Git with a cup of tea, painless self-hosted git service Mirror for internal git.with.parts use https://git.with.parts
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
gitea/modules/markup/html.go

446 lines
14 KiB

// Copyright 2017 The Gitea Authors. All rights reserved.
// SPDX-License-Identifier: MIT
package markup
import (
"bytes"
"io"
"regexp"
"slices"
"strings"
"sync"
"code.gitea.io/gitea/modules/markup/common"
"golang.org/x/net/html"
"golang.org/x/net/html/atom"
"mvdan.cc/xurls/v2"
)
// Issue name styles
const (
IssueNameStyleNumeric = "numeric"
IssueNameStyleAlphanumeric = "alphanumeric"
IssueNameStyleRegexp = "regexp"
)
type globalVarsType struct {
hashCurrentPattern *regexp.Regexp
shortLinkPattern *regexp.Regexp
anyHashPattern *regexp.Regexp
comparePattern *regexp.Regexp
fullURLPattern *regexp.Regexp
emailRegex *regexp.Regexp
blackfridayExtRegex *regexp.Regexp
emojiShortCodeRegex *regexp.Regexp
issueFullPattern *regexp.Regexp
filesChangedFullPattern *regexp.Regexp
codePreviewPattern *regexp.Regexp
tagCleaner *regexp.Regexp
nulCleaner *strings.Replacer
}
var globalVars = sync.OnceValue[*globalVarsType](func() *globalVarsType {
v := &globalVarsType{}
// NOTE: All below regex matching do not perform any extra validation.
// Thus a link is produced even if the linked entity does not exist.
// While fast, this is also incorrect and lead to false positives.
// TODO: fix invalid linking issue
// valid chars in encoded path and parameter: [-+~_%.a-zA-Z0-9/]
// hashCurrentPattern matches string that represents a commit SHA, e.g. d8a994ef243349f321568f9e36d5c3f444b99cae
// Although SHA1 hashes are 40 chars long, SHA256 are 64, the regex matches the hash from 7 to 64 chars in length
// so that abbreviated hash links can be used as well. This matches git and GitHub usability.
v.hashCurrentPattern = regexp.MustCompile(`(?:\s|^|\(|\[)([0-9a-f]{7,64})(?:\s|$|\)|\]|[.,:](\s|$))`)
// shortLinkPattern matches short but difficult to parse [[name|link|arg=test]] syntax
v.shortLinkPattern = regexp.MustCompile(`\[\[(.*?)\]\](\w*)`)
// anyHashPattern splits url containing SHA into parts
v.anyHashPattern = regexp.MustCompile(`https?://(?:\S+/){4,5}([0-9a-f]{40,64})(/[-+~%./\w]+)?(\?[-+~%.\w&=]+)?(#[-+~%.\w]+)?`)
// comparePattern matches "http://domain/org/repo/compare/COMMIT1...COMMIT2#hash"
v.comparePattern = regexp.MustCompile(`https?://(?:\S+/){4,5}([0-9a-f]{7,64})(\.\.\.?)([0-9a-f]{7,64})?(#[-+~_%.a-zA-Z0-9]+)?`)
// fullURLPattern matches full URL like "mailto:...", "https://..." and "ssh+git://..."
v.fullURLPattern = regexp.MustCompile(`^[a-z][-+\w]+:`)
// emailRegex is definitely not perfect with edge cases,
// it is still accepted by the CommonMark specification, as well as the HTML5 spec:
// http://spec.commonmark.org/0.28/#email-address
// https://html.spec.whatwg.org/multipage/input.html#e-mail-state-(type%3Demail)
v.emailRegex = regexp.MustCompile("(?:\\s|^|\\(|\\[)([a-zA-Z0-9.!#$%&'*+\\/=?^_`{|}~-]+@[a-zA-Z0-9](?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?(?:\\.[a-zA-Z0-9]{2,}(?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?)+)(?:\\s|$|\\)|\\]|;|,|\\?|!|\\.(\\s|$))")
// blackfridayExtRegex is for blackfriday extensions create IDs like fn:user-content-footnote
v.blackfridayExtRegex = regexp.MustCompile(`[^:]*:user-content-`)
Support unicode emojis and remove emojify.js (#11032) * Support unicode emojis and remove emojify.js This PR replaces all use of emojify.js and adds unicode emoji support to various areas of gitea. This works in a few ways: First it adds emoji parsing support into gitea itself. This allows us to * Render emojis from valid alias (:smile:) * Detect unicode emojis and let us put them in their own class with proper aria-labels and styling * Easily allow for custom "emoji" * Support all emoji rendering and features without javascript * Uses plain unicode and lets the system render in appropriate emoji font * Doesn't leave us relying on external sources for updates/fixes/features That same list of emoji is also used to create a json file which replaces the part of emojify.js that populates the emoji search tribute. This file is about 35KB with GZIP turned on and I've set it to load after the page renders to not hinder page load time (and this removes loading emojify.js also) For custom "emoji" it uses a pretty simple scheme of just looking for /emojis/img/name.png where name is something a user has put in the "allowed reactions" setting we already have. The gitea reaction that was previously hard coded into a forked copy of emojify.js is included and works as a custom reaction under this method. The emoji data sourced here is from https://github.com/github/gemoji which is the gem library Github uses for their emoji rendering (and a data source for other sites). So we should be able to easily render any emoji and :alias: that Github can, removing any errors from migrated content. They also update it as well, so we can sync when there are new unicode emoji lists released. I've included a slimmed down and slightly modified forked copy of https://github.com/knq/emoji to make up our own emoji module. The code is pretty straight forward and again allows us to have a lot of flexibility in what happens. I had seen a few comments about performance in some of the other threads if we render this ourselves, but there doesn't seem to be any issue here. In a test it can parse, convert, and render 1,000 emojis inside of a large markdown table in about 100ms on my laptop (which is many more emojis than will ever be in any normal issue). This also prevents any flickering and other weirdness from using javascript to render some things while using go for others. Not included here are image fall back URLS. I don't really think they are necessary for anything new being written in 2020. However, managing the emoji ourselves would allow us to add these as a feature later