2021-04-20 06:25:08 +08:00
// Copyright 2017 The Gitea Authors. All rights reserved.
2022-11-27 13:20:29 -05:00
// SPDX-License-Identifier: MIT
2021-04-20 06:25:08 +08:00
package markup
import (
2022-06-09 00:46:39 +03:00
"bytes"
2021-04-20 06:25:08 +08:00
"context"
"errors"
"fmt"
"io"
2022-06-16 11:33:23 +08:00
"net/url"
2021-04-20 06:25:08 +08:00
"path/filepath"
"strings"
"sync"
2025-03-27 19:40:14 +00:00
"forgejo.org/modules/git"
"forgejo.org/modules/setting"
"forgejo.org/modules/util"
2023-04-18 03:05:19 +08:00
"github.com/yuin/goldmark/ast"
)
type RenderMetaMode string
const (
RenderMetaAsDetails RenderMetaMode = "details" // default
RenderMetaAsNone RenderMetaMode = "none"
RenderMetaAsTable RenderMetaMode = "table"
2021-04-20 06:25:08 +08:00
)
2022-10-22 20:15:52 +03:00
type ProcessorHelper struct {
IsUsernameMentionable func ( ctx context . Context , username string ) bool
2024-03-27 18:36:12 +01:00
GetRepoFileBlob func ( ctx context . Context , ownerName , repoName , commitSha , filePath string , language * string ) ( * git . Blob , error )
2023-05-20 23:02:52 +02:00
ElementDir string // the direction of the elements, eg: "ltr", "rtl", "auto", default to no direction attribute
2022-10-22 20:15:52 +03:00
}
2023-05-20 23:02:52 +02:00
var DefaultProcessorHelper ProcessorHelper
2022-10-22 20:15:52 +03:00
2021-04-20 06:25:08 +08:00
// Init initialize regexps for markdown parsing
2022-10-22 20:15:52 +03:00
func Init ( ph * ProcessorHelper ) {
if ph != nil {
2023-05-20 23:02:52 +02:00
DefaultProcessorHelper = * ph
2022-10-22 20:15:52 +03:00
}
2021-04-20 06:25:08 +08:00
NewSanitizer ( )
if len ( setting . Markdown . CustomURLSchemes ) > 0 {
CustomLinkURLSchemes ( setting . Markdown . CustomURLSchemes )
}
// since setting maybe changed extensions, this will reload all renderer extensions mapping
extRenderers = make ( map [ string ] Renderer )
for _ , renderer := range renderers {
for _ , ext := range renderer . Extensions ( ) {
extRenderers [ strings . ToLower ( ext ) ] = renderer
}
}
}
2022-06-08 09:59:16 +01:00
// Header holds the data about a header.
type Header struct {
Level int
Text string
ID string
}
2021-04-20 06:25:08 +08:00
// RenderContext represents a render context
type RenderContext struct {
2022-06-16 11:33:23 +08:00
Ctx context . Context
RelativePath string // relative path from tree root of the branch
Type string
IsWiki bool
2024-01-15 09:49:24 +01:00
Links Links
2022-06-16 11:33:23 +08:00
Metas map [ string ] string
DefaultLink string
GitRepo * git . Repository
ShaExistCache map [ string ] bool
cancelFn func ( )
2023-04-18 03:05:19 +08:00
SidebarTocNode ast . Node
2022-06-16 11:33:23 +08:00
InStandalonePage bool // used by external render. the router "/org/repo/render/..." will output the rendered content in a standalone page
2021-06-20 23:39:12 +01:00
}
2024-01-15 09:49:24 +01:00
type Links struct {
2024-03-13 11:34:58 +01:00
AbsolutePrefix bool
Base string
BranchPath string
TreePath string
}
func ( l * Links ) Prefix ( ) string {
if l . AbsolutePrefix {
return setting . AppURL
}
return setting . AppSubURL
2024-01-15 09:49:24 +01:00
}
func ( l * Links ) HasBranchInfo ( ) bool {
return l . BranchPath != ""
}
func ( l * Links ) SrcLink ( ) string {
return util . URLJoin ( l . Base , "src" , l . BranchPath , l . TreePath )
}
func ( l * Links ) MediaLink ( ) string {
