Make header detection regex more lenient on punctuation and ws

Author: Florent Biville

internal/pkg/core/header_detector.go

@@ -60,12 +60,47 @@ func OpeningLine(styles []CommentStyle) string {
 
 // visible for testing
 func MatchingLine(line string, styles []CommentStyle) string {
-	middleLine := fmt.Sprintf(`[\t\v\f\r ]*%s?[\t\v\f\r ]*\Q%s\E[,.;:?!\t\v\f\r ]*\n?`, combineRegexes(styles, func(style CommentStyle) string { return style.GetString() }), line)
+	openingStyleSymbolRegex := combineRegexes(styles, func(style CommentStyle) string { return style.GetString() })
+	normalizedLine := normalizePunctuation(line)
+	middleLine := fmt.Sprintf(`[\t\v\f\r ]*%s?[\t\v\f\r ]*\Q%s\E[,.;:?!\t\v\f\r ]*\n?`, openingStyleSymbolRegex, normalizedLine)
 	builder := strings.Builder{}
 	builder.WriteString(middleLine)
 	return builder.String()
 }
 
+func normalizePunctuation(line string) string {
+	ignore := `\E.?\Q`
+	normalizedLine := ""
+	// we could use a(n only) slightly better heuristic with a regex matching all dots
+	// not prefixed by a digit or a couple of { and 0-n spaces
+	// but no support for negative lookbehind in Golang regex engine so here we go :(
+	for k, v := range line {
+		if v == ',' || v == ';' || v == ':' || v == '?' || v == '!' {
+			normalizedLine += ignore
+		} else if v == '.' && (k == len(line)-1 || line[k:k+2] == ". ") {
+			// dots in template expressions must and in numerical expressions should be preserved
+			// hence the poor heuristic of the following space or dot as last line's character
+			normalizedLine += ignore
+		} else {
+			normalizedLine += string(v)
+		}
+	}
+	result := strings.NewReplacer(
+		`\t\t`, `\t`,
+		`\v\v`, `\v`,
+		`\f\f`, `\f`,
+		`\r\r`, `\r`,
+		"  ", " ",
+	).Replace(normalizedLine)
+	return strings.NewReplacer(
+		"\t", `\E\t+\Q`,
+		"\v", `\E\v+\Q`,
+		"\f", `\E\f+\Q`,
+		"\r", `\E\r+\Q`,
+		" ", `\E +\Q`,
+	).Replace(result)
+}
+
 // visible for testing
 func ClosingLine(styles []CommentStyle) string {
 	closingLine := fmt.Sprintf(`(?:[\t\v\f\r ]*%s[\t\v\f\r ]*)?`, combineRegexes(styles, func(style CommentStyle) string { return style.GetClosingString() }))

internal/pkg/core/header_detector_test.go

@@ -122,7 +122,7 @@ world`
 		})
 	})
 
-	Context("with punctuation and whitespace variations", func() {
+	Context("with punctuation and whitespace variations in the source file", func() {
 
 
 		const file = `// some multi-line header.!:
@@ -140,6 +140,24 @@ world`
 		})
 	})
 
+	Context("with punctuation and whitespace variations in the license header template", func() {
+
+
+		const file = `// some multi-line header
+// with some text
+hello
+world`
+
+		It("should detect it", func() {
+			regex, err := core.ComputeHeaderDetectionRegex(
+				[]string{"some. multi-line.  header!", "with  some text?"},
+				map[string]string{})
+
+			Expect(err).NotTo(HaveOccurred())
+			Expect(MatchLeftMostPositions(regex, file)).To(Equal([]int{0, 44}))
+		})
+	})
+
 	Context("with a realistic header", func() {
 
 		const template = `Copyright {{.YearRange}} {{.Owner}}