on if it seems necessary. Fixes: https://github.com/go-gitea/gitea/issues/9182 Fixes: https://github.com/go-gitea/gitea/issues/8974 Fixes: https://github.com/go-gitea/gitea/issues/8953 Fixes: https://github.com/go-gitea/gitea/issues/6628 Fixes: https://github.com/go-gitea/gitea/issues/5130 * add new shared function emojiHTML * don't increase emoji size in issue title * Update templates/repo/issue/view_content/add_reaction.tmpl Co-Authored-By: 6543 <6543@obermui.de> * Support for emoji rendering in various templates * Render code and review comments as they should be * Better way to handle mail subjects * insert unicode from tribute selection * Add template helper for plain text when needed * Use existing replace function I forgot about * Don't include emoji greater than Unicode Version 12 Only include emoji and aliases in JSON * Update build/generate-emoji.go * Tweak regex slightly to really match everything including random invisible characters. Run tests for every emoji we have * final updates * code review * code review * hard code gitea custom emoji to match previous behavior * Update .eslintrc Co-Authored-By: silverwind <me@silverwind.io> * disable preempt Co-authored-by: silverwind <me@silverwind.io> Co-authored-by: 6543 <6543@obermui.de> Co-authored-by: Lauris BH <lauris@nix.lv> Co-authored-by: guillep2k <18600385+guillep2k@users.noreply.github.com>
5 years ago
// emojiShortCodeRegex find emoji by alias like :smile:
v.emojiShortCodeRegex = regexp.MustCompile(`:[-+\w]+:`)
// example: https://domain/org/repo/pulls/27#hash
v.issueFullPattern = regexp.MustCompile(`https?://(?:\S+/)[\w_.-]+/[\w_.-]+/(?:issues|pulls)/((?:\w{1,10}-)?[1-9][0-9]*)([\?|#](\S+)?)?\b`)
// example: https://domain/org/repo/pulls/27/files#hash
v.filesChangedFullPattern = regexp.MustCompile(`https?://(?:\S+/)[\w_.-]+/[\w_.-]+/pulls/((?:\w{1,10}-)?[1-9][0-9]*)/files([\?|#](\S+)?)?\b`)
// codePreviewPattern matches "http://domain/.../{owner}/{repo}/src/commit/{commit}/{filepath}#L10-L20"
v.codePreviewPattern = regexp.MustCompile(`https?://\S+/([^\s/]+)/([^\s/]+)/src/commit/([0-9a-f]{7,64})(/\S+)#(L\d+(-L\d+)?)`)
v.tagCleaner = regexp.MustCompile(`<((?:/?\w+/\w+)|(?:/[\w ]+/)|(/?[hH][tT][mM][lL]\b)|(/?[hH][eE][aA][dD]\b))`)
v.nulCleaner = strings.NewReplacer("\000", "")
return v
})
// IsFullURLBytes reports whether link fits valid format.
func IsFullURLBytes(link []byte) bool {
return globalVars().fullURLPattern.Match(link)
}
func IsFullURLString(link string) bool {
return globalVars().fullURLPattern.MatchString(link)
}
func IsNonEmptyRelativePath(link string) bool {
return link != "" && !IsFullURLString(link) && link[0] != '/' && link[0] != '?' && link[0] != '#'
}
// CustomLinkURLSchemes allows for additional schemes to be detected when parsing links within text
func CustomLinkURLSchemes(schemes []string) {
schemes = append(schemes, "http", "https")
withAuth := make([]string, 0, len(schemes))
validScheme := regexp.MustCompile(`^[a-z]+$`)
for _, s := range schemes {
if !validScheme.MatchString(s) {
continue
}
without := false
for _, sna := range xurls.SchemesNoAuthority {
if s == sna {
without = true
break
}
}
if without {
s += ":"
} else {
s += "://"
}
withAuth = append(withAuth, s)
}
common.GlobalVars().LinkRegex, _ = xurls.StrictMatchingScheme(strings.Join(withAuth, "|"))
}
type postProcessError struct {
context string
err error
}
func (p *postProcessError) Error() string {
return "PostProcess: " + p.context + ", " + p.err.Error()
}
type processor func(ctx *RenderContext, node *html.Node)
var defaultProcessors = []processor{
fullIssuePatternProcessor,
comparePatternProcessor,
codePreviewPatternProcessor,
fullHashPatternProcessor,
shortLinkProcessor,
linkProcessor,
mentionProcessor,
issueIndexPatternProcessor,
commitCrossReferencePatternProcessor,
hashCurrentPatternProcessor,
emailAddressProcessor,
Support unicode emojis and remove emojify.js (#11032) * Support unicode emojis and remove emojify.js This PR replaces all use of emojify.js and adds unicode emoji support to various areas of gitea. This works in a few ways: First it adds emoji parsing support into gitea itself. This allows us to * Render emojis from valid alias (:smile:) * Detect unicode emojis and let us put them in their own class with proper aria-labels and styling * Easily allow for custom "emoji" * Support all emoji rendering and features without javascript * Uses plain unicode and lets the system render in appropriate emoji font * Doesn't leave us relying on external sources for updates/fixes/features That same list of emoji is also used to create a json file which replaces the part of emojify.js that populates the emoji search tribute. This file is about 35KB with GZIP turned on and I've set it to load after the page renders to not hinder page load time (and this removes loading emojify.js also) For custom "emoji" it uses a pretty simple scheme of just looking for /emojis/img/name.png where name is something a user has put in the "allowed reactions" setting we already have. The gitea reaction that was previously hard coded into a forked copy of emojify.js is included and works as a custom reaction under this method. The emoji data sourced here is from https://github.com/github/gemoji which is the gem library Github uses for their emoji rendering (and a data source for other sites). So we should be able to easily render any emoji and :alias: that Github can, removing any errors from migrated content. They also update it as well, so we can sync when there are new unicode emoji lists released. I've included a slimmed down and slightly modified forked copy of https://github.com/knq/emoji to make up our own emoji module. The code is pretty straight forward and again allows us to have a lot of flexibility in what happens. I had seen a few comments about performance in some of the other threads if we render this ourselves, but there doesn't seem to be