ralsina/tartrazine
1
0
Fork 0
mirror of https://github.com/ralsina/tartrazine.git synced 2025-06-27 14:47:50 -03:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity

heuristics regexp engine configurable #3, adapt IsVendor optimization & tests

Browse Source 
Regex collation optimization for IsVendor now fails gracefully.
Tests that are affected by non-RE2 syntax are explicitly marked.
This commit is contained in:
Alex Bezzubov
2023-02-16 17:55:57 +01:00
parent 8df9e1ecf2
commit 8246efecce
4 changed files with 79 additions and 76 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
internal/code-generator/assets/vendor.go.tmpl
View File

@ -2,15 +2,21 @@ package data
import "github.com/go-enry/go-enry/v2/regex"
{{define "mustCompile" -}}
{{ if isRE2 . -}}
regex.MustCompile({{ . | stringVal }})
{{- else -}}
regex.MustCompileRuby({{ . | stringVal }})
{{- end -}}
{{end}}
var VendorMatchers = []regex.EnryRegexp{
{{range $re := . -}}
{{ if isRE2 $re -}}
regex.MustCompile({{ $re | stringVal }}),
{{- else -}}
regex.MustCompileRuby({{ $re | stringVal }}),
{{ end }}
{{ template "mustCompile" $re }},
{{end -}}
}
// FastVendorMatcher is equivalent to matching any of the VendorMatchers.
var FastVendorMatcher = regex.MustCompile(`{{ optimize . }}`)
{{with $singleRE := collateAllRegexps . -}}
var FastVendorMatcher = {{template "mustCompile" $singleRE}}
{{end}}

32
internal/code-generator/generator/vendor.go
View File

@ -41,34 +41,14 @@ func Vendor(fileToParse, samplesDir, outPath, tmplPath, tmplName, commit string)
}
func executeVendorTemplate(out io.Writer, regexps []string, tmplPath, tmplName, commit string) error {
funcs := template.FuncMap{"optimize": collateAllMatchers}
funcs := template.FuncMap{"collateAllRegexps": collateAllRegexps}
return executeTemplate(out, tmplName, tmplPath, commit, funcs, regexps)
}
func collateAllMatchers(regexps []string) string {
// We now collate all regexps from VendorMatchers to a single large regexp
// collateAllRegexps all regexps to a single large regexp.
func collateAllRegexps(regexps []string) string {
// which is at least twice as fast to test than simply iterating & matching.
//
// ---
//
// We could test each matcher from VendorMatchers in turn i.e.
//
// func IsVendor(filename string) bool {
// for _, matcher := range data.VendorMatchers {
// if matcher.MatchString(filename) {
// return true
// }
// }
// return false
// }
//
// Or naïvely concatentate all these regexps using groups i.e.
//
// `(regexp1)|(regexp2)|(regexp3)|...`
//
// However, both of these are relatively slow and don't take advantage
// of the inherent structure within our regexps.
//
// Imperical observation: by looking at the regexps, we only have 3 types.
// 1. Those that start with `^`
// 2. Those that start with `(^|/)`
@ -88,12 +68,9 @@ func collateAllMatchers(regexps []string) string {
sort.Strings(regexps)
// Check prefix, group expressions
var caretPrefixed, caretOrSlashPrefixed, theRest []string
// Check prefix, add to the respective group slices
for _, re := range regexps {
if !isRE2(re) {
continue
}
if strings.HasPrefix(re, caret) {
caretPrefixed = append(caretPrefixed, re[len(caret):])
} else if strings.HasPrefix(re, caretOrSlash) {
@ -102,6 +79,7 @@ func collateAllMatchers(regexps []string) string {
theRest = append(theRest, re)
}
}
var sb strings.Builder
appendGroupWithCommonPrefix(&sb, "^", caretPrefixed)
sb.WriteString("|")