lib/storage: optimize convert multiple values regexp filter to composite tag filter (#1610)

* lib/storage: optimize convert multiple values regexp filter to composite tag filter

* Apply suggestions from code review

Co-authored-by: Aliaksandr Valialkin <valyala@gmail.com>

lib/storage/tag_filters.go

@@ -18,34 +18,58 @@ import (
 //
 // This converts `foo{bar="baz",x=~"a.+"}` to `{foo=bar="baz",foo=x=~"a.+"} filter.
 func convertToCompositeTagFilterss(tfss []*TagFilters) []*TagFilters {
-	tfssNew := make([]*TagFilters, len(tfss))
+	tfssNew := make([]*TagFilters, 0, len(tfss))
-	for i, tfs := range tfss {
+	for _, tfs := range tfss {
-		tfssNew[i] = convertToCompositeTagFilters(tfs)
+		tfssNew = append(tfss, convertToCompositeTagFilters(tfs)...)
 	}
 	return tfssNew
 }
 
-func convertToCompositeTagFilters(tfs *TagFilters) *TagFilters {
+func convertToCompositeTagFilters(tfs *TagFilters) []*TagFilters {
-	// Search for metric name filter, which must be used for creating composite filters.
+	tfssCompiled := make([]*TagFilters, 0)
-	var name []byte
+
+	// Search for filters on metric name, which will be used for creating composite filters.
+	var names [][]byte
 	hasPositiveFilter := false
 	for _, tf := range tfs.tfs {
 		if len(tf.key) == 0 && !tf.isNegative && !tf.isRegexp {
-			name = tf.value
+			names = [][]byte{tf.value}
+		} else if len(tf.key) == 0 && !tf.isNegative && tf.isRegexp && len(tf.orSuffixes) > 0 {
+			// Split the filter {__name__=~"name1|...|nameN", other_filters}
+			// into `name1{other_filters}`, ..., `nameN{other_filters}`
+			// and generate composite filters for each of them
+			names = names[:0]
+			for _, orSuffix := range tf.orSuffixes {
+				names = append(names, []byte(orSuffix))
+			}
 		} else if !tf.isNegative && !tf.isEmptyMatch {
 			hasPositiveFilter = true
 		}
 	}
-	if len(name) == 0 {
+	if len(names) == 0 {
+		tfssCompiled = append(tfssCompiled, tfs)
 		atomic.AddUint64(&compositeFilterMissingConversions, 1)
-		return tfs
+		return tfssCompiled
 	}
-	tfsNew := make([]tagFilter, 0, len(tfs.tfs))
+
 	var compositeKey []byte
 	compositeFilters := 0
+	for _, name := range names {
+		tfsNew := make([]tagFilter, 0, len(tfs.tfs))
 		for _, tf := range tfs.tfs {
 			if len(tf.key) == 0 {
-			if !hasPositiveFilter || tf.isNegative || tf.isRegexp || string(tf.value) != string(name) {
+				sameOrSuffixes := true
+				if len(names) != len(tf.orSuffixes) {
+					sameOrSuffixes = false
+				} else {
+					for i, orSuffix := range tf.orSuffixes {
+						if string(names[i]) != orSuffix {
+							sameOrSuffixes = false
+							break
+						}
+					}
+				}
+				if !hasPositiveFilter || tf.isNegative || tf.isRegexp && !sameOrSuffixes || !tf.isRegexp && string(tf.value) != string(name) {
 					tfsNew = append(tfsNew, tf)
 				}
 				continue
@@ -54,6 +78,7 @@ func convertToCompositeTagFilters(tfs *TagFilters) *TagFilters {
 				tfsNew = append(tfsNew, tf)
 				continue
 			}
+
 			compositeKey = marshalCompositeTagKey(compositeKey[:0], name, tf.key)
 			var tfNew tagFilter
 			if err := tfNew.Init(tfs.commonPrefix, compositeKey, tf.value, tf.isNegative, tf.isRegexp); err != nil {
@@ -62,14 +87,17 @@ func convertToCompositeTagFilters(tfs *TagFilters) *TagFilters {
 			tfsNew = append(tfsNew, tfNew)
 			compositeFilters++
 		}
-	if compositeFilters == 0 {
-		atomic.AddUint64(&compositeFilterMissingConversions, 1)
-		return tfs
-	}
 		tfsCompiled := NewTagFilters()
 		tfsCompiled.tfs = tfsNew
+		tfssCompiled = append(tfssCompiled, tfsCompiled)
+	}
+	if compositeFilters == 0 {
+		tfssCompiled = append(tfssCompiled[:0], tfs)
+		atomic.AddUint64(&compositeFilterMissingConversions, 1)
+		return tfssCompiled
+	}
 	atomic.AddUint64(&compositeFilterSuccessConversions, 1)
-	return tfsCompiled
+	return tfssCompiled
 }
 
