mirror of
https://github.com/gohugoio/hugo.git
synced 2024-11-29 06:22:07 -05:00
3cdf19e9b7
This commit is not the smallest in Hugo's history. Some hightlights include: * Page bundles (for complete articles, keeping images and content together etc.). * Bundled images can be processed in as many versions/sizes as you need with the three methods `Resize`, `Fill` and `Fit`. * Processed images are cached inside `resources/_gen/images` (default) in your project. * Symbolic links (both files and dirs) are now allowed anywhere inside /content * A new table based build summary * The "Total in nn ms" now reports the total including the handling of the files inside /static. So if it now reports more than you're used to, it is just **more real** and probably faster than before (see below). A site building benchmark run compared to `v0.31.1` shows that this should be slightly faster and use less memory: ```bash ▶ ./benchSite.sh "TOML,num_langs=.*,num_root_sections=5,num_pages=(500|1000),tags_per_page=5,shortcodes,render" benchmark old ns/op new ns/op delta BenchmarkSiteBuilding/TOML,num_langs=1,num_root_sections=5,num_pages=500,tags_per_page=5,shortcodes,render-4 101785785 78067944 -23.30% BenchmarkSiteBuilding/TOML,num_langs=1,num_root_sections=5,num_pages=1000,tags_per_page=5,shortcodes,render-4 185481057 149159919 -19.58% BenchmarkSiteBuilding/TOML,num_langs=3,num_root_sections=5,num_pages=500,tags_per_page=5,shortcodes,render-4 103149918 85679409 -16.94% BenchmarkSiteBuilding/TOML,num_langs=3,num_root_sections=5,num_pages=1000,tags_per_page=5,shortcodes,render-4 203515478 169208775 -16.86% benchmark old allocs new allocs delta BenchmarkSiteBuilding/TOML,num_langs=1,num_root_sections=5,num_pages=500,tags_per_page=5,shortcodes,render-4 532464 391539 -26.47% BenchmarkSiteBuilding/TOML,num_langs=1,num_root_sections=5,num_pages=1000,tags_per_page=5,shortcodes,render-4 1056549 772702 -26.87% BenchmarkSiteBuilding/TOML,num_langs=3,num_root_sections=5,num_pages=500,tags_per_page=5,shortcodes,render-4 555974 406630 -26.86% BenchmarkSiteBuilding/TOML,num_langs=3,num_root_sections=5,num_pages=1000,tags_per_page=5,shortcodes,render-4 1086545 789922 -27.30% benchmark old bytes new bytes delta BenchmarkSiteBuilding/TOML,num_langs=1,num_root_sections=5,num_pages=500,tags_per_page=5,shortcodes,render-4 53243246 43598155 -18.12% BenchmarkSiteBuilding/TOML,num_langs=1,num_root_sections=5,num_pages=1000,tags_per_page=5,shortcodes,render-4 105811617 86087116 -18.64% BenchmarkSiteBuilding/TOML,num_langs=3,num_root_sections=5,num_pages=500,tags_per_page=5,shortcodes,render-4 54558852 44545097 -18.35% BenchmarkSiteBuilding/TOML,num_langs=3,num_root_sections=5,num_pages=1000,tags_per_page=5,shortcodes,render-4 106903858 86978413 -18.64% ``` Fixes #3651 Closes #3158 Fixes #1014 Closes #2021 Fixes #1240 Updates #3757
492 lines
16 KiB
Go
492 lines
16 KiB
Go
// Copyright 2015 The Hugo Authors. All rights reserved.
|
|
//
|
|
// Licensed under the Apache License, Version 2.0 (the "License");
|
|
// you may not use this file except in compliance with the License.
|
|
// You may obtain a copy of the License at
|
|
// http://www.apache.org/licenses/LICENSE-2.0
|
|
//
|
|
// Unless required by applicable law or agreed to in writing, software
|
|
// distributed under the License is distributed on an "AS IS" BASIS,
|
|
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
// See the License for the specific language governing permissions and
|
|
// limitations under the License.
