchMode – Set to True to run without interactive prompts and without plot windows.

By default, the regression tests require the user to respond if results differ from previous simulations. This method can be used to run the script in batch mode, suppressing all prompts that require the user to enter a response. If run in batch mode, no new results will be stored. To run the regression tests in batch mode, enter

>>> import os
>>> import buildingspy.development.regressiontest as r
>>> r = r.Tester()
>>> r.batchMode(True)
>>> r.run() 

checkPythonModuleAvailability()¶

Check whether all required python modules are installed.

If some modules are missing, then an ImportError is raised.

deleteTemporaryDirectories(delete)¶

Flag, if set to False, then the temporary directories will not be deleted after the regression tests are run.

Parameters: delete – Flag, set to False to avoid the temporary directories to be deleted.

Unless this method is called prior to running the regression tests with delete=False, all temporary directories will be deleted after the regression tests.

static expand_packages(packages)¶: Expand the packages from the form A.{B,C} and return A.B,A.C :param: packages: A list of packages

format_float(value)¶: Return the argument in exponential notation, with non-significant zeros removed.

getLibraryName()¶

Return the name of the library that will be run by this regression test.

Returns: The name of the library that will be run by this regression test.

getModelicaCommand()¶

Return the name of the modelica executable.

Returns: The name of the modelica executable.

get_configuration_data_from_disk()¶: Read the configuration data conf.yml from disk and return it as an arry whose elements are json dictionaries.

get_configuration_file_name()¶: Returns the full name of the conf.yml file.

get_number_of_tests()¶

Returns the number of regression tests that will be run for the current library and configuration.

Note: Needs to be run within the run method (where elements of self._data requiring no simulation are first removed).

static get_plot_variables(line)¶

For a string of the form *y={aa,bb,cc}*, optionally with whitespace characters, return the list [aa, bb, cc]. If the string does not contain y = …, return None.

A usage may be as follows. Note that the second call returns None as it has a different format.

>>> import buildingspy.development.regressiontest as r
>>> r.Tester.get_plot_variables('y = {"a", "b", "c"}')
['a', 'b', 'c']
>>> r.Tester.get_plot_variables('... x}, y = {"a", "b", "c"}, z = {...')
['a', 'b', 'c']
>>> r.Tester.get_plot_variables("y=abc") is None
True

static get_tolerance(library_home, model_name)¶

Return the tolerance as read from the .mo file.

Parameters

library_home – Home directory of the library.
model_name – Name of the model.

get_unit_test_log_files()¶: Return the name of the logs files of the unit tests, such as unitTests-openmodelica.log, unitTests-optimica.log or unitTests-dymola.log and simulator-openmodelica.log etc.

include_fmu_tests(fmu_export)¶

Sets a flag that, if False, does not test the export of FMUs.

Parameters: fmu_export – Set to True to test the export of FMUs (default), or False to not test the FMU export.

To run the unit tests but do not test the export of FMUs, type

>>> import os
>>> import buildingspy.development.regressiontest as r
>>> r = r.Tester()
>>> r.include_fmu_tests(False)
>>> r.run() 

isExecutable(program)¶: Return True if the program is an executable

static isValidLibrary(library_home)¶

Returns true if the regression tester points to a valid library: that implements the scripts for the regression tests.

Parameters: library_home – top-level directory of the library, such as Buildings.
Returns: True if the library implements regression tests, False otherwise.

pedanticModelica(pedanticModelica)¶

Set the pedantic Modelica mode flag.

Parameters: pedanticModelica – Set to True to run the unit tests in the pedantic Modelica mode.

By default, regression tests are run in non-pedantic Modelica mode. This however will be changed in the near future.

>>> import os
>>> import buildingspy.development.regressiontest as r
>>> r = r.Tester()
>>> r.pedanticModelica(True)
>>> r.run() 

printNumberOfClasses()¶: Print the number of models, blocks and functions to the standard output stream

report(timeout=600, browser=None, autoraise=True, comp_file=None)¶

Builds and displays HTML report.

