Merge pull request #21 from dpriskorn/check_relations_for_wikidata_qi…

…d_before_suggesting

Check relations for wikidata qid before suggesting them

pyproject.toml

@@ -16,6 +16,7 @@ questionary = "^1.10.0"
 python-dateutil = "^2.8.2"
 wikibaseintegrator = "^0.12.3"
 pydash = "^7.0.4"
+osmpythontools = "^0.3.5"
 
 
 [tool.poetry.group.dev.dependencies]

pytest.ini

@@ -1,3 +1,3 @@
 [pytest]
 log_cli = 1
-log_cli_level = INFO
+log_cli_level = DEBUG

src/models/trail_item.py

@@ -78,26 +78,28 @@ def waymarked_hiking_trails_search_url(self):
 
     def __convert_waymarked_results_to_choices__(self):
         for result in self.waymarked_results:
-            title = f"{result.name}"
-            if result.id:
-                title += f" ({result.id})"
-            if result.ref:
-                title += f", ref: {result.ref}"
-            if result.number_of_subroutes:
-                title += f", subroutes #: {result.number_of_subroutes}"
-            if result.names_of_subroutes_as_string:
-                title += f", subroutes: {result.names_of_subroutes_as_string}"
-            # if result.description:
-            #     title += f", description: {result.description}"
-            if result.group:
-                title += f", group: {result.group}"
-            if result.itinerary:
-                title += f", itinerary: {', '.join(result.itinerary)}"
-            choice = Choice(
-                title=textwrap.fill(title, 100),
-                value=QuestionaryReturn(osm_id=result.id),
-            )
-            self.choices.append(choice)
+            # We only want choices which are missing a Wikidata tag
+            if not result.already_has_wikidata_tag:
+                title = f"{result.name}"
+                if result.id:
+                    title += f" ({result.id})"
+                if result.ref:
+                    title += f", ref: {result.ref}"
+                if result.number_of_subroutes:
+                    title += f", subroutes #: {result.number_of_subroutes}"
+                if result.names_of_subroutes_as_string:
+                    title += f", subroutes: {result.names_of_subroutes_as_string}"
+                # if result.description:
+                #     title += f", description: {result.description}"
+                if result.group:
+                    title += f", group: {result.group}"
+                if result.itinerary:
+                    title += f", itinerary: {', '.join(result.itinerary)}"
+                choice = Choice(
+                    title=textwrap.fill(title, 100),
+                    value=QuestionaryReturn(osm_id=result.id),
+                )
+                self.choices.append(choice)
 
     def __remove_waymaked_result_duplicates__(self):
         self.waymarked_results = list(set(self.waymarked_results))
@@ -166,7 +168,7 @@ def __parse_not_found_in_osm_last_update_statement__(self):
                     else:
                         print("No not found in-property with a know value found")
         except KeyError:
-            logger.info("No not found in-claims on this item")
+            logger.info("No 'not found in'-claims on this item")
 
     def __lookup_label_on_waymarked_trails_and_ask_user_to_choose_a_match__(
         self,
@@ -253,6 +255,7 @@ def enrich_wikidata(self):
         if self.item:
             if self.chosen_osm_id:
                 self.__add_osm_id_to_item__()
+                self.__remove_not_found_in_osm_claim__()
                 self.summary = (
                     "Added match to OpenStreetMap via "
                     "the [[Wikidata:Tools/hiking trail matcher"
@@ -413,7 +416,9 @@ def __add_osm_id_to_item__(self):
                     prop_nr=Property.OSM_RELATION_ID.value,
                     value=str(self.chosen_osm_id),
                     references=[self.__create_heuristic_reference__()],
-                )
+                ),
+                # Replace no-value statement if it exists
+                action_if_exists=ActionIfExists.REPLACE_ALL,
             )
         else:
             # We got it from OSM Wikidata Link so add a reference
@@ -430,7 +435,9 @@ def __add_osm_id_to_item__(self):
                     prop_nr=Property.OSM_RELATION_ID.value,
                     value=str(self.chosen_osm_id),
                     references=References().add(reference=reference),
-                )
+                ),
+                # Replace no-value statement if it exists
+                action_if_exists=ActionIfExists.REPLACE_ALL,
             )
 
