godoc: update struct field anchor code

Now without regexps and allocations.

And also match comments like:

    // Foo, if non-nil, ...

The comma confused the old pattern.

Updates golang/go#16753

Change-Id: I9016ee7b5933ea343950a39989952804c74a598b
Reviewed-on: https://go-review.googlesource.com/33755
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Chris Broadfoot <cbro@golang.org>

godoc/godoc.go

@@ -233,17 +233,48 @@ func addStructFieldIDAttributes(buf *bytes.Buffer, name string, st *ast.StructTy
 	if st.Fields == nil {
 		return
 	}
-
-	v := buf.Bytes()
-	buf.Reset()
-
+	var scratch bytes.Buffer
 	for _, f := range st.Fields.List {
 		if len(f.Names) == 0 {
 			continue
 		}
 		fieldName := f.Names[0].Name
-		commentStart := []byte("// " + fieldName + " ")
-		if bytes.Contains(v, commentStart) {
+		scratch.Reset()
+		var added bool
+		foreachLine(buf.Bytes(), func(line []byte) {
+			if !added && isLineForStructFieldID(line, fieldName) {
+				added = true
+				fmt.Fprintf(&scratch, `<span id="%s.%s"></span>`, name, fieldName)
+			}
+			scratch.Write(line)
+		})
+		buf.Reset()
+		buf.Write(scratch.Bytes())
+	}
+}
+
+// foreachLine calls fn for each line of in, where a line includes
+// the trailing "\n", except on the last line, if it doesn't exist.
+func foreachLine(in []byte, fn func(line []byte)) {
+	for len(in) > 0 {
+		nl := bytes.IndexByte(in, '\n')
+		if nl == -1 {
+			fn(in)
+			return
+		}
+		fn(in[:nl+1])
+		in = in[nl+1:]
+	}
+}
+
+// commentPrefix is the line prefix for comments after they've been HTMLified.
+var commentPrefix = []byte(`<span class="comment">// `)
+
+// isLineForStructFieldID reports whether line is a line we should
+// add a <span id="#StructName.FieldName"> to. Only the fieldName is provided.
+func isLineForStructFieldID(line []byte, fieldName string) bool {
+	line = bytes.TrimSpace(line)
+
 	// For fields with a doc string of the
 	// conventional form, we put the new span into
 	// the comment instead of the field.
@@ -261,21 +292,35 @@ func addStructFieldIDAttributes(buf *bytes.Buffer, name string, st *ast.StructTy
 	//
 	// TODO: do this better, so it works for all
 	// comments, including unconventional ones.
-			v = bytes.Replace(v, commentStart, []byte(`<span id="`+name+"."+fieldName+`">// `+fieldName+" </span>"), 1)
-		} else {
-			rx := regexp.MustCompile(`(?m)^\s*` + fieldName + `\b`)
-			var matched bool
-			v = rx.ReplaceAllFunc(v, func(sub []byte) []byte {
-				if matched {
-					return sub
+	// For comments
+	if bytes.HasPrefix(line, commentPrefix) {
+		if matchesIdentBoundary(line[len(commentPrefix):], fieldName) {
+			return true
 		}
-				matched = true
-				return []byte(`<span id="` + name + "." + fieldName + `">` + string(sub) + "</span>")
-			})
 	}
+	return matchesIdentBoundary(line, fieldName)
 }
 
-	buf.Write(v)
+// matchesIdentBoundary reports whether line matches /^ident\b/.
+// A boundary is considered either none, or an ASCII non-alphanum.
+func matchesIdentBoundary(line []byte, ident string) bool {
+	if len(line) < len(ident) {
+		return false
+	}
+	if string(line[:len(ident)]) != ident {
+		return false
+	}
+	rest := line[len(ident):]
+	return len(rest) == 0 || !isASCIIWordChar(rest[0])
+}
+
+// isASCIIWordChar reports whether b is an ASCII "word"
+// character. (Matching /\w/ in ASCII mode)
+func isASCIIWordChar(b byte) bool {
+	return 'a' <= b && b <= 'z' ||
+		'A' <= b && b <= 'Z' ||
+		'0' <= b && b <= '0' ||
+		b == '_'
 }
 
 func comment_htmlFunc(comment string) string {

godoc/godoc_test.go

@@ -134,6 +134,9 @@ type T struct {
 
 	// Doc has a comment.
 	Doc string
+
+	// Opt, if non-nil, is an option.
+	Opt *int
 }
 `)
 	fset := token.NewFileSet()
@@ -147,12 +150,15 @@ type T struct {
 	}
 	got := p.node_htmlFunc(pi, genDecl, true)
 	want := `type T struct {
-<span id="T.NoDoc">NoDoc</span> <a href="/pkg/builtin/#string">string</a>
+<span id="T.NoDoc"></span>NoDoc <a href="/pkg/builtin/#string">string</a>
 
-<span class="comment"><span id="T.Doc">// Doc </span>has a comment.</span>
+<span id="T.Doc"></span><span class="comment">// Doc has a comment.</span>
 Doc <a href="/pkg/builtin/#string">string</a>
+
+<span id="T.Opt"></span><span class="comment">// Opt, if non-nil, is an option.</span>
+Opt *<a href="/pkg/builtin/#int">int</a>
 }`
 	if got != want {
-		t.Errorf(" got: %q\nwant: %q\n", got, want)
+		t.Errorf("got: %s\n\nwant: %s\n", got, want)
 	}
 }