[#14,#50] add min/max occurs int support + [#17] fix some tests + fix…

… formatting invalid schema json response

CHANGES.rst

@@ -16,6 +16,7 @@ New Features:
 - Add more functional tests (#11, #17)
 - Add support of ``label`` synonym as ``title`` for inputs and process description
   (`CWL` specifying a ``label`` will set it in `WPS` process) (#31)
+- Add support of input ``minOccurs`` and ``maxOccurs`` as ``int`` while maintaining ``str`` support (#14)
 
 Changes:
 - Use ``bump2version`` and move all config under ``setup.cfg``
@@ -31,6 +32,7 @@ Fixes:
 - Fix case where omitted ``format[s]`` in both `CWL` and `WPS` deploy bodies generated a process description with
   complex I/O (file) without required ``formats`` field. Default ``text/plain`` format is now automatically added.
 - Fix case where ``format[s]`` lists between `CWL` and `WPS` where incorrectly merged.
+- Fix invalid JSON response formatting on failing schema validation of process deployment body.
 
 0.2.2 (2019-05-31)
 ==================

Makefile

@@ -53,6 +53,7 @@ BUILDOUT_FILES := parts eggs develop-eggs bin .installed.cfg .mr.developer.cfg *
 .PHONY: all
 all: help
 
+# FIXME: use autodoc target comments
 .PHONY: help
 help:
 	@echo "Please use \`make <target>' where <target> is one of"
@@ -85,7 +86,11 @@ help:
 	@echo "Testing targets:"
 	@echo "  test-unit          run unit tests (skip long running and online tests)."
 	@echo "  test-func          run funtional tests (online and usage specific)."
+	@echo "  test-online        run online tests (running instance required)."
+	@echo "  test-offline       run offline tests (not marked as online)."
+	@echo "  test-no-tb14       run all tests except ones marked for 'Testbed-14'."
 	@echo "  test-all           run all tests (including long running tests)."
+	@echo "  test               run custom tests from input specification (make TESTS='<spec>' test).
 	@echo "  coverage       	run all tests using coverage analysis."
 	@echo "  pep8               run pep8 code style checks."
 	@echo "Sphinx targets:"
@@ -335,12 +340,37 @@ test-func:
 	bash -c "source $(CONDA_HOME)/bin/activate $(CONDA_ENV); \
 		pytest tests -v -m 'functional' --junitxml $(CURDIR)/tests/results.xml"
 
+.PHONY: test-online
+test-online:
+	@echo "Running online tests (running instance required)..."
+	bash -c "source $(CONDA_HOME)/bin/activate $(CONDA_ENV); \
+		pytest tests -v -m 'online' --junitxml $(CURDIR)/tests/results.xml"
+
+.PHONY: test-offline
+test-offline:
+	@echo "Running offline tests (not marked as online)..."
+	bash -c "source $(CONDA_HOME)/bin/activate $(CONDA_ENV); \
+		pytest tests -v -m 'not online' --junitxml $(CURDIR)/tests/results.xml"
+
+.PHONY: test-no-tb14
+test-no-tb14:
+	@echo "Running all tests except ones marked for 'Testbed-14'..."
+	bash -c "source $(CONDA_HOME)/bin/activate $(CONDA_ENV); \
+		pytest tests -v -m 'not testbed14' --junitxml $(CURDIR)/tests/results.xml"
+
 .PHONY: test-all
 test-all:
 	@echo "Running all tests (including slow and online tests)..."
 	bash -c "source $(CONDA_HOME)/bin/activate $(CONDA_ENV); \
 		pytest tests -v --junitxml $(CURDIR)/tests/results.xml"
 
+.PHONY: test
+test:
+	@echo "Running custom tests from input specification..."
+	@[ "${TESTS}" ] || ( echo ">> 'TESTS' is not set"; exit 1 )
+	bash -c "source $(CONDA_HOME)/bin/activate $(CONDA_ENV); \
+		pytest tests -v -m '${TESTS}' --junitxml $(CURDIR)/tests/results.xml"
+
 .PHONY: coverage
 coverage:
 	@echo "Running coverage analysis..."