|
|
|
|
package helpers
|
|
|
|
import (
|
|
"bytes"
|
|
"html/template"
|
|
"strings"
|
|
"testing"
|
|
|
|
"github.com/spf13/viper"
|
|
|
|
"github.com/miekg/mmark"
|
|
"github.com/russross/blackfriday"
|
|
"github.com/stretchr/testify/assert"
|
|
"github.com/stretchr/testify/require"
|
|
)
|
|
|
|
const tstHTMLContent = "<!DOCTYPE html><html><head><script src=\"http://two/foobar.js\"></script></head><body><nav><ul><li hugo-nav=\"section_0\"></li><li hugo-nav=\"section_1\"></li></ul></nav><article>content <a href=\"http://two/foobar\">foobar</a>. Follow up</article><p>This is some text.<br>And some more.</p></body></html>"
|
|
|
|
func TestStripHTML(t *testing.T) {
|
|
type test struct {
|
|
input, expected string
|
|
}
|
|
data := []test{
|
|
{"<h1>strip h1 tag <h1>", "strip h1 tag "},
|
|
{"<p> strip p tag </p>", " strip p tag "},
|
|
{"</br> strip br<br>", " strip br\n"},
|
|
{"</br> strip br2<br />", " strip br2\n"},
|
|
{"This <strong>is</strong> a\nnewline", "This is a newline"},
|
|
{"No Tags", "No Tags"},
|
|
{`<p>Summary Next Line.
|
|
<figure >
|
|
|
|
<img src="/not/real" />
|
|
|
|
|
|
</figure>
|
|
.
|
|
More text here.</p>
|
|
|
|
<p>Some more text</p>`, "Summary Next Line. . More text here.\nSome more text\n"},
|
|
}
|
|
for i, d := range data {
|
|
output := StripHTML(d.input)
|
|
if d.expected != output {
|
|
t.Errorf("Test %d failed. Expected %q got %q", i, d.expected, output)
|
|
}
|
|
}
|
|
}
|
|
|
|
func BenchmarkStripHTML(b *testing.B) {
|
|
b.ResetTimer()
|
|
for i := 0; i < b.N; i++ {
|
|
StripHTML(tstHTMLContent)
|
|
}
|
|
}
|
|
|
|
func TestStripEmptyNav(t *testing.T) {
|
|
cleaned := stripEmptyNav([]byte("do<nav>\n</nav>\n\nbedobedo"))
|
|
assert.Equal(t, []byte("dobedobedo"), cleaned)
|
|
}
|
|
|
|
func TestBytesToHTML(t *testing.T) {
|
|
assert.Equal(t, template.HTML("dobedobedo"), BytesToHTML([]byte("dobedobedo")))
|
|
}
|
|
|
|
func TestNewContentSpec(t *testing.T) {
|
|
cfg := viper.New()
|
|
assert := require.New(t)
|
|
|
|
cfg.Set("summaryLength", 32)
|
|
cfg.Set("buildFuture", true)
|
|
cfg.Set("buildExpired", true)
|
|
cfg.Set("buildDrafts", true)
|
|
|
|
spec, err := NewContentSpec(cfg)
|
|
|
|
assert.NoError(err)
|
|
assert.Equal(32, spec.summaryLength)
|
|
assert.True(spec.BuildFuture)
|
|
assert.True(spec.BuildExpired)
|
|
assert.True(spec.BuildDrafts)
|
|
|
|
}
|
|
|
|
var benchmarkTruncateString = strings.Repeat("This is a sentence about nothing.", 20)
|
|
|
|
func BenchmarkTestTruncateWordsToWholeSentence(b *testing.B) {
|
|
c := newTestContentSpec()
|
|
b.ResetTimer()
|
|
for i := 0; i < b.N; i++ {
|
|
c.TruncateWordsToWholeSentence(benchmarkTruncateString)
|
|
}
|
|
}
|
|
|
|
func BenchmarkTestTruncateWordsToWholeSentenceOld(b *testing.B) {
|
|
c := newTestContentSpec()
|
|
b.ResetTimer()
|
|
for i := 0; i < b.N; i++ {