return util . URLJoin ( l . Base , "media" , l . BranchPath , l . TreePath )
}
func ( l * Links ) RawLink ( ) string {
return util . URLJoin ( l . Base , "raw" , l . BranchPath , l . TreePath )
}
func ( l * Links ) WikiLink ( ) string {
return util . URLJoin ( l . Base , "wiki" )
}
func ( l * Links ) WikiRawLink ( ) string {
return util . URLJoin ( l . Base , "wiki/raw" )
}
func ( l * Links ) ResolveMediaLink ( isWiki bool ) string {
if isWiki {
return l . WikiRawLink ( )
} else if l . HasBranchInfo ( ) {
return l . MediaLink ( )
}
return l . Base
}
2021-06-20 23:39:12 +01:00
// Cancel runs any cleanup functions that have been registered for this Ctx
func ( ctx * RenderContext ) Cancel ( ) {
if ctx == nil {
return
}
ctx . ShaExistCache = map [ string ] bool { }
if ctx . cancelFn == nil {
return
}
ctx . cancelFn ( )
}
// AddCancel adds the provided fn as a Cleanup for this Ctx
func ( ctx * RenderContext ) AddCancel ( fn func ( ) ) {
if ctx == nil {
return
}
oldCancelFn := ctx . cancelFn
if oldCancelFn == nil {
ctx . cancelFn = fn
return
}
ctx . cancelFn = func ( ) {
defer oldCancelFn ( )
fn ( )
}
2021-04-20 06:25:08 +08:00
}
// Renderer defines an interface for rendering markup file to HTML
type Renderer interface {
Name ( ) string // markup format name
Extensions ( ) [ ] string
2021-06-23 23:09:51 +02:00
SanitizerRules ( ) [ ] setting . MarkupSanitizerRule
2021-04-20 06:25:08 +08:00
Render ( ctx * RenderContext , input io . Reader , output io . Writer ) error
}
2022-06-16 11:33:23 +08:00
// PostProcessRenderer defines an interface for renderers who need post process
type PostProcessRenderer interface {
NeedPostProcess ( ) bool
}
// PostProcessRenderer defines an interface for external renderers
type ExternalRenderer interface {
// SanitizerDisabled disabled sanitize if return true
SanitizerDisabled ( ) bool
// DisplayInIFrame represents whether render the content with an iframe
DisplayInIFrame ( ) bool
}
2022-06-09 00:46:39 +03:00
// RendererContentDetector detects if the content can be rendered
// by specified renderer
type RendererContentDetector interface {
CanRender ( filename string , input io . Reader ) bool
}
2021-04-20 06:25:08 +08:00
var (
extRenderers = make ( map [ string ] Renderer )
renderers = make ( map [ string ] Renderer )
)
// RegisterRenderer registers a new markup file renderer
func RegisterRenderer ( renderer Renderer ) {
renderers [ renderer . Name ( ) ] = renderer
for _ , ext := range renderer . Extensions ( ) {
extRenderers [ strings . ToLower ( ext ) ] = renderer
}
}
// GetRendererByFileName get renderer by filename
func GetRendererByFileName ( filename string ) Renderer {
extension := strings . ToLower ( filepath . Ext ( filename ) )
return extRenderers [ extension ]
}
// GetRendererByType returns a renderer according type
func GetRendererByType ( tp string ) Renderer {
return renderers [ tp ]
}
2022-06-09 00:46:39 +03:00
// DetectRendererType detects the markup type of the content
func DetectRendererType ( filename string , input io . Reader ) string {
buf , err := io . ReadAll ( input )
if err != nil {
return ""
}
for _ , renderer := range renderers {
if detector , ok := renderer . ( RendererContentDetector ) ; ok && detector . CanRender ( filename , bytes . NewReader ( buf ) ) {
return renderer . Name ( )
}
}
return ""
}
2021-04-20 06:25:08 +08:00
// Render renders markup file to HTML with all specific handling stuff.
