fix(query) Regex equals .* must ignore the label and match series eve…

…n without the label (filodb#1639)

core/src/main/scala/filodb.core/memstore/PartKeyLuceneIndex.scala

@@ -817,12 +817,14 @@ class PartKeyLuceneIndex(ref: DatasetRef,
     logger.info(s"Refreshed index searchers to make reads consistent for dataset=$ref shard=$shardNum")
   }
 
+  //scalastyle:off method.length
   private def leafFilter(column: String, filter: Filter): Query = {
     filter match {
       case EqualsRegex(value) =>
         val regex = removeRegexAnchors(value.toString)
-        if (regex.nonEmpty) new RegexpQuery(new Term(column, regex), RegExp.NONE)
-        else leafFilter(column, NotEqualsRegex(".+"))  // value="" means the label is absent or has an empty value.
+        if (regex.replaceAll("\\.\\*", "").nonEmpty) new RegexpQuery(new Term(column, regex), RegExp.NONE)
+        else leafFilter(column, NotEqualsRegex(".+")) // value="" means the label is absent or has an empty value.
+
       case NotEqualsRegex(value) =>
         val term = new Term(column, removeRegexAnchors(value.toString))
         val allDocs = new MatchAllDocsQuery
@@ -864,7 +866,7 @@ class PartKeyLuceneIndex(ref: DatasetRef,
       case _ => throw new UnsupportedOperationException
     }
   }
-
+  //scalastyle:on method.length
   def partIdsFromFilters(columnFilters: Seq[ColumnFilter],
                          startTime: Long,
                          endTime: Long): debox.Buffer[Int] = {

core/src/test/scala/filodb.core/memstore/PartKeyLuceneIndexSpec.scala

@@ -1008,4 +1008,55 @@ class PartKeyLuceneIndexSpec extends AnyFunSpec with Matchers with BeforeAndAfte
     // close CardinalityTracker to avoid leaking of resources
     cardTracker.close()
   }
+
+  it("should match records without label when .* is provided on a non existent label") {
+
+    val pkrs = partKeyFromRecords(dataset6, records(dataset6, readers.take(10)), Some(partBuilder))
+      .zipWithIndex.map { case (addr, i) =>
+      val pk = partKeyOnHeap(dataset6.schema.partKeySchema, ZeroPointer, addr)
+      keyIndex.addPartKey(pk, i, i, i + 10)()
+      PartKeyLuceneIndexRecord(pk, i, i + 10)
+    }
+    keyIndex.refreshReadersBlocking()
+
+
+   // Query with just the existing Label name
+    val filter1 = ColumnFilter("Actor2Code", Equals("GOV".utf8))
+    val result1 = keyIndex.partKeyRecordsFromFilters(Seq(filter1), 0, Long.MaxValue)
+    val expected1 = Seq(pkrs(7), pkrs(8), pkrs(9))
+
+    result1.map(_.partKey.toSeq) shouldEqual expected1.map(_.partKey.toSeq)
+    result1.map(p => (p.startTime, p.endTime)) shouldEqual expected1.map(p => (p.startTime, p.endTime))
+
+    // Query with non existent label name with an empty regex
+    val filter2 = ColumnFilter("dummy", EqualsRegex(".*".utf8))
+    val filter3 = ColumnFilter("Actor2Code", Equals("GOV".utf8))
+    val result2 = keyIndex.partKeyRecordsFromFilters(Seq(filter2, filter3), 0, Long.MaxValue)
+    val expected2 = Seq(pkrs(7), pkrs(8), pkrs(9))
+
+    result2.map(_.partKey.toSeq) shouldEqual expected2.map(_.partKey.toSeq)
+    result2.map(p => (p.startTime, p.endTime)) shouldEqual expected2.map(p => (p.startTime, p.endTime))
+
+    // Query with non existent label name with an regex matching at least 1 character
+    val filter4 = ColumnFilter("dummy", EqualsRegex(".+".utf8))
+    val filter5 = ColumnFilter("Actor2Code", Equals("GOV".utf8))
+    val result3 = keyIndex.partKeyRecordsFromFilters(Seq(filter4, filter5), 0, Long.MaxValue)
+    result3 shouldEqual Seq()
+
+    // Query with non existent label name with an empty regex
+    val filter6 = ColumnFilter("dummy", EqualsRegex("".utf8))
+    val filter7 = ColumnFilter("Actor2Code", Equals("GOV".utf8))
+    val result4 = keyIndex.partKeyRecordsFromFilters(Seq(filter6, filter7), 0, Long.MaxValue)
+    val expected4 = Seq(pkrs(7), pkrs(8), pkrs(9))
+    result4.map(_.partKey.toSeq) shouldEqual expected4.map(_.partKey.toSeq)
+    result4.map(p => (p.startTime, p.endTime)) shouldEqual expected4.map(p => (p.startTime, p.endTime))
+
+    // Query with non existent label name with an empty equals
+    val filter8 = ColumnFilter("dummy", Equals("".utf8))
+    val filter9 = ColumnFilter("Actor2Code", Equals("GOV".utf8))
+    val result5 = keyIndex.partKeyRecordsFromFilters(Seq(filter8, filter9), 0, Long.MaxValue)
+    val expected5 = Seq(pkrs(7), pkrs(8), pkrs(9))
+    result5.map(_.partKey.toSeq) shouldEqual expected5.map(_.partKey.toSeq)
+    result5.map(p => (p.startTime, p.endTime)) shouldEqual expected5.map(p => (p.startTime, p.endTime))
+  }
 }