From 319e630aaf64ca2fc69e9c06ebf1adf170f4f609 Mon Sep 17 00:00:00 2001
From: Alex Bezzubov <bzz@apache.org>
Date: Sun, 25 Dec 2022 11:58:23 +0100
Subject: [PATCH] code-gen: syntax-aware generation of vendor regex

---
 internal/code-generator/assets/vendor.go.tmpl   |  8 ++++++--
 internal/code-generator/generator/generator.go  |  1 +
 internal/code-generator/generator/heuristics.go |  2 +-
 internal/code-generator/generator/vendor.go     | 10 ++++++++++
 4 files changed, 18 insertions(+), 3 deletions(-)

diff --git a/internal/code-generator/assets/vendor.go.tmpl b/internal/code-generator/assets/vendor.go.tmpl
index 4b5274e..a5d0c9f 100644
--- a/internal/code-generator/assets/vendor.go.tmpl
+++ b/internal/code-generator/assets/vendor.go.tmpl
@@ -3,8 +3,12 @@ package data
 import "github.com/go-enry/go-enry/v2/regex"
 
 var VendorMatchers = []regex.EnryRegexp{
-	{{range $regexp := . -}}
-	regex.MustCompile(`{{ $regexp }}`),
+	{{range $re := . -}}
+	{{ if isRE2 $re  -}}
+		regex.MustCompile({{ $re | stringVal }}),
+	{{- else -}}
+		regex.MustCompileRuby({{ $re | stringVal }}),
+	{{ end }}
 	{{end -}}
 }
 
diff --git a/internal/code-generator/generator/generator.go b/internal/code-generator/generator/generator.go
index 3733f5c..ce0bdd4 100644
--- a/internal/code-generator/generator/generator.go
+++ b/internal/code-generator/generator/generator.go
@@ -45,6 +45,7 @@ func executeTemplate(w io.Writer, name, path, commit string, fmap template.FuncM
 	}
 	fmap["getCommit"] = getCommit
 	fmap["stringVal"] = stringVal
+	fmap["isRE2"] = isRE2
 
 	const headerTmpl = "header.go.tmpl"
 	headerPath := filepath.Join(filepath.Dir(path), headerTmpl)
diff --git a/internal/code-generator/generator/heuristics.go b/internal/code-generator/generator/heuristics.go
index ac89b97..65df2a1 100644
--- a/internal/code-generator/generator/heuristics.go
+++ b/internal/code-generator/generator/heuristics.go
@@ -90,7 +90,7 @@ func loadRule(namedPatterns map[string]StringArray, rule *Rule) *LanguagePattern
 	}
 
 	if !isRE2(result.Pattern) {
-		log.Printf("RE2 incompatible rule: language:'%s', rule:'%s'\n", rule.Languages, result.Pattern)
+		log.Printf("RE2 incompatible syntax for heuristic language:'%s', rule:'%s'\n", rule.Languages, result.Pattern)
 	}
 	return result
 }
diff --git a/internal/code-generator/generator/vendor.go b/internal/code-generator/generator/vendor.go
index e239d10..cc3facc 100644
--- a/internal/code-generator/generator/vendor.go
+++ b/internal/code-generator/generator/vendor.go
@@ -5,6 +5,7 @@ import (
 	"fmt"
 	"io"
 	"io/ioutil"
+	"log"
 	"sort"
 	"strings"
 	"text/template"
@@ -25,6 +26,12 @@ func Vendor(fileToParse, samplesDir, outPath, tmplPath, tmplName, commit string)
 		return fmt.Errorf("failed to parse YAML %s, %q", fileToParse, err)
 	}
 
+	for _, re := range regexps {
+		if !isRE2(re) {
+			log.Printf("RE2 incompatible syntax for vendor:'%s'\n", re)
+		}
+	}
+
 	buf := &bytes.Buffer{}
 	if err := executeVendorTemplate(buf, regexps, tmplPath, tmplName, commit); err != nil {
 		return err
@@ -84,6 +91,9 @@ func collateAllMatchers(regexps []string) string {
 	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) {