clair/vendor/golang.org/x/net/html/parse_test.go
2016-06-07 10:08:50 +02:00

389 lines
9.9 KiB
Go

// Copyright 2010 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package html
import (
"bufio"
"bytes"
"errors"
"fmt"
"io"
"io/ioutil"
"os"
"path/filepath"
"runtime"
"sort"
"strings"
"testing"
"golang.org/x/net/html/atom"
)
// readParseTest reads a single test case from r.
func readParseTest(r *bufio.Reader) (text, want, context string, err error) {
line, err := r.ReadSlice('\n')
if err != nil {
return "", "", "", err
}
var b []byte
// Read the HTML.
if string(line) != "#data\n" {
return "", "", "", fmt.Errorf(`got %q want "#data\n"`, line)
}
for {
line, err = r.ReadSlice('\n')
if err != nil {
return "", "", "", err
}
if line[0] == '#' {
break
}
b = append(b, line...)
}
text = strings.TrimSuffix(string(b), "\n")
b = b[:0]
// Skip the error list.
if string(line) != "#errors\n" {
return "", "", "", fmt.Errorf(`got %q want "#errors\n"`, line)
}
for {
line, err = r.ReadSlice('\n')
if err != nil {
return "", "", "", err
}
if line[0] == '#' {
break
}
}
if string(line) == "#document-fragment\n" {
line, err = r.ReadSlice('\n')
if err != nil {
return "", "", "", err
}
context = strings.TrimSpace(string(line))
line, err = r.ReadSlice('\n')
if err != nil {
return "", "", "", err
}
}
// Read the dump of what the parse tree should be.
if string(line) != "#document\n" {
return "", "", "", fmt.Errorf(`got %q want "#document\n"`, line)
}
inQuote := false
for {
line, err = r.ReadSlice('\n')
if err != nil && err != io.EOF {
return "", "", "", err
}
trimmed := bytes.Trim(line, "| \n")
if len(trimmed) > 0 {
if line[0] == '|' && trimmed[0] == '"' {
inQuote = true
}
if trimmed[len(trimmed)-1] == '"' && !(line[0] == '|' && len(trimmed) == 1) {
inQuote = false
}
}
if len(line) == 0 || len(line) == 1 && line[0] == '\n' && !inQuote {
break
}
b = append(b, line...)
}
return text, string(b), context, nil
}
func dumpIndent(w io.Writer, level int) {
io.WriteString(w, "| ")
for i := 0; i < level; i++ {
io.WriteString(w, " ")
}
}
type sortedAttributes []Attribute
func (a sortedAttributes) Len() int {
return len(a)
}
func (a sortedAttributes) Less(i, j int) bool {
if a[i].Namespace != a[j].Namespace {
return a[i].Namespace < a[j].Namespace
}
return a[i].Key < a[j].Key
}
func (a sortedAttributes) Swap(i, j int) {
a[i], a[j] = a[j], a[i]
}
func dumpLevel(w io.Writer, n *Node, level int) error {
dumpIndent(w, level)
switch n.Type {
case ErrorNode:
return errors.New("unexpected ErrorNode")
case DocumentNode:
return errors.New("unexpected DocumentNode")
case ElementNode:
if n.Namespace != "" {
fmt.Fprintf(w, "<%s %s>", n.Namespace, n.Data)
} else {
fmt.Fprintf(w, "<%s>", n.Data)
}
attr := sortedAttributes(n.Attr)
sort.Sort(attr)
for _, a := range attr {
io.WriteString(w, "\n")
dumpIndent(w, level+1)
if a.Namespace != "" {
fmt.Fprintf(w, `%s %s="%s"`, a.Namespace, a.Key, a.Val)
} else {
fmt.Fprintf(w, `%s="%s"`, a.Key, a.Val)
}
}
case TextNode:
fmt.Fprintf(w, `"%s"`, n.Data)
case CommentNode:
fmt.Fprintf(w, "<!-- %s -->", n.Data)
case DoctypeNode:
fmt.Fprintf(w, "<!DOCTYPE %s", n.Data)
if n.Attr != nil {
var p, s string
for _, a := range n.Attr {
switch a.Key {
case "public":
p = a.Val
case "system":
s = a.Val
}
}
if p != "" || s != "" {
fmt.Fprintf(w, ` "%s"`, p)
fmt.Fprintf(w, ` "%s"`, s)
}
}
io.WriteString(w, ">")
case scopeMarkerNode:
return errors.New("unexpected scopeMarkerNode")
default:
return errors.New("unknown node type")
}
io.WriteString(w, "\n")
for c := n.FirstChild; c != nil; c = c.NextSibling {
if err := dumpLevel(w, c, level+1); err != nil {
return err
}
}
return nil
}
func dump(n *Node) (string, error) {
if n == nil || n.FirstChild == nil {
return "", nil
}
var b bytes.Buffer
for c := n.FirstChild; c != nil; c = c.NextSibling {
if err := dumpLevel(&b, c, 0); err != nil {
return "", err
}
}
return b.String(), nil
}
const testDataDir = "testdata/webkit/"
func TestParser(t *testing.T) {
testFiles, err := filepath.Glob(testDataDir + "*.dat")
if err != nil {
t.Fatal(err)
}
for _, tf := range testFiles {
f, err := os.Open(tf)
if err != nil {
t.Fatal(err)
}
defer f.Close()
r := bufio.NewReader(f)
for i := 0; ; i++ {
text, want, context, err := readParseTest(r)
if err == io.EOF {
break
}
if err != nil {
t.Fatal(err)
}
err = testParseCase(text, want, context)
if err != nil {
t.Errorf("%s test #%d %q, %s", tf, i, text, err)
}
}
}
}
// testParseCase tests one test case from the test files. If the test does not
// pass, it returns an error that explains the failure.
