Move the emoji parsing to pageparser

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
This commit is contained in:
Bjørn Erik Pedersen 2018-12-17 21:03:23 +01:00
parent a8853f1c5a
commit 9cd54cab20
13 changed files with 388 additions and 71 deletions

View file

@ -215,7 +215,7 @@ type parsedFile struct {
func parseContentFile(r io.Reader) (parsedFile, error) {
var pf parsedFile
psr, err := pageparser.Parse(r)
psr, err := pageparser.Parse(r, pageparser.Config{})
if err != nil {
return pf, err
}

View file

@ -30,6 +30,12 @@ var (
emojiMaxSize int
)
// Emoji returns the emojy given a key, e.g. ":smile:", nil if not found.
func Emoji(key string) []byte {
emojiInit.Do(initEmoji)
return emojis[key]
}
// Emojify "emojifies" the input source.
// Note that the input byte slice will be modified if needed.
// See http://www.emoji-cheat-sheet.com/

View file

@ -17,6 +17,8 @@ import (
"bytes"
"io"
"github.com/gohugoio/hugo/helpers"
errors "github.com/pkg/errors"
bp "github.com/gohugoio/hugo/bufferpool"
@ -149,6 +151,12 @@ Loop:
result.WriteString(placeHolder)
ordinal++
s.shortcodes.Add(placeHolder, currShortcode)
case it.Type == pageparser.TypeEmoji:
if emoji := helpers.Emoji(it.ValStr()); emoji != nil {
result.Write(emoji)
} else {
result.Write(it.Val)
}
case it.IsEOF():
break Loop
case it.IsError():
@ -170,7 +178,10 @@ Loop:
func (p *Page) parse(reader io.Reader) error {
parseResult, err := pageparser.Parse(reader)
parseResult, err := pageparser.Parse(
reader,
pageparser.Config{EnableEmoji: p.s.Cfg.GetBool("enableEmoji")},
)
if err != nil {
return err
}

View file

@ -1497,6 +1497,45 @@ func TestChompBOM(t *testing.T) {
checkPageTitle(t, p, "Simple")
}
func TestPageWithEmoji(t *testing.T) {
for _, enableEmoji := range []bool{true, false} {
v := viper.New()
v.Set("enableEmoji", enableEmoji)
b := newTestSitesBuilder(t)
b.WithViper(v)
b.WithSimpleConfigFile()
b.WithContent("page-emoji.md", `---
title: "Hugo Smile"
---
This is a :smile:.
<!--more-->
Another :smile: This is :not: an emoji.
`)
b.CreateSites().Build(BuildCfg{})
if enableEmoji {
b.AssertFileContent("public/page-emoji/index.html",
"This is a 😄",
"Another 😄",
"This is :not: an emoji",
)
} else {
b.AssertFileContent("public/page-emoji/index.html",
"This is a :smile:",
"Another :smile:",
"This is :not: an emoji",
)
}
}
}
// https://github.com/gohugoio/hugo/issues/5381
func TestPageManualSummary(t *testing.T) {
b := newTestSitesBuilder(t)

View file

@ -272,10 +272,6 @@ func (c *contentHandlers) handlePageContent() contentHandler {
p := ctx.currentPage
if c.s.Cfg.GetBool("enableEmoji") {
p.workContent = helpers.Emojify(p.workContent)
}
p.workContent = p.renderContent(p.workContent)
tmpContent, tmpTableOfContents := helpers.ExtractTOC(p.workContent)

View file

@ -177,6 +177,16 @@ type shortcode struct {
pos int // the position in bytes in the source file
}
func (s shortcode) innerString() string {
var sb strings.Builder
for _, inner := range s.inner {
sb.WriteString(inner.(string))
}
return sb.String()
}
func (sc shortcode) String() string {
// for testing (mostly), so any change here will break tests!
var params interface{}
@ -363,7 +373,7 @@ func renderShortcode(
if sc.isInline {
templName := path.Join("_inline_shortcode", p.Path(), sc.name)
if sc.isClosing {
templStr := sc.inner[0].(string)
templStr := sc.innerString()
var err error
tmpl, err = p.s.TextTmpl.Parse(templName, templStr)

