[Issue #1455] Implement logic for query box in the search API (#1514)

## Summary Fixes #1455 ### Time to review: __10 mins__ ## Changes proposed Implements logic for the query field in the search API Updated the seed script to allow you to generate more opportunities locally ## Context for reviewers This logic is deliberately very, very simple at the moment. It basically just does a text contains check across several columns. This is the most naive approach, and it is expected that if you tried to search two separate words at the same time, you probably won't get any results. A few options we may consider later (depending on the level of effort we want to put into a non-search-index approach): - Postgres text search / tsvectors: https://www.postgresql.org/docs/current/textsearch.html - Postgres text similarity: https://www.postgresql.org/docs/current/pgtrgm.html ## Additional information The `ids` function I added for the tests was because there are a lot of tests and when test `16` fails, its really hard to figure out which one that is. Now it outputs the request+expected results: ![Screenshot 2024-03-21 at 10 47 34 AM](https://github.com/HHS/simpler-grants-gov/assets/46358556/a7732375-00a8-4e4f-8e16-c00114a80342) --------- Co-authored-by: nava-platform-bot <platform-admins@navapbc.com>
HHS · Mar 22, 2024 · 2b205af · 2b205af
1 parent 7f050b7
commit 2b205af
Show file tree

Hide file tree

Showing 7 changed files with 392 additions and 27 deletions.
diff --git a/api/Makefile b/api/Makefile
@@ -167,7 +167,7 @@ db-migrate-heads: ## Show migrations marked as a head
 	$(alembic_cmd) heads $(args)
 
 db-seed-local:
-	$(PY_RUN_CMD) db-seed-local
+	$(PY_RUN_CMD) db-seed-local $(args)
 
 db-check-migrations: ## Verify the DB schema matches the DB migrations generated
 	$(alembic_cmd) check || (echo -e "\n$(RED)Migrations are not up-to-date, make sure you generate migrations by running 'make db-migrate-create <msg>'$(NO_COLOR)"; exit 1)

diff --git a/api/openapi.generated.yml b/api/openapi.generated.yml
@@ -932,6 +932,8 @@ components:
       properties:
         query:
           type: string
+          minLength: 1
+          maxLength: 100
           description: Query string which searches against several text fields
           example: research
         filters:

diff --git a/api/src/api/opportunities_v0_1/opportunity_schemas.py b/api/src/api/opportunities_v0_1/opportunity_schemas.py
@@ -1,4 +1,4 @@
-from src.api.schemas.extension import Schema, fields
+from src.api.schemas.extension import Schema, fields, validators
 from src.api.schemas.search_schema import StrSearchSchemaBuilder
 from src.constants.lookup_constants import (
     ApplicantType,
@@ -276,7 +276,8 @@ class OpportunitySearchRequestSchema(Schema):
         metadata={
             "description": "Query string which searches against several text fields",
             "example": "research",
-        }
+        },
+        validate=[validators.Length(min=1, max=100)],
     )
 
     filters = fields.Nested(OpportunitySearchFilterSchema())

diff --git a/api/src/services/opportunities_v0_1/search_opportunities.py b/api/src/services/opportunities_v0_1/search_opportunities.py
@@ -12,6 +12,7 @@
     LinkOpportunitySummaryFundingCategory,
     LinkOpportunitySummaryFundingInstrument,
     Opportunity,
+    OpportunityAssistanceListing,
     OpportunitySummary,
 )
 from src.pagination.pagination_models import PaginationInfo, PaginationParams
@@ -51,7 +52,44 @@ def _add_query_filters(stmt: Select[tuple[Any]], query: str | None) -> Select[tu
     if query is None or len(query) == 0:
         return stmt
 
-    # TODO - will implement this in https://github.com/HHS/simpler-grants-gov/issues/1455
+    ilike_query = f"%{query}%"
+
+    # Add a left join to the assistance listing table to filter by any of its values
+    stmt = stmt.outerjoin(
+        OpportunityAssistanceListing,
+        Opportunity.opportunity_id == OpportunityAssistanceListing.opportunity_id,
+    )
+
+    """
+    This adds the following to the inner query (assuming the query value is "example")
+
+    WHERE
+        (opportunity.opportunity_title ILIKE '%example%'
+        OR opportunity.opportunity_number ILIKE '%example%'
+        OR opportunity.agency ILIKE '%example%'
+        OR opportunity_summary.summary_description ILIKE '%example%'
+        OR opportunity_assistance_listing.assistance_listing_number = 'example'
+        OR opportunity_assistance_listing.program_title ILIKE '%example%'))
+
+    Note that SQLAlchemy escapes everything and queries are actually written like:
+
+        opportunity.opportunity_number ILIKE % (opportunity_number_1)
+    """
+    stmt = stmt.where(
+        or_(
+            # Title partial match
+            Opportunity.opportunity_title.ilike(ilike_query),
+            # Number partial match
+            Opportunity.opportunity_number.ilike(ilike_query),
+            # Agency (code) partial match
+            Opportunity.agency.ilike(ilike_query),
+            # Summary description partial match
+            OpportunitySummary.summary_description.ilike(ilike_query),
+            # assistance listing number matches exactly or program title partial match
+            OpportunityAssistanceListing.assistance_listing_number == query,
+            OpportunityAssistanceListing.program_title.ilike(ilike_query),
+        )
+    )
 