any issue here. In a test it can parse, convert, and render 1,000 emojis inside of a large markdown table in about 100ms on my laptop (which is many more emojis than will ever be in any normal issue). This also prevents any flickering and other weirdness from using javascript to render some things while using go for others. Not included here are image fall back URLS. I don't really think they are necessary for anything new being written in 2020. However, managing the emoji ourselves would allow us to add these as a feature later on if it seems necessary. Fixes: https://github.com/go-gitea/gitea/issues/9182 Fixes: https://github.com/go-gitea/gitea/issues/8974 Fixes: https://github.com/go-gitea/gitea/issues/8953 Fixes: https://github.com/go-gitea/gitea/issues/6628 Fixes: https://github.com/go-gitea/gitea/issues/5130 * add new shared function emojiHTML * don't increase emoji size in issue title * Update templates/repo/issue/view_content/add_reaction.tmpl Co-Authored-By: 6543 <6543@obermui.de> * Support for emoji rendering in various templates * Render code and review comments as they should be * Better way to handle mail subjects * insert unicode from tribute selection * Add template helper for plain text when needed * Use existing replace function I forgot about * Don't include emoji greater than Unicode Version 12 Only include emoji and aliases in JSON * Update build/generate-emoji.go * Tweak regex slightly to really match everything including random invisible characters. Run tests for every emoji we have * final updates * code review * code review * hard code gitea custom emoji to match previous behavior * Update .eslintrc Co-Authored-By: silverwind <me@silverwind.io> * disable preempt Co-authored-by: silverwind <me@silverwind.io> Co-authored-by: 6543 <6543@obermui.de> Co-authored-by: Lauris BH <lauris@nix.lv> Co-authored-by: guillep2k <18600385+guillep2k@users.noreply.github.com>
5 years ago
emojiProcessor,
emojiShortCodeProcessor,
}
// PostProcess does the final required transformations to the passed raw HTML
// data, and ensures its validity. Transformations include: replacing links and
// emails with HTML links, parsing shortlinks in the format of [[Link]], like
// MediaWiki, linking issues in the format #ID, and mentions in the format
// @user, and others.
func PostProcess(ctx *RenderContext, input io.Reader, output io.Writer) error {
return postProcess(ctx, defaultProcessors, input, output)
}
var commitMessageProcessors = []processor{
fullIssuePatternProcessor,
comparePatternProcessor,
fullHashPatternProcessor,
linkProcessor,
mentionProcessor,
issueIndexPatternProcessor,
commitCrossReferencePatternProcessor,
hashCurrentPatternProcessor,
emailAddressProcessor,
Support unicode emojis and remove emojify.js (#11032) * Support unicode emojis and remove emojify.js This PR replaces all use of emojify.js and adds unicode emoji support to various areas of gitea. This works in a few ways: First it adds emoji parsing support into gitea itself. This allows us to * Render emojis from valid alias (:smile:) * Detect unicode emojis and let us put them in their own class with proper aria-labels and styling * Easily allow for custom "emoji" * Support all emoji rendering and features without javascript * Uses plain unicode and lets the system render in appropriate emoji font * Doesn't leave us relying on external sources for updates/fixes/features That same list of emoji is also used to create a json file which replaces the part of emojify.js that populates the emoji search tribute. This file is about 35KB with GZIP turned on and I've set it to load after the page renders to not hinder page load time (and this removes loading emojify.js also) For custom "emoji" it uses a pretty simple scheme of just looking for /emojis/img/name.png where name is something a user has put in the "allowed reactions" setting we already have. The gitea reaction that was previously hard coded into a forked copy of emojify.js is included and works as a custom reaction under this method. The emoji data sourced here is from https://github.com/github/gemoji which is the gem library Github uses for their emoji rendering (and a data source for other sites). So we should be able to easily render any emoji and :alias: that Github can, removing any errors from migrated content. They also update it as well, so we can sync when there are new unicode emoji lists released. I've included a slimmed down and slightly modified forked copy of https://github.com/knq/emoji to make up our own emoji module. The code is pretty straight forward and again allows us to have a lot of flexibility in what happens. I had seen a few comments about performance in some of the other threads if we render this ourselves, but there doesn't seem to be any issue here. In a test it can parse, convert, and render 1,000 emojis inside of a large markdown table in about 100ms on my laptop (which is many more emojis than will ever be in any normal issue). This also prevents any flickering and other weirdness from using javascript to render some things while using go for others. Not included here are image fall back URLS. I don't really think they are necessary for anything new being written in 2020. However, managing the emoji ourselves would allow us to add these as a feature later on if it seems necessary. Fixes: https://github.com/go-gitea/gitea/issues/9182 Fixes: https://github.com/go-gitea/gitea/issues/8974 Fixes: https://github.com/go-gitea/gitea/issues/8953 Fixes: https://github.com/go-gitea/gitea/issues/6628 Fixes: https://github.com/go-gitea/gitea/issues/5130 * add new shared function emojiHTML * don't increase emoji size in issue title * Update templates/repo/issue/view_content/add_reaction.tmpl Co-Authored-By: 6543 <6543@obermui.de> * Support for emoji rendering in various templates * Render code and review comments as they should be * Better way to handle mail subjects * insert unicode from tribute selection * Add template helper for plain text when needed * Use existing replace function I forgot about * Don't include emoji greater than Unicode Version 12 Only include emoji and aliases in JSON * Update build/generate-emoji.go * Tweak regex slightly to really match everything including random invisible characters. Run tests for every emoji we have * final updates * code review * code review * hard code gitea custom emoji to match previous behavior * Update .eslintrc Co-Authored-By: silverwind <me@silverwind.io> * disable preempt Co-authored-by: silverwind <me@silverwind.io> Co-authored-by: 6543 <6543@obermui.de> Co-authored-by: Lauris BH <lauris@nix.lv> Co-authored-by: guillep2k <18600385+guillep2k@users.noreply.github.com>
