mirror of
https://github.com/gohugoio/hugo.git
synced 2024-11-14 20:37:55 -05:00
publisher: Exclude comment and doctype elements from writeStats
- Reorder code blocks - Rename cssClassCollectorWriter to htmlElementCollectorWriter, as it just collect html element information - Expand benchmark to test for minified and unminified content Fixes #8396 Fixes #8417
This commit is contained in:
parent
2bb9496ce2
commit
bc80022e03
3 changed files with 328 additions and 191 deletions
|
@ -1113,7 +1113,7 @@ ABC.
|
|||
els := stats.HTMLElements
|
||||
|
||||
b.Assert(els.Classes, qt.HasLen, 3606) // (4 * 900) + 4 +2
|
||||
b.Assert(els.Tags, qt.HasLen, 9)
|
||||
b.Assert(els.Tags, qt.HasLen, 8)
|
||||
b.Assert(els.IDs, qt.HasLen, 1)
|
||||
}
|
||||
}
|
||||
|
|
|
@ -20,22 +20,11 @@ import (
|
|||
"strings"
|
||||
"sync"
|
||||
|
||||
"github.com/gohugoio/hugo/helpers"
|
||||
"golang.org/x/net/html"
|
||||
|
||||
"github.com/gohugoio/hugo/helpers"
|
||||
)
|
||||
|
||||
func newHTMLElementsCollector() *htmlElementsCollector {
|
||||
return &htmlElementsCollector{
|
||||
elementSet: make(map[string]bool),
|
||||
}
|
||||
}
|
||||
|
||||
func newHTMLElementsCollectorWriter(collector *htmlElementsCollector) *cssClassCollectorWriter {
|
||||
return &cssClassCollectorWriter{
|
||||
collector: collector,
|
||||
}
|
||||
}
|
||||
|
||||
// HTMLElements holds lists of tags and attribute values for classes and id.
|
||||
type HTMLElements struct {
|
||||
Tags []string `json:"tags"`
|
||||
|
@ -59,152 +48,6 @@ func (h *HTMLElements) Sort() {
|
|||
sort.Strings(h.IDs)
|
||||
}
|
||||
|
||||
type cssClassCollectorWriter struct {
|
||||
collector *htmlElementsCollector
|
||||
buff bytes.Buffer
|
||||
|
||||
isCollecting bool
|
||||
inPreTag string
|
||||
|
||||
inQuote bool
|
||||
quoteValue byte
|
||||
}
|
||||
|
||||
func (w *cssClassCollectorWriter) Write(p []byte) (n int, err error) {
|
||||
n = len(p)
|
||||
i := 0
|
||||
|
||||
for i < len(p) {
|
||||
if !w.isCollecting {
|
||||
for ; i < len(p); i++ {
|
||||
b := p[i]
|
||||
if b == '<' {
|
||||
w.startCollecting()
|
||||
break
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
if w.isCollecting {
|
||||
for ; i < len(p); i++ {
|
||||
b := p[i]
|
||||
w.toggleIfQuote(b)
|
||||
if !w.inQuote && b == '>' {
|
||||
w.endCollecting()
|
||||
break
|
||||
}
|
||||
w.buff.WriteByte(b)
|
||||
}
|
||||
|
||||
if !w.isCollecting {
|
||||
if w.inPreTag != "" {
|
||||
s := w.buff.String()
|
||||
if tagName, isEnd := w.parseEndTag(s); isEnd && w.inPreTag == tagName {
|
||||
w.inPreTag = ""
|
||||
}
|
||||
w.buff.Reset()
|
||||
continue
|
||||
}
|
||||
|
||||
// First check if we have processed this element before.
|
||||
w.collector.mu.RLock()
|
||||
|
||||
// See https://github.com/dominikh/go-tools/issues/723
|
||||
//lint:ignore S1030 This construct avoids memory allocation for the string.