Serves until timeout (s) or KeyboardInterrupt.

return_new_configuration_data_using_CI_results(configuration_data, simulator_log_file_json, tool, list_of_fmu_exports=None)¶

Compares the entry from the simulator_log_file_json with the configuration_data, and returns a new copy of configuration_data with changes to translate or simulate entries based on the CI tests. This method also removes comments if the translation or simulation was successful.

For models in the list list_of_fmu_exports, no entry about simulation will be added.

run()¶

Run all regression tests and checks the results.

Returns: 0 if no errors and no warnings occurred during the regression tests, otherwise a non-zero value.

This method

creates temporary directories for each processors,
copies the directory CURRENT_DIRECTORY into these temporary directories,
creates run scripts that run all regression tests,
runs these regression tests,
collects the dymola log files from each process,
writes the combined log file unitTests-x.log to the current directory, where x is the name of the Modelica tool,
for Dymola, compares the results of the new simulations with reference results that are stored in Resources/ReferenceResults,
writes the message Regression tests completed successfully. if no error occurred,
returns 0 if no errors and no warnings occurred, or non-zero otherwise.

setDataDictionary(root_package=None)¶

Build the data structures that are needed to parse the output files.

Param: root_package The name of the top-level package for which the files need to be parsed. Separate package names with a period.

setLibraryRoot(rootDir)¶

Set the root directory of the library.

Parameters: rootDir – The top-most directory of the library.

The root directory is the directory that contains the Resources folder and the top-level package.mo file.

Usage: Type

>>> import os
>>> import buildingspy.development.regressiontest as r
>>> rt = r.Tester()
>>> myMoLib = os.path.join("buildingspy", "tests", "MyModelicaLibrary")
>>> rt.setLibraryRoot(myMoLib)

setNumberOfThreads(number)¶

Set the number of parallel threads that are used to run the regression tests.

Parameters: number – The number of parallel threads that are used to run the regression tests.

By default, the number of parallel threads are set to be equal to the number of processors of the computer.

setSinglePackage(packageName)¶

Set the name of one or multiple Modelica package(s) to be tested.

Parameters: packageName – The name of the package(s) to be tested.

Calling this method will cause the regression tests to run only for the examples in the package packageName, and in all its sub-packages.

For example:

If packageName = IBPSA.Controls.Continuous.Examples, then a test of the IBPSA library will run all examples in IBPSA.Controls.Continuous.Examples.
If packageName = IBPSA.Controls.Continuous.Examples,IBPSA.Controls.Continuous.Validation, then a test of the IBPSA library will run all examples in IBPSA.Controls.Continuous.Examples and in IBPSA.Controls.Continuous.Validation.

showGUI(show=True)¶

Call this function to show the GUI of the simulator.

By default, the simulator runs without GUI

useExistingResults(dirs)¶

This function allows to use existing results, as opposed to running a simulation.

Parameters: dirs – A non-empty list of directories that contain existing results.

This method can be used for testing and debugging. If called, then no simulation is run. If the directories ['/tmp/tmp-Buildings-0-zABC44', '/tmp/tmp-Buildings-0-zQNS41'] contain previous results, then this method can be used as

>>> import buildingspy.development.regressiontest as r
>>> l=['/tmp/tmp-Buildings-0-zABC44', '/tmp/tmp-Buildings-0-zQNS41']
>>> rt = r.Tester()
>>> rt.useExistingResults(l)
>>> rt.run() 

writeOpenModelicaResultDictionary()¶

Write in Resources/Scripts/OpenModelica/compareVars files whose name are the name of the example model, and whose content is:

compareVars :=
  {
    "controler.y",
    "sensor.T",
    "heater.Q_flow"
  };

These files are then used in the regression testing that is done by the OpenModelica development team.

Validator of syntax¶

class buildingspy.development.validator.Validator¶

Class that validates .mo files for the correct html syntax.

validateExperimentSetup(root_dir)¶

Validate the experiment setup in .mo and .mos files.