 var (

lib/storage/tag_filters_test.go

@@ -7,7 +7,7 @@ import (
 )
 
 func TestConvertToCompositeTagFilters(t *testing.T) {
-	f := func(tfs, resultExpected []TagFilter) {
+	f := func(tfs []TagFilter, resultExpected [][]TagFilter) {
 		t.Helper()
 		tfsCompiled := NewTagFilters()
 		for _, tf := range tfs {
@@ -15,23 +15,27 @@ func TestConvertToCompositeTagFilters(t *testing.T) {
 				t.Fatalf("cannot add tf=%s: %s", tf.String(), err)
 			}
 		}
-		resultCompiled := convertToCompositeTagFilters(tfsCompiled)
+		resultCompileds := convertToCompositeTagFilters(tfsCompiled)
-		result := make([]TagFilter, len(resultCompiled.tfs))
+		result := make([][]TagFilter, len(resultCompileds))
+		for i, resultCompiled := range resultCompileds {
+			tfs := make([]TagFilter, len(resultCompiled.tfs))
 			for i, tf := range resultCompiled.tfs {
-			result[i] = TagFilter{
+				tfs[i] = TagFilter{
 					Key:        tf.key,
 					Value:      tf.value,
 					IsNegative: tf.isNegative,
 					IsRegexp:   tf.isRegexp,
 				}
 			}
+			result[i] = tfs
+		}
 		if !reflect.DeepEqual(result, resultExpected) {
 			t.Fatalf("unexpected result;\ngot\n%+v\nwant\n%+v", result, resultExpected)
 		}
 	}
 
 	// Empty filters
-	f(nil, []TagFilter{})
+	f(nil, [][]TagFilter{{}})
 
 	// A single non-name filter
 	f([]TagFilter{
@@ -41,13 +45,15 @@ func TestConvertToCompositeTagFilters(t *testing.T) {
 			IsNegative: false,
 			IsRegexp:   false,
 		},
-	}, []TagFilter{
+	}, [][]TagFilter{
+		{
 			{
 				Key:        []byte("foo"),
 				Value:      []byte("bar"),
 				IsNegative: false,
 				IsRegexp:   false,
 			},
+		},
 	})
 
 	// Multiple non-name filters
@@ -64,7 +70,8 @@ func TestConvertToCompositeTagFilters(t *testing.T) {
 			IsNegative: true,
 			IsRegexp:   false,
 		},
-	}, []TagFilter{
+	}, [][]TagFilter{
+		{
 			{
 				Key:        []byte("foo"),
 				Value:      []byte("bar"),
@@ -77,6 +84,7 @@ func TestConvertToCompositeTagFilters(t *testing.T) {
 				IsNegative: true,
 				IsRegexp:   false,
 			},
+		},
 	})
 
 	// A single name filter
@@ -87,13 +95,15 @@ func TestConvertToCompositeTagFilters(t *testing.T) {
 			IsNegative: false,
 			IsRegexp:   false,
 		},
-	}, []TagFilter{
+	}, [][]TagFilter{
+		{
 			{
 				Key:        nil,
 				Value:      []byte("bar"),
 				IsNegative: false,
 				IsRegexp:   false,
 			},
+		},
 	})
 
 	// Two name filters
@@ -110,7 +120,8 @@ func TestConvertToCompositeTagFilters(t *testing.T) {
 			IsNegative: false,
 			IsRegexp:   false,
 		},
-	}, []TagFilter{
+	}, [][]TagFilter{
+		{
 			{
 				Key:        nil,
 				Value:      []byte("bar"),
@@ -123,6 +134,7 @@ func TestConvertToCompositeTagFilters(t *testing.T) {
 				IsNegative: false,
 				IsRegexp:   false,
 			},
+		},
 	})
 
 	// A name filter with non-name filter.
@@ -139,13 +151,15 @@ func TestConvertToCompositeTagFilters(t *testing.T) {
 			IsNegative: false,
 			IsRegexp:   false,
 		},
-	}, []TagFilter{
+	}, [][]TagFilter{
+		{
 			{
 				Key:        []byte("\xfe\x03barfoo"),
 				Value:      []byte("abc"),
 				IsNegative: false,
 				IsRegexp:   false,
 			},
+		},
 	})
 
 	// A name filter with a single negative filter
@@ -162,7 +176,8 @@ func TestConvertToCompositeTagFilters(t *testing.T) {
 			IsNegative: true,
 			IsRegexp:   false,
 		},
-	}, []TagFilter{
+	}, [][]TagFilter{
+		{
 			{
 				Key:        nil,
 				Value:      []byte("bar"),
@@ -175,6 +190,7 @@ func TestConvertToCompositeTagFilters(t *testing.T) {
 				IsNegative: true,
 				IsRegexp:   false,
 			},
+		},
 	})
 
 	// A name filter with a negative and a positive filter
@@ -197,7 +213,8 @@ func TestConvertToCompositeTagFilters(t *testing.T) {
 			IsNegative: false,
 			IsRegexp:   true,
 		},
-	}, []TagFilter{
+	}, [][]TagFilter{
+		{
 			{
 				Key:        []byte("\xfe\x03barfoo"),
 				Value:      []byte("abc"),
@@ -210,6 +227,7 @@ func TestConvertToCompositeTagFilters(t *testing.T) {
 				IsNegative: false,
 				IsRegexp:   true,
 			},
+		},
 	})
 
 	// Two name filters with non-name filter.
@@ -232,7 +250,8 @@ func TestConvertToCompositeTagFilters(t *testing.T) {
 			IsNegative: false,
 			IsRegexp:   false,
 		},
-	}, []TagFilter{
+	}, [][]TagFilter{
+		{
 			{
 				Key:        nil,
 				Value:      []byte("bar"),
@@ -245,6 +264,7 @@ func TestConvertToCompositeTagFilters(t *testing.T) {
 				IsNegative: false,
 				IsRegexp:   false,
 			},
+		},
 	})
 
 	// A name filter with regexp non-name filter, which can be converted to non-regexp.
@@ -261,13 +281,15 @@ func TestConvertToCompositeTagFilters(t *testing.T) {
 			IsNegative: false,
 			IsRegexp:   true,
 		},
-	}, []TagFilter{
+	}, [][]TagFilter{
+		{
 			{
 				Key:        []byte("\xfe\x03barfoo"),
 				Value:      []byte("abc"),
 				IsNegative: false,
 				IsRegexp:   false,
 			},
+		},
 	})
 
 	// A name filter with regexp non-name filter.
@@ -284,13 +306,15 @@ func TestConvertToCompositeTagFilters(t *testing.T) {
 			IsNegative: false,
 			IsRegexp:   true,
 		},
-	}, []TagFilter{
+	}, [][]TagFilter{
+		{
 			{
 				Key:        []byte("\xfe\x03barfoo"),
 				Value:      []byte("abc.+"),
 				IsNegative: false,
 				IsRegexp:   true,
 			},
+		},
 	})
 
 	// A name filter with graphite filter.
@@ -307,7 +331,8 @@ func TestConvertToCompositeTagFilters(t *testing.T) {
 			IsNegative: false,
 			IsRegexp:   false,
 		},
-	}, []TagFilter{
+	}, [][]TagFilter{
+		{
 			{
 				Key:        nil,
 				Value:      []byte("bar"),
@@ -320,6 +345,7 @@ func TestConvertToCompositeTagFilters(t *testing.T) {
 				IsNegative: false,
 				IsRegexp:   false,
 			},
+		},
 	})
 
 	// A name filter with non-name filter and a graphite filter.
@@ -342,7 +368,8 @@ func TestConvertToCompositeTagFilters(t *testing.T) {
 			IsNegative: false,
 			IsRegexp:   false,
 		},
-	}, []TagFilter{
+	}, [][]TagFilter{
+		{
 			{
 				Key:        []byte("\xfe\x03barfoo"),
 				Value:      []byte("abc"),
@@ -355,6 +382,7 @@ func TestConvertToCompositeTagFilters(t *testing.T) {
 				IsNegative: false,
 				IsRegexp:   false,
 			},
+		},
 	})
 
 	// Regexp name filter, which can be converted to non-regexp, with non-name filter.
@@ -371,13 +399,144 @@ func TestConvertToCompositeTagFilters(t *testing.T) {
 			IsNegative: false,
 			IsRegexp:   false,
 		},
-	}, []TagFilter{
+	}, [][]TagFilter{
+		{
 			{
 				Key:        []byte("\xfe\x03barfoo"),
 				Value:      []byte("abc"),
 				IsNegative: false,
 				IsRegexp:   false,
 			},
+		},
+	})
+
+	// Multiple values regexp filter, which can be converted to non-regexp, with non-name filter.
+	f([]TagFilter{
+		{
+			Key:        nil,
+			Value:      []byte("bar|foo"),
+			IsNegative: false,
+			IsRegexp:   true,
+		},
+		{
+			Key:        []byte("foo"),
+			Value:      []byte("abc"),
+			IsNegative: false,
+			IsRegexp:   false,
+		},
+	}, [][]TagFilter{
+		{
+			{
+				Key:        []byte("\xfe\x03barfoo"),
+				Value:      []byte("abc"),
+				IsNegative: false,
+				IsRegexp:   false,
+			},
+		},
+		{
+			{
+				Key:        []byte("\xfe\x03foofoo"),
+				Value:      []byte("abc"),
+				IsNegative: false,
+				IsRegexp:   false,
+			},
+		},
+	})
+
+	// Two multiple values regexp filter, which can be converted to non-regexp, with non-name filter.
+	f([]TagFilter{
+		{
+			Key:        nil,
+			Value:      []byte("bar|foo"),
+			IsNegative: false,
+			IsRegexp:   true,
+		},
+		{
+			Key:        nil,
+			Value:      []byte("abc|def"),
+			IsNegative: false,
+			IsRegexp:   true,
+		},
+		{
+			Key:        []byte("face"),
+			Value:      []byte("air"),
+			IsNegative: false,
+			IsRegexp:   false,
+		},
+	}, [][]TagFilter{
+		{
+			{
+				Key:        nil,
+				Value:      []byte("bar|foo"),
+				IsNegative: false,
+				IsRegexp:   true,
+			},
+			{
+				Key:        []byte("\xfe\x03abcface"),
+				Value:      []byte("air"),
+				IsNegative: false,
+				IsRegexp:   false,
+			},
+		},
+		{
+			{
+				Key:        nil,
+				Value:      []byte("bar|foo"),
+				IsNegative: false,
+				IsRegexp:   true,
+			},
+			{
+				Key:        []byte("\xfe\x03defface"),
+				Value:      []byte("air"),
+				IsNegative: false,
+				IsRegexp:   false,
+			},
+		},
+	})
+
+	// Multiple values regexp filter with a single negative filter
+	f([]TagFilter{
+		{
+			Key:        nil,
+			Value:      []byte("bar|foo"),
+			IsNegative: false,
+			IsRegexp:   true,
+		},
+		{
+			Key:        []byte("foo"),
+			Value:      []byte("abc"),
+			IsNegative: true,
+			IsRegexp:   false,
+		},
+	}, [][]TagFilter{
+		{
+			{
+				Key:        nil,
+				Value:      []byte("bar|foo"),
+				IsNegative: false,
+				IsRegexp:   true,
+			},
+			{
+				Key:        []byte("\xfe\x03barfoo"),
+				Value:      []byte("abc"),
+				IsNegative: true,
+				IsRegexp:   false,
+			},
+		},
+		{
+			{
+				Key:        nil,
+				Value:      []byte("bar|foo"),
+				IsNegative: false,
+				IsRegexp:   true,
+			},
+			{
+				Key:        []byte("\xfe\x03foofoo"),
+				Value:      []byte("abc"),
+				IsNegative: true,
+				IsRegexp:   false,
+			},
+		},
 	})
 
 	// Regexp name filter with non-name filter.
@@ -394,7 +553,8 @@ func TestConvertToCompositeTagFilters(t *testing.T) {
 			IsNegative: true,
 			IsRegexp:   false,
 		},
-	}, []TagFilter{
+	}, [][]TagFilter{
+		{
 			{
 				Key:        nil,
 				Value:      []byte("bar.+"),
@@ -407,6 +567,7 @@ func TestConvertToCompositeTagFilters(t *testing.T) {
 				IsNegative: true,
 				IsRegexp:   false,
 			},
+		},
 	})
 
 	// Regexp non-name filter, which matches anything.
@@ -423,13 +584,15 @@ func TestConvertToCompositeTagFilters(t *testing.T) {
 			IsNegative: false,
 			IsRegexp:   true,
 		},
-	}, []TagFilter{
+	}, [][]TagFilter{
+		{
 			{
 				Key:        nil,
 				Value:      []byte("bar"),
 				IsNegative: false,
 				IsRegexp:   false,
 			},
+		},
 	})
 }