|
||||
seen := w.collector.elementSet[string(w.buff.Bytes())]
|
||||
w.collector.mu.RUnlock()
|
||||
if seen {
|
||||
w.buff.Reset()
|
||||
continue
|
||||
}
|
||||
|
||||
s := w.buff.String()
|
||||
|
||||
w.buff.Reset()
|
||||
|
||||
if strings.HasPrefix(s, "</") {
|
||||
continue
|
||||
}
|
||||
|
||||
key := s
|
||||
|
||||
s, tagName := w.insertStandinHTMLElement(s)
|
||||
el := parseHTMLElement(s)
|
||||
el.Tag = tagName
|
||||
if w.isPreFormatted(tagName) {
|
||||
w.inPreTag = tagName
|
||||
}
|
||||
|
||||
w.collector.mu.Lock()
|
||||
w.collector.elementSet[key] = true
|
||||
if el.Tag != "" {
|
||||
w.collector.elements = append(w.collector.elements, el)
|
||||
}
|
||||
w.collector.mu.Unlock()
|
||||
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
return
|
||||
}
|
||||
|
||||
// No need to look inside these for HTML elements.
|
||||
func (c *cssClassCollectorWriter) isPreFormatted(s string) bool {
|
||||
return s == "pre" || s == "textarea" || s == "script"
|
||||
}
|
||||
|
||||
// The net/html parser does not handle single table elements as input, e.g. tbody.
|
||||
// We only care about the element/class/ids, so just store away the original tag name
|
||||
// and pretend it's a <div>.
|
||||
func (c *cssClassCollectorWriter) insertStandinHTMLElement(el string) (string, string) {
|
||||
tag := el[1:]
|
||||
spacei := strings.Index(tag, " ")
|
||||
if spacei != -1 {
|
||||
tag = tag[:spacei]
|
||||
}
|
||||
tag = strings.Trim(tag, "\n ")
|
||||
newv := strings.Replace(el, tag, "div", 1)
|
||||
return newv, strings.ToLower(tag)
|
||||
}
|
||||
|
||||
func (c *cssClassCollectorWriter) parseEndTag(s string) (string, bool) {
|
||||
if !strings.HasPrefix(s, "</") {
|
||||
return "", false
|
||||
}
|
||||
s = strings.TrimPrefix(s, "</")
|
||||
s = strings.TrimSuffix(s, ">")
|
||||
return strings.ToLower(strings.TrimSpace(s)), true
|
||||
}
|
||||
|
||||
func (c *cssClassCollectorWriter) endCollecting() {
|
||||
c.isCollecting = false
|
||||
c.inQuote = false
|
||||
|
||||
}
|
||||
|
||||
func (c *cssClassCollectorWriter) startCollecting() {
|
||||
c.isCollecting = true
|
||||
|
||||
}
|
||||
|
||||
func (c *cssClassCollectorWriter) toggleIfQuote(b byte) {
|
||||
if isQuote(b) {
|
||||
if c.inQuote && b == c.quoteValue {
|
||||
c.inQuote = false
|
||||
} else if !c.inQuote {
|
||||
c.inQuote = true
|
||||
c.quoteValue = b
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
type htmlElement struct {
|
||||
Tag string
|
||||
Classes []string
|
||||
IDs []string
|
||||
}
|
||||
|
||||
type htmlElementsCollector struct {
|
||||
// Contains the raw HTML string. We will get the same element
|
||||
// several times, and want to avoid costly reparsing when this
|
||||
|
@ -216,6 +59,12 @@ type htmlElementsCollector struct {
|
|||
mu sync.RWMutex
|
||||
}
|
||||
|
||||
func newHTMLElementsCollector() *htmlElementsCollector {
|
||||
return &htmlElementsCollector{
|
||||
elementSet: make(map[string]bool),
|
||||
}
|
||||
}
|
||||
|
||||
func (c *htmlElementsCollector) getHTMLElements() HTMLElements {
|
||||
var (
|
||||
classes []string
|
||||
|
@ -242,21 +91,205 @@ func (c *htmlElementsCollector) getHTMLElements() HTMLElements {
|
|||
return els
|
||||
}
|
||||
|
