Merge pull request #215 from microbiomedata/issue-214

Issue 214: get_nmdc_jsonschema_dict not returning correct data

CHANGELOG.md

@@ -11,6 +11,17 @@ Changes to the schema are documented in this file.
 ### Removed
   - N/A
 
+## [2021.11.22rc1](https://github.com/microbiomedata/nmdc-schema/releases/tag/2021.11.12rc1)
+### Added
+  - N/A
+### Fixed
+  - get_nmdc_jsonschema_dict not returning correct data (#214)
+  - typo in NMDC Schema description
+### Changed 
+  - N/A
+### Removed
+  - N/A
+
 ## [2021.11.19rc3](https://github.com/microbiomedata/nmdc-schema/releases/tag/2021.11.19rc3)
 ### Added
   - N/A

jsonld-context/nmdc.context.jsonld

@@ -1,5 +1,5 @@
 {
-   "_comments": "Auto generated from nmdc.yaml by jsonldcontextgen.py version: 0.1.1\n    Generation date: 2021-11-15 12:55\n    Schema: NMDC\n    \n    id: https://microbiomedata/schema\n    description: Schema for National Microbiome Data Collaborative (NMDC). This schem is organized into 3 separate modules:\n  \nThis schema is organized into distinct modules:\n  \n * a set of core types for representing data values\n * the mixs schema (auto-translated from mixs excel)\n * annotation schema\n * the NMDC schema itself\n    license: https://creativecommons.org/publicdomain/zero/1.0/\n    ",
+   "_comments": "Auto generated from nmdc.yaml by jsonldcontextgen.py version: 0.1.1\n    Generation date: 2021-11-22T18:20:03\n    Schema: NMDC\n    \n    id: https://microbiomedata/schema\n    description: Schema for National Microbiome Data Collaborative (NMDC).\n  \nThis schema is organized into distinct modules:\n  \n * a set of core types for representing data values\n * the mixs schema (auto-translated from mixs excel)\n * annotation schema\n * the NMDC schema itself\n    license: https://creativecommons.org/publicdomain/zero/1.0/\n    ",
    "@context": {
       "CAS": "http://identifiers.org/cas/",
       "CATH": "http://identifiers.org/cath/",

python/nmdc.py

@@ -1,12 +1,11 @@
 # Auto generated from nmdc.yaml by pythongen.py version: 0.9.0
-# Generation date: 2021-11-15 12:56
+# Generation date: 2021-11-22T18:20:38
 # Schema: NMDC
 #
 # id: https://microbiomedata/schema
-# description: Schema for National Microbiome Data Collaborative (NMDC). This schem is organized into 3 separate
-#              modules: This schema is organized into distinct modules: * a set of core types for representing
-#              data values * the mixs schema (auto-translated from mixs excel) * annotation schema * the NMDC
-#              schema itself
+# description: Schema for National Microbiome Data Collaborative (NMDC). This schema is organized into distinct
+#              modules: * a set of core types for representing data values * the mixs schema (auto-translated from
+#              mixs excel) * annotation schema * the NMDC schema itself
 # license: https://creativecommons.org/publicdomain/zero/1.0/
 
 import dataclasses
@@ -3110,6 +3109,24 @@ def _addvals(cls):
         setattr(cls, "QC Statistics",
                 PermissibleValue(text="QC Statistics",
                                  description="Reads QC summary statistics") )
+        setattr(cls, "TIGRFam Annotation GFF",
+                PermissibleValue(text="TIGRFam Annotation GFF",
+                                 description="GFF3 format file with TIGRfam") )
+        setattr(cls, "Clusters of Orthologous Groups (COG) Annotation GFF",
+                PermissibleValue(text="Clusters of Orthologous Groups (COG) Annotation GFF",
+                                 description="GFF3 format file with COGs") )
+        setattr(cls, "CATH FunFams (Functional Families) Annotation GFF",
+                PermissibleValue(text="CATH FunFams (Functional Families) Annotation GFF",
+                                 description="GFF3 format file with CATH FunFams") )
+        setattr(cls, "SUPERFam Annotation GFF",
+                PermissibleValue(text="SUPERFam Annotation GFF",
+                                 description="GFF3 format file with SUPERFam") )
+        setattr(cls, "SMART Annotation GFF",
+                PermissibleValue(text="SMART Annotation GFF",
+                                 description="GFF3 format file with SMART") )
+        setattr(cls, "Pfam Annotation GFF",
+                PermissibleValue(text="Pfam Annotation GFF",
+                                 description="GFF3 format file with Pfam") )
 
 class CreditEnum(EnumDefinitionImpl):
 
@@ -6245,6 +6262,10 @@ class slots:
 slots.reaction_participant_stoichiometry = Slot(uri=NMDC.stoichiometry, name="reaction participant_stoichiometry", curie=NMDC.curie('stoichiometry'),
                    model_uri=NMDC.reaction_participant_stoichiometry, domain=ReactionParticipant, range=Optional[int])
 