5 years ago
emojiProcessor,
emojiShortCodeProcessor,
}
// RenderCommitMessage will use the same logic as PostProcess, but will disable
// the shortLinkProcessor and will add a defaultLinkProcessor if defaultLink is
// set, which changes every text node into a link to the passed default link.
func RenderCommitMessage(ctx *RenderContext, content string) (string, error) {
procs := commitMessageProcessors
return renderProcessString(ctx, procs, content)
}
var commitMessageSubjectProcessors = []processor{
fullIssuePatternProcessor,
comparePatternProcessor,
fullHashPatternProcessor,
linkProcessor,
mentionProcessor,
issueIndexPatternProcessor,
commitCrossReferencePatternProcessor,
hashCurrentPatternProcessor,
Support unicode emojis and remove emojify.js (#11032) * Support unicode emojis and remove emojify.js This PR replaces all use of emojify.js and adds unicode emoji support to various areas of gitea. This works in a few ways: First it adds emoji parsing support into gitea itself. This allows us to * Render emojis from valid alias (:smile:) * Detect unicode emojis and let us put them in their own class with proper aria-labels and styling * Easily allow for custom "emoji" * Support all emoji rendering and features without javascript * Uses plain unicode and lets the system render in appropriate emoji font * Doesn't leave us relying on external sources for updates/fixes/features That same list of emoji is also used to create a json file which replaces the part of emojify.js that populates the emoji search tribute. This file is about 35KB with GZIP turned on and I've set it to load after the page renders to not hinder page load time (and this removes loading emojify.js also) For custom "emoji" it uses a pretty simple scheme of just looking for /emojis/img/name.png where name is something a user has put in the "allowed reactions" setting we already have. The gitea reaction that was previously hard coded into a forked copy of emojify.js is included and works as a custom reaction under this method. The emoji data sourced here is from https://github.com/github/gemoji which is the gem library Github uses for their emoji rendering (and a data source for other sites). So we should be able to easily render any emoji and :alias: that Github can, removing any errors from migrated content. They also update it as well, so we can sync when there are new unicode emoji lists released. I've included a slimmed down and slightly modified forked copy of https://github.com/knq/emoji to make up our own emoji module. The code is pretty straight forward and again allows us to have a lot of flexibility in what happens. I had seen a few comments about performance in some of the other threads if we render this ourselves, but there doesn't seem to be any issue here. In a test it can parse, convert, and render 1,000 emojis inside of a large markdown table in about 100ms on my laptop (which is many more emojis than will ever be in any normal issue). This also prevents any flickering and other weirdness from using javascript to render some things while using go for others. Not included here are image fall back URLS. I don't really think they are necessary for anything new being written in 2020. However, managing the emoji ourselves would allow us to add these as a feature later on if it seems necessary. Fixes: https://github.com/go-gitea/gitea/issues/9182 Fixes: https://github.com/go-gitea/gitea/issues/8974 Fixes: https://github.com/go-gitea/gitea/issues/8953 Fixes: https://github.com/go-gitea/gitea/issues/6628 Fixes: https://github.com/go-gitea/gitea/issues/5130 * add new shared function emojiHTML * don't increase emoji size in issue title * Update templates/repo/issue/view_content/add_reaction.tmpl Co-Authored-By: 6543 <6543@obermui.de> * Support for emoji rendering in various templates * Render code and review comments as they should be * Better way to handle mail subjects * insert unicode from tribute selection * Add template helper for plain text when needed * Use existing replace function I forgot about * Don't include emoji greater than Unicode Version 12 Only include emoji and aliases in JSON * Update build/generate-emoji.go * Tweak regex slightly to really match everything including random invisible characters. Run tests for every emoji we have * final updates * code review * code review * hard code gitea custom emoji to match previous behavior * Update .eslintrc Co-Authored-By: silverwind <me@silverwind.io> * disable preempt Co-authored-by: silverwind <me@silverwind.io> Co-authored-by: 6543 <6543@obermui.de> Co-authored-by: Lauris BH <lauris@nix.lv> Co-authored-by: guillep2k <18600385+guillep2k@users.noreply.github.com>
5 years ago
emojiShortCodeProcessor,
emojiProcessor,
}
var emojiProcessors = []processor{
emojiShortCodeProcessor,
emojiProcessor,
}
// RenderCommitMessageSubject will use the same logic as PostProcess and
// RenderCommitMessage, but will disable the shortLinkProcessor and
// emailAddressProcessor, will add a defaultLinkProcessor if defaultLink is set,
// which changes every text node into a link to the passed default link.
