feat(combine-service): added options for validation of COMBINE archives

apps/combine-service/src/handlers/combine/get_manifest.py

@@ -69,7 +69,7 @@ def handler(body, file=None):
 
     manifest_filename = os.path.join(temp_dirname, 'manifest.xml')
     reader = CombineArchiveReader()
-    contents = reader.read_manifest(manifest_filename)
+    contents = reader.read_manifest(manifest_filename, archive_filename)
     if reader.errors:
         raise BadRequestException(
             title='COMBINE/OMEX archive does not contain a valid manifest.\n  {}'.format(

apps/combine-service/src/handlers/combine/validate.py

@@ -3,27 +3,46 @@
 from biosimulators_utils.combine.data_model import CombineArchiveContentFormat
 from biosimulators_utils.combine.io import CombineArchiveReader
 from biosimulators_utils.combine.validation import validate
+from biosimulators_utils.config import Config
 from biosimulators_utils.omex_meta.data_model import OmexMetaSchema
 import os
 import requests
 import requests.exceptions
 
 
-def handler(body, file=None):
+def handler(body, file=None,
+            validateOmexManifest=True,
+            validateSedml=True,
+            validateSedmlModels=True,
+            validateOmexMetadata=True,
+            validateImages=True):
     ''' Validate a COMBINE/OMEX archive
 
     Args:
         body (:obj:`dict`): dictionary with keys
 
-            * ``url`` whose value
-            has schema ``#/components/schemas/Url`` with the
-            URL for a COMBINE/OMEX archive
+            * ``url`` whose value has schema ``#/components/schemas/Url`` with the URL for a COMBINE/OMEX archive
+
+        validateOmexManifest (:obj:`bool`, optional): Whether to validate the OMEX manifest file in the archive
+        validateSedml (:obj:`bool`, optional): Whether to validate the SED-ML files in the archive
+        validateSedmlModels (:obj:`bool`, optional): Whether to validate the sources of the models in the SED-ML files in the archive
+        validateOmexMetadata (:obj:`bool`, optional): Whether to validate the OMEX metdata files in the archive according to
+            `BioSimulators' conventions <https://biosimulators.org/conventions/metadata>`_
+        validateImages (:obj:`bool`, optional): Whether to validate the images (BMP, GIF, JPEG, PNG, TIFF WEBP) files in the archive
         file (:obj:`werkzeug.datastructures.FileStorage`): COMBINE/OMEX archive file
 
     Returns:
         ``#/components/schemas/ValidationReport``: information about the validity or
             lack thereof of a COMBINE/OMEX archive
     '''
+    config = Config(
+        VALIDATE_OMEX_MANIFESTS=validateOmexManifest,
+        VALIDATE_SEDML=validateSedml,
+        VALIDATE_SEDML_MODELS=validateSedmlModels,
+        VALIDATE_OMEX_METADATA=validateOmexMetadata,
+        VALIDATE_IMAGES=validateImages,
+    )
+
     archive_file = file
     archive_url = body.get('url', None)
     if archive_url and archive_file:
@@ -62,7 +81,7 @@ def handler(body, file=None):
     errors = []
     warnings = []
     try:
-        archive = reader.run(archive_filename, archive_dirname)
+        archive = reader.run(archive_filename, archive_dirname, config=config)
     except Exception as exception:
         errors = [['The file could not be parsed as a COMBINE/OMEX archive.', [[str(exception)]]]]
 
@@ -71,6 +90,7 @@ def handler(body, file=None):
             archive, archive_dirname,
             formats_to_validate=list(CombineArchiveContentFormat.__members__.values()),
             metadata_schema=OmexMetaSchema.biosimulations,
+            config=config,
         )
 
     return make_validation_report(errors, warnings, filenames=[archive_filename])

