Merge pull request #247 from YoheiZuho/features/security

Features/security
This commit is contained in:
YoheiZuho 2019-08-13 02:10:48 +09:00 committed by GitHub
commit 2ff9998c0b
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
13 changed files with 949 additions and 1 deletions

View File

@ -96,6 +96,8 @@ gem 'json-ld', git: 'https://github.com/ruby-rdf/json-ld.git', ref: '345b7a57333
gem 'json-ld-preloaded', '~> 3.0'
gem 'rdf-normalize', '~> 0.3'
gem 'redcarpet', "~> 3.4.0"
group :development, :test do
gem 'fabrication', '~> 2.20'
gem 'fuubar', '~> 2.4'

View File

@ -496,6 +496,7 @@ GEM
link_header (~> 0.0, >= 0.0.8)
rdf-normalize (0.3.3)
rdf (>= 2.2, < 4.0)
redcarpet (3.4.0)
redis (4.1.2)
redis-actionpack (5.0.2)
actionpack (>= 4.0, < 6)
@ -755,6 +756,7 @@ DEPENDENCIES
rails-i18n (~> 5.1)
rails-settings-cached (~> 0.6)
rdf-normalize (~> 0.3)
redcarpet (~> 3.4.0)
redis (~> 4.1)
redis-namespace (~> 1.5)
redis-rails (~> 5.0)

View File

@ -0,0 +1,90 @@
import React from 'react';
import Motion from 'react-motion/lib/Motion';
import spring from 'react-motion/lib/spring';
import PropTypes from 'prop-types';
class IconButton extends React.PureComponent {
static propTypes = {
className: PropTypes.string,
title: PropTypes.string.isRequired,
icon: PropTypes.string.isRequired,
onClick: PropTypes.func,
size: PropTypes.number,
active: PropTypes.bool,
style: PropTypes.object,
activeStyle: PropTypes.object,
disabled: PropTypes.bool,
inverted: PropTypes.bool,
animate: PropTypes.bool,
overlay: PropTypes.bool,
};
static defaultProps = {
size: 18,
active: false,
disabled: false,
animate: false,
overlay: false,
};
handleClick = (e) => {
e.preventDefault();
if (!this.props.disabled) {
this.props.onClick(e);
}
}
render () {
const style = {
fontSize: `${this.props.size}px`,
width: `${this.props.size * 1.28571429}px`,
height: `${this.props.size * 1.28571429}px`,
lineHeight: `${this.props.size}px`,
...this.props.style,
...(this.props.active ? this.props.activeStyle : {}),
};
const classes = ['icon-button'];
if (this.props.active) {
classes.push('active');
}
if (this.props.disabled) {
classes.push('disabled');
}
if (this.props.inverted) {
classes.push('inverted');
}
if (this.props.overlay) {
classes.push('overlayed');
}
if (this.props.className) {
classes.push(this.props.className);
}
return (
<Motion defaultStyle={{ rotate: this.props.active ? 180 : 0 }} style={{ rotate: this.props.animate ? spring(this.props.active ? 0 : 180) : 0 }}>
{({ rotate }) =>
<button
aria-label={this.props.title}
title={this.props.title}
className={classes.join(' ')}
onClick={this.handleClick}
style={style}
>
<i style={{ transform: `rotate(${rotate}deg)` }} className={`fa fa-fw fa-${this.props.icon}`} aria-hidden='true' />
</button>
}
</Motion>
);
}
}
export default IconButton;

