mirror of
https://github.com/gohugoio/hugo.git
synced 2024-11-21 20:46:30 -05:00
9cd54cab20
This avoids double parsing the page content when `enableEmoji=true`. This commit also adds some general improvements to the parser, making it in general much faster: ```bash benchmark old ns/op new ns/op delta BenchmarkShortcodeLexer-4 90258 101730 +12.71% BenchmarkParse-4 148940 15037 -89.90% benchmark old allocs new allocs delta BenchmarkShortcodeLexer-4 456 700 +53.51% BenchmarkParse-4 28 33 +17.86% benchmark old bytes new bytes delta BenchmarkShortcodeLexer-4 69875 81014 +15.94% BenchmarkParse-4 8128 8304 +2.17% ``` Running some site benchmarks with Emoji support turned on: ```bash benchmark old ns/op new ns/op delta BenchmarkSiteBuilding/TOML,num_langs=3,num_pages=5000,tags_per_page=5,shortcodes,render-4 924556797 818115620 -11.51% benchmark old allocs new allocs delta BenchmarkSiteBuilding/TOML,num_langs=3,num_pages=5000,tags_per_page=5,shortcodes,render-4 4112613 4133787 +0.51% benchmark old bytes new bytes delta BenchmarkSiteBuilding/TOML,num_langs=3,num_pages=5000,tags_per_page=5,shortcodes,render-4 426982864 424363832 -0.61% ``` Fixes #5534
127 lines
5.5 KiB
Go
127 lines
5.5 KiB
Go
// Copyright 2018 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 pageparser
|
|
|
|
import (
|
|
"fmt"
|
|
"reflect"
|
|
"strings"
|
|
"testing"
|
|
)
|
|
|
|
type lexerTest struct {
|
|
name string
|
|
input string
|
|
items []Item
|
|
}
|
|
|
|
func nti(tp ItemType, val string) Item {
|
|
return Item{tp, 0, []byte(val)}
|
|
}
|
|
|
|
var (
|
|
tstJSON = `{ "a": { "b": "\"Hugo\"}" } }`
|
|
tstFrontMatterTOML = nti(TypeFrontMatterTOML, "foo = \"bar\"\n")
|
|
tstFrontMatterYAML = nti(TypeFrontMatterYAML, "foo: \"bar\"\n")
|
|
tstFrontMatterYAMLCRLF = nti(TypeFrontMatterYAML, "foo: \"bar\"\r\n")
|
|
tstFrontMatterJSON = nti(TypeFrontMatterJSON, tstJSON+"\r\n")
|
|
tstSomeText = nti(tText, "\nSome text.\n")
|
|
tstSummaryDivider = nti(TypeLeadSummaryDivider, "<!--more-->\n")
|
|
tstHtmlStart = nti(TypeHTMLStart, "<")
|
|
tstNewline = nti(tText, "\n")
|
|
|
|
tstORG = `
|
|
#+TITLE: T1
|
|
#+AUTHOR: A1
|
|
#+DESCRIPTION: D1
|
|
`
|
|
tstFrontMatterORG = nti(TypeFrontMatterORG, tstORG)
|
|
)
|
|
|
|
var crLfReplacer = strings.NewReplacer("\r", "#", "\n", "$")
|
|
|
|
// TODO(bep) a way to toggle ORG mode vs the rest.
|
|
var frontMatterTests = []lexerTest{
|
|
{"empty", "", []Item{tstEOF}},
|
|
{"Byte order mark", "\ufeff\nSome text.\n", []Item{nti(TypeIgnore, "\ufeff"), tstSomeText, tstEOF}},
|
|
{"HTML Document", ` <html> `, []Item{nti(tText, " "), tstHtmlStart, nti(tText, "html> "), tstEOF}},
|
|
{"HTML Document with shortcode", `<html>{{< sc1 >}}</html>`, []Item{tstHtmlStart, nti(tText, "html>"), tstLeftNoMD, tstSC1, tstRightNoMD, nti(tText, "</html>"), tstEOF}},
|
|
{"No front matter", "\nSome text.\n", []Item{tstSomeText, tstEOF}},
|
|
{"YAML front matter", "---\nfoo: \"bar\"\n---\n\nSome text.\n", []Item{tstFrontMatterYAML, tstSomeText, tstEOF}},
|
|
{"YAML empty front matter", "---\n---\n\nSome text.\n", []Item{nti(TypeFrontMatterYAML, ""), tstSomeText, tstEOF}},
|
|
{"YAML commented out front matter", "<!--\n---\nfoo: \"bar\"\n---\n-->\nSome text.\n", []Item{nti(TypeIgnore, "<!--\n"), tstFrontMatterYAML, nti(TypeIgnore, "-->"), tstSomeText, tstEOF}},