templates/weaver.ini

@@ -87,11 +87,13 @@ handlers = console
 level =  ${settings:log-level}
 handlers = console
 qualname = weaver
+propagate = 0
 
 [logger_celery]
 level = ${settings:log-level}
 handlers = console
 qualname = celery
+propagate = 0
 
 [handler_console]
 class = StreamHandler

tests/functional/test_ems_end2end.py

@@ -129,6 +129,7 @@ def setUpClass(cls):
                 .format(cls.logger_separator_cases, cls.current_case_name(), now(), cls.logger_separator_cases))
 
         # test execution configs
+        cls.WEAVER_TEST_REQUEST_TIMEOUT = int(os.getenv("WEAVER_TEST_JOB_ACCEPTED_MAX_TIMEOUT", 60))
         cls.WEAVER_TEST_JOB_ACCEPTED_MAX_TIMEOUT = int(os.getenv("WEAVER_TEST_JOB_ACCEPTED_MAX_TIMEOUT", 30))
         cls.WEAVER_TEST_JOB_RUNNING_MAX_TIMEOUT = int(os.getenv("WEAVER_TEST_JOB_RUNNING_MAX_TIMEOUT", 6000))
         cls.WEAVER_TEST_JOB_GET_STATUS_INTERVAL = int(os.getenv("WEAVER_TEST_JOB_GET_STATUS_INTERVAL", 5))
@@ -433,7 +434,7 @@ def request(cls, method, url, ignore_errors=False, force_requests=False, log_ena
             cls.log(trace)
 
         if with_requests:
-            kw.update({"verify": False})
+            kw.update({"verify": False, "timeout": cls.WEAVER_TEST_REQUEST_TIMEOUT})
             resp = requests.request(method, url, json=json_body, data=data_body, **kw)
 
             # add some properties similar to `webtest.TestApp`
@@ -550,7 +551,7 @@ def validate_test_server(cls):
             servers.append(cls.WEAVER_TEST_WSO2_URL)
             servers.append(cls.WEAVER_TEST_MAGPIE_URL)
         for server_url in servers:
-            cls.request("GET", server_url, headers=cls.headers, status=HTTPOk.code, timeout=10)
+            cls.request("GET", server_url, headers=cls.headers, status=HTTPOk.code)
         # verify that EMS configuration requirement is met
         resp = cls.request("GET", cls.WEAVER_RESTAPI_URL, headers=cls.headers, status=HTTPOk.code)
         cls.assert_test(lambda: resp.json.get("configuration") == WEAVER_CONFIGURATION_EMS,

tests/functional/test_wps_package.py

@@ -21,6 +21,7 @@
 from tests import resources
 import pytest
 import unittest
+import colander
 import six
 
 EDAM_PLAIN = EDAM_NAMESPACE + ":" + EDAM_MAPPING[CONTENT_TYPE_TEXT_PLAIN]
@@ -684,7 +685,7 @@ def test_resolution_io_min_max_occurs(self):
         assert desc["process"]["inputs"][9]["minOccurs"] == "0"
         assert desc["process"]["inputs"][9]["maxOccurs"] == "1"
         assert desc["process"]["inputs"][10]["id"] == "required_array_min_fixed_from_wps"
-        # FIXME:
+        # FIXME: https://github.com/crim-ca/weaver/issues/50
         #   `maxOccurs=1` not updated to `maxOccurs="unbounded"` as it is evaluated as a single value,
         #   but it should be considered an array since `minOccurs>1`
         #   (see: https://github.com/crim-ca/weaver/issues/17)
@@ -739,6 +740,106 @@ def test_resolution_io_min_max_occurs(self):
         assert pkg["inputs"][13]["id"] == "optional_array_max_fixed_from_wps"
         # assert pkg["inputs"][13]["type"] == "string[]?"
 