||||
type htmlElementsCollectorWriter struct {
|
||||
collector *htmlElementsCollector
|
||||
buff bytes.Buffer
|
||||
|
||||
isCollecting bool
|
||||
inPreTag string
|
||||
|
||||
inQuote bool
|
||||
quoteValue byte
|
||||
}
|
||||
|
||||
func newHTMLElementsCollectorWriter(collector *htmlElementsCollector) *htmlElementsCollectorWriter {
|
||||
return &htmlElementsCollectorWriter{
|
||||
collector: collector,
|
||||
}
|
||||
}
|
||||
|
||||
// Write splits the incoming stream into single html element and writes these into elementSet
|
||||
func (w *htmlElementsCollectorWriter) Write(p []byte) (n int, err error) {
|
||||
n = len(p)
|
||||
i := 0
|
||||
|
||||
for i < len(p) {
|
||||
// if is not collecting, cycle through byte stream until start bracket "<" is found
|
||||
if !w.isCollecting {
|
||||
for ; i < len(p); i++ {
|
||||
b := p[i]
|
||||
if b == '<' {
|
||||
w.startCollecting()
|
||||
break
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
if w.isCollecting {
|
||||
// if is collecting, cycle through byte stream until end bracket ">" is found
|
||||
// disregard any ">" if within a quote
|
||||
// write bytes until found to buffer
|
||||
for ; i < len(p); i++ {
|
||||
b := p[i]
|
||||
w.toggleIfQuote(b)
|
||||
w.buff.WriteByte(b)
|
||||
|
||||
if !w.inQuote && b == '>' {
|
||||
w.endCollecting()
|
||||
break
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// if no end bracket ">" is found while collecting, but the stream ended
|
||||
// this could mean we received chunks of a stream from e.g. the minify functionality
|
||||
// next if loop will be skipped
|
||||
|
||||
// at this point we have collected an element line between angle brackets "<" and ">"
|
||||
if !w.isCollecting {
|
||||
s := w.buff.String()
|
||||
w.buff.Reset()
|
||||
|
||||
// filter out unwanted tags
|
||||
// empty string, just in case
|
||||
// if within preformatted code blocks <pre>, <textarea>, <script>, <style>
|
||||
// comments and doctype tags
|
||||
// end tags
|
||||
switch {
|
||||
case s == "": // empty string
|
||||
continue
|
||||
case w.inPreTag != "": // within preformatted code block
|
||||
if tagName, isEnd := parseEndTag(s); isEnd && w.inPreTag == tagName {
|
||||
w.inPreTag = ""
|
||||
}
|
||||
continue
|
||||
case strings.HasPrefix(s, "<!"): // comment or doctype tag
|
||||
continue
|
||||
case strings.HasPrefix(s, "</"): // end tag
|
||||
continue
|
||||
}
|
||||
|
||||
// check if we have processed this element before.
|
||||
w.collector.mu.RLock()
|
||||
seen := w.collector.elementSet[s]
|
||||
w.collector.mu.RUnlock()
|
||||
if seen {
|
||||
continue
|
||||
}
|
||||
|
||||
// check if a preformatted code block started
|
||||
if tagName, isStart := parseStartTag(s); isStart && isPreFormatted(tagName) {
|
||||
w.inPreTag = tagName
|
||||
}
|
||||
|
||||
// parse each collected element
|
||||
el, err := parseHTMLElement(s)
|
||||
if err != nil {
|
||||
return n, err
|
||||
}
|
||||
|
||||
// write this tag to the element set
|
||||
w.collector.mu.Lock()
|
||||
w.collector.elementSet[s] = true
|
||||
w.collector.elements = append(w.collector.elements, el)
|
||||
w.collector.mu.Unlock()
|
||||
}
|
||||
}
|
||||
|
||||
return
|
||||
}
|
||||
|
||||
