refactors search_predicate

WolfgangFahl · Mar 7, 2024 · cdddf9d · cdddf9d
1 parent e0f79de
commit cdddf9d
Show file tree

Hide file tree

Showing 4 changed files with 18 additions and 18 deletions.
diff --git a/ez_wikidata/__init__.py b/ez_wikidata/__init__.py
@@ -1 +1 @@
-__version__ = "0.1.5"
+__version__ = "0.2.0"
diff --git a/ez_wikidata/trulytabular.py b/ez_wikidata/trulytabular.py
@@ -28,7 +28,7 @@ def __init__(
         itemQid,
         propertyLabels: list = [],
         propertyIds: list = [],
-        subclassPredicate="wdt:P31",
+        search_predicate="wdt:P31",
         where: str = None,
         endpointConf=None,
         lang="en",
@@ -41,7 +41,7 @@ def __init__(
             itemQid(str): wikidata id of the type to analyze
             propertyLabels(list): a list of labels of properties to be considered
             propertyIds(list): a list of ids of properties to be considered
-            subclassPredicate(str): the subclass Predicate to use
+            search_predicate(str): the search predicate to use e.g. instanceof / subclass of
             where(str): extra where clause for instance selection (if any)
             endpoint(str): the url of the SPARQL endpoint to be used
         """
@@ -53,7 +53,7 @@ def __init__(
         self.wpm=WikidataPropertyManager.get_instance(endpoint_url=endpointConf.endpoint,lang=lang)
         self.sparql = SPARQL(endpointConf.endpoint, method=self.endpointConf.method)
         self.sparql.debug = self.debug
-        self.subclassPredicate = subclassPredicate
+        self.search_predicate = search_predicate
         self.where = f"\n  {where}" if where is not None else ""
         self.lang = lang
         self.item = WikidataItem(
@@ -88,7 +88,7 @@ def count(self):
 WHERE
 {{
   # instance of {self.item.qlabel}
-  ?item {self.subclassPredicate} wd:{self.item.qid}.{self.where}
+  ?item {self.search_predicate} wd:{self.item.qid}.{self.where}
 }}"""
         try:
             count = self.sparql.getValue(query, "count")
@@ -196,7 +196,7 @@ def generateSparqlQuery(
         sparqlQuery += f"""
 WHERE {{
   # instanceof {item.qid}:{item.qlabel}
-  ?{item.itemVarname} {self.subclassPredicate} wd:{item.qid}.
+  ?{item.itemVarname} {self.search_predicate} wd:{item.qid}.
   # label
   ?{item.itemVarname} rdfs:label ?{item.labelVarname}.  
   FILTER (LANG(?{item.labelVarname}) = "{lang}").
@@ -254,7 +254,7 @@ def mostFrequentPropertiesQuery(self, whereClause: str = None, minCount: int = 0
             whereClause(str): an extra WhereClause to use
         """
         if whereClause is None:
-            whereClause = f"?item {self.subclassPredicate} wd:{self.itemQid}"
+            whereClause = f"?item {self.search_predicate} wd:{self.itemQid}"
             if self.endpointConf.database != "qlever":
                 whereClause += ";?p ?id"
         whereClause += "."
@@ -322,7 +322,7 @@ def noneTabularQuery(self, wdProperty: WikidataProperty, asFrequency: bool = Tru
 WHERE
 {{
   # instance of {self.item.qlabel}
-  ?item {self.subclassPredicate} wd:{self.itemQid}.{self.where}
+  ?item {self.search_predicate} wd:{self.itemQid}.{self.where}
   ?item rdfs:label ?itemLabel.
   FILTER (LANG(?itemLabel) = "{self.lang}").
   # {propertyLabel}

diff --git a/ez_wikidata/version.py b/ez_wikidata/version.py
@@ -17,7 +17,7 @@ class Version(object):
     name = "py_ez_wikidata"
     version = ez_wikidata.__version__
     date = "2024-03-01"
-    updated = "2024-03-06"
+    updated = "2024-03-07"
     description = "Mapping for Wikidata allows creation of wikidata entries from dicts"
 
     authors = "Tim Holzheim, Wolfgang Fahl"

diff --git a/tests/testTrulyTabular.py b/tests/testTrulyTabular.py
@@ -389,7 +389,7 @@ def testGenerateSparqlQuery(self):
             {
                 "naive": True,
                 "qid": "Q2020153",  # academic conference
-                "subclassPredicate": "wdt:P31",
+                "search_predicate": "wdt:P31",
                 "propertyIdMap": {
                     "P1813": ["label"],
                     "P17": ["label"],
@@ -400,7 +400,7 @@ def testGenerateSparqlQuery(self):
             {
                 "naive": False,
                 "qid": "Q2020153",  # academic conference
-                "subclassPredicate": "wdt:P31",
+                "search_predicate": "wdt:P31",
                 "propertyIdMap": {
                     "P1813": ["sample"],
                     "P17": ["sample"],
@@ -411,7 +411,7 @@ def testGenerateSparqlQuery(self):
             {
                 "naive": False,
                 "qid": "Q2020153",  # academic conference
-                "subclassPredicate": "wdt:P31",
+                "search_predicate": "wdt:P31",
                 "propertyIdMap": {
                     "P1813": ["count", "list"],
                     "P17": ["sample", "ignore"],
@@ -427,7 +427,7 @@ def testGenerateSparqlQuery(self):
             {
                 "naive": False,
                 "qid": "Q1667921",  # novel series
-                "subclassPredicate": "wdt:P31",
+                "search_predicate": "wdt:P31",
                 "propertyIdMap": {
                     "P50": ["sample", "ignore"],  # author
                     "P136": ["sample", "ignore"],  # genre
@@ -438,7 +438,7 @@ def testGenerateSparqlQuery(self):
             {
                 "naive": False,
                 "qid": "Q1667921",  # novel series
-                "subclassPredicate": "wdt:P31",
+                "search_predicate": "wdt:P31",
                 "propertyIdMap": {
                     "P50": ["sample", "ignore", "label"],  # author
                     "P136": ["sample", "ignore", "label"],  # genre
@@ -448,7 +448,7 @@ def testGenerateSparqlQuery(self):
             },
             {
                 "naive": False,
-                "subclassPredicate": "wdt:P279*/wdt:P31*",
+                "search_predicate": "wdt:P279*/wdt:P31*",
                 "qid": "Q8063",  # rock
                 "propertyIdMap": {
                     "P18": ["sample"],  # image
@@ -464,14 +464,14 @@ def testGenerateSparqlQuery(self):
             qid = config["qid"]
             naive = config["naive"]
             propertyIdMap = config["propertyIdMap"]
-            subclassPredicate = config["subclassPredicate"]
+            search_predicate = config["search_predicate"]
             expectedList = config["expected"]
 
             # create a truly tabular analysis
             tt = TrulyTabular(
                 qid,
                 propertyIds=list(propertyIdMap.keys()),
-                subclassPredicate=subclassPredicate,
+                search_predicate=search_predicate,
             )
             varname = tt.item.itemVarname
             # generate a SPARQL Query
@@ -481,7 +481,7 @@ def testGenerateSparqlQuery(self):
                 pprint(config)
                 print(f"{sparqlQuery}")
             # all queries should have basic graph patterns for the subclass
-            self.assertTrue(f"?{varname} {subclassPredicate} wd:{qid}." in sparqlQuery)
+            self.assertTrue(f"?{varname} {search_predicate} wd:{qid}." in sparqlQuery)
             # and for the properties
             for pid in propertyIdMap.keys():
                 self.assertTrue(f"?{varname} wdt:{pid}" in sparqlQuery)