Added transforms to search payload, id2uri function for classes and p…

…roperties

pyproject.toml

@@ -28,6 +28,7 @@ loguru = "^0.6.0"
 orjson = "^3.8.2"
 drepr = "^2.10.0"
 rsoup = "^2.5.1"
+nh3 = "^0.2.13"
 
 lat_lon_parser = "^1.3.0"
 

sand/config.py

@@ -31,6 +31,7 @@
     "ont_classes": {
         "constructor": "sand.extensions.wikidata.get_ontclass_db",
         "uri2id": "sand.extensions.wikidata.uri2id",
+        "id2uri": "sm.namespaces.prelude.WikidataNamespace.get_entity_abs_uri",
         "args": {
             "dbfile": "/tmp/wdclasses.db",
             "proxy": True,
@@ -41,6 +42,7 @@
     "ont_props": {
         "constructor": "sand.extensions.wikidata.get_ontprop_db",
         "uri2id": "sand.extensions.wikidata.uri2id",
+        "id2uri": "sm.namespaces.prelude.WikidataNamespace.get_prop_abs_uri",
         "args": {
             "dbfile": "/tmp/wdprops.db",
             "proxy": True,

sand/extensions/search/wikidata_search.py

@@ -1,12 +1,15 @@
 import requests
 from flask import request, jsonify
+import nh3
 from sand.controllers.search import ISearch
-
+from sand.models.entity import Entity
+from sand.models.ontology import OntClass, OntProperty
 
 class WikidataSearch(ISearch):
 
     def __init__(self):
         self.wikidata_url = "https://www.wikidata.org/w/api.php"
+        self.local_class_idsearch_uri = "http://0.0.0.0:5525/api/classes/"
         self.PARAMS = {
             "action": "query",
             "format": "json",
@@ -17,13 +20,23 @@ def __init__(self):
             "srlimit": 10,
             "srprop": "snippet|titlesnippet"
         }
+        self.search_item_template = {
+            "label": "",
+            "id": "",
+            "description": "",
+            "uri": "",
+        }
 
     def get_class_search_params(self, search_text):
         class_params = self.PARAMS.copy()
         class_params["srnamespace"] = 0
         class_params['srsearch'] = f"haswbstatement:P279 {search_text}"
         return class_params
 
+    def get_local_class_properties(self, id):
+        api_data = requests.get(self.local_class_idsearch_uri+str(id))
+        return api_data.json()
+
     def get_entity_search_params(self, search_text):
         entity_params = self.PARAMS.copy()
         entity_params["srnamespace"] = 0
@@ -38,15 +51,43 @@ def get_props_search_params(self, search_text):
 
     def find_class_by_name(self, search_text):
         request_params = self.get_class_search_params(search_text)
-        data = requests.get(self.wikidata_url, request_params)
-        return data.json()
+        api_data = requests.get(self.wikidata_url, request_params)
+        search_items = api_data.json()['query']['search']
+        payload = {"items": []}
+        for search_item in search_items:
+            item = self.search_item_template.copy()
+            item['id'] = search_item['title']
+            local_class_props = self.get_local_class_properties(item['id'])
+            item['label'] = local_class_props['label']
+            item['description'] = local_class_props['description']
+            item['uri'] = OntClass.id2uri(item['id'])
+            payload['items'].append(item)
+        return payload
 
     def find_entity_by_name(self, search_text):
         request_params = self.get_entity_search_params(search_text)
-        data = requests.get(self.wikidata_url, request_params)
-        return data.json()['query']['search']
+        api_data = requests.get(self.wikidata_url, request_params)
+        search_items = api_data.json()['query']['search']
+        payload = {"items":[]}
+        for search_item in search_items:
+            item = self.search_item_template.copy()
+            item['label'] = nh3.clean(search_item['titlesnippet'], tags=set())
+            item['id'] = search_item['title']
+            item['description'] = nh3.clean(search_item['snippet'], tags=set())
+            item['uri'] = Entity.id2uri(item['id'])
+            payload['items'].append(item)
+        return payload
 
     def find_props_by_name(self, search_text):
         request_params = self.get_props_search_params(search_text)
-        data = requests.get(self.wikidata_url, request_params)
-        return data.json()['query']['search']
+        api_data = requests.get(self.wikidata_url, request_params)
+        search_items = api_data.json()['query']['search']
+        payload = {"items": []}
+        for search_item in search_items:
+            item = self.search_item_template.copy()
+            item['label'] = nh3.clean(search_item['titlesnippet'], tags=set())
+            item['id'] = search_item['title'].split(":")[1]
+            item['description'] = nh3.clean(search_item['snippet'], tags=set())
+            item['uri'] = OntProperty.id2uri(item['id'])
+            payload['items'].append(item)
+        return payload

sand/models/ontology.py

@@ -24,9 +24,15 @@ def readable_label(self):
     def uri2id(uri: str) -> str:
         """Convert class URI to entity ID."""
         raise NotImplementedError(
-            "The method is set when its store is initialized. Check the call order to ensure `OntClassAR` is called first"
+            "The method is set when its store is initialized. Check the call order to ensure `OntClassAR`is called first"
         )
 
+    @staticmethod
+    def id2uri(id: str) -> str:
+        """Convert class ID to class URI."""
+        raise NotImplementedError(
+            "The method is set when its store is initialized. Check the call order to ensure `OntClassAR` is called first"
+        )
 
 OntPropertyDataType = Literal[
     "monolingualtext",
@@ -62,6 +68,12 @@ def uri2id(uri: str) -> str:
             "The method is set when its store is initialized. Check the call order to ensure `OntPropertyAR` is called first"
         )
 
+    @staticmethod
+    def id2uri(id: str) -> str:
+        """Convert property ID to property URI."""
+        raise NotImplementedError(
+            "The method is set when its store is initialized. Check the call order to ensure `OntPropertyAR` is called first"
+        )
 
 PROP_AR = None
 CLASS_AR = None
@@ -89,6 +101,7 @@ def OntPropertyAR() -> Mapping[str, OntProperty]:
         func = import_func(cfg["constructor"])
         PROP_AR = func(**cfg["args"])
         OntProperty.uri2id = import_func(cfg["uri2id"])
+        OntProperty.id2uri = import_func(cfg["id2uri"])
 
     return PROP_AR
 
@@ -102,5 +115,5 @@ def OntClassAR() -> Mapping[str, OntClass]:
         func = import_func(cfg["constructor"])
         CLASS_AR = func(**cfg["args"])
         OntClass.uri2id = import_func(cfg["uri2id"])
-
+        OntClass.id2uri = import_func(cfg["id2uri"])
     return CLASS_AR