func (c *htmlElementsCollectorWriter) startCollecting() {
|
||||
c.isCollecting = true
|
||||
}
|
||||
|
||||
func (c *htmlElementsCollectorWriter) endCollecting() {
|
||||
c.isCollecting = false
|
||||
c.inQuote = false
|
||||
}
|
||||
|
||||
func (c *htmlElementsCollectorWriter) toggleIfQuote(b byte) {
|
||||
if isQuote(b) {
|
||||
if c.inQuote && b == c.quoteValue {
|
||||
c.inQuote = false
|
||||
} else if !c.inQuote {
|
||||
c.inQuote = true
|
||||
c.quoteValue = b
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
func isQuote(b byte) bool {
|
||||
return b == '"' || b == '\''
|
||||
}
|
||||
|
||||
func parseStartTag(s string) (string, bool) {
|
||||
if strings.HasPrefix(s, "</") || strings.HasPrefix(s, "<!") {
|
||||
return "", false
|
||||
}
|
||||
|
||||
s = strings.TrimPrefix(s, "<")
|
||||
s = strings.TrimSuffix(s, ">")
|
||||
|
||||
spaceIndex := strings.Index(s, " ")
|
||||
if spaceIndex != -1 {
|
||||
s = s[:spaceIndex]
|
||||
}
|
||||
|
||||
return strings.ToLower(strings.TrimSpace(s)), true
|
||||
}
|
||||
|
||||
func parseEndTag(s string) (string, bool) {
|
||||
if !strings.HasPrefix(s, "</") {
|
||||
return "", false
|
||||
}
|
||||
|
||||
s = strings.TrimPrefix(s, "</")
|
||||
s = strings.TrimSuffix(s, ">")
|
||||
|
||||
return strings.ToLower(strings.TrimSpace(s)), true
|
||||
}
|
||||
|
||||
// No need to look inside these for HTML elements.
|
||||
func isPreFormatted(s string) bool {
|
||||
return s == "pre" || s == "textarea" || s == "script" || s == "style"
|
||||
}
|
||||
|
||||
type htmlElement struct {
|
||||
Tag string
|
||||
Classes []string
|
||||
IDs []string
|
||||
}
|
||||
|
||||
var (
|
||||
htmlJsonFixer = strings.NewReplacer(", ", "\n")
|
||||
jsonAttrRe = regexp.MustCompile(`'?(.*?)'?:.*`)
|
||||
classAttrRe = regexp.MustCompile(`(?i)^class$|transition`)
|
||||
)
|
||||
|
||||
func parseHTMLElement(elStr string) (el htmlElement) {
|
||||
elStr = strings.TrimSpace(elStr)
|
||||
if !strings.HasSuffix(elStr, ">") {
|
||||
elStr += ">"
|
||||
func parseHTMLElement(elStr string) (el htmlElement, err error) {
|
||||
var tagBuffer string = ""
|
||||
exceptionList := map[string]bool{
|
||||
"thead": true,
|
||||
"tbody": true,
|
||||
"tfoot": true,
|
||||
"td": true,
|
||||
"tr": true,
|
||||
}
|
||||
|
||||
tagName, ok := parseStartTag(elStr)
|
||||
if !ok {
|
||||
return
|
||||
}
|
||||
|
||||
// The net/html parser does not handle single table elements as input, e.g. tbody.
|
||||
// We only care about the element/class/ids, so just store away the original tag name
|
||||
// and pretend it's a <div>.
|
||||
if exceptionList[tagName] {
|
||||
tagBuffer = tagName
|
||||
elStr = strings.Replace(elStr, tagName, "div", 1)
|
||||
}
|
||||
|
||||
n, err := html.Parse(strings.NewReader(elStr))
|
||||
if err != nil {
|
||||
return
|
||||
|
@ -287,7 +320,6 @@ func parseHTMLElement(elStr string) (el htmlElement) {
|
|||
val = strings.Join(lines, "\n")
|
||||
val = jsonAttrRe.ReplaceAllString(val, "$1")
|
||||
el.Classes = append(el.Classes, strings.Fields(val)...)
|
||||
|
||||
}
|
||||
}
|
||||
}
|
||||
|
@ -301,5 +333,10 @@ func parseHTMLElement(elStr string) (el htmlElement) {
|
|||
|
||||
walk(n)
|
||||
|
||||
// did we replaced the start tag?
