github
/
prometheus
mirror of https://github.com/prometheus/prometheus
1
0
Fork 0
Code Issues Projects Releases Wiki Activity
Browse Source

Merge pull request #14173 from pracucci/fastregexmatcher-optimize-contains 

Improve contains check done by FastRegexMatcher
pull/14252/head
Bryan Boreham 6 months ago committed by GitHub
parent
687b9c2d62 d966ae6400
commit
6fb738af51
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 70 additions and 33 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Split View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 43
      model/labels/regexp.go
  2. 60
      model/labels/regexp_test.go

43
model/labels/regexp.go
Unescape View File

@ -42,7 +42,7 @@ type FastRegexMatcher struct {
stringMatcher StringMatcher
prefix string
suffix string
contains string
contains []string
// matchString is the "compiled" function to run by MatchString().
matchString func(string) bool
@ -87,7 +87,7 @@ func NewFastRegexMatcher(v string) (*FastRegexMatcher, error) {
// compileMatchStringFunction returns the function to run by MatchString().
func (m *FastRegexMatcher) compileMatchStringFunction() func(string) bool {
// If the only optimization available is the string matcher, then we can just run it.
if len(m.setMatches) == 0 && m.prefix == "" && m.suffix == "" && m.contains == "" && m.stringMatcher != nil {
if len(m.setMatches) == 0 && m.prefix == "" && m.suffix == "" && len(m.contains) == 0 && m.stringMatcher != nil {
return m.stringMatcher.Matches
}
@ -106,7 +106,7 @@ func (m *FastRegexMatcher) compileMatchStringFunction() func(string) bool {
if m.suffix != "" && !strings.HasSuffix(s, m.suffix) {
return false
}
if m.contains != "" && !strings.Contains(s, m.contains) {
if len(m.contains) > 0 && !containsInOrder(s, m.contains) {
return false
}
if m.stringMatcher != nil {
@ -119,7 +119,7 @@ func (m *FastRegexMatcher) compileMatchStringFunction() func(string) bool {
// IsOptimized returns true if any fast-path optimization is applied to the
// regex matcher.
func (m *FastRegexMatcher) IsOptimized() bool {
return len(m.setMatches) > 0 || m.stringMatcher != nil || m.prefix != "" || m.suffix != "" || m.contains != ""
return len(m.setMatches) > 0 || m.stringMatcher != nil || m.prefix != "" || m.suffix != "" || len(m.contains) > 0
}
// findSetMatches extract equality matches from a regexp.
@ -361,8 +361,9 @@ func optimizeAlternatingLiterals(s string) (StringMatcher, []string) {
// optimizeConcatRegex returns literal prefix/suffix text that can be safely
// checked against the label value before running the regexp matcher.
func optimizeConcatRegex(r *syntax.Regexp) (prefix, suffix, contains string) {
func optimizeConcatRegex(r *syntax.Regexp) (prefix, suffix string, contains []string) {
sub := r.Sub
clearCapture(sub...)
// We can safely remove begin and end text matchers respectively
// at the beginning and end of the regexp.
@ -387,13 +388,11 @@ func optimizeConcatRegex(r *syntax.Regexp) (prefix, suffix, contains string) {
suffix = string(sub[last].Rune)
}
// If contains any literal which is not a prefix/suffix, we keep the
// 1st one. We do not keep the whole list of literals to simplify the
// fast path.
// If contains any literal which is not a prefix/suffix, we keep track of
// all the ones which are case-sensitive.
for i := 1; i < len(sub)-1; i++ {
if sub[i].Op == syntax.OpLiteral && (sub[i].Flags&syntax.FoldCase) == 0 {
contains = string(sub[i].Rune)
break
contains = append(contains, string(sub[i].Rune))
}
}
@ -940,3 +939,27 @@ func hasPrefixCaseInsensitive(s, prefix string) bool {
func hasSuffixCaseInsensitive(s, suffix string) bool {
return len(s) >= len(suffix) && strings.EqualFold(s[len(s)-len(suffix):], suffix)
}
func containsInOrder(s string, contains []string) bool {
// Optimization for the case we only have to look for 1 substring.
if len(contains) == 1 {
return strings.Contains(s, contains[0])
}
return containsInOrderMulti(s, contains)
}
func containsInOrderMulti(s string, contains []string) bool {
offset := 0
for _, substr := range contains {
at := strings.Index(s[offset:], substr)
if at == -1 {
return false
}
offset += at + len(substr)
}
return true
}

