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

Add check for non-backtracking subexpressions

Browse Source
This commit is contained in:
Lauris BH
2022-03-21 13:54:11 +02:00
parent 8eac4cab85
commit ae2b0576a7
4 changed files with 11 additions and 38 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
internal/code-generator/generator/heuristics.go
View File

@ -39,7 +39,7 @@ func GenHeuristics(fileToParse, _, outPath, tmplPath, tmplName, commit string) e
// loadHeuristics transforms parsed YAML to map[".ext"]->IR for code generation.
func loadHeuristics(yaml *Heuristics) (map[string][]*LanguagePattern, error) {
var patterns = make(map[string][]*LanguagePattern)
patterns := make(map[string][]*LanguagePattern)
for _, disambiguation := range yaml.Disambiguations {
var rules []*LanguagePattern
for _, rule := range disambiguation.Rules {
@ -161,13 +161,14 @@ func parseYaml(file string) (*Heuristics, error) {
// isUnsupportedRegexpSyntax filters regexp syntax that is not supported by RE2.
// In particular, we stumbled up on usage of next cases:
// - lookbehind & lookahead
// - non-backtracking subexpressions
// - named & numbered capturing group/after text matching
// - backreference
// - possessive quantifier
// For referece on supported syntax see https://github.com/google/re2/wiki/Syntax
func isUnsupportedRegexpSyntax(reg string) bool {
return strings.Contains(reg, `(?<`) || strings.Contains(reg, `(?=`) || strings.Contains(reg, `(?!`) ||
strings.Contains(reg, `\1`) || strings.Contains(reg, `*+`) ||
strings.Contains(reg, `(?>`) || strings.Contains(reg, `\1`) || strings.Contains(reg, `*+`) ||
// See https://github.com/github/linguist/pull/4243#discussion_r246105067
(strings.HasPrefix(reg, multilinePrefix+`/`) && strings.HasSuffix(reg, `/`))
}

21
internal/code-generator/generator/test_files/content.gold
View File

@ -697,12 +697,7 @@ var ContentHeuristics = map[string]*Heuristics{
regexp.MustCompile(`(?m)^(\s*namespace\s*[\w\.]+\s*{|\s*\/\/)`),
),
},
".csc": &Heuristics{
rule.Or(
rule.MatchingLanguages("GSC"),
regexp.MustCompile(`(?m)^\s*#\s*(?:using|insert|include|define|namespace)[ \t]+\w|^\s*(?>(?:autoexec|private)\s+){0,2}function\s+(?>(?:autoexec|private)\s+){0,2}\w+\s*\(|\b(?:level|self)[ \t]+thread[ \t]+(?:\[\[[ \t]*(?>\w+\.)*\w+[ \t]*\]\]|\w+)[ \t]*\([^\r\n\)]*\)[ \t]*;|^[ \t]*#[ \t]*(?:precache|using_animtree)[ \t]*\(`),
),
},
".csc": &Heuristics{},
".csl": &Heuristics{
rule.Or(
rule.MatchingLanguages("XML"),
@ -879,18 +874,8 @@ var ContentHeuristics = map[string]*Heuristics{
regexp.MustCompile(`(?m)^\[indent=[0-9]+\]`),
),
},
".gsc": &Heuristics{
rule.Or(
rule.MatchingLanguages("GSC"),
regexp.MustCompile(`(?m)^\s*#\s*(?:using|insert|include|define|namespace)[ \t]+\w|^\s*(?>(?:autoexec|private)\s+){0,2}function\s+(?>(?:autoexec|private)\s+){0,2}\w+\s*\(|\b(?:level|self)[ \t]+thread[ \t]+(?:\[\[[ \t]*(?>\w+\.)*\w+[ \t]*\]\]|\w+)[ \t]*\([^\r\n\)]*\)[ \t]*;|^[ \t]*#[ \t]*(?:precache|using_animtree)[ \t]*\(`),
),
},
".gsh": &Heuristics{
rule.Or(
rule.MatchingLanguages("GSC"),
regexp.MustCompile(`(?m)^\s*#\s*(?:using|insert|include|define|namespace)[ \t]+\w|^\s*(?>(?:autoexec|private)\s+){0,2}function\s+(?>(?:autoexec|private)\s+){0,2}\w+\s*\(|\b(?:level|self)[ \t]+thread[ \t]+(?:\[\[[ \t]*(?>\w+\.)*\w+[ \t]*\]\]|\w+)[ \t]*\([^\r\n\)]*\)[ \t]*;|^[ \t]*#[ \t]*(?:precache|using_animtree)[ \t]*\(`),
),
},
".gsc": &Heuristics{},
".gsh": &Heuristics{},
".h": &Heuristics{
rule.Or(
rule.MatchingLanguages("Objective-C"),