|
|
c.truncateWordsToWholeSentenceOld(benchmarkTruncateString)
|
|
}
|
|
}
|
|
|
|
func TestTruncateWordsToWholeSentence(t *testing.T) {
|
|
c := newTestContentSpec()
|
|
type test struct {
|
|
input, expected string
|
|
max int
|
|
truncated bool
|
|
}
|
|
data := []test{
|
|
{"a b c", "a b c", 12, false},
|
|
{"a b c", "a b c", 3, false},
|
|
{"a", "a", 1, false},
|
|
{"This is a sentence.", "This is a sentence.", 5, false},
|
|
{"This is also a sentence!", "This is also a sentence!", 1, false},
|
|
{"To be. Or not to be. That's the question.", "To be.", 1, true},
|
|
{" \nThis is not a sentence\nAnd this is another", "This is not a sentence", 4, true},
|
|
{"", "", 10, false},
|
|
{"This... is a more difficult test?", "This... is a more difficult test?", 1, false},
|
|
}
|
|
for i, d := range data {
|
|
c.summaryLength = d.max
|
|
output, truncated := c.TruncateWordsToWholeSentence(d.input)
|
|
if d.expected != output {
|
|
t.Errorf("Test %d failed. Expected %q got %q", i, d.expected, output)
|
|
}
|
|
|
|
if d.truncated != truncated {
|
|
t.Errorf("Test %d failed. Expected truncated=%t got %t", i, d.truncated, truncated)
|
|
}
|
|
}
|
|
}
|
|
|
|
func TestTruncateWordsByRune(t *testing.T) {
|
|
c := newTestContentSpec()
|
|
type test struct {
|
|
input, expected string
|
|
max int
|
|
truncated bool
|
|
}
|
|
data := []test{
|
|
{"", "", 1, false},
|
|
{"a b c", "a b c", 12, false},
|
|
{"a b c", "a b c", 3, false},
|
|
{"a", "a", 1, false},
|
|
{"Hello 中国", "", 0, true},
|
|
{"这是中文,全中文。", "这是中文,", 5, true},
|
|
{"Hello 中国", "Hello 中", 2, true},
|
|
{"Hello 中国", "Hello 中国", 3, false},
|
|
{"Hello中国 Good 好的", "Hello中国 Good 好", 9, true},
|
|
{"This is a sentence.", "This is", 2, true},
|
|
{"This is also a sentence!", "This", 1, true},
|
|
{"To be. Or not to be. That's the question.", "To be. Or not", 4, true},
|
|
{" \nThis is not a sentence\n ", "This is not", 3, true},
|
|
}
|
|
for i, d := range data {
|
|
c.summaryLength = d.max
|
|
output, truncated := c.TruncateWordsByRune(strings.Fields(d.input))
|
|
if d.expected != output {
|
|
t.Errorf("Test %d failed. Expected %q got %q", i, d.expected, output)
|
|
}
|
|
|
|
if d.truncated != truncated {
|
|
t.Errorf("Test %d failed. Expected truncated=%t got %t", i, d.truncated, truncated)
|
|
}
|
|
}
|
|
}
|
|
|
|
func TestGetHTMLRendererFlags(t *testing.T) {
|
|
c := newTestContentSpec()
|
|
ctx := &RenderingContext{Cfg: c.cfg, Config: c.BlackFriday}
|
|
renderer := c.getHTMLRenderer(blackfriday.HTML_USE_XHTML, ctx)
|
|
flags := renderer.GetFlags()
|
|
if flags&blackfriday.HTML_USE_XHTML != blackfriday.HTML_USE_XHTML {
|
|
t.Errorf("Test flag: %d was not found amongs set flags:%d; Result: %d", blackfriday.HTML_USE_XHTML, flags, flags&blackfriday.HTML_USE_XHTML)
|
|
}
|
|
}
|
|
|
|
func TestGetHTMLRendererAllFlags(t *testing.T) {
