better overall search results for type ahead. Still dealing with stop words failing.

crc/api/workflow.py

@@ -201,15 +201,6 @@ def lookup(workflow_id, task_id, field_id, query, limit):
     if not field:
         raise ApiError("unknown_field", "No field named %s in task %s" % (task_id, spiff_task.task_spec.name))
 
-    lookup_model = WorkflowService.get_lookup_table(spiff_task, field)
-    db_query = LookupDataModel.query.filter(LookupDataModel.lookup_file_model == lookup_model)
-
-    query = query.strip()
-    if(len(query) > 1):
-        if(' ' in query):
-            query = ':* ||'.join(query.split(' '))
-            db_query = db_query.filter(LookupDataModel.label.match("%s:*" % query)).limit(limit)
-        else:
-            db_query = db_query.filter(LookupDataModel.label.match("%s:*" % query)).limit(limit)
-
-    return LookupDataSchema(many=True).dump(db_query.all())
+    lookup_table = WorkflowService.get_lookup_table(spiff_task, field)
+    lookup_data = WorkflowService.run_lookup_query(lookup_table.id, query, limit)
+    return LookupDataSchema(many=True).dump(lookup_data)

crc/models/file.py

@@ -127,7 +127,7 @@ class LookupDataModel(db.Model):
     __table_args__ = (
         Index(
             'ix_lookupdata_tsv',
-            func.to_tsvector('english', label),
+            func.to_tsvector('simple', label),  # Use simple, not english to keep stop words in place.
             postgresql_using='gin'
             ),
         )

crc/services/workflow_service.py

@@ -5,6 +5,7 @@ from SpiffWorkflow.bpmn.workflow import BpmnWorkflow
 from SpiffWorkflow.dmn.specs.BuisnessRuleTask import BusinessRuleTask
 from SpiffWorkflow.specs import CancelTask, StartTask
 from pandas import ExcelFile
+from sqlalchemy import func
 
 from crc import db
 from crc.api.common import ApiError
@@ -205,3 +206,24 @@ class WorkflowService(object):
             db.session.commit()
 
         return lookup_model
+
+    @staticmethod
+    def run_lookup_query(lookup_file_id, query, limit):
+        db_query = LookupDataModel.query.filter(LookupDataModel.lookup_file_model_id == lookup_file_id)
+
+        query = query.strip()
+        if len(query) > 1:
+            if ' ' in query:
+                terms = query.split(' ')
+                query = ""
+                new_terms = []
+                for t in terms:
+                    new_terms.append(t + ":*")
+                query = '|'.join(new_terms)
+            else:
+                query = "%s:*" % query
+            db_query = db_query.filter(LookupDataModel.label.match(query))
+
+#            db_query = db_query.filter(text("lookup_data.label @@ to_tsquery('simple', '%s')" % query))
+
+        return db_query.limit(limit).all()

tests/test_workflow_service.py

@@ -129,5 +129,51 @@ class TestWorkflowService(BaseTest):
         self.assertIsNotNone(lookup_records)
         self.assertEqual(2, len(lookup_records))
 
-    def test_some_queries(self):
-        pass
+    def test_some_full_text_queries(self):
+        self.load_test_spec('enum_options_from_file')
+        file_model = session.query(FileModel).filter(FileModel.name == "customer_list.xls").first()
+        file_data_model = session.query(FileDataModel).filter(FileDataModel.file_model == file_model).first()
+        lookup_table = WorkflowService._get_lookup_table_from_data_model(file_data_model, "CUSTOMER_NUMBER", "CUSTOMER_NAME")
+        lookup_data = session.query(LookupDataModel).filter(LookupDataModel.lookup_file_model == lookup_table).all()
+
+        results = WorkflowService.run_lookup_query(lookup_table.id, "medicines", limit=10)
+        self.assertEquals(1, len(results), "words in the middle of label are detected.")
+        self.assertEquals("The Medicines Company", results[0].label)
+
+        results = WorkflowService.run_lookup_query(lookup_table.id, "", limit=10)
+        self.assertEquals(10, len(results), "Blank queries return everything, to the limit")
+
+        results = WorkflowService.run_lookup_query(lookup_table.id, "UVA", limit=10)
+        self.assertEquals(1, len(results), "Beginning of label is found.")
+        self.assertEquals("UVA - INTERNAL - GM USE ONLY", results[0].label)
+
+        results = WorkflowService.run_lookup_query(lookup_table.id, "uva", limit=10)
+        self.assertEquals(1, len(results), "case does not matter.")
+        self.assertEquals("UVA - INTERNAL - GM USE ONLY", results[0].label)
+
+
+
+        results = WorkflowService.run_lookup_query(lookup_table.id, "medici", limit=10)
+        self.assertEquals(1, len(results), "partial words are picked up.")
+        self.assertEquals("The Medicines Company", results[0].label)
+
+        results = WorkflowService.run_lookup_query(lookup_table.id, "Genetics Savings", limit=10)
+        self.assertEquals(1, len(results), "multiple terms are picked up..")
+        self.assertEquals("Genetics Savings & Clone, Inc.", results[0].label)
+
+        results = WorkflowService.run_lookup_query(lookup_table.id, "Genetics Sav", limit=10)
+        self.assertEquals(1, len(results), "prefix queries still work with partial terms")
+        self.assertEquals("Genetics Savings & Clone, Inc.", results[0].label)
+
+        results = WorkflowService.run_lookup_query(lookup_table.id, "Gen Sav", limit=10)
+        self.assertEquals(1, len(results), "prefix queries still work with ALL the partial terms")
+        self.assertEquals("Genetics Savings & Clone, Inc.", results[0].label)
+
+        results = WorkflowService.run_lookup_query(lookup_table.id, "Inc", limit=10)
+        self.assertEquals(7, len(results), "short terms get multiple correct results.")
+        self.assertEquals("Genetics Savings & Clone, Inc.", results[0].label)
+
+        # Fixme:  Stop words are taken into account on the query side, and haven't found a fix yet.
+        #results = WorkflowService.run_lookup_query(lookup_table.id, "in", limit=10)
+        #self.assertEquals(7, len(results), "stop words are not removed.")
+        #self.assertEquals("Genetics Savings & Clone, Inc.", results[0].label)