Refactor markup render system (#32533)

Remove unmaintainable sanitizer rules. No need to add special "class"
regexp rules anymore, use RenderInternal.SafeAttr instead, more details
(and examples) are in the tests
This commit is contained in:
wxiaoguang 2024-11-18 13:25:42 +08:00 committed by GitHub
parent 4f879a00df
commit 8a20fba8eb
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
42 changed files with 568 additions and 508 deletions

View file

@ -16,37 +16,12 @@ import (
func (st *Sanitizer) createDefaultPolicy() *bluemonday.Policy {
policy := bluemonday.UGCPolicy()
// For JS code copy and Mermaid loading state
policy.AllowAttrs("class").Matching(regexp.MustCompile(`^code-block( is-loading)?$`)).OnElements("pre")
// NOTICE: DO NOT add special "class" regexp rules here anymore, use RenderInternal.SafeAttr instead
// For code preview
policy.AllowAttrs("class").Matching(regexp.MustCompile(`^code-preview-[-\w]+( file-content)?$`)).Globally()
policy.AllowAttrs("class").Matching(regexp.MustCompile(`^lines-num$`)).OnElements("td")
policy.AllowAttrs("data-line-number").OnElements("span")
policy.AllowAttrs("class").Matching(regexp.MustCompile(`^lines-code chroma$`)).OnElements("td")
policy.AllowAttrs("class").Matching(regexp.MustCompile(`^code-inner$`)).OnElements("div")
// For code preview (unicode escape)
policy.AllowAttrs("class").Matching(regexp.MustCompile(`^file-view( unicode-escaped)?$`)).OnElements("table")
policy.AllowAttrs("class").Matching(regexp.MustCompile(`^lines-escape$`)).OnElements("td")
policy.AllowAttrs("class").Matching(regexp.MustCompile(`^toggle-escape-button btn interact-bg$`)).OnElements("a") // don't use button, button might submit a form
policy.AllowAttrs("class").Matching(regexp.MustCompile(`^(ambiguous-code-point|escaped-code-point|broken-code-point)$`)).OnElements("span")
policy.AllowAttrs("class").Matching(regexp.MustCompile(`^char$`)).OnElements("span")
policy.AllowAttrs("data-tooltip-content", "data-escaped").OnElements("span")
// For color preview
policy.AllowAttrs("class").Matching(regexp.MustCompile(`^color-preview$`)).OnElements("span")
// For attention
policy.AllowAttrs("class").Matching(regexp.MustCompile(`^attention-header attention-\w+$`)).OnElements("blockquote")
policy.AllowAttrs("class").Matching(regexp.MustCompile(`^attention-\w+$`)).OnElements("strong")
policy.AllowAttrs("class").Matching(regexp.MustCompile(`^attention-icon attention-\w+ svg octicon-[\w-]+$`)).OnElements("svg")
policy.AllowAttrs("viewBox", "width", "height", "aria-hidden").OnElements("svg")
// General safe SVG attributes
policy.AllowAttrs("viewBox", "width", "height", "aria-hidden", "data-attr-class").OnElements("svg")
policy.AllowAttrs("fill-rule", "d").OnElements("path")
// For Chroma markdown plugin
policy.AllowAttrs("class").Matching(regexp.MustCompile(`^(chroma )?language-[\w-]+( display)?( is-loading)?$`)).OnElements("code")
// Checkboxes
policy.AllowAttrs("type").Matching(regexp.MustCompile(`^checkbox$`)).OnElements("input")
policy.AllowAttrs("checked", "disabled", "data-source-position").OnElements("input")
@ -66,28 +41,15 @@ func (st *Sanitizer) createDefaultPolicy() *bluemonday.Policy {
policy.AllowURLSchemeWithCustomPolicy("data", disallowScheme)
}
// Allow classes for anchors
policy.AllowAttrs("class").Matching(regexp.MustCompile(`ref-issue( ref-external-issue)?`)).OnElements("a")
// Allow classes for task lists
policy.AllowAttrs("class").Matching(regexp.MustCompile(`task-list-item`)).OnElements("li")
// Allow classes for org mode list item status.
policy.AllowAttrs("class").Matching(regexp.MustCompile(`^(unchecked|checked|indeterminate)$`)).OnElements("li")
// Allow icons
policy.AllowAttrs("class").Matching(regexp.MustCompile(`^icon(\s+[\p{L}\p{N}_-]+)+$`)).OnElements("i")
// Allow classes for emojis
policy.AllowAttrs("class").Matching(regexp.MustCompile(`emoji`)).OnElements("img")
// Allow icons, emojis, chroma syntax and keyword markup on span
policy.AllowAttrs("class").Matching(regexp.MustCompile(`^((icon(\s+[\p{L}\p{N}_-]+)+)|(emoji)|(language-math display)|(language-math inline))$|^([a-z][a-z0-9]{0,2})$|^` + keywordClass + `$`)).OnElements("span")
// Allow 'color' and 'background-color' properties for the style attribute on text elements.
policy.AllowStyles("color", "background-color").OnElements("span", "p")
// Allow generally safe attributes
policy.AllowAttrs("src", "autoplay", "controls").OnElements("video")
// Allow generally safe attributes (reference: https://github.com/jch/html-pipeline)
generalSafeAttrs := []string{
"abbr", "accept", "accept-charset",
"accesskey", "action", "align", "alt",
@ -106,10 +68,9 @@ func (st *Sanitizer) createDefaultPolicy() *bluemonday.Policy {
"selected", "shape", "size", "span",
"start", "summary", "tabindex", "target",
"title", "type", "usemap", "valign", "value",
"vspace", "width", "itemprop",
"data-markdown-generated-content",
"vspace", "width", "itemprop", "itemscope", "itemtype",
"data-markdown-generated-content", "data-attr-class",
}
generalSafeElements := []string{
"h1", "h2", "h3", "h4", "h5", "h6", "h7", "h8", "br", "b", "i", "strong", "em", "a", "pre", "code", "img", "tt",
"div", "ins", "del", "sup", "sub", "p", "ol", "ul", "table", "thead", "tbody", "tfoot", "blockquote", "label",
@ -117,14 +78,8 @@ func (st *Sanitizer) createDefaultPolicy() *bluemonday.Policy {
"details", "caption", "figure", "figcaption",
"abbr", "bdo", "cite", "dfn", "mark", "small", "span", "time", "video", "wbr",
}
policy.AllowAttrs(generalSafeAttrs...).OnElements(generalSafeElements...)
policy.AllowAttrs("src", "autoplay", "controls").OnElements("video")
policy.AllowAttrs("itemscope", "itemtype").OnElements("div")
// FIXME: Need to handle longdesc in img but there is no easy way to do it
policy.AllowAttrs(generalSafeAttrs...).OnElements(generalSafeElements...)
// Custom keyword markup
defaultSanitizer.addSanitizerRules(policy, setting.ExternalSanitizerRules)