More OWL stuff.HEADmaster

5 files changed, 70 insertions, 31 deletions

diff --git a/trygvis/eda/__init__.py b/trygvis/eda/__init__.py
index 408bd7c..26181d4 100644
--- a/trygvis/eda/__init__.py
+++ b/trygvis/eda/__init__.py
@@ -1,2 +1,7 @@
+from rdflib import Namespace
+
+EDA = Namespace("https://trygvis.io/purl/eda#")
+
+
 class EdaException(Exception):
     pass
diff --git a/trygvis/eda/cli/__init__.py b/trygvis/eda/cli/__init__.py
index 422cf48..3807fb3 100644
--- a/trygvis/eda/cli/__init__.py
+++ b/trygvis/eda/cli/__init__.py
@@ -5,13 +5,14 @@ from os import mkdir
 from os.path import dirname, isdir
 from typing import Callable
 
-from rdflib import store, ConjunctiveGraph, Graph, RDF, RDFS
+from rdflib import store, ConjunctiveGraph, Graph, RDF, RDFS, OWL
 from rdflib.plugins.sparql import prepareQuery
 import rdflib.plugins.stores.sparqlstore as sparqlstore
 import configparser
 # noinspection PyUnresolvedReferences
 import argparse
 
+from .. import EDA
 from ..digikey import rdf as digikey_rdf
 from ..kicad import rdf as kicad_rdf
 
@@ -84,6 +85,9 @@ def with_database(tx):
 def create_graph(digikey=False, kicad=False) -> Graph:
     g = Graph()
 
+    g.bind("owl", OWL)
+    g.bind("eda", EDA)
+
     if digikey:
         g.bind("dk", digikey_rdf.DIGIKEY)
         g.bind("dk-part", digikey_rdf.DIGIKEY_PART)
@@ -124,6 +128,8 @@ def write_graph(gen_g: Callable[[], Graph], filename: str = None, force_write: b
 _initNs = {
     "rdf": RDF,
     "rdfs": RDFS,
+    "owl": OWL,
+    "eda": EDA,
     "dk": digikey_rdf.DIGIKEY,
     "dk-attr-type": digikey_rdf.DIGIKEY_ATTRIBUTE_TYPE,
     "dk-attr-value": digikey_rdf.DIGIKEY_ATTRIBUTE_VALUE,
diff --git a/trygvis/eda/cli/digikey_download_metadata.py b/trygvis/eda/cli/digikey_download_metadata.py
index 60880ab..59c8099 100755
--- a/trygvis/eda/cli/digikey_download_metadata.py
+++ b/trygvis/eda/cli/digikey_download_metadata.py
@@ -1,4 +1,6 @@
-from trygvis.eda import cli
+from rdflib.term import URIRef
+
+from trygvis.eda import cli, EDA
 from trygvis.eda.digikey import *
 
 
@@ -19,11 +21,15 @@ def do_categories(db: DigikeyDatabase, client: DigikeyClient, output: str):
 
 
 def do_category(client: DigikeyClient, sc: DigikeyProductCategory, output: str):
-    cli.info('Downloading categories for %s' % sc.label)
+    cli.info('Downloading attributes for %s' % sc.label)
 
     def make_graph():
         attributes = download_attribute_types_from_category(sc, client)
-        g = cli.create_graph()
+        g = cli.create_graph(digikey=True)
+
+        g.add((sc.url(), RDF.type, OWL.Ontology))
+        g.add((sc.url(), OWL.imports, URIRef(rdf.DIGIKEY)))
+
         for a in attributes:
             [g.add(node) for node in a.to_nodes()]
 
@@ -31,6 +37,7 @@ def do_category(client: DigikeyClient, sc: DigikeyProductCategory, output: str):
 
     cli.write_graph(make_graph, output, force_write=True)
 
+
 def run(args):
     db = DigikeyDatabase()
     client = DigikeyClient()
diff --git a/trygvis/eda/digikey/__init__.py b/trygvis/eda/digikey/__init__.py
index 8fe18c8..2f93858 100644
--- a/trygvis/eda/digikey/__init__.py
+++ b/trygvis/eda/digikey/__init__.py
@@ -7,7 +7,7 @@ from cachecontrol.caches.file_cache import FileCache
 from cachecontrol.heuristics import ExpiresAfter
 from lxml import html
 from rdflib import Literal
-from rdflib.namespace import RDF, RDFS
+from rdflib.namespace import RDF, RDFS, OWL
 
 import trygvis.eda.digikey.rdf
 
@@ -41,8 +41,8 @@ class DigikeyDatabase(object):
                     return sc
         return None
 
-    def merge_attribute_types(self, attributeTypes):
-        for a in attributeTypes:
+    def merge_attribute_types(self, attribute_types):
+        for a in attribute_types:
             if a.id in self.attributeTypes:
                 # TODO: implement merging
                 continue
@@ -78,11 +78,12 @@ class DigikeyProductCategory(object):
         node = self.url()
         nodes = [
             (node, RDF.type, rdf.DIGIKEY.productCategory),
+            # (node, RDF.type, OWL.Class),
             (node, RDFS.label, Literal(self.label)),
         ]
         if self.parent is not None:
-            parentUrl = rdf.DIGIKEY_PRODUCT_CATEGORY[self.parent.id]
-            nodes.append((node, rdf.DIGIKEY.parent, parentUrl))
+            parent_url = rdf.DIGIKEY_PRODUCT_CATEGORY[self.parent.id]
+            nodes.append((node, rdf.DIGIKEY.parent, parent_url))
         if self.digikey_url is not None:
             nodes.append((node, rdf.DIGIKEY.url, Literal(self.digikey_url)))
         return nodes
@@ -101,17 +102,34 @@ class DigikeyAttributeType(object):
         assert self.options is not None
 