+    # FIXME: https://github.com/crim-ca/weaver/issues/50
+    #   'unbounded' value should not override literal 2/'2'
+    @pytest.mark.xfail(reason="MinOccurs/MaxOccurs values in response should be preserved as defined in deploy body")
+    def test_valid_io_min_max_occurs_as_str_or_int(self):
+        """
+        Test validates that I/O definitions with ``minOccurs`` and/or ``maxOccurs`` are permitted as both integer
+        and string definitions in order to support (1, "1", "unbounded") variations.
+
+        .. seealso::
+            - :meth:`test_invalid_io_min_max_occurs_wrong_format`
+        """
+        cwl = {
+            "cwlVersion": "v1.0",
+            "class": "CommandLineTool",
+            "inputs": [
+                {"id": "io_min_int_max_int", "type": "string"},
+                {"id": "io_min_int_max_str", "type": "string"},
+                {"id": "io_min_str_max_int", "type": "string"},
+                {"id": "io_min_str_max_str", "type": "string"},
+                {"id": "io_min_int_max_unbounded", "type": "string"},
+                {"id": "io_min_str_max_unbounded", "type": "string"},
+            ],
+            "outputs": {"values": {"type": "string"}}
+        }
+        body = {
+            "processDescription": {
+                "process": {
+                    "id": self._testMethodName,
+                    "title": "some title",
+                    "abstract": "this is a test",
+                },
+                "inputs": [
+                    {"id": "io_min_int_max_int", "minOccurs": 1, "maxOccurs": 2},
+                    {"id": "io_min_int_max_str", "minOccurs": 1, "maxOccurs": "2"},
+                    {"id": "io_min_str_max_int", "minOccurs": "1", "maxOccurs": 2},
+                    {"id": "io_min_str_max_str", "minOccurs": "1", "maxOccurs": "2"},
+                    {"id": "io_min_int_max_unbounded", "minOccurs": 1, "maxOccurs": "unbounded"},
+                    {"id": "io_min_str_max_unbounded", "minOccurs": "1", "maxOccurs": "unbounded"},
+                ]
+            },
+            "deploymentProfileName": "http://www.opengis.net/profiles/eoc/wpsApplication",
+            "executionUnit": [{"unit": cwl}],
+        }
+        try:
+            desc, _ = self.deploy_process(body)
+        except colander.Invalid:
+            self.fail("MinOccurs/MaxOccurs values defined as valid int/str should not raise an invalid schema error")
+
+        inputs = body["processDescription"]["inputs"]
+        assert isinstance(desc["process"]["inputs"], list)
+        assert len(desc["process"]["inputs"]) == len(inputs)
+        for i, input in enumerate(inputs):
+            assert desc["process"]["inputs"][i]["id"] == input["id"]
+            for field in ["minOccurs", "maxOccurs"]:
+                assert desc["process"]["inputs"][i][field] in (input[field], str(input[field]))
+
+    # FIXME: test not working
+    #   same payloads sent directly to running weaver properly raise invalid schema -> bad request error
+    #   somehow they don't work within this test (not raised)...
+    @pytest.mark.xfail(reason="MinOccurs/MaxOccurs somehow fail validation here, but s")
+    def test_invalid_io_min_max_occurs_wrong_format(self):
+        """
+        Test verifies that ``minOccurs`` and/or ``maxOccurs`` definitions other than allowed formats are
+        raised as invalid schemas.
+
+        .. seealso::
+            :meth:`test_valid_io_min_max_occurs_as_str_or_int`
+        """
+        cwl = {
+            "cwlVersion": "v1.0",
+            "class": "CommandLineTool",
+            "inputs": [{}],   # updated after
+            "outputs": {"values": {"type": "string"}}
+        }
+        body = {
+            "processDescription": {
+                "process": {
+                    "id": self._testMethodName,
+                    "title": "some title",
+                    "abstract": "this is a test",
+                },
+                "inputs": [{}]    # updated after
+            },
+            "deploymentProfileName": "http://www.opengis.net/profiles/eoc/wpsApplication",
+            "executionUnit": [{"unit": cwl}],
+        }
+
+        # replace by invalid min/max and check that it raises
+        cwl["inputs"][0] = {"id": "test", "type": {"type": "array", "items": "string"}}
+        body["processDescription"]["inputs"][0] = {"id": "test", "minOccurs": [1], "maxOccurs": 1}
+        with self.assertRaises(colander.Invalid):
+            desc, _ = self.deploy_process(body)
+            self.fail("Invalid input minOccurs schema definition should have been raised")
+
+        cwl["inputs"][0] = {"id": "test", "type": {"type": "array", "items": "string"}}
+        body["processDescription"]["inputs"][0] = {"id": "test", "minOccurs": 1, "maxOccurs": 3.1416}
+        with self.assertRaises(HTTPBadRequest):
+            desc, _ = self.deploy_process(body)
+            self.fail("Invalid input maxOccurs schema definition should have been raised")
+
     def test_complex_io_from_package(self):
         """
         Test validates that complex I/O definitions *only* defined in the `CWL` package as `JSON` within the

tests/test_opensearch.py

@@ -24,7 +24,7 @@
 COLLECTION_IDS = {
     "sentinel2": "EOP:IPT:Sentinel2",
     "probav": "EOP:VITO:PROBAV_P_V001",
-    # "deimos": "DE2_PS3_L1C",
+    "deimos": "DE2_PS3_L1C",
 }
 
 
@@ -116,7 +116,7 @@ def make_deque(id_, value):
             make_deque(OPENSEARCH_START_DATE, "2018-01-30T00:00:00.000Z"),
             make_deque(OPENSEARCH_END_DATE, "2018-01-31T23:59:59.999Z"),
             make_deque(OPENSEARCH_AOI, "100.4,15.3,104.6,19.3"),
-            make_deque("files", "EOP:IPT:Sentinel2"),
+            make_deque("files", COLLECTION_IDS["sentinel2"]),
             make_deque("output_file_type", "GEOTIFF"),
             make_deque("output_name", "stack_result.tif"),
         ]
@@ -134,7 +134,7 @@ def make_deque(id_, value):
     )
 
     with mock.patch.object(opensearch.OpenSearchQuery, "query_datasets", return_value=mocked_query):
-        eo_image_source_info = make_eo_image_source_info("files", "EOP:IPT:Sentinel2")
+        eo_image_source_info = make_eo_image_source_info("files", COLLECTION_IDS["sentinel2"])
         mime_types = {'files': eo_image_source_info['files']['mime_types']}
         transformed = opensearch.query_eo_images_from_wps_inputs(inputs, eo_image_source_info, mime_types)
 
@@ -241,19 +241,17 @@ def test_get_additional_parameters():
     assert ("UniqueTOI", ["true"]) in params
 
 
-@pytest.mark.online
-def test_get_template_urls():
+def get_template_urls(collection_id):
     all_fields = set()
-    for name, collection_id in COLLECTION_IDS.items():
-        o = opensearch.OpenSearchQuery(collection_id, osdd_url=OSDD_URL)
-        template = o.get_template_url()
-        params = parse_qsl(urlparse(template).query)
-        param_names = list(sorted(p[0] for p in params))
-        if all_fields:
-            # noinspection PyUnresolvedReferences
-            all_fields = all_fields.intersection(param_names)
-        else:
-            all_fields.update(param_names)
+    o = opensearch.OpenSearchQuery(collection_id, osdd_url=OSDD_URL)
+    template = o.get_template_url()
+    params = parse_qsl(urlparse(template).query)
+    param_names = list(sorted(p[0] for p in params))
+    if all_fields:
+        # noinspection PyUnresolvedReferences
+        all_fields = all_fields.intersection(param_names)
+    else:
+        all_fields.update(param_names)
 
     fields_in_all_queries = list(sorted(all_fields))
     expected = [
@@ -274,6 +272,17 @@ def test_get_template_urls():
     assert not set(expected) - set(fields_in_all_queries)
 
 
+@pytest.mark.online
+def test_get_template_sentinel2():
+    get_template_urls(COLLECTION_IDS["sentinel2"])
+
+
+@pytest.mark.xfail(reason="Collection 'probav' dataset series cannot be found.")
+@pytest.mark.online
+def test_get_template_probav():
+    get_template_urls(COLLECTION_IDS["probav"])
+
+
 def inputs_unique_aoi_toi(files_id):
     return {
         OPENSEARCH_AOI: deque([LiteralInput(OPENSEARCH_AOI, "Area", data_type="string")]),
@@ -318,13 +327,13 @@ def sentinel2_inputs(unique_aoi_toi=True):
     else:
         inputs = inputs_non_unique_aoi_toi(sentinel_id)
 
-    inputs[sentinel_id][0].data = "EOP:IPT:Sentinel2"
+    inputs[sentinel_id][0].data = COLLECTION_IDS["sentinel2"]
     inputs[end_date][0].data = u"2018-01-31T23:59:59.999Z"
     inputs[start_date][0].data = u"2018-01-30T00:00:00.000Z"
     # inputs[aoi][0].data = u"POLYGON ((100 15, 104 15, 104 19, 100 19, 100 15))"
     inputs[aoi][0].data = u"100.0, 15.0, 104.0, 19.0"
 
-    eo_image_source_info = make_eo_image_source_info(sentinel_id, "EOP:IPT:Sentinel2")
+    eo_image_source_info = make_eo_image_source_info(sentinel_id, COLLECTION_IDS["sentinel2"])
     return inputs, eo_image_source_info
 
 
@@ -336,14 +345,14 @@ def probav_inputs(unique_aoi_toi=True):
     else:
         inputs = inputs_non_unique_aoi_toi(probav_id)
 
-    inputs[probav_id][0].data = "EOP:VITO:PROBAV_P_V001"
+    inputs[probav_id][0].data = COLLECTION_IDS["probav"]
     inputs[end_date][0].data = u"2018-01-31T23:59:59.999Z"
     inputs[start_date][0].data = u"2018-01-30T00:00:00.000Z"
     # inputs[aoi][0].data = u"POLYGON ((100 15, 104 15, 104 19, 100 19, 100 15))"
     inputs[aoi][0].data = u"100.0, 15.0, 104.0, 19.0"
 
     eo_image_source_info = make_eo_image_source_info(
-        probav_id, "EOP:VITO:PROBAV_P_V001"
+        probav_id, COLLECTION_IDS["probav"]
     )
 
     return inputs, eo_image_source_info
@@ -367,13 +376,13 @@ def deimos_inputs(unique_aoi_toi=True):
     end_date, start_date, aoi = query_param_names(unique_aoi_toi, deimos_id)
     inputs = inputs_unique_aoi_toi(deimos_id)
 
-    inputs[deimos_id][0].data = "DE2_PS3_L1C"
+    inputs[deimos_id][0].data = COLLECTION_IDS["deimos"]
     inputs[start_date][0].data = u"2008-01-01T00:00:00Z"
     inputs[end_date][0].data = u"2009-01-01T00:00:00Z"
     # inputs[aoi][0].data = u"MULTIPOINT ((-117 32), (-115 34))"
     inputs[aoi][0].data = u"-117, 32, -115, 34"
 
-    eo_image_source_info = make_eo_image_source_info(deimos_id, "DE2_PS3_L1C")
+    eo_image_source_info = make_eo_image_source_info(deimos_id, COLLECTION_IDS["deimos"])
     return inputs, eo_image_source_info