func RenderCommitMessageSubject(ctx *RenderContext, defaultLink, content string) (string, error) {
procs := slices.Clone(commitMessageSubjectProcessors)
procs = append(procs, func(ctx *RenderContext, node *html.Node) {
ch := &html.Node{Parent: node, Type: html.TextNode, Data: node.Data}
node.Type = html.ElementNode
node.Data = "a"
node.DataAtom = atom.A
node.Attr = []html.Attribute{{Key: "href", Val: defaultLink}, {Key: "class", Val: "muted"}}
node.FirstChild, node.LastChild = ch, ch
})
return renderProcessString(ctx, procs, content)
}
// RenderIssueTitle to process title on individual issue/pull page
func RenderIssueTitle(ctx *RenderContext, title string) (string, error) {
// do not render other issue/commit links in an issue's title - which in most cases is already a link.
return renderProcessString(ctx, []processor{
emojiShortCodeProcessor,
emojiProcessor,
}, title)
}
func renderProcessString(ctx *RenderContext, procs []processor, content string) (string, error) {
var buf strings.Builder
if err := postProcess(ctx, procs, strings.NewReader(content), &buf); err != nil {
return "", err
}
return buf.String(), nil
}
// RenderDescriptionHTML will use similar logic as PostProcess, but will
// use a single special linkProcessor.
func RenderDescriptionHTML(ctx *RenderContext, content string) (string, error) {
return renderProcessString(ctx, []processor{
descriptionLinkProcessor,
emojiShortCodeProcessor,
emojiProcessor,
}, content)
}
Support unicode emojis and remove emojify.js (#11032) * Support unicode emojis and remove emojify.js This PR replaces all use of emojify.js and adds unicode emoji support to various areas of gitea. This works in a few ways: First it adds emoji parsing support into gitea itself. This allows us to * Render emojis from valid alias (:smile:) * Detect unicode emojis and let us put them in their own class with proper aria-labels and styling * Easily allow for custom "emoji" * Support all emoji rendering and features without javascript * Uses plain unicode and lets the system render in appropriate emoji font * Doesn't leave us relying on external sources for updates/fixes/features That same list of emoji is also used to create a json file which replaces the part of emojify.js that populates the emoji search tribute. This file is about 35KB with GZIP turned on and I've set it to load after the page renders to not hinder page load time (and this removes loading emojify.js also) For custom "emoji" it uses a pretty simple scheme of just looking for /emojis/img/name.png where name is something a user has put in the "allowed reactions" setting we already have. The gitea reaction that was previously hard coded into a forked copy of emojify.js is included and works as a custom reaction under this method. The emoji data sourced here is from https://github.com/github/gemoji which is the gem library Github uses for their emoji rendering (and a data source for other sites). So we should be able to easily render any emoji and :alias: that Github can, removing any errors from migrated content. They also update it as well, so we can sync when there are new unicode emoji lists released. I've included a slimmed down and slightly modified forked copy of https://github.com/knq/emoji to make up our own emoji module. The code is pretty straight forward and again allows us to have a lot of flexibility in what happens. I had seen a few comments about performance in some of the other threads if we render this ourselves, but there doesn't seem to be any issue here. In a test it can parse, convert, and render 1,000 emojis inside of a large markdown table in about 100ms on my laptop (which is many more emojis than will ever be in any normal issue). This also prevents any flickering and other weirdness from using javascript to render some things while using go for others. Not included here are image fall back URLS. I don't really think they are necessary for anything new being written in 2020. However, managing the emoji ourselves would allow us to add these as a feature later on if it seems necessary. Fixes: https://github.com/go-gitea/gitea/issues/9182 Fixes: https://github.com/go-gitea/gitea/issues/8974 Fixes: https://github.com/go-gitea/gitea/issues/8953 Fixes: https://github.com/go-gitea/gitea/issues/6628 Fixes: https://github.com/go-gitea/gitea/issues/5130 * add new shared function emojiHTML * don't increase emoji size in issue title * Update templates/repo/issue/view_content/add_reaction.tmpl Co-Authored-By: 6543 <6543@obermui.de> * Support for emoji rendering in various templates * Render code and review comments as they should be * Better way to handle mail subjects * insert unicode from tribute selection * Add template helper for plain text when needed * Use existing replace function I forgot about * Don't include emoji greater than Unicode Version 12 Only include emoji and aliases in JSON * Update build/generate-emoji.go * Tweak regex slightly to really match everything including random invisible characters. Run tests for every emoji we have * final updates * code review * code review * hard code gitea custom emoji to match previous behavior * Update .eslintrc Co-Authored-By: silverwind <me@silverwind.io> * disable preempt Co-authored-by: silverwind <me@silverwind.io> Co-authored-by: 6543 <6543@obermui.de> Co-authored-by: Lauris BH <lauris@nix.lv> Co-authored-by: guillep2k <18600385+guillep2k@users.noreply.github.com>