apps/combine-service/src/spec/spec.yml

@@ -155,6 +155,64 @@ paths:
         required: true
       tags:
       - COMBINE
+      parameters:
+      - examples:
+          "True":
+            value: "true"
+        name: validateOmexManifest
+        description: |-
+          Whether to validate the OMEX manifest file in the archive.
+
+          Default: `true`.
+        schema:
+          type: boolean
+        in: query
+        required: false
+      - examples:
+          "True":
+            value: "true"
+        name: validateSedml
+        description: |-
+          Whether to validate the SED-ML files in the archive.
+
+          Default: `true`.
+        schema:
+          type: boolean
+        in: query
+      - examples:
+          "True":
+            value: "true"
+        name: validateSedmlModels
+        description: |-
+          Whether to validate the source (e.g., CellML, SBML file) of each model of each SED-ML file in the archive.
+
+          Default: `true`.
+        schema:
+          type: boolean
+        in: query
+      - examples:
+          "True":
+            value: "true"
+        name: validateOmexMetadata
+        description: |-
+          Whether to validate the OMEX Metadata files in the archive according to [BioSimulators' conventions](https://biosimulators.org/conventions/metadata).
+
+          Default: `true`.
+        schema:
+          type: boolean
+        in: query
+        required: false
+      - examples:
+          "True":
+            value: "true"
+        name: validateImages
+        description: |-
+          Whether to validate the image (BMP, GIF, PNG, JPEG, TIFF, WEBP) files in the archive.
+
+          Default: `true`.
+        schema:
+          type: boolean
+        in: query
       responses:
         "200":
           content:

apps/combine-service/tests/test_app.py

@@ -1410,6 +1410,69 @@ def test_validate_is_not_an_archive(self):
         self.assertEqual(validation_report['status'], "invalid")
         self.assertIn('is not a valid COMBINE/OMEX archive', json.dumps(validation_report['errors']))
 
+    def test_validate_omex_manifest_option(self):
+        # no manifest
+        archive_filename = os.path.join(
+            self.FIXTURES_DIR, 'no-manifest.omex')
+
+        fid = open(archive_filename, 'rb')
+        data = MultiDict([
+            ('file', fid),
+        ])
+        endpoint = '/combine/validate'
+        with app.app.app.test_client() as client:
+            response = client.post(endpoint, data=data, content_type="multipart/form-data")
+        fid.close()
+        self.assertEqual(response.status_code, 200, response.json)
+        validation_report = response.json
+
+        self.assertEqual(validation_report['status'], "invalid")
+        self.assertIn('is not a valid COMBINE/OMEX archive', json.dumps(validation_report['errors']))
+
+        fid = open(archive_filename, 'rb')
+        data = MultiDict([
+            ('file', fid),
+        ])
+        endpoint = '/combine/validate?validateOmexManifest=false'
+        with app.app.app.test_client() as client:
+            response = client.post(endpoint, data=data, content_type="multipart/form-data")
+        fid.close()
+        self.assertEqual(response.status_code, 200, response.json)
+        validation_report = response.json
+
+        self.assertEqual(validation_report['status'], "warnings")
+
+        # bad manifest
+        archive_filename = os.path.join(
+            self.FIXTURES_DIR, 'bad-manifest.omex')
+
+        fid = open(archive_filename, 'rb')
+        data = MultiDict([
+            ('file', fid),
+        ])
+        endpoint = '/combine/validate'
+        with app.app.app.test_client() as client:
+            response = client.post(endpoint, data=data, content_type="multipart/form-data")
+        fid.close()
+        self.assertEqual(response.status_code, 200, response.json)
+        validation_report = response.json
+
+        self.assertEqual(validation_report['status'], "invalid")
+        self.assertIn('is not a valid COMBINE/OMEX archive', json.dumps(validation_report['errors']))
+
+        fid = open(archive_filename, 'rb')
+        data = MultiDict([
+            ('file', fid),
+        ])
+        endpoint = '/combine/validate?validateOmexManifest=false'
+        with app.app.app.test_client() as client:
+            response = client.post(endpoint, data=data, content_type="multipart/form-data")
+        fid.close()
+        self.assertEqual(response.status_code, 200, response.json)
+        validation_report = response.json
+
+        self.assertEqual(validation_report['status'], "warnings")
+
     def test_get_similar_algorithms(self):
         endpoint = '/kisao/get-similar-algorithms?algorithms=KISAO_0000088'
         with app.app.app.test_client() as client: