Issue #49: Replace pythesint with metvocab tool (#84)

* Add metvocab as a dependency
* Replace CF standard lookup
* Replace vocabulary lookup
* Remove pythesint

dmci/tools/check_mmd.py

@@ -18,7 +18,8 @@
 """
 
 import logging
-import pythesint as pti
+
+from metvocab import CFStandard, MMDVocab
 
 from lxml import etree
 from urllib.parse import urlparse
@@ -29,9 +30,31 @@
 class CheckMMD():
 
     def __init__(self):
+
         self._status_pass = []
         self._status_fail = []
         self._status_ok = True
+
+        self._cf_standard = CFStandard()
+        self._cf_standard.init_vocab()
+        self._status_ok &= self._cf_standard.is_initialised
+
+        self._access_constraing = MMDVocab("mmd", "https://vocab.met.no/mmd/Access_Constraint")
+        self._access_constraing.init_vocab()
+        self._status_ok &= self._access_constraing.is_initialised
+
+        self._activity_type = MMDVocab("mmd", "https://vocab.met.no/mmd/Activity_Type")
+        self._activity_type.init_vocab()
+        self._status_ok &= self._activity_type.is_initialised
+
+        self._operational_status = MMDVocab("mmd", "https://vocab.met.no/mmd/Operational_Status")
+        self._operational_status.init_vocab()
+        self._status_ok &= self._operational_status.is_initialised
+
+        self._use_constraint = MMDVocab("mmd", "https://vocab.met.no/mmd/Use_Constraint")
+        self._use_constraint.init_vocab()
+        self._status_ok &= self._use_constraint.is_initialised
+
         return
 
     def clear(self):
@@ -172,19 +195,22 @@ def check_cf(self, xmldoc):
             cf_list = [elem.text for elem in cf_elements[0]]
             if len(cf_list) > 1:
                 err.append("Only one CF name should be provided, got %d." % len(cf_list))
-                ok = False
+                ok &= False
 
             # Check CF names even if more than one provided
             for cf_name in cf_list:
                 try:
-                    pti.get_cf_standard_name(cf_name)
-                except IndexError:
-                    err.append("Keyword '%s' is not a CF standard name." % cf_name)
-                    ok = False
+                    cf_ok = self._cf_standard.check_standard_name(cf_name)
+                    if not cf_ok:
+                        err.append("Keyword '%s' is not a CF standard name." % cf_name)
+                        ok &= False
+                except Exception:
+                    err.append("Internal Error: CF standard name lookup failed.")
+                    ok &= False
 
         elif n_cf > 1:
             err.append("More than one CF entry found. Only one is allowed.")
-            ok = False
+            ok &= False
 
         if n_cf > 0:
             self._log_result("Climate and Forecast Standard Names Check", ok, err)
@@ -218,10 +244,10 @@ def check_vocabulary(self, xmldoc):
             List of errors
         """
         vocabularies = {
-            "access_constraint":  pti.get_mmd_access_constraints,
-            "activity_type":      pti.get_mmd_activity_type,
-            "operational_status": pti.get_mmd_operstatus,
-            "use_constraint":     pti.get_mmd_use_constraint_type,
+            "access_constraint":  self._access_constraing.check_concept_value,
+            "activity_type":      self._activity_type.check_concept_value,
+            "operational_status": self._operational_status.check_concept_value,
+            "use_constraint":     self._use_constraint.check_concept_value,
         }
         ok = True
         err = []
@@ -237,12 +263,15 @@ def check_vocabulary(self, xmldoc):
                 for rep in elems_found:
                     num += 1
                     try:
-                        f_name(rep.text)
-                    except IndexError:
-                        err.append("Incorrect vocabulary '%s' for element '%s'." % (
-                            rep.text, element_name
-                        ))
-                        ok = False
+                        v_ok = f_name(rep.text)
+                        if not v_ok:
+                            err.append("Incorrect vocabulary '%s' for element '%s'." % (
+                                rep.text, element_name
+                            ))
+                            ok &= False
+                    except Exception:
+                        err.append("Internal Error: '%s' vocabulary lookup failed." % element_name)
+                        ok &= False
 
         if num > 0:
             self._log_result("Controlled Vocabularies Check", ok, err)

requirements.txt

@@ -1,4 +1,4 @@
-pythesint @ git+/~https://github.com/metno/py-thesaurus-interface@f43f9ec70131fcbb46f129ea566a0ca241b3ba67
+metvocab @ git+/~https://github.com/metno/met-vocab-tools@c6d51ad3370b65ae1ebd805cd14b2a9e69d8922f
 requests>=2.22
 pyyaml>=5.1
 flask>=1.0

setup.cfg

@@ -31,7 +31,7 @@ include_package_data = True
 packages = find:
 scripts = dmci_start_api.py
 install_requires =
-    pythesint @ git+/~https://github.com/metno/py-thesaurus-interface@f43f9ec70131fcbb46f129ea566a0ca241b3ba67
+    metvocab @ git+/~https://github.com/metno/met-vocab-tools@c6d51ad3370b65ae1ebd805cd14b2a9e69d8922f
     requests>=2.22
     pyyaml>=5.1
     flask>=1.0

tests/test_mmd_tools/test_mmd_tools.py

@@ -21,6 +21,9 @@
 import pytest
 
 from lxml import etree
+from metvocab import CFStandard, MMDVocab
+
+from tools import causeOSError
 
 from dmci.tools import CheckMMD
 
@@ -249,7 +252,7 @@ def testMMDTools_CheckURLs():
 
 
 @pytest.mark.tools
-def testMMDTools_CheckCF():
+def testMMDTools_CheckCF(monkeypatch):
     """Test the check_cf function."""
     chkMMD = CheckMMD()
 
@@ -301,11 +304,25 @@ def testMMDTools_CheckCF():
     assert err == ["More than one CF entry found. Only one is allowed."]
     assert n == 2
 
+    # Check Exception
+    with monkeypatch.context() as mp:
+        mp.setattr(CFStandard, "check_standard_name", causeOSError)
+        ok, err, n = chkMMD.check_cf(etree.ElementTree(etree.XML(
+            "<root>"
+            "  <keywords vocabulary='Climate and Forecast Standard Names'>"
+            "    <keyword>sea_surface_temperature</keyword>"
+            "  </keywords>"
+            "</root>"
+        )))
+        assert ok is False
+        assert err == ["Internal Error: CF standard name lookup failed."]
+        assert n == 1
+
 # END Test testMMDTools_CheckCF
 
 
 @pytest.mark.tools
-def testMMDTools_CheckVocabulary():
+def testMMDTools_CheckVocabulary(monkeypatch):
     """Test the check_vocabulary function."""
     chkMMD = CheckMMD()
     ok, err = chkMMD.check_vocabulary(etree.ElementTree(etree.XML(
@@ -320,6 +337,15 @@ def testMMDTools_CheckVocabulary():
     assert ok is False
     assert err == ["Incorrect vocabulary 'OOperational' for element 'operational_status'."]
 
+    # Check Exception
+    with monkeypatch.context() as mp:
+        mp.setattr(MMDVocab, "check_concept_value", causeOSError)
+        ok, err = chkMMD.check_vocabulary(etree.ElementTree(etree.XML(
+            "<root><operational_status>Operational</operational_status></root>"
+        )))
+        assert ok is False
+        assert err == ["Internal Error: 'operational_status' vocabulary lookup failed."]
+
 # END Test testMMDTools_CheckVocabulary