Merge pull request #27 from shashank-boyapally/lookback-size

lookback size

fmatch/matcher.py

@@ -1,5 +1,4 @@
-""" metadata matcher
-"""
+"""metadata matcher"""
 
 # pylint: disable = invalid-name, invalid-unary-operand-type, no-member
 import os
@@ -23,10 +22,10 @@ class Matcher:
 
     def __init__(
         self,
-        index: str ="ospst-perf-scale-ci",
-        level: int =logging.INFO,
-        ES_URL: str =os.getenv("ES_SERVER"),
-        verify_certs: bool =True,
+        index: str = "ospst-perf-scale-ci",
+        level: int = logging.INFO,
+        ES_URL: str = os.getenv("ES_SERVER"),
+        verify_certs: bool = True,
     ):
         self.index = index
         self.es_url = ES_URL
@@ -68,9 +67,29 @@ def query_index(self, index: str, search: Search) -> Response:
         return search.execute()
 
     def get_uuid_by_metadata(
-        self, meta: Dict[str, Any], index: str = None, lookback_date: datetime = None
+        self,
+        meta: Dict[str, Any],
+        index: str = None,
+        lookback_date: datetime = None,
+        lookback_size: int = 10000,
     ) -> List[Dict[str, str]]:
-        """get_uuid_by_metadata"""
+        """gets uuid by metadata
+
+        Args:
+            meta (Dict[str, Any]): metadata of the runs
+            index (str, optional): Index to search. Defaults to None.
+            lookback_date (datetime, optional): 
+            The cutoff date to get the uuids from. Defaults to None.
+            lookback_size (int, optional): 
+            Maximum number of runs to get, gets the latest. Defaults to 10000.
+
+            lookback_size and lookback_date get the data on the 
+            precedency of whichever cutoff comes first.
+            Similar to a car manufacturer's warranty limits.
+
+        Returns:
+            List[Dict[str, str]]: _description_
+        """
         if index is None:
             index = self.index
         version = meta["ocpVersion"][:4]
@@ -98,7 +117,12 @@ def get_uuid_by_metadata(
             must=must_clause,
             filter=filter_clause,
         )
-        s = Search(using=self.es, index=index).query(query).extra(size=self.search_size)
+        s = (
+            Search(using=self.es, index=index)
+            .query(query)
+            .sort({"timestamp": {"order": "desc"}})
+            .extra(size=lookback_size)
+        )
         result = self.query_index(index, s)
         hits = result.hits.hits
         uuids_docs = [

fmatch/tests/test_matcher.py

@@ -100,6 +100,30 @@ def test_get_uuid_by_metadata_lookback(matcher_instance):
                 "buildUrl":"buildUrl1"}]
     assert result == expected
 
+def test_get_uuid_by_metadata_lookback_size(matcher_instance):
+    matcher_instance.es.search = lambda *args, **kwargs: {
+        "hits": {
+            "hits": [{"_source": {"uuid": "uuid1",
+                                  "buildUrl":"buildUrl1",
+                                  "timestamp":"2024-07-10T13:46:24Z"}}, 
+                    {"_source": {"uuid": "uuid2",
+                                  "buildUrl":"buildUrl1",
+                                  "timestamp":"2024-07-08T13:46:24Z"}}]
+        }
+    }
+    meta = {
+        "field1": "value1",
+        "ocpVersion": "4.15",
+    }
+    date= datetime.datetime.strptime("2024-07-07T13:46:24Z","%Y-%m-%dT%H:%M:%SZ")
+    result = matcher_instance.get_uuid_by_metadata(meta=meta, lookback_date=date, lookback_size=2)
+    print(result)
+    expected= [{"uuid": "uuid1",
+                "buildUrl":"buildUrl1"},
+               {"uuid": "uuid2",
+                "buildUrl":"buildUrl1"}]
+    assert result == expected
+
 
 def test_match_kube_burner(matcher_instance):
     result = matcher_instance.match_kube_burner(["uuid1"],index="ospst-*")