View file

@ -113,6 +113,7 @@ const (
TypeFrontMatterTOML
TypeFrontMatterJSON
TypeFrontMatterORG
TypeEmoji
TypeIgnore // // The BOM Unicode byte order marker and possibly others
// shortcode items

View file

@ -37,6 +37,12 @@ type pageLexer struct {
start int // item start position
width int // width of last element
// Contains lexers for shortcodes and other main section
// elements.
sectionHandlers *sectionHandlers
cfg Config
// The summary divider to look for.
summaryDivider []byte
// Set when we have parsed any summary divider
@ -60,13 +66,17 @@ func (l *pageLexer) Input() []byte {
}
type Config struct {
EnableEmoji bool
}
// note: the input position here is normally 0 (start), but
// can be set if position of first shortcode is known
func newPageLexer(input []byte, inputPosition int, stateStart stateFunc) *pageLexer {
func newPageLexer(input []byte, stateStart stateFunc, cfg Config) *pageLexer {
lexer := &pageLexer{
input: input,
pos: inputPosition,
stateStart: stateStart,
cfg: cfg,
lexerShortcodeState: lexerShortcodeState{
currLeftDelimItem: tLeftDelimScNoMarkup,
currRightDelimItem: tRightDelimScNoMarkup,
@ -75,6 +85,8 @@ func newPageLexer(input []byte, inputPosition int, stateStart stateFunc) *pageLe
items: make([]Item, 0, 5),
}
lexer.sectionHandlers = createSectionHandlers(lexer)
return lexer
}
@ -100,6 +112,8 @@ var (
delimOrg = []byte("#+")
htmlCommentStart = []byte("<!--")
htmlCommentEnd = []byte("-->")
emojiDelim = byte(':')
)
func (l *pageLexer) next() rune {
@ -132,6 +146,10 @@ func (l *pageLexer) emit(t ItemType) {
l.start = l.pos
}
func (l *pageLexer) isEOF() bool {
return l.pos >= len(l.input)
}
// special case, do not send '\\' back to client
func (l *pageLexer) ignoreEscapesAndEmit(t ItemType) {
val := bytes.Map(func(r rune) rune {
@ -193,30 +211,80 @@ func (l *pageLexer) consumeSpace() {
}
}
func lexMainSection(l *pageLexer) stateFunc {
if l.isInHTMLComment {
return lexEndFromtMatterHTMLComment
}
// lex a string starting at ":"
func lexEmoji(l *pageLexer) stateFunc {
pos := l.pos + 1
valid := false
// Fast forward as far as possible.
var l1, l2 int
if !l.summaryDividerChecked && l.summaryDivider != nil {
l1 = l.index(l.summaryDivider)
if l1 == -1 {
l.summaryDividerChecked = true
for i := pos; i < len(l.input); i++ {
if i > pos && l.input[i] == emojiDelim {
pos = i + 1
valid = true
break
}
r, _ := utf8.DecodeRune(l.input[i:])
if !isAlphaNumeric(r) {
break
}
}
l2 = l.index(leftDelimSc)
skip := minIndex(l1, l2)
if skip > 0 {
l.pos += skip
if valid {
l.pos = pos
l.emit(TypeEmoji)
} else {
l.pos++
l.emit(tText)
}
for {
if l.isShortCodeStart() {
return lexMainSection
}
type sectionHandlers struct {
l *pageLexer
// Set when none of the sections are found so we
// can safely stop looking and skip to the end.
skipAll bool
handlers []*sectionHandler
skipIndexes []int
}
func (s *sectionHandlers) skip() int {
if s.skipAll {
return -1
}
s.skipIndexes = s.skipIndexes[:0]
var shouldSkip bool
for _, skipper := range s.handlers {
idx := skipper.skip()
if idx != -1 {
shouldSkip = true
s.skipIndexes = append(s.skipIndexes, idx)
}
}
if !shouldSkip {
s.skipAll = true
return -1
}
return minIndex(s.skipIndexes...)
}
func createSectionHandlers(l *pageLexer) *sectionHandlers {
shortCodeHandler := &sectionHandler{
l: l,
skipFunc: func(l *pageLexer) int {
return l.index(leftDelimSc)
},
lexFunc: func(origin stateFunc, l *pageLexer) (stateFunc, bool) {
if !l.isShortCodeStart() {
return origin, false
}
if l.isInline {
// If we're inside an inline shortcode, the only valid shortcode markup is
// the markup which closes it.
@ -225,14 +293,11 @@ func lexMainSection(l *pageLexer) stateFunc {
if end != len(l.input)-1 {
b = bytes.TrimSpace(b[end+1:])
