Merge pull request #38 from INCATools/safe-rg

make ontology safe for relation graph; add view2table

Author: cmungall

poetry.lock

@@ -6,7 +6,7 @@ authors = ["cmungall <cjm@berkeleybop.org>"]
 
 [tool.poetry.dependencies]
 python = "^3.8"
-linkml = "^1.2.10"
+linkml = "^1.2.11"
 SQLAlchemy-Utils = "^0.38.2"
 click = "^8.1.3"
 

pyproject.toml

@@ -38,6 +38,9 @@ $(TEMPLATE): $(THIS_DIR)/sql_schema/semsql.sql
 	mv $@.tmp $@
 .PRECIOUS: $(TEMPLATE)
 
+%-min.owl: %.owl
+	robot remove -i $< --axioms "equivalent disjoint annotation" -o $@
+
 # -- MAIN TARGET --
 # A db is constructed from
 # (1) triples loaded using rdftab
@@ -61,7 +64,7 @@ $(TEMPLATE): $(THIS_DIR)/sql_schema/semsql.sql
 # will be simplified in future. See:
 #  - https://github.com/balhoff/relation-graph/issues/123
 #  - https://github.com/balhoff/relation-graph/issues/25
-%-$(RGSUFFIX).tsv: %.owl
+%-$(RGSUFFIX).tsv: %-min.owl
 	$(RG) --disable-owl-nothing true \
                        --ontology-file $<\
                        --output-file $@.ttl.tmp \

src/semsql/builder/build.Makefile

@@ -3,6 +3,9 @@
 import click
 import logging
 import semsql.builder.builder as builder
+from linkml_runtime import SchemaView
+from linkml_runtime.utils.formatutils import underscore
+from semsql.sqlutils.viewgen import get_viewdef
 from sqlalchemy import text
 
 
@@ -68,6 +71,40 @@ def query(input, query):
         print(row)
 
 
+@main.command()
+@click.argument('inputs', nargs=-1)
+@click.option('--index/--no-index', default=True, help='Create indexes on each column')
+@click.option('--name', '-n', help='Name of class/view to materialize. If blank, will perform for ALL')
+def view2table(inputs, name: str, index: bool):
+    """
+    Generates a command that turns a view into a table
+
+    See https://github.com/cmungall/semantic-sql/issues/9
+
+    Example usage:
+    ```
+    semsql view2table src/linkml/rdf.yaml -n rdfs_label_statement | sqlite3 db/pato.db
+    ```
+    """
+    for input in inputs:
+        with open(input, 'r') as stream:
+            sv = SchemaView(input)
+            schema = sv.schema
+            for cn, c in sv.all_classes().items():
+                tn = underscore(cn)
+                if name is None or str(cn) == name or tn == name:
+                    view = get_viewdef(c)
+                    if view is not None:
+                        print(f'DROP VIEW {tn};')
+                        print(f'CREATE TABLE {tn} AS {view};')
+                        if index:
+                            for sn in sv.class_slots(cn):
+                                colname = underscore(sn)
+                                print(f'CREATE INDEX {tn}_{colname} ON {tn}({colname});')
+                    else:
+                        logging.error(f'No view for {cn}')
+
+
 
 if __name__ == '__main__':
     main()

src/semsql/builder/cli.py

@@ -1,19 +1,13 @@
 import click
 from typing import List
+
+from linkml_runtime import SchemaView
 from linkml_runtime.linkml_model import SchemaDefinition, ClassDefinitionName, SlotDefinitionName
-from linkml.utils.formatutils import underscore
-from linkml.utils.schemaloader import SchemaLoader
-from semsql import get_viewdef
-
-def all_slots(schema: SchemaDefinition, cn: ClassDefinitionName) -> List[SlotDefinitionName]:
-    c = schema.classes[cn]
-    slots = list(c.slots)
-    if c.is_a:
-        slots += all_slots(schema, c.is_a)
-    if c.mixins:
-        for m in c.mixins:
-            slots += all_slots(schema, m)
-    return slots
+from linkml_runtime.utils.formatutils import underscore
+from semsql.sqlutils.viewgen import get_viewdef
+
+
+
 
 @click.command()
 @click.argument('inputs', nargs=-1)
@@ -32,16 +26,17 @@ def cli(inputs, name: str, index: bool):
     """
     for input in inputs:
         with open(input, 'r') as stream:
-            schema = SchemaLoader(input).schema
-            for cn, c in schema.classes.items():
+            sv = SchemaView(input)
+            schema = sv.schema
+            for cn, c in sv.all_classes().items():
                 tn = underscore(cn)
                 if name is None or str(cn) == name or tn == name:
-                    view = get_viewdef(schema, c)
+                    view = get_viewdef(c)
                     if view is not None:
                         print(f'DROP VIEW {tn};')
                         print(f'CREATE TABLE {tn} AS {view};')
                         if index:
-                            for sn in all_slots(schema, cn):
+                            for sn in sv.class_slots(cn):
                                 colname = underscore(sn)
                                 print(f'CREATE INDEX {tn}_{colname} ON {tn}({colname});')
 

src/semsql/sqlutils/view2table.py

@@ -1,12 +1,11 @@
 import click
-from typing import List
 from linkml_runtime.linkml_model import SchemaDefinition, ClassDefinition
 from linkml.utils.schemaloader import load_raw_schema, SchemaLoader
 from linkml_runtime.utils.formatutils import underscore
 
 VIEW_CODE = 'sqlview>>'
 
-def get_viewdef(schema: SchemaDefinition, c: ClassDefinition) -> str:
+def get_viewdef(c: ClassDefinition) -> str:
     """
     Return all VIEW definitions for a class
     :param schema: