Merge pull request #52 from biosimulators/import-copasi

jonrkarr · web-flow · commit f26a924b8886 · 2021-12-09T08:31:27.000-05:00
Allow the COPASI Biosimulator to simulate files exported by COPASI
diff --git a/.gitignore b/.gitignore
@@ -120,6 +120,7 @@ venv.bak/
 # Temp directory
 temp/
 tests/results/
+tests/out/
 
 *.tmp
 
diff --git a/biosimulators_copasi/__main__.py b/biosimulators_copasi/__main__.py
@@ -10,14 +10,91 @@
 from . import get_simulator_version
 from ._version import __version__
 from .core import exec_sedml_docs_in_combine_archive
+from .utils import fix_copasi_generated_combine_archive as fix_copasi_generated_combine_archive_func
+from biosimulators_utils.config import get_config
 from biosimulators_utils.simulator.cli import build_cli
+from biosimulators_utils.simulator.data_model import EnvironmentVariable
+from biosimulators_utils.simulator.environ import ENVIRONMENT_VARIABLES as DEFAULT_ENVIRONMENT_VARIABLES
+import cement
+import termcolor
+
+ENVIRONMENT_VARIABLES = list(DEFAULT_ENVIRONMENT_VARIABLES.values())
+ENVIRONMENT_VARIABLES.append(
+    EnvironmentVariable(
+        name='FIX_COPASI_GENERATED_COMBINE_ARCHIVE',
+        description=(
+            'Whether to make COPASI-generated COMBINE archives compatible with the '
+            'specifications of the OMEX manifest and SED-ML standards.'
+        ),
+        options=['0', '1'],
+        default='0',
+        more_info_url='https://docs.biosimulators.org/Biosimulators_COPASI/source/biosimulators_copasi.html',
+    )
+)
 
 App = build_cli('biosimulators-copasi', __version__,
                 'COPASI', get_simulator_version(), 'http://copasi.org',
                 exec_sedml_docs_in_combine_archive,
+                environment_variables=ENVIRONMENT_VARIABLES,
                 )
 
 
 def main():
     with App() as app:
         app.run()
+
+
+class FixCopasiGeneratedCombineArchiveController(cement.Controller):
+    """ Controller for fixing COPASI-generated COMBINE archives """
+
+    class Meta:
+        label = 'base'
+        help = 'Fix a COPASI-generated COMBINE/OMEX archive'
+        description = (
+            'Correct a COPASI-generated COMBINE/OMEX archive to be consistent with '
+            'the specifications of the OMEX manifest and SED-ML formats'
+        )
+        arguments = [
+            (
+                ['-i', '--in-file'],
+                dict(
+                    type=str,
+                    required=True,
+                    help='Path to COMBINE/OMEX file to correct',
+                ),
+            ),
+            (
+                ['-o', '--out-file'],
+                dict(
+                    type=str,
+                    required=True,
+                    help='Path to save the corrected archive',
+                ),
+            ),
+        ]
+
+    @cement.ex(hide=True)
+    def _default(self):
+        args = self.app.pargs
+        config = get_config()
+        try:
+            fix_copasi_generated_combine_archive_func(args.in_file, args.out_file)
+        except Exception as exception:
+            if config.DEBUG:
+                raise
+            raise SystemExit(termcolor.colored(str(exception), 'red')) from exception
+
+
+class FixCopasiGeneratedCombineArchiveApp(cement.App):
+    """ Command line application for fixing COPASI-generated COMBINE/OMEX archives """
+    class Meta:
+        label = 'fix-copasi-generated-combine-archive'
+        base_controller = 'base'
+        handlers = [
+            FixCopasiGeneratedCombineArchiveController,
+        ]
+
+
+def fix_copasi_generated_combine_archive():
+    with FixCopasiGeneratedCombineArchiveApp() as app:
+        app.run()
diff --git a/biosimulators_copasi/core.py b/biosimulators_copasi/core.py
@@ -22,17 +22,19 @@
 from kisao.data_model import AlgorithmSubstitutionPolicy, ALGORITHM_SUBSTITUTION_POLICY_LEVELS
 from .data_model import KISAO_ALGORITHMS_MAP, Units
 from .utils import (get_algorithm_id, set_algorithm_parameter_value,
-                    get_copasi_model_object_by_sbml_id, get_copasi_model_obj_sbml_ids)
+                    get_copasi_model_object_by_sbml_id, get_copasi_model_obj_sbml_ids,
+                    fix_copasi_generated_combine_archive as fix_copasi_generated_combine_archive_func)
 import COPASI
 import lxml
 import math
 import numpy
-
+import os
+import tempfile
 