|
||||
if tagBuffer != "" {
|
||||
el.Tag = tagBuffer
|
||||
}
|
||||
|
||||
return
|
||||
}
|
||||
|
|
|
@ -14,17 +14,17 @@
|
|||
package publisher
|
||||
|
||||
import (
|
||||
"bytes"
|
||||
"fmt"
|
||||
"strings"
|
||||
"testing"
|
||||
|
||||
"github.com/gohugoio/hugo/minifiers"
|
||||
|
||||
"github.com/gohugoio/hugo/media"
|
||||
"github.com/gohugoio/hugo/minifiers"
|
||||
"github.com/gohugoio/hugo/output"
|
||||
"github.com/spf13/viper"
|
||||
|
||||
qt "github.com/frankban/quicktest"
|
||||
"github.com/spf13/viper"
|
||||
)
|
||||
|
||||
func TestClassCollector(t *testing.T) {
|
||||
|
@ -50,7 +50,6 @@ func TestClassCollector(t *testing.T) {
|
|||
|
||||
skipMinifyTest := map[string]bool{
|
||||
"Script tags content should be skipped": true, // https://github.com/tdewolff/minify/issues/396
|
||||
|
||||
}
|
||||
|
||||
for _, test := range []struct {
|
||||
|
@ -62,56 +61,57 @@ func TestClassCollector(t *testing.T) {
|
|||
{"duplicates", `<div class="b a b"></div>`, f("div", "a b", "")},
|
||||
{"single quote", `<body class='b a'></body>`, f("body", "a b", "")},
|
||||
{"no quote", `<body class=b id=myelement></body>`, f("body", "b", "myelement")},
|
||||
{"thead", `
|
||||
https://github.com/gohugoio/hugo/issues/7318
|
||||
<table class="cl1">
|
||||
// https://github.com/gohugoio/hugo/issues/7318
|
||||
{"thead", `<table class="cl1">
|
||||
<thead class="cl2"><tr class="cl3"><td class="cl4"></td></tr></thead>
|
||||
<tbody class="cl5"><tr class="cl6"><td class="cl7"></td></tr></tbody>
|
||||
</table>`, f("table tbody td thead tr", "cl1 cl2 cl3 cl4 cl5 cl6 cl7", "")},
|
||||
// https://github.com/gohugoio/hugo/issues/7161
|
||||
{"minified a href", `<a class="b a" href=/></a>`, f("a", "a b", "")},
|
||||
|
||||
{"AlpineJS bind 1", `<body>
|
||||
<div x-bind:class="{
|
||||
<div x-bind:class="{
|
||||
'class1': data.open,
|
||||
'class2 class3': data.foo == 'bar'
|
||||
}">
|
||||
</div>
|
||||
</body>`, f("body div", "class1 class2 class3", "")},
|
||||
|
||||
{
|
||||
"Alpine bind 2", `<div x-bind:class="{ 'bg-black': filter.checked }"
|
||||
class="inline-block mr-1 mb-2 rounded bg-gray-300 px-2 py-2">FOO</div>`,
|
||||
</div>
|
||||
</body>`, f("body div", "class1 class2 class3", "")},
|
||||
{"AlpineJS bind 2", `<div x-bind:class="{ 'bg-black': filter.checked }" class="inline-block mr-1 mb-2 rounded bg-gray-300 px-2 py-2">FOO</div>`,
|
||||
f("div", "bg-black bg-gray-300 inline-block mb-2 mr-1 px-2 py-2 rounded", ""),
|
||||
},
|
||||
|
||||
{"Alpine bind 3", `<div x-bind:class="{ 'text-gray-800': !checked, 'text-white': checked }"></div>`, f("div", "text-gray-800 text-white", "")},
|
||||
{"Alpine bind 4", `<div x-bind:class="{ 'text-gray-800': !checked,
|
||||
{"AlpineJS bind 3", `<div x-bind:class="{ 'text-gray-800': !checked, 'text-white': checked }"></div>`, f("div", "text-gray-800 text-white", "")},
|
||||
{"AlpineJS bind 4", `<div x-bind:class="{ 'text-gray-800': !checked,
|
||||