// text is the HTML to be parsed, want is a dump of the correct parse tree,
// and context is the name of the context node, if any.
func testParseCase(text, want, context string) (err error) {
defer func() {
if x := recover(); x != nil {
switch e := x.(type) {
case error:
err = e
default:
err = fmt.Errorf("%v", e)
}
}
}()
var doc *Node
if context == "" {
doc, err = Parse(strings.NewReader(text))
if err != nil {
return err
}
} else {
contextNode := &Node{
Type: ElementNode,
DataAtom: atom.Lookup([]byte(context)),
Data: context,
}
nodes, err := ParseFragment(strings.NewReader(text), contextNode)
if err != nil {
return err
}
doc = &Node{
Type: DocumentNode,
}
for _, n := range nodes {
doc.AppendChild(n)
}
}
if err := checkTreeConsistency(doc); err != nil {
return err
}
got, err := dump(doc)
if err != nil {
return err
}
// Compare the parsed tree to the #document section.
if got != want {
return fmt.Errorf("got vs want:\n----\n%s----\n%s----", got, want)
}
if renderTestBlacklist[text] || context != "" {
return nil
}
// Check that rendering and re-parsing results in an identical tree.
pr, pw := io.Pipe()
go func() {
pw.CloseWithError(Render(pw, doc))
}()
doc1, err := Parse(pr)
if err != nil {
return err
}
got1, err := dump(doc1)
if err != nil {
return err
}
if got != got1 {
return fmt.Errorf("got vs got1:\n----\n%s----\n%s----", got, got1)
}
return nil
}
// Some test input result in parse trees are not 'well-formed' despite
// following the HTML5 recovery algorithms. Rendering and re-parsing such a
// tree will not result in an exact clone of that tree. We blacklist such
// inputs from the render test.
var renderTestBlacklist = map[string]bool{
// The second <a> will be reparented to the first <table>'s parent. This
// results in an <a> whose parent is an <a>, which is not 'well-formed'.
`<a><table><td><a><table></table><a></tr><a></table><b>X</b>C<a>Y`: true,
// The same thing with a <p>:
`<p><table></p>`: true,
// More cases of <a> being reparented:
`<a href="blah">aba<table><a href="foo">br<tr><td></td></tr>x</table>aoe`: true,
`<a><table><a></table><p><a><div><a>`: true,
`<a><table><td><a><table></table><a></tr><a></table><a>`: true,
// A similar reparenting situation involving <nobr>:
`<!DOCTYPE html><body><b><nobr>1<table><nobr></b><i><nobr>2<nobr></i>3`: true,
// A <plaintext> element is reparented, putting it before a table.
// A <plaintext> element can't have anything after it in HTML.
`<table><plaintext><td>`: true,
`<!doctype html><table><plaintext></plaintext>`: true,
`<!doctype html><table><tbody><plaintext></plaintext>`: true,
`<!doctype html><table><tbody><tr><plaintext></plaintext>`: true,
// A form inside a table inside a form doesn't work either.
`<!doctype html><form><table></form><form></table></form>`: true,
// A script that ends at EOF may escape its own closing tag when rendered.
`<!doctype html><script><!--<script `: true,
`<!doctype html><script><!--<script <`: true,
`<!doctype html><script><!--<script <a`: true,
`<!doctype html><script><!--<script </`: true,
`<!doctype html><script><!--<script </s`: true,
`<!doctype html><script><!--<script </script`: true,
`<!doctype html><script><!--<script </scripta`: true,
`<!doctype html><script><!--<script -`: true,
`<!doctype html><script><!--<script -a`: true,
`<!doctype html><script><!--<script -<`: true,
`<!doctype html><script><!--<script --`: true,
`<!doctype html><script><!--<script --a`: true,
`<!doctype html><script><!--<script --<`: true,
`<script><!--<script `: true,
`<script><!--<script <a`: true,
`<script><!--<script </script`: true,
`<script><!--<script </scripta`: true,
`<script><!--<script -`: true,
`<script><!--<script -a`: true,
`<script><!--<script --`: true,
`<script><!--<script --a`: true,
`<script><!--<script <`: true,
`<script><!--<script </`: true,
`<script><!--<script </s`: true,
// Reconstructing the active formatting elements results in a <plaintext>
// element that contains an <a> element.
`<!doctype html><p><a><plaintext>b`: true,
}
func TestNodeConsistency(t *testing.T) {
// inconsistentNode is a Node whose DataAtom and Data do not agree.
inconsistentNode := &Node{
Type: ElementNode,
DataAtom: atom.Frameset,
Data: "table",
}
_, err := ParseFragment(strings.NewReader("<p>hello</p>"), inconsistentNode)
if err == nil {
t.Errorf("got nil error, want non-nil")
}
}
func BenchmarkParser(b *testing.B) {
buf, err := ioutil.ReadFile("testdata/go1.html")
if err != nil {
b.Fatalf("could not read testdata/go1.html: %v", err)
}
b.SetBytes(int64(len(buf)))
runtime.GC()
b.ReportAllocs()
b.ResetTimer()
for i := 0; i < b.N; i++ {
Parse(bytes.NewBuffer(buf))
}
}