 __all__ = ['exec_sedml_docs_in_combine_archive', 'exec_sed_doc', 'exec_sed_task', 'preprocess_sed_task']
 
 
-def exec_sedml_docs_in_combine_archive(archive_filename, out_dir, config=None):
+def exec_sedml_docs_in_combine_archive(archive_filename, out_dir, config=None, fix_copasi_generated_combine_archive=None):
     """ Execute the SED tasks defined in a COMBINE/OMEX archive and save the outputs
 
     Args:
@@ -45,16 +47,31 @@ def exec_sedml_docs_in_combine_archive(archive_filename, out_dir, config=None):
               with reports at keys ``{ relative-path-to-SED-ML-file-within-archive }/{ report.id }`` within the HDF5 file
 
         config (:obj:`Config`, optional): BioSimulators common configuration
+        fix_copasi_generated_combine_archive (:obj:`bool`, optional): Whether to make COPASI-generated COMBINE archives
+            compatible with the specifications of the OMEX manifest and SED-ML standards
 
     Returns:
         :obj:`tuple`:
 
             * :obj:`SedDocumentResults`: results
             * :obj:`CombineArchiveLog`: log
     """
-    return exec_sedml_docs_in_archive(exec_sed_doc, archive_filename, out_dir,
-                                      apply_xml_model_changes=True,
-                                      config=config)
+    if fix_copasi_generated_combine_archive is None:
+        fix_copasi_generated_combine_archive = os.getenv('FIX_COPASI_GENERATED_COMBINE_ARCHIVE', '0').lower() in ['1', 'true']
+
+    if fix_copasi_generated_combine_archive:
+        temp_archive_file, temp_archive_filename = tempfile.mkstemp()
+        os.close(temp_archive_file)
+        fix_copasi_generated_combine_archive_func(archive_filename, temp_archive_filename)
+        archive_filename = temp_archive_filename
+
+    result = exec_sedml_docs_in_archive(exec_sed_doc, archive_filename, out_dir,
+                                        apply_xml_model_changes=True,
+                                        config=config)
+    if fix_copasi_generated_combine_archive:
+        os.remove(temp_archive_filename)
+
+    return result
 
 
 def exec_sed_doc(doc, working_dir, base_out_path, rel_out_path=None,
diff --git a/biosimulators_copasi/utils.py b/biosimulators_copasi/utils.py
@@ -7,20 +7,28 @@
 """
 
 from .data_model import KISAO_ALGORITHMS_MAP, KISAO_PARAMETERS_MAP, Units
-from biosimulators_utils.config import Config  # noqa: F401
+from biosimulators_utils.combine.data_model import CombineArchiveContentFormat
+from biosimulators_utils.combine.io import CombineArchiveReader, CombineArchiveWriter
+from biosimulators_utils.config import get_config, Config  # noqa: F401
 from biosimulators_utils.data_model import ValueType
 from biosimulators_utils.simulator.utils import get_algorithm_substitution_policy
 from biosimulators_utils.utils.core import validate_str_value, parse_value
 from kisao.data_model import AlgorithmSubstitutionPolicy, ALGORITHM_SUBSTITUTION_POLICY_LEVELS
 from kisao.utils import get_preferred_substitute_algorithm_by_ids
 import COPASI
 import itertools
+import libsedml
+import lxml
+import os
+import shutil
+import tempfile
 
 __all__ = [
     'get_algorithm_id',
     'set_algorithm_parameter_value',
     'get_copasi_model_object_by_sbml_id',
     'get_copasi_model_obj_sbml_ids',
+    'fix_copasi_generated_combine_archive',
 ]
 