|
|
{"YAML commented out front matter, no end", "<!--\n---\nfoo: \"bar\"\n---\nSome text.\n", []Item{nti(TypeIgnore, "<!--\n"), tstFrontMatterYAML, nti(tError, "starting HTML comment with no end")}},
|
|
// Note that we keep all bytes as they are, but we need to handle CRLF
|
|
{"YAML front matter CRLF", "---\r\nfoo: \"bar\"\r\n---\n\nSome text.\n", []Item{tstFrontMatterYAMLCRLF, tstSomeText, tstEOF}},
|
|
{"TOML front matter", "+++\nfoo = \"bar\"\n+++\n\nSome text.\n", []Item{tstFrontMatterTOML, tstSomeText, tstEOF}},
|
|
{"JSON front matter", tstJSON + "\r\n\nSome text.\n", []Item{tstFrontMatterJSON, tstSomeText, tstEOF}},
|
|
{"ORG front matter", tstORG + "\nSome text.\n", []Item{tstFrontMatterORG, tstSomeText, tstEOF}},
|
|
{"Summary divider ORG", tstORG + "\nSome text.\n# more\nSome text.\n", []Item{tstFrontMatterORG, tstSomeText, nti(TypeLeadSummaryDivider, "# more\n"), nti(tText, "Some text.\n"), tstEOF}},
|
|
{"Summary divider", "+++\nfoo = \"bar\"\n+++\n\nSome text.\n<!--more-->\nSome text.\n", []Item{tstFrontMatterTOML, tstSomeText, tstSummaryDivider, nti(tText, "Some text.\n"), tstEOF}},
|
|
{"Summary divider same line", "+++\nfoo = \"bar\"\n+++\n\nSome text.<!--more-->Some text.\n", []Item{tstFrontMatterTOML, nti(tText, "\nSome text."), nti(TypeLeadSummaryDivider, "<!--more-->"), nti(tText, "Some text.\n"), tstEOF}},
|
|
// https://github.com/gohugoio/hugo/issues/5402
|
|
{"Summary and shortcode, no space", "+++\nfoo = \"bar\"\n+++\n\nSome text.\n<!--more-->{{< sc1 >}}\nSome text.\n", []Item{tstFrontMatterTOML, tstSomeText, nti(TypeLeadSummaryDivider, "<!--more-->"), tstLeftNoMD, tstSC1, tstRightNoMD, tstSomeText, tstEOF}},
|
|
// https://github.com/gohugoio/hugo/issues/5464
|
|
{"Summary and shortcode only", "+++\nfoo = \"bar\"\n+++\n{{< sc1 >}}\n<!--more-->\n{{< sc2 >}}", []Item{tstFrontMatterTOML, tstLeftNoMD, tstSC1, tstRightNoMD, tstNewline, tstSummaryDivider, tstLeftNoMD, tstSC2, tstRightNoMD, tstEOF}},
|
|
}
|
|
|
|
func TestFrontMatter(t *testing.T) {
|
|
t.Parallel()
|
|
for i, test := range frontMatterTests {
|
|
items := collect([]byte(test.input), false, lexIntroSection)
|
|
if !equal(items, test.items) {
|
|
got := crLfReplacer.Replace(fmt.Sprint(items))
|
|
expected := crLfReplacer.Replace(fmt.Sprint(test.items))
|
|
t.Errorf("[%d] %s: got\n\t%v\nexpected\n\t%v", i, test.name, got, expected)
|
|
}
|
|
}
|
|
}
|
|
|
|
func collectWithConfig(input []byte, skipFrontMatter bool, stateStart stateFunc, cfg Config) (items []Item) {
|
|
l := newPageLexer(input, stateStart, cfg)
|
|
l.run()
|
|
t := l.newIterator()
|
|
|
|
for {
|
|
item := t.Next()
|
|
items = append(items, item)
|
|
if item.Type == tEOF || item.Type == tError {
|
|
break
|
|
}
|
|
}
|
|
return
|
|
}
|
|
|
|
func collect(input []byte, skipFrontMatter bool, stateStart stateFunc) (items []Item) {
|
|
var cfg Config
|
|
|
|
return collectWithConfig(input, skipFrontMatter, stateStart, cfg)
|
|
|
|
}
|
|
|
|
// no positional checking, for now ...
|
|
func equal(i1, i2 []Item) bool {
|
|
if len(i1) != len(i2) {
|
|
return false
|
|
}
|
|
for k := range i1 {
|
|
if i1[k].Type != i2[k].Type {
|
|
return false
|
|
}
|
|
if !reflect.DeepEqual(i1[k].Val, i2[k].Val) {
|
|
return false
|
|
}
|
|
}
|
|
return true
|
|
}
|