code-gen: make content heuristics regexp engine configurable & generation syntax-aware

regex/oniguruma.go

@@ -1,3 +1,4 @@
+//go:build oniguruma
 // +build oniguruma
 
 package regex
@@ -8,8 +9,17 @@ import (
 
 type EnryRegexp = *rubex.Regexp
 
-func MustCompile(str string) EnryRegexp {
-	return rubex.MustCompileASCII(str)
+func MustCompile(s string) EnryRegexp {
+	return rubex.MustCompileASCII(s)
+}
+
+// MustCompileMultiline matches in multi-line mode by default with Oniguruma.
+func MustCompileMultiline(s string) EnryRegexp {
+	return MustCompile(s)
+}
+
+func MustCompileRuby(s string) EnryRegexp {
+	return MustCompile(s)
 }
 
 func QuoteMeta(s string) string {

regex/standard.go

@@ -1,3 +1,4 @@
+//go:build !oniguruma
 // +build !oniguruma
 
 package regex
@@ -12,6 +13,20 @@ func MustCompile(str string) EnryRegexp {
 	return regexp.MustCompile(str)
 }
 
+// MustCompileMultiline mimics Ruby defaults for regexp, where ^$ matches begin/end of line.
+// I.e. it converts Ruby regexp syntaxt to RE2 equivalent
+func MustCompileMultiline(s string) EnryRegexp {
+	const multilineModeFlag = "(?m)"
+	return regexp.MustCompile(multilineModeFlag + s)
+}
+
+// MustCompileRuby used for expressions with syntax not supported by RE2.
+func MustCompileRuby(s string) EnryRegexp {
+	// TODO(bzz): find a bettee way?
+	// This will only trigger a panic on .Match() for the clients
+	return nil
+}
+
 func QuoteMeta(s string) string {
 	return regexp.QuoteMeta(s)
 }