'text-white': checked }"></div>`, f("div", "text-gray-800 text-white", "")},
|
||||
|
||||
{"Alpine bind 5", `<a x-bind:class="{
|
||||
{"AlpineJS bind 5", `<a x-bind:class="{
|
||||
'text-a': a && b,
|
||||
'text-b': !a && b || c,
|
||||
'pl-3': a === 1,
|
||||
pl-2: b == 3,
|
||||
'text-gray-600': (a > 1)
|
||||
|
||||
}" class="block w-36 cursor-pointer pr-3 no-underline capitalize"></a>`, f("a", "block capitalize cursor-pointer no-underline pl-2 pl-3 pr-3 text-a text-b text-gray-600 w-36", "")},
|
||||
|
||||
{"Alpine transition 1", `<div x-transition:enter-start="opacity-0 transform mobile:-translate-x-8 sm:-translate-y-8">`, f("div", "mobile:-translate-x-8 opacity-0 sm:-translate-y-8 transform", "")},
|
||||
{"AlpineJS transition 1", `<div x-transition:enter-start="opacity-0 transform mobile:-translate-x-8 sm:-translate-y-8">`, f("div", "mobile:-translate-x-8 opacity-0 sm:-translate-y-8 transform", "")},
|
||||
{"Vue bind", `<div v-bind:class="{ active: isActive }"></div>`, f("div", "active", "")},
|
||||
// Issue #7746
|
||||
{"Apostrophe inside attribute value", `<a class="missingclass" title="Plus d'information">my text</a><div></div>`, f("a div", "missingclass", "")},
|
||||
// Issue #7567
|
||||
{"Script tags content should be skipped", `<script><span>foo</span><span>bar</span></script><div class="foo"></div>`, f("div script", "foo", "")},
|
||||
{"Style tags content should be skipped", `<style>p{color: red;font-size: 20px;}</style><div class="foo"></div>`, f("div style", "foo", "")},
|
||||
{"Pre tags content should be skipped", `<pre class="preclass"><span>foo</span><span>bar</span></pre><div class="foo"></div>`, f("div pre", "foo preclass", "")},
|
||||
{"Textare tags content should be skipped", `<textarea class="textareaclass"><span>foo</span><span>bar</span></textarea><div class="foo"></div>`, f("div textarea", "foo textareaclass", "")},
|
||||
{"Textarea tags content should be skipped", `<textarea class="textareaclass"><span>foo</span><span>bar</span></textarea><div class="foo"></div>`, f("div textarea", "foo textareaclass", "")},
|
||||
{"DOCTYPE should beskipped", `<!DOCTYPE html>`, f("", "", "")},
|
||||
{"Comments should be skipped", `<!-- example comment -->`, f("", "", "")},
|
||||
// Issue #8417
|
||||
{"Tabs inline", `<hr id="a" class="foo"><div class="bar">d</div>`, f("div hr", "bar foo", "a")},
|
||||
{"Tabs on multiple rows", `<form
|
||||
id="a"
|
||||
action="www.example.com"
|
||||
method="post"
|
||||
></form>
|
||||
<div id="b" class="foo">d</div>`, f("div form", "foo", "a b")},
|
||||
} {
|
||||
|
||||
for _, minify := range []bool{false, true} {
|
||||
c.Run(fmt.Sprintf("%s--minify-%t", test.name, minify), func(c *qt.C) {
|
||||
w := newHTMLElementsCollectorWriter(newHTMLElementsCollector())
|
||||
|
||||
if minify {
|
||||
if skipMinifyTest[test.name] {
|
||||
c.Skip("skip minify test")
|
||||
|
@ -152,6 +152,106 @@ func BenchmarkClassCollectorWriter(b *testing.B) {
|
|||
for i := 0; i < b.N; i++ {