Parameters: root_dir – Root directory.

validateHTMLInPackage(rootDir)¶

This function recursively validates all .mo files in a package.

If there is malformed html code in the info or the revision section, then this function write the error message of tidy to the standard output.

Note that the line number correspond to an intermediate format (e.g., the output format of tidy), which may be different from the .mo file.

Parameters: rootDir – The root directory of the package.
Returns: str[] Warning/error messages from tidylib.

Usage: Type

>>> import os
>>> import buildingspy.development.validator as v
>>> val = v.Validator()
>>> myMoLib = os.path.join(                    "buildingspy", "tests", "MyModelicaLibrary")
>>> # Get a list whose elements are the error strings
>>> errStr = val.validateHTMLInPackage(myMoLib)

Merging Modelica IBPSA Library¶

class buildingspy.development.merger.IBPSA(ibpsa_dir, dest_dir)¶

Class that merges the Modelica IBPSA Library with other Modelica libraries.

Both libraries need to have the same package structure.

By default, the top-level packages Experimental and Obsolete are not included in the merge. This can be overwritten with the function set_excluded_directories().

static filter_files(file_list, pattern)¶

Return file_list for those that match pattern.

Currently, pattern can contain at most one wildchar (‘*’) character.

Parameters

file_list – List of files.
pattern – Pattern that is used to match files.

A typical usage is

>>> import buildingspy.development.merger as m
>>> m.IBPSA.filter_files(['a.txt', 'aa/b.txt', 'aa/bb/c.txt'], '*.txt')
['a.txt']
>>> m.IBPSA.filter_files(['a.txt', 'aa/b.txt', 'aa/bb/c.txt'], 'aa/*.txt')
['aa/b.txt']
>>> m.IBPSA.filter_files(['a.txt', 'aa/b1.txt', 'aa/b2.txt', 'aa/bb/c.txt'], 'aa/*.txt')
['aa/b1.txt', 'aa/b2.txt']

merge(overwrite_reference_results=False)¶

Merge all files except the license file and the top-level package.mo

Warning

This method is experimental. Do not use it without having a backup of your code.

This function merges the IBPSA library into other Modelica libraries.

In the top-level directory of the destination library, this function creates the file .copiedFiles.txt that lists all files that have been copied. In subsequent calls, this function deletes all files listed in .copiedFiles.txt, then merges the libraries, and creates a new version of .copiedFiles.txt. Therefore, if a model is moved in the IBPSA library, it will also be moved in the target library by deleting the old file and copying the new file.

This function will merge all Modelica files, Modelica scripts, regression results and images. An exception is the file IBPSA/package.mo, as libraries typically have their own top-level package file that contains their release notes and version information.

When copying the files, all references and file names that contain the string IBPSA will be renamed with the name of the top-level directory of the destination library. Afterwards, the package.order files will be regenerated, which allows libraries to have Modelica classes in the same directory as are used by the IBPSA library, as long as their name differs.

Parameters: overwrite_reference_results – Boolean True causes the reference results to be overwritten, False causes the reference results to be skipped if they already exist. Note that if the reference result does not yet exist, it will be merged regardless of the setting of this parameter.

A typical usage is

>>> import buildingspy.development.merger as m
>>> import os
>>> home = os.path.expanduser("~")
>>> root = os.path.join(home, "test")
>>> ibpsa_dir = os.path.join(root, "modelica", "IBPSA")
>>> dest_dir = os.path.join(root, "modelica-buildings", "Buildings")
>>> mer = m.IBPSA(ibpsa_dir, dest_dir) 
>>> mer.merge()                            

static remove_library_specific_documentation(file_lines, library_name)¶

Remove library specific content.

For example, for the Buildings and IDEAS libraries, include the section in the commented block below, but keep the comment as an html-comment for other libraries.

Parameters: file_lines – The lines of the file to be merged.
Returns: The lines of the files, with comments removed as indicated by the tag(s) in the comment line.

set_excluded_directories(directories)¶

Set the directories that are excluded from the merge.

