ref: replace search_dataset with search_dataset_via_api

Author: paulmueller

CHANGELOG

@@ -2,6 +2,8 @@
  - fix: error handling when connection to server fails on download (#75)
  - enh: add `CKANAPI.hostname`
  - ref: replace `user_list` with `user_autocomplete`
+ - ref: replace `search_dataset` with `search_dataset_via_api`
+   in `APIInterrogator`; `search_dataset` is now only free text search.
 0.17.4
  - fix: correctly handle dataset IDs and names in `DBExtract`
  - fix: `get_datasets_user_shared` failed when many circles/collections exist

dcoraid/dbmodel/db_api.py

@@ -63,7 +63,7 @@ def get_datasets_user_following(self):
     def get_datasets_user_owned(self):
         """Return datasets the user created"""
         assert self.mode == "user"
-        dbextract = self.search_dataset(
+        dbextract = self.search_dataset_via_api(
             filter_queries=[f"+creator_user_id:{self.api.user_id}"],
             limit=0,
         )
@@ -78,14 +78,14 @@ def get_datasets_user_shared(self):
         dbextract = DBExtract()
 
         for circles_batch in batched(self.get_circles(), 20):
-            dbextract += self.search_dataset(
+            dbextract += self.search_dataset_via_api(
                 circles=list(circles_batch),
                 filter_queries=[f"-creator_user_id:{self.api.user_id}"],
                 limit=0,
                 )
 
         for collections_batch in batched(self.get_collections(), 20):
-            dbextract += self.search_dataset(
+            dbextract += self.search_dataset_via_api(
                 collections=list(collections_batch),
                 filter_queries=[f"-creator_user_id:{self.api.user_id}"],
                 limit=0,
@@ -117,9 +117,27 @@ def get_users(self, ret_fullnames=False):
         else:
             return user_list
 
-    def search_dataset(self, query="*:*", filter_queries=None, circles=None,
-                       collections=None, circle_collection_union=False,
-                       limit=100):
+    def search_dataset(self,
+                       text: str,
+                       limit: int = 100):
+        """Free text search for a dataset in the database
+
+        Parameters
+        ----------
+        text: str
+            text to search for
+        limit: int
+            number of results to return
+        """
+        return self.search_dataset_via_api(query=text, limit=limit)
+
+    def search_dataset_via_api(self,
+                               query: str = "*:*",
+                               filter_queries: list[str] = None,
+                               circles: list[str] = None,
+                               collections: list[str] = None,
+                               circle_collection_union: bool = False,
+                               limit: int = 100):
         """Search datasets via the CKAN API
 
         Parameters

dcoraid/dbmodel/db_core.py

@@ -65,7 +65,8 @@ def get_users(self):
         pass
 
     @abc.abstractmethod
-    def search_dataset(self, query, circles=None, collections=None):
+    def search_dataset(self, text):
+        """Free text search for a dataset in the database"""
         pass
 
     @property

dcoraid/gui/browse_public/widget_browse_public.py

@@ -33,8 +33,8 @@ def on_public_search(self):
         api = get_ckan_api(
             public=not self.checkBox_public_include_private.isChecked())
         try:
-            ai = APIInterrogator(api=api)
-            dbextract = ai.search_dataset(
+            db = APIInterrogator(api=api)
+            dbextract = db.search_dataset(
                 self.lineEdit_public_search.text(),
                 limit=self.spinBox_public_rows.value())
             self.public_filter_chain.set_db_extract(dbextract)

tests/test_dbmodel_api.py

@@ -78,21 +78,21 @@ def test_search_dataset_basic():
                    activate=True)
     db = db_api.APIInterrogator(api=api)
     # Positive test
-    data = db.search_dataset(query="dcoraid",
-                             circles=[defaults["circle"]],
-                             collections=[defaults["collection"]],
-                             )
+    data = db.search_dataset_via_api(query="dcoraid",
+                                     circles=[defaults["circle"]],
+                                     collections=[defaults["collection"]],
+                                     )
     assert len(data) >= 1
     for dd in data:
         if dd["name"].endswith(ranstr):
             break
     else:
         assert False, "{} not found!".format(defaults["dataset"])
     # Negative test
-    data = db.search_dataset(query="cliauwenlc_should_never_exist",
-                             circles=[defaults["circle"]],
-                             collections=[defaults["collection"]],
-                             )
+    data = db.search_dataset_via_api(query="cliauwenlc_should_never_exist",
+                                     circles=[defaults["circle"]],
+                                     collections=[defaults["collection"]],
+                                     )
     assert len(data) == 0, "search result for non-existent dataset?"
 
 
@@ -115,19 +115,21 @@ def test_search_dataset_limit():
             activate=True)
         dataset_ids.append(ds_dict["id"])
     db = db_api.APIInterrogator(api=api)
-    data_limited = db.search_dataset(query=ranstr,
-                                     circles=[defaults["circle"]],
-                                     collections=[defaults["collection"]],
-                                     circle_collection_union=True,
-                                     limit=2
-                                     )
+    data_limited = db.search_dataset_via_api(
+        query=ranstr,
+        circles=[defaults["circle"]],
+        collections=[defaults["collection"]],
+        circle_collection_union=True,
+        limit=2
+        )
     assert len(data_limited) == 2
-    data_unlimited = db.search_dataset(query=ranstr,
-                                       circles=[defaults["circle"]],
-                                       collections=[defaults["collection"]],
-                                       circle_collection_union=True,
-                                       limit=0
-                                       )
+    data_unlimited = db.search_dataset_via_api(
+        query=ranstr,
+        circles=[defaults["circle"]],
+        collections=[defaults["collection"]],
+        circle_collection_union=True,
+        limit=0
+        )
     assert len(data_unlimited) == 3
 
 
@@ -148,12 +150,13 @@ def test_search_dataset_limit_negative_error():
         activate=True)
     db = db_api.APIInterrogator(api=api)
     with pytest.raises(ValueError, match="must be 0 or >0"):
-        db.search_dataset(query=ranstr,
-                          circles=[defaults["circle"]],
-                          collections=[defaults["collection"]],
-                          circle_collection_union=True,
-                          limit=-1
-                          )
+        db.search_dataset_via_api(
+            query=ranstr,
+            circles=[defaults["circle"]],
+            collections=[defaults["collection"]],
+            circle_collection_union=True,
+            limit=-1
+            )
 
 
 @pytest.mark.skipif(not HAS_FIGSHARE_ACCESS,
@@ -162,7 +165,8 @@ def test_search_dataset_only_one_filter_query():
     # The figshare circle must have the testing user as a member
     api = common.get_api()
     db = db_api.APIInterrogator(api=api)
-    ds = db.search_dataset(filter_queries=[f"-creator_user_id:{api.user_id}"])
+    ds = db.search_dataset_via_api(
+        filter_queries=[f"-creator_user_id:{api.user_id}"])
     for di in ds:
         if di["name"] == "figshare-7771184-v2":
             break