updated tests

src/poprox_recommender/topics.py

@@ -63,10 +63,10 @@ def find_locality(past_articles: list[Article], article_id: UUID):
 
 
 def normalized_category_count(counts: dict[str, int]):
-    if isinstance(counts, dict):
+    try:
         total_count = sum(counts.values())
         normalized_counts = {key: value / total_count for key, value in counts.items()}
-    else:
+    except Exception as _:
         normalized_counts = {}
     return normalized_counts
 

tests/components/test_calibration.py

@@ -4,38 +4,48 @@
 
 import logging
 
-from pytest import mark, skip
+from pytest import skip, xfail
 
 from poprox_concepts import ArticleSet
 from poprox_concepts.api.recommendations import RecommendationRequest
 from poprox_recommender.config import allow_data_test_failures
 from poprox_recommender.paths import project_root
 from poprox_recommender.recommenders import PipelineLoadError, select_articles
+from poprox_recommender.topics import user_locality_preference, user_topic_preference
 
 logger = logging.getLogger(__name__)
 logger.setLevel(logging.DEBUG)
 
 
 def test_request_with_topic_calibrator():
     test_dir = project_root() / "tests"
-    req_f = test_dir / "request_data" / "medium_request.json"
+    req_f = test_dir / "request_data" / "request_body.json"
     if allow_data_test_failures() and not req_f.exists():
         skip("request file does not exist")
 
     req = RecommendationRequest.model_validate_json(req_f.read_text())
-
-    base_outputs = select_articles(
-        ArticleSet(articles=req.todays_articles),
-        ArticleSet(articles=req.past_articles),
-        req.interest_profile,
-    )
-    topic_calibrated_outputs = select_articles(
-        ArticleSet(articles=req.todays_articles),
-        ArticleSet(articles=req.past_articles),
-        req.interest_profile,
-        pipeline_params={"pipeline": "topic-cali"},
+    req.interest_profile.click_topic_counts = user_topic_preference(
+        req.past_articles, req.interest_profile.click_history
     )
 
+    try:
+        base_outputs = select_articles(
+            ArticleSet(articles=req.todays_articles),
+            ArticleSet(articles=req.past_articles),
+            req.interest_profile,
+        )
+        topic_calibrated_outputs = select_articles(
+            ArticleSet(articles=req.todays_articles),
+            ArticleSet(articles=req.past_articles),
+            req.interest_profile,
+            pipeline_params={"pipeline": "topic-cali"},
+        )
+    except PipelineLoadError as e:
+        if allow_data_test_failures():
+            xfail("data not pulled")
+        else:
+            raise e
+
     # do we get recommendations?
     tco_recs = topic_calibrated_outputs.default.articles
     bo_recs = base_outputs.default.articles
@@ -51,23 +61,31 @@ def test_request_with_topic_calibrator():
 
 def test_request_with_locality_calibrator():
     test_dir = project_root() / "tests"
-    req_f = test_dir / "request_data" / "medium_request.json"
+    req_f = test_dir / "request_data" / "request_body.json"
     if allow_data_test_failures() and not req_f.exists():
         skip("request file does not exist")
 
     req = RecommendationRequest.model_validate_json(req_f.read_text())
-
-    base_outputs = select_articles(
-        ArticleSet(articles=req.todays_articles),
-        ArticleSet(articles=req.past_articles),
-        req.interest_profile,
-    )
-    locality_calibrated_outputs = select_articles(
-        ArticleSet(articles=req.todays_articles),
-        ArticleSet(articles=req.past_articles),
-        req.interest_profile,
-        pipeline_params={"pipeline": "locality-cali"},
+    req.interest_profile.click_locality_counts = user_locality_preference(
+        req.past_articles, req.interest_profile.click_history
     )
+    try:
+        base_outputs = select_articles(
+            ArticleSet(articles=req.todays_articles),
+            ArticleSet(articles=req.past_articles),
+            req.interest_profile,
+        )
+        locality_calibrated_outputs = select_articles(
+            ArticleSet(articles=req.todays_articles),
+            ArticleSet(articles=req.past_articles),
+            req.interest_profile,
+            pipeline_params={"pipeline": "locality-cali"},
+        )
+    except PipelineLoadError as e:
+        if allow_data_test_failures():
+            xfail("data not pulled")
+        else:
+            raise e
 
     # do we get recommendations?
     tco_recs = locality_calibrated_outputs.default.articles

tests/test_topic_classification.py

@@ -30,6 +30,7 @@ def load_test_articles():
     return candidate, past, click_history, num_recs
 
 
+@pytest.mark.skip(reason="too time intensive and not used by current prod logic")
 def test_topic_classification():
     candidate, _, _, _ = load_test_articles()
     topic_matched_dict, todays_article_matched_topics = match_news_topics_to_general(candidate.articles)