 
@@ -241,3 +249,61 @@ def get_copasi_model_obj_sbml_ids(model):
         ids.append(object.getSBMLId())
 
     return ids
+
+
+def fix_copasi_generated_combine_archive(in_filename, out_filename, config=None):
+    """ Utility function that corrects COMBINE/OMEX archives generated by COPASI so they are compatible
+    with other tools.
+
+    All currently released versions of COPASI export COMBINE archive files. However, these archives
+    presently diverge from the specifications of the SED-ML format.
+
+    * Format in OMEX manifests is not a valid PURL media type URI
+    * SED-ML files lack namespaces for SBML
+
+    Args:
+        in_filename (:obj:`str`): path to a COMBINE archive to correct
+        out_filename (:obj:`str`): path to save correctd COMBINE archive
+        config (:obj:`Config`, optional): BioSimulators-utils configuration
+    """
+    config = config or get_config()
+    archive_dirname = tempfile.mkdtemp()
+    try:
+        archive = CombineArchiveReader().run(in_filename, archive_dirname, config=config)
+    except Exception:
+        shutil.rmtree(archive_dirname)
+        raise
+
+    # correct URI for COPASI application format
+    for content in archive.contents:
+        if content.format == 'application/x-copasi':
+            content.format = CombineArchiveContentFormat.CopasiML
+            # potentially issue warning messages if needed
+            break
+
+    # add SBML namespace to SED-ML file
+    ns = None
+    for content in archive.contents:
+        if content.format == 'http://identifiers.org/combine.specifications/sbml':
+            with open(os.path.join(archive_dirname, content.location), 'rb') as sbml:
+                root = lxml.etree.parse(sbml)
+                # get default ns
+                ns = root.getroot().nsmap[None]
+                break
+
+    if ns:
+        for content in archive.contents:
+            if content.format == 'http://identifiers.org/combine.specifications/sed-ml':
+                sedml_file = os.path.join(archive_dirname, content.location)
+                doc = libsedml.readSedMLFromFile(sedml_file)
+                sedml_ns = doc.getSedNamespaces().getNamespaces()
+                if not sedml_ns.hasPrefix('sbml'):
+                    sedml_ns.add(ns, 'sbml')
+                    libsedml.writeSedMLToFile(doc, sedml_file)
+                    # potentially issue warning message here, that the sedml file had no SBML prefix and it was added
+                    break
+
+    try:
+        CombineArchiveWriter().run(archive, archive_dirname, out_filename)
+    finally:
+        shutil.rmtree(archive_dirname)
diff --git a/docs-src/tutorial.rst b/docs-src/tutorial.rst
@@ -24,8 +24,8 @@ New values of parameter KISAO_0000534 (list of deterministic reactions) of KISAO
     </algorithm>
 
 
-Command-line program
---------------------
+Command-line program for executing COMBINE/OMEX archives
+--------------------------------------------------------
 
 The command-line program can be used to execute COMBINE/OMEX archives that describe simulations as illustrated below.
 