     return stmt
 

diff --git a/api/tests/lib/seed_local_db.py b/api/tests/lib/seed_local_db.py
@@ -1,5 +1,6 @@
 import logging
 
+import click
 from sqlalchemy import func
 
 import src.adapters.db as db
@@ -13,7 +14,7 @@
 logger = logging.getLogger(__name__)
 
 
-def _build_opportunities(db_session: db.Session) -> None:
+def _build_opportunities(db_session: db.Session, iterations: int) -> None:
     # Just create a variety of opportunities for local testing
     # we can eventually look into creating more specific scenarios
 
@@ -26,23 +27,27 @@ def _build_opportunities(db_session: db.Session) -> None:
     logger.info(f"Creating opportunities starting with opportunity_id {max_opportunity_id + 1}")
     factories.OpportunityFactory.reset_sequence(value=max_opportunity_id + 1)
 
-    # Create a few opportunities in various scenarios
-    factories.OpportunityFactory.create_batch(size=5, is_forecasted_summary=True)
-    factories.OpportunityFactory.create_batch(size=5, is_posted_summary=True)
-    factories.OpportunityFactory.create_batch(size=5, is_closed_summary=True)
-    factories.OpportunityFactory.create_batch(size=5, is_archived_non_forecast_summary=True)
-    factories.OpportunityFactory.create_batch(size=5, is_archived_forecast_summary=True)
-    factories.OpportunityFactory.create_batch(size=5, no_current_summary=True)
-
-    # generate a few opportunities with mostly null values
-    all_null_opportunities = factories.OpportunityFactory.create_batch(size=5, all_fields_null=True)
-    for all_null_opportunity in all_null_opportunities:
-        summary = factories.OpportunitySummaryFactory.create(
-            all_fields_null=True, opportunity=all_null_opportunity
-        )
-        factories.CurrentOpportunitySummaryFactory.create(
-            opportunity=all_null_opportunity, opportunity_summary=summary
+    for i in range(iterations):
+        logger.info(f"Creating opportunity batch number  {i}")
+        # Create a few opportunities in various scenarios
+        factories.OpportunityFactory.create_batch(size=5, is_forecasted_summary=True)
+        factories.OpportunityFactory.create_batch(size=5, is_posted_summary=True)
+        factories.OpportunityFactory.create_batch(size=5, is_closed_summary=True)
+        factories.OpportunityFactory.create_batch(size=5, is_archived_non_forecast_summary=True)
+        factories.OpportunityFactory.create_batch(size=5, is_archived_forecast_summary=True)
+        factories.OpportunityFactory.create_batch(size=5, no_current_summary=True)
+
+        # generate a few opportunities with mostly null values
+        all_null_opportunities = factories.OpportunityFactory.create_batch(
+            size=5, all_fields_null=True
         )
+        for all_null_opportunity in all_null_opportunities:
+            summary = factories.OpportunitySummaryFactory.create(
+                all_fields_null=True, opportunity=all_null_opportunity
+            )
+            factories.CurrentOpportunitySummaryFactory.create(
+                opportunity=all_null_opportunity, opportunity_summary=summary
+            )
 
     logger.info("Finished creating opportunities")
 
@@ -57,7 +62,13 @@ def _build_opportunities(db_session: db.Session) -> None:
     logger.info("Finished creating records in the transfer_topportunity table")
 
 
-def seed_local_db() -> None:
+@click.command()
+@click.option(
+    "--iterations",
+    default=1,
+    help="Number of sets of opportunities to create, note that several are created per iteration",
+)
+def seed_local_db(iterations: int) -> None:
     with src.logging.init("seed_local_db"):
         logger.info("Running seed script for local DB")
         error_if_not_local()
@@ -67,7 +78,7 @@ def seed_local_db() -> None:
         with db_client.get_session() as db_session:
             factories._db_session = db_session
 
-            _build_opportunities(db_session)
+            _build_opportunities(db_session, iterations)
             # Need to commit to force any updates made
             # after factories created objects
             db_session.commit()