|
|
c := newTestContentSpec()
|
|
|
|
type data struct {
|
|
testFlag int
|
|
}
|
|
|
|
allFlags := []data{
|
|
{blackfriday.HTML_USE_XHTML},
|
|
{blackfriday.HTML_FOOTNOTE_RETURN_LINKS},
|
|
{blackfriday.HTML_USE_SMARTYPANTS},
|
|
{blackfriday.HTML_SMARTYPANTS_QUOTES_NBSP},
|
|
{blackfriday.HTML_SMARTYPANTS_ANGLED_QUOTES},
|
|
{blackfriday.HTML_SMARTYPANTS_FRACTIONS},
|
|
{blackfriday.HTML_HREF_TARGET_BLANK},
|
|
{blackfriday.HTML_SMARTYPANTS_DASHES},
|
|
{blackfriday.HTML_SMARTYPANTS_LATEX_DASHES},
|
|
}
|
|
defaultFlags := blackfriday.HTML_USE_XHTML
|
|
ctx := &RenderingContext{Cfg: c.cfg, Config: c.BlackFriday}
|
|
ctx.Config.AngledQuotes = true
|
|
ctx.Config.Fractions = true
|
|
ctx.Config.HrefTargetBlank = true
|
|
ctx.Config.LatexDashes = true
|
|
ctx.Config.PlainIDAnchors = true
|
|
ctx.Config.SmartDashes = true
|
|
ctx.Config.Smartypants = true
|
|
ctx.Config.SmartypantsQuotesNBSP = true
|
|
renderer := c.getHTMLRenderer(defaultFlags, ctx)
|
|
actualFlags := renderer.GetFlags()
|
|
var expectedFlags int
|
|
//OR-ing flags together...
|
|
for _, d := range allFlags {
|
|
expectedFlags |= d.testFlag
|
|
}
|
|
if expectedFlags != actualFlags {
|
|
t.Errorf("Expected flags (%d) did not equal actual (%d) flags.", expectedFlags, actualFlags)
|
|
}
|
|
}
|
|
|
|
func TestGetHTMLRendererAnchors(t *testing.T) {
|
|
c := newTestContentSpec()
|
|
ctx := &RenderingContext{Cfg: c.cfg, Config: c.BlackFriday}
|
|
ctx.DocumentID = "testid"
|
|
ctx.Config.PlainIDAnchors = false
|
|
|
|
actualRenderer := c.getHTMLRenderer(0, ctx)
|
|
headerBuffer := &bytes.Buffer{}
|
|
footnoteBuffer := &bytes.Buffer{}
|
|
expectedFootnoteHref := []byte("href=\"#fn:testid:href\"")
|
|
expectedHeaderID := []byte("<h1 id=\"id:testid\"></h1>\n")
|
|
|
|
actualRenderer.Header(headerBuffer, func() bool { return true }, 1, "id")
|
|
actualRenderer.FootnoteRef(footnoteBuffer, []byte("href"), 1)
|
|
|
|
if !bytes.Contains(footnoteBuffer.Bytes(), expectedFootnoteHref) {
|
|
t.Errorf("Footnote anchor prefix not applied. Actual:%s Expected:%s", footnoteBuffer.String(), expectedFootnoteHref)
|
|
}
|
|
|
|
if !bytes.Equal(headerBuffer.Bytes(), expectedHeaderID) {
|
|
t.Errorf("Header Id Postfix not applied. Actual:%s Expected:%s", headerBuffer.String(), expectedHeaderID)
|
|
}
|
|
}
|
|
|
|
func TestGetMmarkHTMLRenderer(t *testing.T) {
|
|
c := newTestContentSpec()
|
|
ctx := &RenderingContext{Cfg: c.cfg, Config: c.BlackFriday}
|
|
ctx.DocumentID = "testid"
|
|
ctx.Config.PlainIDAnchors = false
|
|
actualRenderer := c.getMmarkHTMLRenderer(0, ctx)
|
|
|
|
headerBuffer := &bytes.Buffer{}
|
|
footnoteBuffer := &bytes.Buffer{}
|
|
expectedFootnoteHref := []byte("href=\"#fn:testid:href\"")
|
|
expectedHeaderID := []byte("<h1 id=\"id\"></h1>")
|
|
|
|
actualRenderer.FootnoteRef(footnoteBuffer, []byte("href"), 1)