5 years ago
// RenderEmoji for when we want to just process emoji and shortcodes
// in various places it isn't already run through the normal markdown processor
func RenderEmoji(ctx *RenderContext, content string) (string, error) {
return renderProcessString(ctx, emojiProcessors, content)
Support unicode emojis and remove emojify.js (#11032) * Support unicode emojis and remove emojify.js This PR replaces all use of emojify.js and adds unicode emoji support to various areas of gitea. This works in a few ways: First it adds emoji parsing support into gitea itself. This allows us to * Render emojis from valid alias (:smile:) * Detect unicode emojis and let us put them in their own class with proper aria-labels and styling * Easily allow for custom "emoji" * Support all emoji rendering and features without javascript * Uses plain unicode and lets the system render in appropriate emoji font * Doesn't leave us relying on external sources for updates/fixes/features That same list of emoji is also used to create a json file which replaces the part of emojify.js that populates the emoji search tribute. This file is about 35KB with GZIP turned on and I've set it to load after the page renders to not hinder page load time (and this removes loading emojify.js also) For custom "emoji" it uses a pretty simple scheme of just looking for /emojis/img/name.png where name is something a user has put in the "allowed reactions" setting we already have. The gitea reaction that was previously hard coded into a forked copy of emojify.js is included and works as a custom reaction under this method. The emoji data sourced here is from https://github.com/github/gemoji which is the gem library Github uses for their emoji rendering (and a data source for other sites). So we should be able to easily render any emoji and :alias: that Github can, removing any errors from migrated content. They also update it as well, so we can sync when there are new unicode emoji lists released. I've included a slimmed down and slightly modified forked copy of https://github.com/knq/emoji to make up our own emoji module. The code is pretty straight forward and again allows us to have a lot of flexibility in what happens. I had seen a few comments about performance in some of the other threads if we render this ourselves, but there doesn't seem to be any issue here. In a test it can parse, convert, and render 1,000 emojis inside of a large markdown table in about 100ms on my laptop (which is many more emojis than will ever be in any normal issue). This also prevents any flickering and other weirdness from using javascript to render some things while using go for others. Not included here are image fall back URLS. I don't really think they are necessary for anything new being written in 2020. However, managing the emoji ourselves would allow us to add these as a feature later on if it seems necessary. Fixes: https://github.com/go-gitea/gitea/issues/9182 Fixes: https://github.com/go-gitea/gitea/issues/8974 Fixes: https://github.com/go-gitea/gitea/issues/8953 Fixes: https://github.com/go-gitea/gitea/issues/6628 Fixes: https://github.com/go-gitea/gitea/issues/5130 * add new shared function emojiHTML * don't increase emoji size in issue title * Update templates/repo/issue/view_content/add_reaction.tmpl Co-Authored-By: 6543 <6543@obermui.de> * Support for emoji rendering in various templates * Render code and review comments as they should be * Better way to handle mail subjects * insert unicode from tribute selection * Add template helper for plain text when needed * Use existing replace function I forgot about * Don't include emoji greater than Unicode Version 12 Only include emoji and aliases in JSON * Update build/generate-emoji.go * Tweak regex slightly to really match everything including random invisible characters. Run tests for every emoji we have * final updates * code review * code review * hard code gitea custom emoji to match previous behavior * Update .eslintrc Co-Authored-By: silverwind <me@silverwind.io> * disable preempt Co-authored-by: silverwind <me@silverwind.io> Co-authored-by: 6543 <6543@obermui.de> Co-authored-by: Lauris BH <lauris@nix.lv> Co-authored-by: guillep2k <18600385+guillep2k@users.noreply.github.com>
5 years ago
}
func postProcess(ctx *RenderContext, procs []processor, input io.Reader, output io.Writer) error {
defer ctx.Cancel()
// FIXME: don't read all content to memory
rawHTML, err := io.ReadAll(input)
if err != nil {
return err
}
// parse the HTML
node, err := html.Parse(io.MultiReader(
// prepend "<html><body>"
strings.NewReader("<html><body>"),
// Strip out nuls - they're always invalid
bytes.NewReader(globalVars().tagCleaner.ReplaceAll([]byte(globalVars().nulCleaner.Replace(string(rawHTML))), []byte("&lt;$1"))),
// close the tags
strings.NewReader("</body></html>"),
))
if err != nil {
return &postProcessError{"invalid HTML", err}
}
if node.Type == html.DocumentNode {
node = node.FirstChild
}
visitNode(ctx, procs, node)
newNodes := make([]*html.Node, 0, 5)
if node.Data == "html" {
node = node.FirstChild
for node != nil && node.Data != "body" {
node = node.NextSibling
}
}
if node != nil {
if node.Data == "body" {
child := node.FirstChild
for child != nil {
newNodes = append(newNodes, child)
child = child.NextSibling
}
} else {
newNodes = append(newNodes, node)
}
}
// Render everything to buf.
for _, node := range newNodes {
3 years ago
if err := html.Render(output, node); err != nil {
return &postProcessError{"error rendering processed HTML", err}
}
}
return nil
}
func isEmojiNode(node *html.Node) bool {
if node.Type == html.ElementNode && node.Data == atom.Span.String() {
for _, attr := range node.Attr {
if (attr.Key == "class" || attr.Key == "data-attr-class") && strings.Contains(attr.Val, "emoji") {
return true
}
}
}
return false
}
func visitNode(ctx *RenderContext, procs []processor, node *html.Node) *html.Node {
// Add user-content- to IDs and "#" links if they don't already have them
for idx, attr := range node.Attr {
val := strings.TrimPrefix(attr.Val, "#")
notHasPrefix := !(strings.HasPrefix(val, "user-content-") || globalVars().blackfridayExtRegex.MatchString(val))
if attr.Key == "id" && notHasPrefix {
node.Attr[idx].Val = "user-content-" + attr.Val
}
Support unicode emojis and remove emojify.js (#11032) * Support unicode emojis and remove emojify.js This PR replaces all use of emojify.js and adds unicode emoji support to various areas of gitea. This works in a few ways: First it adds emoji parsing support into gitea itself. This allows us to * Render emojis from valid alias (:smile:) * Detect unicode emojis and let us put them in their own class with proper aria-labels and styling * Easily allow for custom "emoji" * Support all emoji rendering and features without javascript * Uses plain unicode and lets the system render in appropriate emoji font * Doesn't leave us relying on external sources for updates/fixes/features That same list of emoji is also used to create a json file which replaces the part of emojify.js that populates the emoji search tribute. This file is about 35KB with GZIP turned on and I've set it to load after the page renders to not hinder page load time (and this removes loading emojify.js also) For custom "emoji" it uses a pretty simple scheme of just looking for /emojis/img/name.png where name is something a user has put in the "allowed reactions" setting we already have. The gitea reaction that was previously hard coded into a forked copy of emojify.js is included and works as a custom reaction under this method. The emoji data sourced here is from https://github.com/github/gemoji which is the gem library Github uses for their emoji rendering (and a data source for other sites). So we should be able to easily render any emoji and :alias: that Github can, removing any errors from migrated content. They also update it as well, so we can sync when there are new unicode emoji lists released. I've included a slimmed down and slightly modified forked copy of https://github.com/knq/emoji to make up our own emoji module. The code is pretty straight forward and again allows us to have a lot of flexibility in what happens. I had seen a few comments about performance in some of the other threads if we render this ourselves, but there doesn't seem to be any issue here. In a test it can parse, convert, and render 1,000 emojis inside of a large markdown table in about 100ms on my laptop (which is many more emojis than will ever be in any normal issue). This also prevents any flickering and other weirdness from using javascript to render some things while using go for others. Not included here are image fall back URLS. I don't really think they are necessary for anything new being written in 2020. However, managing the emoji ourselves would allow us to add these as a feature later on if it seems necessary. Fixes: https://github.com/go-gitea/gitea/issues/9182 Fixes: https://github.com/go-gitea/gitea/issues/8974 Fixes: https://github.com/go-gitea/gitea/issues/8953 Fixes: https://github.com/go-gitea/gitea/issues/6628 Fixes: https://github.com/go-gitea/gitea/issues/5130 * add new shared function emojiHTML * don't increase emoji size in issue title * Update templates/repo/issue/view_content/add_reaction.tmpl Co-Authored-By: 6543 <6543@obermui.de> * Support for emoji rendering in various templates * Render code and review comments as they should be * Better way to handle mail subjects * insert unicode from tribute selection * Add template helper for plain text when needed * Use existing replace function I forgot about * Don't include emoji greater than Unicode Version 12 Only include emoji and aliases in JSON * Update build/generate-emoji.go * Tweak regex slightly to really match everything including random invisible characters. Run tests for every emoji we have * final updates * code review * code review * hard code gitea custom emoji to match previous behavior * Update .eslintrc Co-Authored-By: silverwind <me@silverwind.io> * disable preempt Co-authored-by: silverwind <me@silverwind.io> Co-authored-by: 6543 <6543@obermui.de> Co-authored-by: Lauris BH <lauris@nix.lv> Co-authored-by: guillep2k <18600385+guillep2k@users.noreply.github.com>
5 years ago
if attr.Key == "href" && strings.HasPrefix(attr.Val, "#") && notHasPrefix {
node.Attr[idx].Val = "#user-content-" + val
}
}
Support unicode emojis and remove emojify.js (#11032) * Support unicode emojis and remove emojify.js This PR replaces all use of emojify.js and adds unicode emoji support to various areas of gitea. This works in a few ways: First it adds emoji parsing support into gitea itself. This allows us to * Render emojis from valid alias (:smile:) * Detect unicode emojis and let us put them in their own class with proper aria-labels and styling * Easily allow for custom "emoji" * Support all emoji rendering and features without javascript * Uses plain unicode and lets the system render in appropriate emoji font * Doesn't leave us relying on external sources for updates/fixes/features That same list of emoji is also used to create a json file which replaces the part of emojify.js that populates the emoji search tribute. This file is about 35KB with GZIP turned on and I've set it to load after the page renders to not hinder page load time (and