Merge pull request #180 from NCEAS/feature-178-project-fts

Project full text search

Author: regetz

helm/db/migrations/V2.2__enable_project_fts.sql

@@ -0,0 +1,56 @@
+/* -----------------------------
+    Project Full Text Search
+----------------------------- */
+
+-- Add name/description search column (and index) to project table
+ALTER TABLE project
+  ADD COLUMN search_vector tsvector;
+
+CREATE INDEX idx_project_search
+  ON project USING gin(search_vector);
+
+-- Create function to construct project search vector for a given project
+CREATE OR REPLACE FUNCTION build_project_search_vector(p_project_id INTEGER)
+RETURNS tsvector AS $$
+DECLARE
+  result tsvector;
+BEGIN
+  SELECT
+    setweight(to_tsvector('simple', COALESCE(pj.projectname, '')), 'A') ||
+    setweight(to_tsvector('simple', COALESCE(pj.projectdescription, '')), 'B')
+  INTO result
+  FROM project pj
+  WHERE pj.project_id = p_project_id;
+
+  RETURN result;
+END;
+$$ LANGUAGE plpgsql STABLE;
+
+-- Create function to populate the project search column with a
+-- constructed search vector for a single project record
+CREATE OR REPLACE FUNCTION update_project_search_vector(p_project_id INTEGER)
+RETURNS void AS $$
+BEGIN
+  UPDATE project
+  SET search_vector = build_project_search_vector(p_project_id)
+  WHERE project_id = p_project_id;
+END;
+$$ LANGUAGE plpgsql;
+
+-- Trigger for any changes to the project record itself
+CREATE OR REPLACE FUNCTION project_search_trigger()
+RETURNS trigger AS $$
+BEGIN
+  PERFORM update_project_search_vector(NEW.project_id);
+  RETURN NEW;
+END;
+$$ LANGUAGE plpgsql;
+
+CREATE TRIGGER trg_project_search
+AFTER INSERT OR UPDATE OF projectname, projectdescription ON project
+FOR EACH ROW
+EXECUTE FUNCTION project_search_trigger();
+
+-- Populate search vectors for all current projects
+UPDATE project
+  SET search_vector = build_project_search_vector(project_id);

vegbank/operators/Project.py

@@ -29,6 +29,10 @@ def __init__(self, params):
 
     def configure_query(self, *args, **kwargs):
         base_columns = {'*': "*"}
+        base_columns_search = {
+            'search_rank': "TS_RANK(pj.search_vector, " +
+                           "WEBSEARCH_TO_TSQUERY('simple', %s))"
+        }
         main_columns = {}
         main_columns['full'] = {
             'pj_code': "'pj.' || project_id",
@@ -53,6 +57,10 @@ def configure_query(self, *args, **kwargs):
                     'columns': base_columns,
                     'params': []
                 },
+                'search': {
+                    'columns': base_columns_search,
+                    'params': ['search']
+                },
             },
             'from': {
                 'sql': "FROM project AS pj",
@@ -63,6 +71,12 @@ def configure_query(self, *args, **kwargs):
                     'sql': None,
                     'params': []
                 },
+                'search': {
+                    'sql': """\
+                         pj.search_vector @@ WEBSEARCH_TO_TSQUERY('simple', %s)
+                    """,
+                    'params': ['search']
+                },
                 "pj": {
                     'sql': "pj.project_id = %s",
                     'params': ['vb_id']
@@ -78,6 +92,10 @@ def configure_query(self, *args, **kwargs):
                 'columns': main_columns[self.detail],
                 'params': []
             },
+            'search': {
+                'columns': {'search_rank': 'pj.search_rank'},
+                'params': []
+            },
         }
         self.query['from'] = {
             'sql': from_sql,
@@ -106,8 +124,13 @@ def validate_query_params(self, request_args):
         if request_args.get("detail", self.default_detail) not in ("full"):
             raise QueryParameterError("When provided, 'detail' must be 'full'.")
 
-        # now dispatch to the base validation method
-        return super().validate_query_params(request_args)
+        # dispatch to the base validation method
+        params = super().validate_query_params(request_args)
+
+        # capture search parameter, if it exists
+        params['search'] = request_args.get('search')
+
+        return params
 
 
     def upload_project(self, request, params):

vegbank/vegbankapi.py

@@ -446,6 +446,7 @@ def projects(pj_code):
             retrieved. If None, retrieves all projects.
 
     GET Query Parameters:
+        search (str, optional): Project name search query.
         detail (str, optional): Level of detail for the response.
             Only 'full' is defined for this method. Defaults to 'full'.
         limit (int, optional): Maximum number of records to return.