|
|
actualRenderer.Header(headerBuffer, func() bool { return true }, 1, "id")
|
|
|
|
if !bytes.Contains(footnoteBuffer.Bytes(), expectedFootnoteHref) {
|
|
t.Errorf("Footnote anchor prefix not applied. Actual:%s Expected:%s", footnoteBuffer.String(), expectedFootnoteHref)
|
|
}
|
|
|
|
if bytes.Equal(headerBuffer.Bytes(), expectedHeaderID) {
|
|
t.Errorf("Header Id Postfix applied. Actual:%s Expected:%s", headerBuffer.String(), expectedHeaderID)
|
|
}
|
|
}
|
|
|
|
func TestGetMarkdownExtensionsMasksAreRemovedFromExtensions(t *testing.T) {
|
|
c := newTestContentSpec()
|
|
ctx := &RenderingContext{Cfg: c.cfg, Config: c.BlackFriday}
|
|
ctx.Config.Extensions = []string{"headerId"}
|
|
ctx.Config.ExtensionsMask = []string{"noIntraEmphasis"}
|
|
|
|
actualFlags := getMarkdownExtensions(ctx)
|
|
if actualFlags&blackfriday.EXTENSION_NO_INTRA_EMPHASIS == blackfriday.EXTENSION_NO_INTRA_EMPHASIS {
|
|
t.Errorf("Masked out flag {%v} found amongst returned extensions.", blackfriday.EXTENSION_NO_INTRA_EMPHASIS)
|
|
}
|
|
}
|
|
|
|
func TestGetMarkdownExtensionsByDefaultAllExtensionsAreEnabled(t *testing.T) {
|
|
type data struct {
|
|
testFlag int
|
|
}
|
|
c := newTestContentSpec()
|
|
ctx := &RenderingContext{Cfg: c.cfg, Config: c.BlackFriday}
|
|
ctx.Config.Extensions = []string{""}
|
|
ctx.Config.ExtensionsMask = []string{""}
|
|
allExtensions := []data{
|
|
{blackfriday.EXTENSION_NO_INTRA_EMPHASIS},
|
|
{blackfriday.EXTENSION_TABLES},
|
|
{blackfriday.EXTENSION_FENCED_CODE},
|
|
{blackfriday.EXTENSION_AUTOLINK},
|
|
{blackfriday.EXTENSION_STRIKETHROUGH},
|
|
// {blackfriday.EXTENSION_LAX_HTML_BLOCKS},
|
|
{blackfriday.EXTENSION_SPACE_HEADERS},
|
|
// {blackfriday.EXTENSION_HARD_LINE_BREAK},
|
|
// {blackfriday.EXTENSION_TAB_SIZE_EIGHT},
|
|
{blackfriday.EXTENSION_FOOTNOTES},
|
|
// {blackfriday.EXTENSION_NO_EMPTY_LINE_BEFORE_BLOCK},
|
|
{blackfriday.EXTENSION_HEADER_IDS},
|
|
// {blackfriday.EXTENSION_TITLEBLOCK},
|
|
{blackfriday.EXTENSION_AUTO_HEADER_IDS},
|
|
{blackfriday.EXTENSION_BACKSLASH_LINE_BREAK},
|
|
{blackfriday.EXTENSION_DEFINITION_LISTS},
|
|
}
|
|
|
|
actualFlags := getMarkdownExtensions(ctx)
|
|
for _, e := range allExtensions {
|
|
if actualFlags&e.testFlag != e.testFlag {
|
|
t.Errorf("Flag %v was not found in the list of extensions.", e)
|
|
}
|
|
}
|
|
}
|
|
|
|
func TestGetMarkdownExtensionsAddingFlagsThroughRenderingContext(t *testing.T) {
|
|
c := newTestContentSpec()
|
|
ctx := &RenderingContext{Cfg: c.cfg, Config: c.BlackFriday}
|
|
ctx.Config.Extensions = []string{"definitionLists"}
|
|
ctx.Config.ExtensionsMask = []string{""}
|
|
|
|
actualFlags := getMarkdownExtensions(ctx)
|
|
if actualFlags&blackfriday.EXTENSION_DEFINITION_LISTS != blackfriday.EXTENSION_DEFINITION_LISTS {
|
|
t.Errorf("Masked out flag {%v} found amongst returned extensions.", blackfriday.EXTENSION_DEFINITION_LISTS)
|
|
}
|
|
}
|
|
|
|
func TestGetMarkdownRenderer(t *testing.T) {
|
|
c := newTestContentSpec()
|
|
ctx := &RenderingContext{Cfg: c.cfg, Config: c.BlackFriday}
|
|
ctx.Content = []byte("testContent")
|
|
actualRenderedMarkdown := c.markdownRender(ctx)
|
|
expectedRenderedMarkdown := []byte("<p>testContent</p>\n")
|
|
if !bytes.Equal(actualRenderedMarkdown, expectedRenderedMarkdown) {
|
|
t.Errorf("Actual rendered Markdown (%s) did not match expected markdown (%s)", actualRenderedMarkdown, expectedRenderedMarkdown)
|
|
}
|
|
}
|
|
|
|
func TestGetMarkdownRendererWithTOC(t *testing.T) {
|
|
c := newTestContentSpec()
|
|
ctx := &RenderingContext{RenderTOC: true, Cfg: c.cfg, Config: c.BlackFriday}
|
|
ctx.Content = []byte("testContent")
|
|
actualRenderedMarkdown := c.markdownRender(ctx)
|
|
expectedRenderedMarkdown := []byte("<nav>\n</nav>\n\n<p>testContent</p>\n")
|
|
if !bytes.Equal(actualRenderedMarkdown, expectedRenderedMarkdown) {
|
|
t.Errorf("Actual rendered Markdown (%s) did not match expected markdown (%s)", actualRenderedMarkdown, expectedRenderedMarkdown)
|
|
}
|
|
}
|
|
|
|
func TestGetMmarkExtensions(t *testing.T) {
|
|
//TODO: This is doing the same just with different marks...
|
|
type data struct {
|
|
testFlag int
|
|
}
|
|
c := newTestContentSpec()
|
|
ctx := &RenderingContext{Cfg: c.cfg, Config: c.BlackFriday}
|
|
ctx.Config.Extensions = []string{"tables"}
|
|
ctx.Config.ExtensionsMask = []string{""}
|
|
allExtensions := []data{
|
|
{mmark.EXTENSION_TABLES},
|
|
{mmark.EXTENSION_FENCED_CODE},
|
|
{mmark.EXTENSION_AUTOLINK},
|
|
{mmark.EXTENSION_SPACE_HEADERS},
|
|
{mmark.EXTENSION_CITATION},
|
|
{mmark.EXTENSION_TITLEBLOCK_TOML},
|
|
{mmark.EXTENSION_HEADER_IDS},
|
|
{mmark.EXTENSION_AUTO_HEADER_IDS},
|
|
{mmark.EXTENSION_UNIQUE_HEADER_IDS},
|
|
{mmark.EXTENSION_FOOTNOTES},
|
|
{mmark.EXTENSION_SHORT_REF},
|
|
{mmark.EXTENSION_NO_EMPTY_LINE_BEFORE_BLOCK},
|
|
{mmark.EXTENSION_INCLUDE},
|
|
}
|
|
|
|
actualFlags := getMmarkExtensions(ctx)
|
|
for _, e := range allExtensions {
|
|
if actualFlags&e.testFlag != e.testFlag {
|
|
t.Errorf("Flag %v was not found in the list of extensions.", e)
|
|
}
|
|
}
|
|
}
|
|
|
|
func TestMmarkRender(t *testing.T) {
|
|
c := newTestContentSpec()
|
|
ctx := &RenderingContext{Cfg: c.cfg, Config: c.BlackFriday}
|
|
ctx.Content = []byte("testContent")
|
|
actualRenderedMarkdown := c.mmarkRender(ctx)
|
|
expectedRenderedMarkdown := []byte("<p>testContent</p>\n")
|
|
if !bytes.Equal(actualRenderedMarkdown, expectedRenderedMarkdown) {
|
|
t.Errorf("Actual rendered Markdown (%s) did not match expected markdown (%s)", actualRenderedMarkdown, expectedRenderedMarkdown)
|
|
}
|
|
}
|
|
|
|
func TestExtractTOCNormalContent(t *testing.T) {
|
|