+slots.functional_annotation_has_function = Slot(uri="str(uriorcurie)", name="functional annotation_has function", curie=None,
+                   model_uri=NMDC.functional_annotation_has_function, domain=FunctionalAnnotation, range=Optional[str],
+                   pattern=re.compile(r'^(KEGG.PATHWAY:\w{2,4}\d{5}|KEGG.REACTION:R\d+|RHEA:\d{5}|MetaCyc:[A-Za-z0-9+_.%-:]+|EC:\d{1,2}(\.\d{0,3}){0,3}|GO:\d{7}|MetaNetX:(MNXR\d+|EMPTY)|SEED:\w+|KEGG\.ORTHOLOGY:K\d+|EGGNOG:\w+|PFAM:PF\d{5}|TIGRFAM:TIGR\d+|SUPFAM:\w+|CATH:[1-6]\.[0-9]+\.[0-9]+\.[0-9]+|PANTHER.FAMILY:PTHR\d{5}(\:SF\d{1,3})?)$'))
+
 slots.functional_annotation_type = Slot(uri=NMDC.type, name="functional annotation_type", curie=NMDC.curie('type'),
                    model_uri=NMDC.functional_annotation_type, domain=FunctionalAnnotation, range=Optional[Union[str, OntologyClassId]])
 

src/schema/nmdc.yaml

@@ -4,7 +4,6 @@ title: NMDC Schema
 
 description: >-
   Schema for National Microbiome Data Collaborative (NMDC).
-  This schem is organized into 3 separate modules:
     
   This schema is organized into distinct modules:
     
@@ -14,7 +13,7 @@ description: >-
    * the NMDC schema itself
   
 license: https://creativecommons.org/publicdomain/zero/1.0/
-version: 2021-11-19
+version: 2021-11-22
 
 prefixes:
   nmdc: https://microbiomedata/meta/

util/nmdc_data.py

@@ -52,22 +52,53 @@ def get_nmdc_yaml_string() -> str:
     return nmdc_yaml.decode("utf-8")
 
 
-def get_nmdc_dict() -> Dict:
-    """Parses the nmdc.yaml file into a dict.
+def get_nmdc_jsonschema_bytesIO() -> io.BytesIO:
+    """Returns the nmdc.schema.json file as bytes steam.
+    This function is not intended to be used directly, but it used by other functions
 
     Returns
     -------
-    dict
-        The dict of the keys and value in the nmdc.yaml file.
+    BytesIO
+        A bytes stream of nmdc.schema.json file.
     """
     # get nmdc.yaml file from the package data
-    nmdc_yaml = get_nmdc_yaml_bytesIO()
+    return io.BytesIO(pkgutil.get_data("nmdc_schema", "nmdc.schema.json"))
+
+
+def get_nmdc_jsonschema_bytes() -> bytes:
+    """Retruns the nmdc.schema.json file as bytes.
+
+    Returns
+    -------
+    bytes
+        The bytes of the nmdc.schema.json file.
+    """
+    nmdc_json = get_nmdc_jsonschema_bytesIO()
+    return nmdc_json.getvalue()
+
 
-    # convert yaml to dict
-    nmdc_dict = yaml.load(nmdc_yaml, Loader=yaml.CLoader)
+def get_nmdc_jsonschema_string() -> str:
+    """Retruns the nmdc.schema.json file as a string.
 
-    # return dict
-    return nmdc_dict
+    Returns
+    -------
+    str
+        A string containing the contents of nmdc.schema.json file.
+    """
+    nmdc_json = get_nmdc_jsonschema_bytes()
+    return nmdc_json.decode("utf-8")
+
+
+def get_nmdc_jsonschema_dict() -> Dict:
+    """Parses the nmdc.schema.json file into a dict.
+
+    Returns
+    -------
+    dict
+        The dict of the keys and value in the nmdc.schema.json file.
+    """
+    nmdc_json = get_nmdc_jsonschema_bytes()
+    return json.loads(nmdc_json)
 
 
 def get_nmdc_jsonschema() -> str:
@@ -79,12 +110,12 @@ def get_nmdc_jsonschema() -> str:
     str
         JSON string representation of the NMDC jsonschema (nmdc.schema.json).
     """
-    nmdc_schema = get_nmdc_dict()
+    nmdc_schema = get_nmdc_jsonschema_dict()
     return json.dumps(nmdc_schema, indent=2)
 
 
 def get_nmdc_schema_definition() -> SchemaDefinition:
-    """Returns a SchemaDefintion object created from the nmdc.yaml file.
+    """Returns a LinkML SchemaDefintion object created from the nmdc.yaml file.
 
     Returns
     -------
@@ -180,7 +211,7 @@ def get_gold_sssom() -> str:
     yaml            returns the nmdc.yaml file as a string
     jsonschema      returns the NMDC jsonschema as json
     dict            returns the NMDC jsonschema as a dict
-    schemadef       returns the SchemaDefintion created from the nmdc.yaml file
+    schemadef       returns the LinkML SchemaDefintion created from the nmdc.yaml file
     filetypeenums   returns informaton about the NMDC file type enums as json
     goldsssom       returns the gold-to-mixs.sssom.tsv file contents
     """,
@@ -195,7 +226,7 @@ def cli(ctx, fetch):
     elif "jsonschema" == fetch:
         click.echo(get_nmdc_jsonschema())
     elif "dict" == fetch:
-        click.echo(get_nmdc_dict())
+        click.echo(get_nmdc_jsonschema_dict())
     elif "schemadef" == fetch:
         click.echo(get_nmdc_schema_definition())
     elif "filetypeenums" == fetch:

util/validate_nmdc_json.py

@@ -3,7 +3,7 @@
 """Provides CLI to validate json files against the NMDC jsonschema."""
 
 import json, jsonschema, io, click
-from .nmdc_data import get_nmdc_jsonschema, get_nmdc_dict
+from .nmdc_data import get_nmdc_jsonschema, get_nmdc_jsonschema_dict
 from deprecated import deprecated
 
 
@@ -18,7 +18,7 @@ def get_nmdc_schema() -> dict:
     dict
         Dict representation of the nmdc.schema.json package data file.
     """
-    return get_nmdc_dict()
+    return get_nmdc_jsonschema_dict()
 
 
 @deprecated(reason="functionality moved to nmdc_data.get_nmdc_jsonschema()")