@@ -69,3 +69,31 @@ For example, the following command could be used to use the Docker image to exec
         ghcr.io/biosimulators/copasi:latest \
             -i /tmp/working-dir/modeling-study.omex \
             -o /tmp/working-dir
+
+Command-line program for correcting COMBINE/OMEX archives created by COPASI
+---------------------------------------------------------------------------
+
+The ``fix-copasi-generated-combine-archive`` command-line program can be used to align COMBINE/OMEX archives created by COPASI with the specifications of the OMEX manifest and SED-ML formats.
+
+.. code-block:: text
+
+    usage: fix-copasi-generated-combine-archive [-h] [-d] [-q] -i IN_FILE -o OUT_FILE
+
+    Correct a COPASI-generated COMBINE/OMEX archive to be consistent with the specifications of the OMEX manifest and SED-ML formats
+
+    optional arguments:
+      -h, --help            show this help message and exit
+      -d, --debug           full application debug mode
+      -q, --quiet           suppress all console output
+      -i IN_FILE, --in-file IN_FILE
+                            Path to COMBINE/OMEX file to correct
+      -o OUT_FILE, --out-file OUT_FILE
+                            Path to save the corrected archive
+
+For example, the following command could be used to correct the example COPASI-generated archive in the ``tests/fixtures`` directory:
+
+.. code-block:: text
+
+    fix-copasi-generated-combine-archive \
+        -i tests/fixtures/copasi-34-export.omex \
+        -o tests/fixtures/copasi-34-export-fixed.omex
diff --git a/setup.py b/setup.py
@@ -48,6 +48,7 @@
     entry_points={
         'console_scripts': [
             'biosimulators-copasi = biosimulators_copasi.__main__:main',
+            'fix-copasi-generated-combine-archive = biosimulators_copasi.__main__:fix_copasi_generated_combine_archive',
         ],
     },
 )
diff --git a/tests/fixtures/copasi-34-export.omex b/tests/fixtures/copasi-34-export.omex
diff --git a/tests/test_fix_copasi_generated_combine_archive.py b/tests/test_fix_copasi_generated_combine_archive.py
@@ -0,0 +1,49 @@
+import shutil
+import tempfile
+import unittest
+import os
+from unittest import mock
+
+from biosimulators_utils.combine.io import CombineArchiveReader, CombineArchiveWriter
+from biosimulators_utils.config import get_config
+from biosimulators_utils.log.data_model import Status
+
+import biosimulators_copasi
+import biosimulators_copasi.utils
+
+
+class FixCopasiGeneratedCombineArchiveTestCase(unittest.TestCase):
+    def setUp(self):
+        dirname = os.path.dirname(__file__)
+        self.filename = os.path.join(dirname, 'fixtures', 'copasi-34-export.omex')
+        self.assertTrue(os.path.exists(self.filename))
+        self.archive_tmp_dir = tempfile.mkdtemp()
+        self.out_dir = tempfile.mkdtemp()
+
+    def tearDown(self):
+        shutil.rmtree(self.archive_tmp_dir)
+        shutil.rmtree(self.out_dir)
+
+    @unittest.expectedFailure
+    def test_execute_copasi_generated_archive(self):
+        # ideally this shouldn't fail
+        results, log = biosimulators_copasi.exec_sedml_docs_in_combine_archive(self.filename, self.out_dir)
+        self.assertEqual(log.status, Status.SUCCEEDED)
+
+    def test_executed_fixed_copasi_generated_archive(self):
+        # fix the archive
+        corrected_filename = os.path.join(self.archive_tmp_dir, 'archive.omex')
+        biosimulators_copasi.utils.fix_copasi_generated_combine_archive(self.filename, corrected_filename)
+
+        # check that the corrected archive can be executed
+        results, log = biosimulators_copasi.exec_sedml_docs_in_combine_archive(corrected_filename, self.out_dir)
+        self.assertEqual(log.status, Status.SUCCEEDED)
+
+        results, log = biosimulators_copasi.exec_sedml_docs_in_combine_archive(
+            self.filename, self.out_dir, fix_copasi_generated_combine_archive=True)
+        self.assertEqual(log.status, Status.SUCCEEDED)
+
+        with mock.patch.dict(os.environ, {'FIX_COPASI_GENERATED_COMBINE_ARCHIVE': '1'}):
+            results, log = biosimulators_copasi.exec_sedml_docs_in_combine_archive(
+                self.filename, self.out_dir)
+        self.assertEqual(log.status, Status.SUCCEEDED)

Original file line number	Diff line number	Diff line change
`@@ -48,6 +48,7 @@`
`48`	`48`	`entry_points={`
`49`	`49`	`'console_scripts': [`
`50`	`50`	`'biosimulators-copasi = biosimulators_copasi.__main__:main',`
	`51`	`+ 'fix-copasi-generated-combine-archive = biosimulators_copasi.__main__:fix_copasi_generated_combine_archive',`
`51`	`52`	`],`
`52`	`53`	`},`
`53`	`54`	`)`