View File

@ -0,0 +1,127 @@
import React from 'react';
import Immutable from 'immutable';
import PropTypes from 'prop-types';
import Link from 'react-router-dom/Link';
import { defineMessages, injectIntl } from 'react-intl';
import IconButton from '../../../components/announcement_icon_button';
import Motion from 'react-motion/lib/Motion';
import spring from 'react-motion/lib/spring';
const Collapsable = ({ fullHeight, minHeight, isVisible, children }) => (
<Motion defaultStyle={{ height: isVisible ? fullHeight : minHeight }} style={{ height: spring(!isVisible ? minHeight : fullHeight) }}>
{({ height }) =>
<div style={{ height: `${height}px`, overflow: 'hidden' }}>
{children}
</div>
}
</Motion>
);
Collapsable.propTypes = {
fullHeight: PropTypes.number.isRequired,
minHeight: PropTypes.number.isRequired,
isVisible: PropTypes.bool.isRequired,
children: PropTypes.node.isRequired,
};
const messages = defineMessages({
toggle_visible: { id: 'media_gallery.toggle_visible', defaultMessage: 'Toggle visibility' },
welcome: { id: 'welcome.message', defaultMessage: '{domain}へようこそ!' },
markdown: { id: 'markdown.list', defaultMessage: 'markdown一覧' },
});
const hashtags = Immutable.fromJS([
'神崎ドン自己紹介',
]);
class Announcements extends React.PureComponent {
static propTypes = {
intl: PropTypes.object.isRequired,
homeSize: PropTypes.number,
isLoading: PropTypes.bool,
};
state = {
showId: null,
isLoaded: false,
};
onClick = (announcementId, currentState) => {
this.setState({ showId: currentState.showId === announcementId ? null : announcementId });
}
nl2br (text) {
return text.split(/(\n)/g).map((line, i) => {
if (line.match(/(\n)/g)) {
return React.createElement('br', { key: i });
}
return line;
});
}
render () {
const { intl } = this.props;
return (
<ul className='announcements'>
<li>
<Collapsable isVisible={this.state.showId === 'markdown'} fullHeight={1240} minHeight={20} >
<div className='announcements__body'>
<p>{ this.nl2br(intl.formatMessage(messages.markdown, { domain: document.title }))}<br />
<br />
(半角)は半角スペースを入力する必要がある場所です(半角)だけの列は半角スペースのみが入力された列が必要であるを指します<br /><br />
見出し<br /><br />
#(半角)見出しテキスト<br /><br />
#は16個重ねることができます<br /><br />
コードブロック<br /><br />
`コード`<br /><br />
引用<br /><br />
>引用文<br />
(半角)<br />
ここから先は引用が切れます<br />
引用は複数回重ねることが可能です<br /><br />
リスト<br /><br />
(半角)<br />
+(半角)内容1<br />
+(半角)内容2<br />
(半角)<br /><br />
内容の数に制限はありません<br />
投稿トップにリストを持ってくる場合に限り1行目の(半角)は必要ありません<br />
+(半角)を1.(半角)に置き換えることで数字付きリストになります<br /><br />
上付き文字<br /><br />
_上付き文字_<br /><br />
下付き文字<br /><br />
__下付き文字__<br /><br />
小さい文字<br /><br />
___小さい文字___<br /><br />
取り消し線<br /><br />
~~取り消したい文字列~~<br /><br />
横罫線<br /><br />
___<br /><br />
リンク<br /><br />
[リンク文章](https://・・・)<br /><br />
画像<br /><br />
![画像説明](https://・・・)<br /><br />
リンク画像ともにURLにはhttps://から始まる物のみご利用可能です。
</p>
</div>
</Collapsable>
<div className='announcements__icon'>
<IconButton title={intl.formatMessage(messages.toggle_visible)} icon='caret-up' onClick={() => this.onClick('markdown', this.state)} size={20} animate active={this.state.showId === 'markdown'} />
</div>
</li>
</ul>
);
}
componentWillReceiveProps (nextProps) {
if (!this.state.isLoaded) {
if (!nextProps.isLoading && (nextProps.homeSize === 0 || this.props.homeSize !== nextProps.homeSize)) {
this.setState({ isLoaded: true });
}
}
}
}
export default injectIntl(Announcements);

View File

@ -0,0 +1,11 @@
import { connect } from 'react-redux';
import Announcements from '../components/announcements';
const mapStateToProps = state => {
return {
homeSize: state.getIn(['timelines', 'home', 'items']).size,
isLoading: state.getIn(['timelines', 'home', 'isLoading']),
};
};
export default connect(mapStateToProps)(Announcements);

View File

@ -15,6 +15,7 @@ import { changeComposing } from '../../actions/compose';
import elephantUIPlane from '../../../images/elephant_ui_plane.svg';
import { mascot } from '../../initial_state';
import Icon from 'mastodon/components/icon';
import AnnouncementsContainer from './containers/announcements_container';
const messages = defineMessages({
start: { id: 'getting_started.heading', defaultMessage: 'Getting started' },
@ -108,6 +109,7 @@ class Compose extends React.PureComponent {
<NavigationContainer onClose={this.onBlur} />
<ComposeFormContainer />
<AnnouncementsContainer />
<div className='drawer__inner__mastodon'>
<img alt='' draggable='false' src={mascot || elephantUIPlane} />

View File

@ -26,3 +26,6 @@
@import 'mastodon/dashboard';
@import 'mastodon/rtl';
@import 'mastodon/accessibility';
@import 'markdown';
@import 'astarte';

View File

@ -0,0 +1,59 @@
.announcements {
padding: 0 10px;
li {
display: flex;
padding: 10px;
color: #282c37;
background: darken($white, 10%);
border-radius: 4px;
& + li {
margin-top: 10px;
}
}
}
.announcements__admin {
width: 100%;
position: relative;
p {
padding: 0 5px;
font-size: 14px;
}
}
.announcements__icon {
display: inline-block;
position: absolute;
margin: -5px 5px;
right: 10px;
}
.announcements__body {
width: 100%;
position: relative;
p {
padding: 0 5px;
font-size: 14px;
}
a {
display: inline-block;
float: right;
clear: both;
color: #282c37;
background: darken($white, 5%);
text-decoration: none;
padding: 1px 10px 0;
border: solid 1px #282c37;
font-size: 10px;
font-weight: 600;
border-radius: 4px;
margin-top: 5px;
}
}

View File

@ -1,3 +1,5 @@
@import 'contrast/variables';
@import 'application';
@import 'contrast/diff';
@import 'markdown';

View File

@ -0,0 +1,266 @@
.status__content {
font-family: inherit;
h1{
color: #ec840d;
font-weight: bold;
font-size: 1.6em;
padding: 0.5em;
display: inline-block;
line-height: 1.3;
background: #dbebf8;
vertical-align: middle;
border-radius: 25px 25px 25px 25px;
text-align: center;
border-bottom: solid 3px #ff0000;
}
h2{
color: #ec840d;
font-weight: bold;
font-size: 1.5em;
padding: 0.5em;
display: inline-block;
line-height: 1.3;
background: #dbebf8;
vertical-align: middle;
border-radius: 25px 25px 25px 25px;
text-align: center;
border-bottom: solid 3px #fffb00;
}
h3{
color: #ec840d;
font-weight: bold;
font-size: 1.4em;
padding: 0.5em;
display: inline-block;
line-height: 1.3;
background: #dbebf8;
vertical-align: middle;
border-radius: 25px 25px 25px 25px;
text-align: center;
border-bottom: solid 3px #2bff00;
}
h4{
color: #ec840d;
font-weight: bold;
font-size: 1.3em;
padding: 0.5em;
display: inline-block;
line-height: 1.3;
background: #dbebf8;
vertical-align: middle;
border-radius: 25px 25px 25px 25px;
text-align: center;
border-bottom: solid 3px #00ffea;
}
h5{
color: #ec840d;
font-weight: bold;
font-size: 1.2em;
padding: 0.5em;
display: inline-block;
line-height: 1.3;
background: #dbebf8;
vertical-align: middle;
border-radius: 25px 25px 25px 25px;
text-align: center;
border-bottom: solid 3px #0004ff;
}
h6{
color: #ec840d;
font-weight: bold;
font-size: 1.1em;
padding: 0.5em;
display: inline-block;
line-height: 1.3;
background: #dbebf8;
vertical-align: middle;
border-radius: 25px 25px 25px 25px;
text-align: center;
border-bottom: solid 3px #7700ff;
}
em{
font-style: italic;
}
strong{
font-weight: bold;
}
code{
display: block;
border-left: 2px solid;
border-color: #079903;
color: $white;
padding-left: 10px;
margin-top: 5px;
margin-bottom: 5px;
margin-left: 5px;
background-color: #000000;
.positive{
color: #5bda57;
}
.negative{
color: #ff4949;
}
.rust-fanc{
color: #ba7eff;
}
.ruby-func{
color: #24a8e6;
}
.rust-macro{
color: #d2ff6a;
}
.contents{
color: #ff9925;
}
}
pre{
display: inline-block;
font-family: 'Noto Sans Japanese', sans-serif;
font-weight: 400;
}
p ~ blockquote {
margin-top: -8px;
}
blockquote{
padding-left: 8px;
margin-top: 0.5em;
margin-bottom: 0.5em;
color: $primary-text-color;
background-color: $ui-base-color;
display: block;
border-left: 4px solid $classic-highlight-color;
}
ul.md-contents {
border: double 4px #21b384;
padding: 0.5em 1em 0.5em 2.3em;
position: relative;
}
ul li.md-contents {
line-height: 1.5;
padding: 0.2em 0;
list-style-type: none!important;
}
ul li.md-contents:before {
font-family: FontAwesome;
content: "\f0a4";
position: absolute;
left : 1em;
color: #21b384;
}
ol.md-contents {
border: double 4px #ff954f;
padding: 0.5em 1em 0.5em 1em;
position: relative;
list-style: inside decimal;
}
ol li.md-contents {
line-height: 1.5;
padding: 0.2em 0;
}
hr {
border-width: 2px 0px 0px 0px;
border-style: dashed;
border-color: #ff7676;
height: 1px;
}
p>a>img{
width: 100%;
height: 100%;
object-fit: contain;
}
a>img{
width: 100%;
height: 100%;
object-fit: contain;
}
p>a{
color: #1FBFF9;
}
a{
color: #1FBFF9;
}
sup{
font-size: 75.5%;
vertical-align: top;
position: relative;
top: -0.5em;
}
sub{
font-size: 75.5%;
vertical-align: bottom;
position: relative;
top: 0.5em;
}
small{
font-size: 50.0%;
vertical-align: bottom;
position: relative;
}
table {
margin-top: 0.5em;
margin-bottom: 0.5em;
color: $classic-highlight-color;
display: block;
width: 100%;
overflow: auto;
border-spacing: 0;
border-collapse: collapse;
}
table tr{
background-color: #000000;
}
table th, table td{
padding: 6px 13px;
border: 1px solid $classic-highlight-color;
}
table th{
font-weight: 600;
}
table thead tr{
background-color: $black;
}
td, th{
padding: 0;
}
span.img_FTL {
display: none;
}
}

View File

@ -1,3 +1,5 @@
@import 'mastodon-light/variables';
@import 'application';
@import 'mastodon-light/diff';
@import 'markdown';

View File

@ -1,6 +1,7 @@
# frozen_string_literal: true
require 'singleton'
require_relative './formatter_markdown'
require_relative './sanitize_config'
class Formatter
@ -35,12 +36,21 @@ class Formatter
linkable_accounts << status.account
html = raw_content
mdFormatter = Formatter_Markdown.new(html)
html = mdFormatter.formatted
html = "RT @#{prepend_reblog} #{html}" if prepend_reblog
html = encode_and_link_urls(html, linkable_accounts)
html = encode_custom_emojis(html, status.emojis, options[:autoplay]) if options[:custom_emojify]
html = simple_format(html, {}, sanitize: false)
html = html.delete("\n")
mdLinkDecoder = MDLinkDecoder.new(html)
html = mdLinkDecoder.decode
html.gsub!(/(&amp;)/){"&"}
html.html_safe # rubocop:disable Rails/OutputSafety
end
@ -111,13 +121,18 @@ class Formatter
def encode_and_link_urls(html, accounts = nil, options = {})
entities = utf8_friendly_extractor(html, extract_url_without_protocol: false)
mdExtractor = MDExtractor.new(html)
entities.concat(mdExtractor.extractEntities)
if accounts.is_a?(Hash)
options = accounts
accounts = nil
end
rewrite(html.dup, entities) do |entity|
if entity[:url]
if entity[:markdown]
html[entity[:indices][0]...entity[:indices][1]]
elsif entity[:url]
link_to_url(entity, options)
elsif entity[:hashtag]
link_to_hashtag(entity)

View File

@ -0,0 +1,367 @@
require 'uri'
require 'redcarpet'
require 'redcarpet/render_strip'
class Formatter_Markdown
def initialize(html)
@html = html.dup
end
def formatted
mdRenderer = CustomMDRenderer.new(
strikethrough: true,
hard_wrap: true,
autolink: false,
superscript:false,
fenced_link: true,
fenced_image: true,
no_intra_emphasis: true,
no_links: true,
no_styles: true,
no_images: true,
filter_html: true,
escape_html: true,
safe_links_only: true,
with_toc_data: true,
xhtml: false,
prettify: true,
link_attributes: true
)
md = Redcarpet::Markdown.new(
mdRenderer,
strikethrough: true,
hard_wrap: true,
superscript:false,
autolink: false,
space_after_headers: true,
no_intra_emphasis: true,
no_links: true,
no_styles: true,
no_images: true,
filter_html: true,
escape_html: true,
safe_links_only: true,
with_toc_data: true,
xhtml: false,
prettify: true,
link_attributes: true
)
s = @html
s.gsub!(/\n[\n]+/) {"\n \n"}# 改行周りの問題を修正
s.gsub!(/`[ ]+`/) {" "}# code内が半角スペースのみだとHTMLが壊れるのでそれの回避
renderedMD = md.render(s)
result = renderedMD
result.gsub!(/(<\w+)([^>]*>)/) { "#{$1} data-md='true' #{$2}" }# ToDo data-md="true" を認識して他鯖の人にmarkdownの使用を伝える機能の実装
result.gsub!(/(https?:\/\/[^<>"\[\]  ]+)/){"#{$1} "}#URLの後ろにスペースをねじ込む奴 mastodonのURL認識がゆるいのをmarkdownで対処
result
end
class CustomMDRenderer < Redcarpet::Render::HTML
#基本的な実装の流れ
#URLの削除(mastodonの機能上URLとして認識されると十中八九HTMLが壊れるので)
#markdownコンテンツ内でのmarkdownコンテンツの禁止(意図しないHTMLタグの生成によってHTMLの不正出力を防ぐ目的)
#最後にHTMLに出力される際にHTML的にヤバイ子たちのエスケープ
def paragraph(text)
%(#{text.strip})
end
def linebreak()
%(<br>)
end
def block_quote(quote)
urlRemoved = "#{remove_url(quote)}"
escapedContents = "#{blockquote_markdown_escape(urlRemoved)}"
%(<blockquote>#{escapedContents.strip}</blockquote>)
end
def header(text, header_level)
urlRemoved = "#{remove_url(text)}"
mdContentsRemoved = "#{markdown_escape(urlRemoved)}"
%(<h#{header_level}>#{encode(mdContentsRemoved)}</h#{header_level}>\n)
end
def block_code(code, language)
%(<br>#{code.strip})
end
def codespan(code)
urlRemoved = "#{remove_url(code)}"
escapedCode = "#{escape_bbcode(urlRemoved)}"
encoded = "#{encode(escapedCode)}"
%(<code>#{code_contents(encoded)}</code>)
end
def list(contents, list_type)
if list_type == :unordered
%(<ul class='md-contents'>#{contents.strip}</ul>)
elsif list_type == :ordered
%(<ol class='md-contents'>#{contents.strip}</ol>)
else
%(<#{list_type} class='md-contents'>#{contents.strip}</#{list_type}>)
end
end
def list_item(text, list_type)
urlRemoved = "#{remove_url(text)}"
mdContentsRemoved = "#{markdown_escape(urlRemoved)}"
%(<li class='md-contents'>#{encode(mdContentsRemoved)}</li>)
end
def emphasis(text)
urlRemoved = "#{remove_url(text)}"
mdContentsRemoved = "#{markdown_escape(urlRemoved)}"
%(<sup>#{encode(mdContentsRemoved)}</sup>)
end
def double_emphasis(text)
urlRemoved = "#{remove_url(text)}"
mdContentsRemoved = "#{markdown_escape(urlRemoved)}"
%(<sub>#{encode(mdContentsRemoved)}</sub>)
end
def triple_emphasis(text)
urlRemoved = "#{remove_url(text)}"
mdContentsRemoved = "#{markdown_escape(urlRemoved)}"
%(<small>#{encode(mdContentsRemoved)}</small>)
end