content := []byte("<nav>\n<ul>\nTOC<li><a href=\"#")
|
|
|
|
actualTocLessContent, actualToc := ExtractTOC(content)
|
|
expectedTocLess := []byte("TOC<li><a href=\"#")
|
|
expectedToc := []byte("<nav id=\"TableOfContents\">\n<ul>\n")
|
|
|
|
if !bytes.Equal(actualTocLessContent, expectedTocLess) {
|
|
t.Errorf("Actual tocless (%s) did not equal expected (%s) tocless content", actualTocLessContent, expectedTocLess)
|
|
}
|
|
|
|
if !bytes.Equal(actualToc, expectedToc) {
|
|
t.Errorf("Actual toc (%s) did not equal expected (%s) toc content", actualToc, expectedToc)
|
|
}
|
|
}
|
|
|
|
func TestExtractTOCGreaterThanSeventy(t *testing.T) {
|
|
content := []byte("<nav>\n<ul>\nTOC This is a very long content which will definitely be greater than seventy, I promise you that.<li><a href=\"#")
|
|
|
|
actualTocLessContent, actualToc := ExtractTOC(content)
|
|
//Because the start of Toc is greater than 70+startpoint of <li> content and empty TOC will be returned
|
|
expectedToc := []byte("")
|
|
|
|
if !bytes.Equal(actualTocLessContent, content) {
|
|
t.Errorf("Actual tocless (%s) did not equal expected (%s) tocless content", actualTocLessContent, content)
|
|
}
|
|
|
|
if !bytes.Equal(actualToc, expectedToc) {
|
|
t.Errorf("Actual toc (%s) did not equal expected (%s) toc content", actualToc, expectedToc)
|
|
}
|
|
}
|
|
|
|
func TestExtractNoTOC(t *testing.T) {
|
|
content := []byte("TOC")
|
|
|
|
actualTocLessContent, actualToc := ExtractTOC(content)
|
|
expectedToc := []byte("")
|
|
|
|
if !bytes.Equal(actualTocLessContent, content) {
|
|
t.Errorf("Actual tocless (%s) did not equal expected (%s) tocless content", actualTocLessContent, content)
|
|
}
|
|
|
|
if !bytes.Equal(actualToc, expectedToc) {
|
|
t.Errorf("Actual toc (%s) did not equal expected (%s) toc content", actualToc, expectedToc)
|
|
}
|
|
}
|
|
|
|
var totalWordsBenchmarkString = strings.Repeat("Hugo Rocks ", 200)
|
|
|
|
func TestTotalWords(t *testing.T) {
|
|
|
|
for i, this := range []struct {
|
|
s string
|
|
words int
|
|
}{
|
|
{"Two, Words!", 2},
|
|
{"Word", 1},
|
|
{"", 0},
|
|
{"One, Two, Three", 3},
|
|
{totalWordsBenchmarkString, 400},
|
|
} {
|
|
actualWordCount := TotalWords(this.s)
|
|
|
|
if actualWordCount != this.words {
|
|
t.Errorf("[%d] Actual word count (%d) for test string (%s) did not match %d", i, actualWordCount, this.s, this.words)
|
|
}
|
|
}
|
|
}
|
|
|
|
func BenchmarkTotalWords(b *testing.B) {
|
|
b.ResetTimer()
|
|
for i := 0; i < b.N; i++ {
|
|
wordCount := TotalWords(totalWordsBenchmarkString)
|
|
if wordCount != 400 {
|
|
b.Fatal("Wordcount error")
|
|
}
|
|
}
|
|
}
|
|
|
|
func BenchmarkTotalWordsOld(b *testing.B) {
|
|
b.ResetTimer()
|
|
for i := 0; i < b.N; i++ {
|
|
wordCount := totalWordsOld(totalWordsBenchmarkString)
|
|
if wordCount != 400 {
|
|
b.Fatal("Wordcount error")
|
|
}
|
|
}
|
|
}
|