     def time_to_check_again(self, testing: bool = False) -> bool:
@@ -497,3 +504,21 @@ def __remove_osm_relation_no_value_claim__(self):
                 self.item.claims.remove(Property.OSM_RELATION_ID.value)
         except KeyError:
             logger.debug("No OSM_RELATION_ID found on this item to clean up")
+
+    def __remove_not_found_in_osm_claim__(self):
+        try:
+            claims = self.item.claims.get(Property.NOT_FOUND_IN.value)
+            if claims:
+                if len(claims) > 1:
+                    # todo iterate and remove only the right one
+                    console.print(claims)
+                    raise NotImplementedError(
+                        "removing only one of "
+                        "multiple not-found-in-"
+                        "statements is not supported yet"
+                    )
+                else:
+                    logger.info("Removing 'not found in'-claim")
+                    self.item.claims.remove(Property.OSM_RELATION_ID.value)
+        except KeyError:
+            logger.debug("No NOT_FOUND_IN found on this item to remove")

src/models/waymarked_result.py

@@ -1,6 +1,7 @@
 import logging
 from typing import Dict, List
 
+from OSMPythonTools.api import Api  # type: ignore
 from pydantic import BaseModel
 from requests import Session
 
@@ -78,3 +79,15 @@ def names_of_subroutes_as_string(self) -> str:
             return ", ".join(self.__names_of_subroutes__)
         else:
             return ""
+
+    @property
+    def already_has_wikidata_tag(self) -> bool:
+        """This check uses the Openstreetmap API because it is fast"""
+        api = Api()
+        relation = api.query(f"relation/{self.id}")
+        wikidata = relation.tag("wikidata")
+        logging.debug(f"wikidata tag: {wikidata}")
+        if wikidata is None:
+            return False
+        else:
+            return True

tests/test_trail_item.py

@@ -31,12 +31,11 @@ def test_remove_duplicates(self):
 
     def test_convert_to_choices(self):
         trail_item = TrailItem(wbi=WikibaseIntegrator())
-        trail_item.waymarked_results.append(WaymarkedResult(name="test", id=1))
-        trail_item.waymarked_results.append(WaymarkedResult(name="test2", id=2))
+        trail_item.waymarked_results.append(WaymarkedResult(name="test", id=10528596))
         trail_item.__convert_waymarked_results_to_choices__()
-        assert len(trail_item.choices) == 2
-        assert trail_item.choices[0].title == "test (1)"
-        assert trail_item.choices[0].value.osm_id == 1
+        assert len(trail_item.choices) == 1
+        assert trail_item.choices[0].title == "test (10528596)"
+        assert trail_item.choices[0].value.osm_id == 10528596
 
     def test_lookup_in_osm_wikidata_link_api_no_match(self):
         trail_item = TrailItem(wbi=WikibaseIntegrator(), qid="Q820225")

tests/test_waymarked_result.py

@@ -71,3 +71,13 @@ def test_number_of_subroutes(self):
         wr = WaymarkedResult(id=1014050, name="skåneleden")  # skåneleden
         wr.get_details()
         assert wr.number_of_subroutes == 6
+
+    def test_already_has_wikidata_tag_true(self):
+        wr = WaymarkedResult(id=1014050, name="skåneleden")  # skåneleden
+        wr.get_details()
+        assert wr.already_has_wikidata_tag is True
+
+    def test_already_has_wikidata_tag_false(self):
+        wr = WaymarkedResult(id=10528596, name="test")  # östra leden på kebnekaise
+        wr.get_details()
+        assert wr.already_has_wikidata_tag is False