def strikethrough(text)
urlRemoved = "#{remove_url(text)}"
mdContentsRemoved = "#{markdown_escape(urlRemoved)}"
%(<s>#{encode(mdContentsRemoved)}</s>)
end
def superscript(text)
urlRemoved = "#{remove_url(text)}"
mdContentsRemoved = "#{markdown_escape(urlRemoved)}"
%(<sup>#{encode(mdContentsRemoved)}</sup>)
end
def underline(text)
urlRemoved = "#{remove_url(text)}"
mdContentsRemoved = "#{markdown_escape(urlRemoved)}"
%(<u>#{encode(mdContentsRemoved)}</u>)
end
def highlight(text)
urlRemoved = "#{remove_url(text)}"
mdContentsRemoved = "#{markdown_escape(urlRemoved)}"
%(<mark>#{encode(mdContentsRemoved)}</mark>)
end
#オートリンクはmastodonとの相性が悪いので基本的には使わない
def autolink(link, link_type)
%(<a herf="#{link}">リンク</a>)
end
#https以外の物がURLとして記入された時にTextをHTML的に考えて安全に表示するように変更
def image(link, title, alt_text)
if alt_text =~ /[<>"\[\]  ]+/
alt_text = "設定なし"
end
imgcheck = "#{link}"
if imgcheck !~ /\Ahttps:\/\/[^<>"\[\]  ]+\z/
%(#{encode(alt_text)})
else
%(<span class="img_FTL">画像が添付されています。</span><img src="#{URI.encode_www_form_component(link)}">)
end
end
def link(link, title, content)
if content =~ /([<>"\[\]  ]+|https?:\/\/|#|@)/
content = "リンク"
elsif content !~ /.+/
content = "リンク"
end
linkcheck = "#{link}"
if linkcheck !~ /\Ahttps:\/\/[^<>"\[\]  ]+\z/
%(#{encode(content)})
else
%(<a href="#{URI.encode_www_form_component(link)}">#{encode(content)}</a>)
end
end
#ここから下はいろいろエスケープするための奴
#HTML的に考えてよろしくない子たちをエスケープする奴
def encode(html)
HTMLEntities.new.encode(html)
end
#markdownコンテンツないでURLが生成されるのを防ぐためのエスケープする奴
def remove_url(string)
url = string.gsub(/https?:\/\//){ "URL:" }
reply = url.gsub(/@/){ "" }
hashTag = reply.gsub(/#/){ "" }
end
#前々から要望があったcode内でBBCodeを無効化するための奴
def escape_bbcode(string)
string.gsub(/\[/){ "&#091;" }
end
#markdownの中でmarkdownを展開させないためのエスケープする奴
#blockquote以外は下のが使える
def markdown_escape(string)
string.gsub(/<[^>]+>/) { "" }
end
#blockquoteコンテンツ内でblockquoteタグだけを許可するためのエスケープ
def blockquote_markdown_escape(string)
string.gsub(/<([\/]?a[^>]*|[\/]?img[^>]*|[\/]?code[^>]*|[\/]?h[1-6][^>]*|[\/]?sup[^>]*|[\/]?sub[^>]*|[\/]?small[^>]*|[\/]?ul[^>]*|[\/]?ol[^>]*|[\/]?li[^>]*|[\/]?hr[^>]*|[\/]?s[^>]*|[\/]?u[^>]*|[\/]?mark[^>]*)>/) { "" }
end
#code内の一部を色分けするための変更
def code_contents(string)
simple = string.gsub(/(true|error|false|failed|def|puts|end|fn|let|mut|use|String|println!)/ ,
"true" => "<span class='positive'>#{:true}</span>",
"error" => "<span class='negative'>#{:error}</span>",
"false" => "<span class='negative'>#{:false}</span>",
"failed" => "<span class='negative'>#{:failed}</span>",
"def" => "<span class='ruby-func'>#{:def}</span>",
"puts" => "<span class='ruby-func'>#{:puts}</span>",
"end" => "<span class='ruby-func'>#{:end}</span>",
"fn" => "<span class='rust-fanc'>#{:fn}</span>",
"let" => "<span class='rust-fanc'>#{:let}</span>",
"mut" => "<span class='rust-fanc'>#{:mut}</span>",
"use" => "<span class='rust-fanc'>#{:use}</span>",
"String" => "<span class='rust-macro'>#{:String}</span>",
"println!" => "<span class='rust-macro'>#{:println!}</span>",
)
simple.gsub(/(&quot;[a-zA-Z0-9_ ,]+&quot;)/){ "<span class='contents'>#{$1}</span>" }
# "" => "<span class=''>#{:}</span>",
end
#テストで書きなぐった奴
def html_escape(string)
string.gsub(/['&\"<>\/]/, {
'&' => '&amp;',
'<' => '&lt;',
'>' => '&gt;',
'"' => '&quot;',
"'" => '&#x27;',
"/" => '&#x2F;',
})
end
end
end
#URLとかいう人類には早すぎたやばい子たちを大人しくするために必要な機構
class MDLinkDecoder
def initialize(html)
@html = html.dup
end
def decode
imageDecoded = @html.gsub(/<img data-md='true'\s+src="([^"]+)"([^>]*)>/) { "<a href=\"" + URI.decode_www_form_component($1) + "\"" + $2 + "><img data-md='true' src=\"" + URI.decode_www_form_component($1) + "\"" + $2 + "></a>" }
imageDecoded.gsub(/<a data-md='true'\s+href="([^"]+)"([^>]*)>/) { "<a data-md='true' href=\"" + URI.decode_www_form_component($1) + "\"" + $2 + ">" }
end
end
#エスケープを回避するHTMLタグの設定とかその他
class MDExtractor
def initialize(html)
@html = html.dup
end
def extractEntities
[
extractByHTMLTagName("h1"),
extractByHTMLTagName("h2"),
extractByHTMLTagName("h3"),
extractByHTMLTagName("h4"),
extractByHTMLTagName("h5"),
extractByHTMLTagName("h6"),
extractByHTMLTagName("em"),
extractByHTMLTagName("sup"),
extractByHTMLTagName("sub"),
extractByHTMLTagName("small"),
extractByHTMLTagName("u"),
extractByHTMLTagName("strong"),
extractByHTMLTagName("ul", false, false, "li"),
extractByHTMLTagName("ol", false, false, "li"),
extractByHTMLTagName("code"),
extractByHTMLTagName("blockquote", false),
extractByHTMLTagName("hr", false, true),
extractByHTMLTagName("br", false, true),
extractByHTMLTagName("a"),
extractByHTMLTagName("img", false, true),
extractByHTMLTagName("s"),
extractByHTMLTagName("span")
].flatten.compact
end
def extractByHTMLTagName(tagName, isNoNest = true, isSingle = false, itemTagName = nil)
entities = []
@html.to_s.scan(htmlTagPatternByCond(tagName, isNoNest, isSingle, itemTagName)) do
match = $~
beginPos = match.char_begin(0)
endPos = match.char_end(0)
#puts "MDExtractor extracted with:\n" + @html + "\nbeginPos: " + beginPos.to_s + ", endPos: " + endPos.to_s + ", length: " + @html.length.to_s
entity = {
:markdown => true,
:indices => [beginPos, endPos]
}
entities.push(entity)
end
entities
end
def htmlTagPatternByCond(tagName, isNoNest, isSingle, itemTagName)
if isSingle
htmlTagPatternSingle(tagName)
elsif isNoNest
htmlTagPatternNoNest(tagName)
elsif itemTagName && itemTagName.length > 0
htmlTagPatternOuterMostWithItem(tagName, itemTagName)
else
htmlTagPatternOuterMost(tagName)
end
end
def htmlTagPattern(tagName)
Regexp.compile("<#{tagName} data-md=[^>]*>(?:[^<]|<#{tagName} data-md=[^>]*>|<\\/#{tagName}>)*(?:<\\/#{tagName}>)*")
end
def htmlTagPatternNoNest(tagName)
Regexp.compile("<#{tagName} data-md=[^>]*>(?:.|\n)*?<\\/#{tagName}>")
end
def htmlTagPatternSingle(tagName)
Regexp.compile("<#{tagName} data-md=[^>]*>")
end
# https://stackoverflow.com/questions/546433/regular-expression-to-match-outer-brackets
def htmlTagPatternOuterMost(tagName)
Regexp.compile("<#{tagName} data-md=[^>]*>(?:[^<>]|(\\g<0>))*<\/#{tagName}>")
end
def htmlTagPatternOuterMostWithItem(tagName, itemTagName)
Regexp.compile("<#{tagName} data-md=[^>]*>(?:[^<>]|<#{itemTagName} data-md=[^>]*>|<\\/#{itemTagName}>|(\\g<0>))*<\/#{tagName}>")
end
end