|
||||
w := newHTMLElementsCollectorWriter(newHTMLElementsCollector())
|
||||
fmt.Fprint(w, benchHTML)
|
||||
|
||||
}
|
||||
}
|
||||
|
||||
const benchHTML = `
|
||||
<!DOCTYPE html>
|
||||
<html>
|
||||
<head>
|
||||
<title>title</title>
|
||||
<style>
|
||||
a {color: red;}
|
||||
.c {color: blue;}
|
||||
</style>
|
||||
</head>
|
||||
<body id="i1" class="a b c d">
|
||||
<a class="c d e"></a>
|
||||
<hr>
|
||||
<a class="c d e"></a>
|
||||
<a class="c d e"></a>
|
||||
<hr>
|
||||
<a id="i2" class="c d e f"></a>
|
||||
<a id="i3" class="c d e"></a>
|
||||
<a class="c d e"></a>
|
||||
<p>To force<br> line breaks<br> in a text,<br> use the br<br> element.</p>
|
||||
<hr>
|
||||
<a class="c d e"></a>
|
||||
<a class="c d e"></a>
|
||||
<a class="c d e"></a>
|
||||
<a class="c d e"></a>
|
||||
<table>
|
||||
<thead class="ch">
|
||||
<tr>
|
||||
<th>Month</th>
|
||||
<th>Savings</th>
|
||||
</tr>
|
||||
</thead>
|
||||
<tbody class="cb">
|
||||
<tr>
|
||||
<td>January</td>
|
||||
<td>$100</td>
|
||||
</tr>
|
||||
<tr>
|
||||
<td>February</td>
|
||||
<td>$200</td>
|
||||
</tr>
|
||||
</tbody>
|
||||
<tfoot class="cf">
|
||||
<tr>
|
||||
<td></td>
|
||||
<td>$300</td>
|
||||
</tr>
|
||||
</tfoot>
|
||||
</table>
|
||||
</body>
|
||||
</html>
|
||||
`
|
||||
|
||||
func BenchmarkElementsCollectorWriter(b *testing.B) {
|
||||
b.ReportAllocs()
|
||||
for i := 0; i < b.N; i++ {
|
||||
w := newHTMLElementsCollectorWriter(newHTMLElementsCollector())
|
||||
fmt.Fprint(w, benchHTML)
|
||||
}
|
||||
}
|
||||
|
||||
func BenchmarkElementsCollectorWriterMinified(b *testing.B) {
|
||||
b.ReportAllocs()
|
||||
v := viper.New()
|
||||
m, _ := minifiers.New(media.DefaultTypes, output.DefaultFormats, v)
|
||||
var buf bytes.Buffer
|
||||
m.Minify(media.HTMLType, &buf, strings.NewReader(benchHTML))
|
||||
b.ResetTimer()
|
||||
|
||||
for i := 0; i < b.N; i++ {
|
||||
w := newHTMLElementsCollectorWriter(newHTMLElementsCollector())
|
||||
fmt.Fprint(w, buf.String())
|
||||
}
|
||||
}
|
||||
|
||||
func BenchmarkElementsCollectorWriterWithMinifyStream(b *testing.B) {
|
||||
b.ReportAllocs()
|
||||
v := viper.New()
|
||||
m, _ := minifiers.New(media.DefaultTypes, output.DefaultFormats, v)
|
||||
b.ResetTimer()
|
||||
|
||||
for i := 0; i < b.N; i++ {
|
||||
w := newHTMLElementsCollectorWriter(newHTMLElementsCollector())
|
||||
m.Minify(media.HTMLType, w, strings.NewReader(benchHTML))
|
||||
}
|
||||
}
|
||||
|
||||
func BenchmarkElementsCollectorWriterWithMinifyString(b *testing.B) {
|
||||
b.ReportAllocs()
|
||||
v := viper.New()
|
||||
m, _ := minifiers.New(media.DefaultTypes, output.DefaultFormats, v)
|
||||
b.ResetTimer()
|
||||
|
||||
for i := 0; i < b.N; i++ {
|
||||
var buf bytes.Buffer
|
||||
m.Minify(media.HTMLType, &buf, strings.NewReader(benchHTML))
|
||||
w := newHTMLElementsCollectorWriter(newHTMLElementsCollector())
|
||||
fmt.Fprint(w, buf.String())
|
||||
}
|
||||
}
|
||||
|
|
Loading…
Reference in a new issue