     def to_nodes(self):
-        nodes = []
         node = rdf.DIGIKEY_ATTRIBUTE_TYPE[self.id]
-        nodes.append((node, RDF.type, rdf.DIGIKEY.attributeType))
-        nodes.append((node, RDFS.label, Literal(self.label)))
+        node_value = rdf.DIGIKEY_ATTRIBUTE_TYPE[self.id + "-value"]
+        node_option = rdf.DIGIKEY_ATTRIBUTE_TYPE[self.id + "-option"]
+        nodes = [
+            (node, RDF.type, rdf.DIGIKEY.attributeType),
+            (node, RDF.type, OWL.Class),
+            (node, RDFS.subClassOf, rdf.DIGIKEY['AttributeType']),
+            (node, RDFS.label, Literal(self.label)),
+
+            (node_option, RDF.type, OWL.ObjectProperty),
+            (node_option, RDFS.domain, node),
+            (node_option, RDFS.range, node_value),
+            (node_option, RDFS.label, Literal('Options for ' + self.label)),
+
+            (node_value, RDF.type, OWL.Class),
+            (node_value, RDFS.subClassOf, rdf.DIGIKEY['AttributeValue']),
+            (node_value, RDFS.label, Literal(self.label + ' values')),
+        ]
 
         for o in self.options:
-            optionNode = rdf.DIGIKEY_ATTRIBUTE_VALUE[self.id + '-' + o.id]
+            option_node = rdf.DIGIKEY_ATTRIBUTE_VALUE[self.id + '-' + o.id]
             nodes.extend([
-                (optionNode, rdf.DIGIKEY.id, Literal(o.id)),
-                (optionNode, RDFS.label, Literal(o.label)),
-                (node, rdf.DIGIKEY.value, optionNode)])
+                (option_node, RDF.type, OWL.Class),
+                (option_node, RDFS.subClassOf, node_value),
+                (option_node, rdf.DIGIKEY.id, Literal(o.id)),
+                (option_node, RDFS.label, Literal(o.label)),
+                # (node, rdf.DIGIKEY.value, option_node)])
+                (node, node_option, option_node)])
         return nodes
 
 
@@ -190,17 +208,20 @@ def download_category_tree(database: DigikeyDatabase, client: DigikeyClient,
     for h2 in items:
         label = _to_string(h2)
         #        print(h2)
-        pcId = None
+        if not label == 'Resistors':
+            continue
+
+        pc_id = None
         for a in h2.getchildren():
             url = a.get('href')
-            pcId = _id_from_url(url)
-            if pcId is None:
+            pc_id = _id_from_url(url)
+            if pc_id is None:
                 continue
 
-        if pcId is None:
+        if pc_id is None:
             continue
 
-        pc = DigikeyProductCategory(pcId, label)
+        pc = DigikeyProductCategory(pc_id, label)
         n = h2.getnext()
         if n.tag == 'span':
             n = n.getnext()
@@ -230,13 +251,13 @@ def download_attribute_types_from_category(category: DigikeyProductCategory, cli
             td = select.getparent()
             index = td.getparent().index(td)
             try:
-                attributeLabel = headers[index]
+                attribute_label = headers[index]
             except:
                 continue
-            attributeId = select.get('name')
-            print("label: " + attributeLabel + ", id: " + attributeId)
+            attribute_id = select.get('name')
+            print("label: " + attribute_label + ", id: " + attribute_id)
             options = []
-            type = DigikeyAttributeType(category, attributeId, attributeLabel, options)
+            type = DigikeyAttributeType(category, attribute_id, attribute_label, options)
             for o in select.xpath("./option"):
                 id = o.get('value')
                 label = _to_string(o)
diff --git a/trygvis/eda/digikey/rdf.py b/trygvis/eda/digikey/rdf.py
index 7a21275..5f1dede 100644
--- a/trygvis/eda/digikey/rdf.py
+++ b/trygvis/eda/digikey/rdf.py
@@ -1,7 +1,7 @@
-import rdflib
+from rdflib import Namespace
 
-DIGIKEY = rdflib.Namespace("https://trygvis.io/purl/digikey#")
-DIGIKEY_ATTRIBUTE_TYPE = rdflib.Namespace("https://trygvis.io/purl/digikey-attribute-type#")
-DIGIKEY_ATTRIBUTE_VALUE = rdflib.Namespace("https://trygvis.io/purl/digikey-attribute-value#")
-DIGIKEY_PART = rdflib.Namespace("https://trygvis.io/purl/digikey-part#")
-DIGIKEY_PRODUCT_CATEGORY = rdflib.Namespace("https://trygvis.io/purl/digikey-product-category#")
+DIGIKEY = Namespace("https://trygvis.io/purl/digikey#")
+DIGIKEY_ATTRIBUTE_TYPE = Namespace("https://trygvis.io/purl/digikey-attribute-type#")
+DIGIKEY_ATTRIBUTE_VALUE = Namespace("https://trygvis.io/purl/digikey-attribute-value#")
+DIGIKEY_PART = Namespace("https://trygvis.io/purl/digikey-part#")
+DIGIKEY_PRODUCT_CATEGORY = Namespace("https://trygvis.io/purl/digikey-product-category#")