60
model/labels/regexp_test.go
Unescape View File

@ -81,10 +81,15 @@ var (
".*foo.?",
".?foo.+",
"foo.?|bar",
// Concat of literals and wildcards.
".*-.*-.*-.*-.*",
"(.+)-(.+)-(.+)-(.+)-(.+)",
"((.*))(?i:f)((.*))o((.*))o((.*))",
"((.*))f((.*))(?i:o)((.*))o((.*))",
}
values = []string{
"foo", " foo bar", "bar", "buzz\nbar", "bar foo", "bfoo", "\n", "\nfoo", "foo\n", "hello foo world", "hello foo\n world", "",
"FOO", "Foo", "OO", "Oo", "\nfoo\n", strings.Repeat("f", 20), "prometheus", "prometheus_api_v1", "prometheus_api_v1_foo",
"FOO", "Foo", "fOo", "foO", "OO", "Oo", "\nfoo\n", strings.Repeat("f", 20), "prometheus", "prometheus_api_v1", "prometheus_api_v1_foo",
"10.0.1.20", "10.0.2.10", "10.0.3.30", "10.0.4.40",
"foofoo0", "foofoo", "😀foo0",
@ -132,29 +137,29 @@ func TestOptimizeConcatRegex(t *testing.T) {
regex string
prefix string
suffix string
contains string
contains []string
}{
{regex: "foo(hello|bar)", prefix: "foo", suffix: "", contains: ""},
{regex: "foo(hello|bar)world", prefix: "foo", suffix: "world", contains: ""},
{regex: "foo.*", prefix: "foo", suffix: "", contains: ""},
{regex: "foo.*hello.*bar", prefix: "foo", suffix: "bar", contains: "hello"},
{regex: ".*foo", prefix: "", suffix: "foo", contains: ""},
{regex: "^.*foo$", prefix: "", suffix: "foo", contains: ""},
{regex: ".*foo.*", prefix: "", suffix: "", contains: "foo"},
{regex: ".*foo.*bar.*", prefix: "", suffix: "", contains: "foo"},
{regex: ".*(foo|bar).*", prefix: "", suffix: "", contains: ""},
{regex: ".*[abc].*", prefix: "", suffix: "", contains: ""},
{regex: ".*((?i)abc).*", prefix: "", suffix: "", contains: ""},
{regex: ".*(?i:abc).*", prefix: "", suffix: "", contains: ""},
{regex: "(?i:abc).*", prefix: "", suffix: "", contains: ""},
{regex: ".*(?i:abc)", prefix: "", suffix: "", contains: ""},
{regex: ".*(?i:abc)def.*", prefix: "", suffix: "", contains: "def"},
{regex: "(?i).*(?-i:abc)def", prefix: "", suffix: "", contains: "abc"},
{regex: ".*(?msU:abc).*", prefix: "", suffix: "", contains: "abc"},
{regex: "[aA]bc.*", prefix: "", suffix: "", contains: "bc"},
{regex: "^5..$", prefix: "5", suffix: "", contains: ""},
{regex: "^release.*", prefix: "release", suffix: "", contains: ""},
{regex: "^env-[0-9]+laio[1]?[^0-9].*", prefix: "env-", suffix: "", contains: "laio"},
{regex: "foo(hello|bar)", prefix: "foo", suffix: "", contains: nil},
{regex: "foo(hello|bar)world", prefix: "foo", suffix: "world", contains: nil},
{regex: "foo.*", prefix: "foo", suffix: "", contains: nil},
{regex: "foo.*hello.*bar", prefix: "foo", suffix: "bar", contains: []string{"hello"}},
{regex: ".*foo", prefix: "", suffix: "foo", contains: nil},
{regex: "^.*foo$", prefix: "", suffix: "foo", contains: nil},
{regex: ".*foo.*", prefix: "", suffix: "", contains: []string{"foo"}},
{regex: ".*foo.*bar.*", prefix: "", suffix: "", contains: []string{"foo", "bar"}},
{regex: ".*(foo|bar).*", prefix: "", suffix: "", contains: nil},
{regex: ".*[abc].*", prefix: "", suffix: "", contains: nil},
{regex: ".*((?i)abc).*", prefix: "", suffix: "", contains: nil},
{regex: ".*(?i:abc).*", prefix: "", suffix: "", contains: nil},
{regex: "(?i:abc).*", prefix: "", suffix: "", contains: nil},
{regex: ".*(?i:abc)", prefix: "", suffix: "", contains: nil},
{regex: ".*(?i:abc)def.*", prefix: "", suffix: "", contains: []string{"def"}},
{regex: "(?i).*(?-i:abc)def", prefix: "", suffix: "", contains: []string{"abc"}},
{regex: ".*(?msU:abc).*", prefix: "", suffix: "", contains: []string{"abc"}},
{regex: "[aA]bc.*", prefix: "", suffix: "", contains: []string{"bc"}},
{regex: "^5..$", prefix: "5", suffix: "", contains: nil},
{regex: "^release.*", prefix: "release", suffix: "", contains: nil},
{regex: "^env-[0-9]+laio[1]?[^0-9].*", prefix: "env-", suffix: "", contains: []string{"laio"}},
}
for _, c := range cases {
@ -1089,6 +1094,15 @@ func TestHasSuffixCaseInsensitive(t *testing.T) {
require.False(t, hasSuffixCaseInsensitive("marco", "abcdefghi"))
}
func TestContainsInOrder(t *testing.T) {
require.True(t, containsInOrder("abcdefghilmno", []string{"ab", "cd", "no"}))
require.True(t, containsInOrder("abcdefghilmno", []string{"def", "hil"}))
require.False(t, containsInOrder("abcdefghilmno", []string{"ac"}))
require.False(t, containsInOrder("abcdefghilmno", []string{"ab", "cd", "de"}))
require.False(t, containsInOrder("abcdefghilmno", []string{"cd", "ab"}))
}
func getTestNameFromRegexp(re string) string {
if len(re) > 32 {
return re[:32]

Write Preview
Loading…
Cancel
Save