this removes loading emojify.js also) For custom "emoji" it uses a pretty simple scheme of just looking for /emojis/img/name.png where name is something a user has put in the "allowed reactions" setting we already have. The gitea reaction that was previously hard coded into a forked copy of emojify.js is included and works as a custom reaction under this method. The emoji data sourced here is from https://github.com/github/gemoji which is the gem library Github uses for their emoji rendering (and a data source for other sites). So we should be able to easily render any emoji and :alias: that Github can, removing any errors from migrated content. They also update it as well, so we can sync when there are new unicode emoji lists released. I've included a slimmed down and slightly modified forked copy of https://github.com/knq/emoji to make up our own emoji module. The code is pretty straight forward and again allows us to have a lot of flexibility in what happens. I had seen a few comments about performance in some of the other threads if we render this ourselves, but there doesn't seem to be any issue here. In a test it can parse, convert, and render 1,000 emojis inside of a large markdown table in about 100ms on my laptop (which is many more emojis than will ever be in any normal issue). This also prevents any flickering and other weirdness from using javascript to render some things while using go for others. Not included here are image fall back URLS. I don't really think they are necessary for anything new being written in 2020. However, managing the emoji ourselves would allow us to add these as a feature later on if it seems necessary. Fixes: https://github.com/go-gitea/gitea/issues/9182 Fixes: https://github.com/go-gitea/gitea/issues/8974 Fixes: https://github.com/go-gitea/gitea/issues/8953 Fixes: https://github.com/go-gitea/gitea/issues/6628 Fixes: https://github.com/go-gitea/gitea/issues/5130 * add new shared function emojiHTML * don't increase emoji size in issue title * Update templates/repo/issue/view_content/add_reaction.tmpl Co-Authored-By: 6543 <6543@obermui.de> * Support for emoji rendering in various templates * Render code and review comments as they should be * Better way to handle mail subjects * insert unicode from tribute selection * Add template helper for plain text when needed * Use existing replace function I forgot about * Don't include emoji greater than Unicode Version 12 Only include emoji and aliases in JSON * Update build/generate-emoji.go * Tweak regex slightly to really match everything including random invisible characters. Run tests for every emoji we have * final updates * code review * code review * hard code gitea custom emoji to match previous behavior * Update .eslintrc Co-Authored-By: silverwind <me@silverwind.io> * disable preempt Co-authored-by: silverwind <me@silverwind.io> Co-authored-by: 6543 <6543@obermui.de> Co-authored-by: Lauris BH <lauris@nix.lv> Co-authored-by: guillep2k <18600385+guillep2k@users.noreply.github.com>
5 years ago
switch node.Type {
case html.TextNode:
for _, proc := range procs {
proc(ctx, node) // it might add siblings
}
case html.ElementNode:
if isEmojiNode(node) {
// TextNode emoji will be converted to `<span class="emoji">`, then the next iteration will visit the "span"
// if we don't stop it, it will go into the TextNode again and create an infinite recursion
return node.NextSibling
} else if node.Data == "code" || node.Data == "pre" {
return node.NextSibling // ignore code and pre nodes
} else if node.Data == "img" {
return visitNodeImg(ctx, node)
} else if node.Data == "video" {
return visitNodeVideo(ctx, node)
} else if node.Data == "a" {
procs = emojiProcessors // Restrict text in links to emojis
}
for n := node.FirstChild; n != nil; {
n = visitNode(ctx, procs, n)
}
default:
}
return node.NextSibling
}
// createKeyword() renders a highlighted version of an action keyword
func createKeyword(ctx *RenderContext, content string) *html.Node {
// CSS class for action keywords (e.g. "closes: #1")
const keywordClass = "issue-keyword"
span := &html.Node{
Type: html.ElementNode,
Data: atom.Span.String(),
Attr: []html.Attribute{},
}
span.Attr = append(span.Attr, ctx.RenderInternal.NodeSafeAttr("class", keywordClass))
text := &html.Node{
Type: html.TextNode,
Data: content,
}
span.AppendChild(text)
return span
}
func createLink(ctx *RenderContext, href, content, class string) *html.Node {
a := &html.Node{
Type: html.ElementNode,
Data: atom.A.String(),
Attr: []html.Attribute{{Key: "href", Val: href}},
}
if !RenderBehaviorForTesting.DisableInternalAttributes {
a.Attr = append(a.Attr, html.Attribute{Key: "data-markdown-generated-content"})
}
if class != "" {
a.Attr = append(a.Attr, ctx.RenderInternal.NodeSafeAttr("class", class))
}
text := &html.Node{
Type: html.TextNode,
Data: content,
}
a.AppendChild(text)
return a
}
// replaceContent takes text node, and in its content it replaces a section of
// it with the specified newNode.
func replaceContent(node *html.Node, i, j int, newNode *html.Node) {
replaceContentList(node, i, j, []*html.Node{newNode})
}
// replaceContentList takes text node, and in its content it replaces a section of
// it with the specified newNodes. An example to visualize how this can work can
// be found here: https://play.golang.org/p/5zP8NnHZ03s
func replaceContentList(node *html.Node, i, j int, newNodes []*html.Node) {
// get the data before and after the match
before := node.Data[:i]
after := node.Data[j:]
// Replace in the current node the text, so that it is only what it is
// supposed to have.
node.Data = before
// Get the current next sibling, before which we place the replaced data,
// and after that we place the new text node.
nextSibling := node.NextSibling
for _, n := range newNodes {
node.Parent.InsertBefore(n, nextSibling)
}
if after != "" {
node.Parent.InsertBefore(&html.Node{
Type: html.TextNode,
Data: after,
}, nextSibling)
}
}