Parameters: packages – A list of directories to be excluded.

Error dictionary¶

class buildingspy.development.error_dictionary.ErrorDictionary¶

Class that contains data fields needed for the error checking of the regression tests.

If additional error messages need to be checked, then they should be added to the constructor of this class.

get_dictionary()¶: Return the dictionary with all error data

increment_counter(key)¶

Increment the error counter by one for the error type defined by key.

Parameters: key – The json key of the error type.

keys()¶: Return a copy of the dictionary’s list of keys.

tool_messages()¶: Return a copy of the tool messages as a list.

Comparing simulation performance¶

class buildingspy.development.simulationCompare.Comparator(tools, branches, package, repo, nPro=0, simulate=True, tolAbsTime=0.1, tolRelTime=0.1, postCloneCommand=None)¶

Class that compares various simulation statistics across tools or branches.

This class allows comparing various simulation performance indicators (CPU time, number of state events, number of Jacobian evaluations) across Modelica simulation tools and across git branches. The tests can be run across a whole library, or across an individual Modelica package. The results will be summarized in a table format that compares the performance across tools and branches.

Initiate with the following optional arguments:

Parameters

tools – A list of tools to compare, such as ['openmodelica', 'dymola'].
branches – A list of branches to compare, such as ['master', 'issueXXX'].
package – Name of top-level package to compare, such as Buildings or Buildings.Examples.
repo – Name of repository, such as https://github.com/lbl-srg/modelica-buildings.
nPro – Number of threads that are used to run the translations and simulations. Set to 0 to use all processors.
tolAbsTim – float (default 0.1). Absolute tolerance in time, if exceeded, results will be flagged in summary table.
tolRelTim – float (default 0.1). Relative tolerance in time, if exceeded, results will be flagged in summary table.
postCloneCommand – list. A list of a command and its arguments that is run after cloning the repository. The command is run from the root folder inside the repository, e.g., the folder that contains the .git folder.

This class can be used to compare translation and simulation statistics across tools and branches. Note that only one simulation is done, hence the simulation time can vary from one run to another, and therefore indicates trends rather than exact comparison of computing time.

To run the comparison, type

>>> import os
>>> import buildingspy.development.simulationCompare as sc
>>> s = sc.Comparator(
...   tools=['dymola', 'openmodelica'],
...   branches=['master'],
...   package='Buildings',
...   repo='https://github.com/lbl-srg/modelica-buildings',
...   postCloneCommand=[
...      "python",
...      "Buildings/Resources/src/ThermalZones/install.py",
...      "--binaries-for-os-only"])
>>> s.run()                                           

To change the comparison for different tolerances without running the simulations again, type

>>> import os
>>> import buildingspy.development.simulationCompare as sc
>>> s = sc.Comparator(
...   tools=['dymola', 'openmodelica'],
...   branches=['master'],
...   package='Buildings',
...   repo='https://github.com/lbl-srg/modelica-buildings',
...   postCloneCommand=[
...      "python",
...      "Buildings/Resources/src/ThermalZones/install.py",
...      "--binaries-for-os-only"])
>>> s.post_process(tolAbsTime=0.2, tolRelTime=0.2)     

post_process(tolAbsTime=None, tolRelTime=None)¶

Generate the html tables.

This function post-processes the simulations, generates the overview tables, and writes the tables to the directory results.

Parameters

tolAbsTime – float. Optional argument for absolute tolerance in time, if exceeded, results will be flagged in summary table.
tolRelTime – float. Optional argument for relative tolerance in time, if exceeded, results will be flagged in summary table.

run()¶

Run the comparison and generate the output.

The output files will be in the directory results, and the raw test data are in the directories with the same names as specified by the parameter tools.

Table of Contents

Previous topic

Next topic

Development¶

Refactoring code¶

Regression tests¶

Validator of syntax¶

Merging Modelica IBPSA Library¶

Error dictionary¶

Comparing simulation performance¶