if end == -1 || !bytes.HasPrefix(b, []byte(l.currShortcodeName+" ")) {
return l.errorf("inline shortcodes do not support nesting")
return l.errorf("inline shortcodes do not support nesting"), true
}
}
}
if l.pos > l.start {
l.emit(tText)
}
if l.hasPrefix(leftDelimScWithMarkup) {
l.currLeftDelimItem = tLeftDelimScWithMarkup
l.currRightDelimItem = tRightDelimScWithMarkup
@ -240,32 +305,139 @@ func lexMainSection(l *pageLexer) stateFunc {
l.currLeftDelimItem = tLeftDelimScNoMarkup
l.currRightDelimItem = tRightDelimScNoMarkup
}
return lexShortcodeLeftDelim
}
if !l.summaryDividerChecked && l.summaryDivider != nil {
if l.hasPrefix(l.summaryDivider) {
if l.pos > l.start {
l.emit(tText)
}
l.summaryDividerChecked = true
l.pos += len(l.summaryDivider)
// This makes it a little easier to reason about later.
l.consumeSpace()
l.emit(TypeLeadSummaryDivider)
// We have already moved to the next.
continue
}
}
r := l.next()
if r == eof {
break
}
return lexShortcodeLeftDelim, true
},
}
summaryDividerHandler := &sectionHandler{
l: l,
skipFunc: func(l *pageLexer) int {
if l.summaryDividerChecked || l.summaryDivider == nil {
return -1
}
return l.index(l.summaryDivider)
},
lexFunc: func(origin stateFunc, l *pageLexer) (stateFunc, bool) {
if !l.hasPrefix(l.summaryDivider) {
return origin, false
}
l.summaryDividerChecked = true
l.pos += len(l.summaryDivider)
// This makes it a little easier to reason about later.
l.consumeSpace()
l.emit(TypeLeadSummaryDivider)
return origin, true
},
}
handlers := []*sectionHandler{shortCodeHandler, summaryDividerHandler}
if l.cfg.EnableEmoji {
emojiHandler := &sectionHandler{
l: l,
skipFunc: func(l *pageLexer) int {
return l.indexByte(emojiDelim)
},
lexFunc: func(origin stateFunc, l *pageLexer) (stateFunc, bool) {
return lexEmoji, true
},
}
handlers = append(handlers, emojiHandler)
}
return &sectionHandlers{
l: l,
handlers: handlers,
skipIndexes: make([]int, len(handlers)),
}
}
func (s *sectionHandlers) lex(origin stateFunc) stateFunc {
if s.skipAll {
return nil
}
if s.l.pos > s.l.start {
s.l.emit(tText)
}
for _, handler := range s.handlers {
if handler.skipAll {
continue
}
next, handled := handler.lexFunc(origin, handler.l)
if next == nil || handled {
return next
}
}
// Not handled by the above.
s.l.pos++
return origin
}
type sectionHandler struct {
l *pageLexer
// No more sections of this type.
skipAll bool
// Returns the index of the next match, -1 if none found.
skipFunc func(l *pageLexer) int
// Lex lexes the current section and returns the next state func and
// a bool telling if this section was handled.
// Note that returning nil as the next state will terminate the
// lexer.
lexFunc func(origin stateFunc, l *pageLexer) (stateFunc, bool)
}
func (s *sectionHandler) skip() int {
if s.skipAll {
return -1
}
idx := s.skipFunc(s.l)
if idx == -1 {
s.skipAll = true
}
return idx
}
func lexMainSection(l *pageLexer) stateFunc {
if l.isEOF() {
return lexDone
}
if l.isInHTMLComment {
return lexEndFromtMatterHTMLComment
}
// Fast forward as far as possible.
skip := l.sectionHandlers.skip()
if skip == -1 {
l.pos = len(l.input)
return lexDone
} else if skip > 0 {
l.pos += skip
}
next := l.sectionHandlers.lex(lexMainSection)
if next != nil {
return next
}
l.pos = len(l.input)
return lexDone
}
@ -297,10 +469,22 @@ func (l *pageLexer) index(sep []byte) int {
return bytes.Index(l.input[l.pos:], sep)
}
func (l *pageLexer) indexByte(sep byte) int {
return bytes.IndexByte(l.input[l.pos:], sep)
}
func (l *pageLexer) hasPrefix(prefix []byte) bool {
return bytes.HasPrefix(l.input[l.pos:], prefix)
}
func (l *pageLexer) hasPrefixByte(prefix byte) bool {
b := l.input[l.pos:]
if len(b) == 0 {
return false
}
return b[0] == prefix
}
// helper functions
// returns the min index >= 0

View file

@ -27,7 +27,7 @@ import (
// Result holds the parse result.
type Result interface {
// Iterator returns a new Iterator positioned at the benning of the parse tree.
// Iterator returns a new Iterator positioned at the beginning of the parse tree.
Iterator() *Iterator
// Input returns the input to Parse.
Input() []byte
@ -35,27 +35,21 @@ type Result interface {
var _ Result = (*pageLexer)(nil)
// Parse parses the page in the given reader.
func Parse(r io.Reader) (Result, error) {
// Parse parses the page in the given reader according to the given Config.
func Parse(r io.Reader, cfg Config) (Result, error) {
b, err := ioutil.ReadAll(r)
if err != nil {
return nil, errors.Wrap(err, "failed to read page content")
}
return parseBytes(b)
return parseBytes(b, cfg)
}
func parseBytes(b []byte) (Result, error) {
lexer := newPageLexer(b, 0, lexIntroSection)
func parseBytes(b []byte, cfg Config) (Result, error) {
lexer := newPageLexer(b, lexIntroSection, cfg)
lexer.run()
return lexer, nil
}
func parseMainSection(input []byte, from int) Result {
lexer := newPageLexer(input, from, lexMainSection)
lexer.run()
return lexer
}
// An Iterator has methods to iterate a parsed page with support going back
// if needed.
type Iterator struct {

View file

@ -88,8 +88,8 @@ func TestFrontMatter(t *testing.T) {
}
}
func collect(input []byte, skipFrontMatter bool, stateStart stateFunc) (items []Item) {
l := newPageLexer(input, 0, stateStart)
func collectWithConfig(input []byte, skipFrontMatter bool, stateStart stateFunc, cfg Config) (items []Item) {
l := newPageLexer(input, stateStart, cfg)
l.run()
t := l.newIterator()
@ -103,6 +103,13 @@ func collect(input []byte, skipFrontMatter bool, stateStart stateFunc) (items []
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) {

View file

@ -0,0 +1,40 @@
// 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"
"testing"
)
func TestMain(t *testing.T) {
t.Parallel()
var mainTests = []lexerTest{
{"emoji #1", "Some text with :emoji:", []Item{nti(tText, "Some text with "), nti(TypeEmoji, ":emoji:"), tstEOF}},
{"emoji #2", "Some text with :emoji: and some text.", []Item{nti(tText, "Some text with "), nti(TypeEmoji, ":emoji:"), nti(tText, " and some text."), tstEOF}},
{"looks like an emoji #1", "Some text and then :emoji", []Item{nti(tText, "Some text and then "), nti(tText, ":"), nti(tText, "emoji"), tstEOF}},
{"looks like an emoji #2", "Some text and then ::", []Item{nti(tText, "Some text and then "), nti(tText, ":"), nti(tText, ":"), tstEOF}},
{"looks like an emoji #3", ":Some :text", []Item{nti(tText, ":"), nti(tText, "Some "), nti(tText, ":"), nti(tText, "text"), tstEOF}},
}
for i, test := range mainTests {
items := collectWithConfig([]byte(test.input), false, lexMainSection, Config{EnableEmoji: true})
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)
}
}
}

View file

@ -152,7 +152,8 @@ var shortCodeLexerTests = []lexerTest{
{"basic inline", `{{< sc1.inline >}}Hello World{{< /sc1.inline >}}`, []Item{tstLeftNoMD, tstSC1Inline, tstRightNoMD, tstText, tstLeftNoMD, tstSCClose, tstSC1Inline, tstRightNoMD, tstEOF}},
{"basic inline with space", `{{< sc1.inline >}}Hello World{{< / sc1.inline >}}`, []Item{tstLeftNoMD, tstSC1Inline, tstRightNoMD, tstText, tstLeftNoMD, tstSCClose, tstSC1Inline, tstRightNoMD, tstEOF}},
{"inline self closing", `{{< sc1.inline >}}Hello World{{< /sc1.inline >}}Hello World{{< sc1.inline />}}`, []Item{tstLeftNoMD, tstSC1Inline, tstRightNoMD, tstText, tstLeftNoMD, tstSCClose, tstSC1Inline, tstRightNoMD, tstText, tstLeftNoMD, tstSC1Inline, tstSCClose, tstRightNoMD, tstEOF}},
{"inline with nested shortcode (not supported)", `{{< sc1.inline >}}Hello World{{< sc1 >}}{{< /sc1.inline >}}`, []Item{tstLeftNoMD, tstSC1Inline, tstRightNoMD, nti(tError, "inline shortcodes do not support nesting")}},
{"inline with template syntax", `{{< sc1.inline >}}{{ .Get 0 }}{{ .Get 1 }}{{< /sc1.inline >}}`, []Item{tstLeftNoMD, tstSC1Inline, tstRightNoMD, nti(tText, "{{ .Get 0 }}"), nti(tText, "{{ .Get 1 }}"), tstLeftNoMD, tstSCClose, tstSC1Inline, tstRightNoMD, tstEOF}},
{"inline with nested shortcode (not supported)", `{{< sc1.inline >}}Hello World{{< sc1 >}}{{< /sc1.inline >}}`, []Item{tstLeftNoMD, tstSC1Inline, tstRightNoMD, tstText, nti(tError, "inline shortcodes do not support nesting")}},
{"inline case mismatch", `{{< sc1.Inline >}}Hello World{{< /sc1.Inline >}}`, []Item{tstLeftNoMD, nti(tError, "period in shortcode name only allowed for inline identifiers")}},
}
@ -171,10 +172,11 @@ func BenchmarkShortcodeLexer(b *testing.B) {
for i, input := range shortCodeLexerTests {
testInputs[i] = []byte(input.input)
}
var cfg Config
b.ResetTimer()
for i := 0; i < b.N; i++ {
for _, input := range testInputs {
items := collect(input, true, lexMainSection)
items := collectWithConfig(input, true, lexMainSection, cfg)
if len(items) == 0 {
}

View file

@ -34,10 +34,37 @@ This is some summary. This is some summary. This is some summary. This is some s
`
input := []byte(start + strings.Repeat(strings.Repeat("this is text", 30)+"{{< myshortcode >}}This is some inner content.{{< /myshortcode >}}", 10))
cfg := Config{EnableEmoji: false}
b.ResetTimer()
for i := 0; i < b.N; i++ {
if _, err := parseBytes(input); err != nil {
if _, err := parseBytes(input, cfg); err != nil {
b.Fatal(err)
}
}
}
func BenchmarkParseWithEmoji(b *testing.B) {
start := `
---
title: "Front Matters"
description: "It really does"
---
This is some summary. This is some summary. This is some summary. This is some summary.
<!--more-->
`
input := []byte(start + strings.Repeat("this is not emoji: ", 50) + strings.Repeat("some text ", 70) + strings.Repeat("this is not: ", 50) + strings.Repeat("but this is a :smile: ", 3) + strings.Repeat("some text ", 70))
cfg := Config{EnableEmoji: true}
b.ResetTimer()
for i := 0; i < b.N; i++ {
if _, err := parseBytes(input, cfg); err != nil {
b.Fatal(err)
}
}