func Render ( ctx * RenderContext , input io . Reader , output io . Writer ) error {
if ctx . Type != "" {
return renderByType ( ctx , input , output )
2022-06-16 11:33:23 +08:00
} else if ctx . RelativePath != "" {
2021-04-20 06:25:08 +08:00
return renderFile ( ctx , input , output )
}
return errors . New ( "Render options both filename and type missing" )
}
// RenderString renders Markup string to HTML with all specific handling stuff and return string
func RenderString ( ctx * RenderContext , content string ) ( string , error ) {
var buf strings . Builder
if err := Render ( ctx , strings . NewReader ( content ) , & buf ) ; err != nil {
return "" , err
}
return buf . String ( ) , nil
}
2022-03-06 16:41:54 +08:00
type nopCloser struct {
io . Writer
}
func ( nopCloser ) Close ( ) error { return nil }
2022-06-16 11:33:23 +08:00
func renderIFrame ( ctx * RenderContext , output io . Writer ) error {
2025-09-06 16:23:01 +02:00
// set height="300", otherwise if the postMessage mechanism breaks, we are left with a 0-height iframe
2022-06-16 11:33:23 +08:00
// at the moment, only "allow-scripts" is allowed for sandbox mode.
// "allow-same-origin" should never be used, it leads to XSS attack, and it makes the JS in iframe can access parent window's config and CSRF token
// TODO: when using dark theme, if the rendered content doesn't have proper style, the default text color is black, which is not easy to read
_ , err := io . WriteString ( output , fmt . Sprintf ( `
< iframe src = "%s/%s/%s/render/%s/%s"
2025-09-06 16:23:01 +02:00
class = "external-render"
width = "100%%" height = "300" frameborder = "0"
2022-06-16 11:33:23 +08:00
sandbox = "allow-scripts"
> < / iframe > ` ,
setting . AppSubURL ,
url . PathEscape ( ctx . Metas [ "user" ] ) ,
url . PathEscape ( ctx . Metas [ "repo" ] ) ,
ctx . Metas [ "BranchNameSubURL" ] ,
url . PathEscape ( ctx . RelativePath ) ,
) )
return err
}
2021-06-07 06:50:07 +08:00
func render ( ctx * RenderContext , renderer Renderer , input io . Reader , output io . Writer ) error {
2021-04-20 06:25:08 +08:00
var wg sync . WaitGroup
var err error
pr , pw := io . Pipe ( )
defer func ( ) {
_ = pr . Close ( )
_ = pw . Close ( )
} ( )
2022-03-06 16:41:54 +08:00
var pr2 io . ReadCloser
var pw2 io . WriteCloser
2022-06-16 11:33:23 +08:00
var sanitizerDisabled bool
if r , ok := renderer . ( ExternalRenderer ) ; ok {
sanitizerDisabled = r . SanitizerDisabled ( )
}
if ! sanitizerDisabled {
2022-03-06 16:41:54 +08:00
pr2 , pw2 = io . Pipe ( )
defer func ( ) {
_ = pr2 . Close ( )
_ = pw2 . Close ( )
} ( )
wg . Add ( 1 )
go func ( ) {
err = SanitizeReader ( pr2 , renderer . Name ( ) , output )
_ = pr2 . Close ( )
wg . Done ( )
} ( )
} else {
pw2 = nopCloser { output }
}
2021-06-23 23:09:51 +02:00
wg . Add ( 1 )
go func ( ) {
2022-06-16 11:33:23 +08:00
if r , ok := renderer . ( PostProcessRenderer ) ; ok && r . NeedPostProcess ( ) {
2021-06-07 06:50:07 +08:00
err = PostProcess ( ctx , pr , pw2 )
2021-06-23 23:09:51 +02:00
} else {
_ , err = io . Copy ( pw2 , pr )
}
_ = pr . Close ( )
_ = pw2 . Close ( )
wg . Done ( )
} ( )
2021-06-07 06:50:07 +08:00
if err1 := renderer . Render ( ctx , input , pw ) ; err1 != nil {
2021-04-20 06:25:08 +08:00
return err1
}
2025-09-06 16:23:01 +02:00
if r , ok := renderer . ( ExternalRenderer ) ; ok && r . DisplayInIFrame ( ) {
// Append a short script to the iframe's contents, which will communicate the scroll height of the embedded document via postMessage, either once loaded (in case the containing page loads first) in response to a postMessage from external.js, in case the iframe loads first
// We use '*' as a target origin for postMessage, because can be certain we are embedded on the same domain, due to X-Frame-Options configured elsewhere. (Plus, the offsetHeight of an embedded document is likely not sensitive data anyway.)
_ , _ = pw . Write ( [ ] byte ( "<script>{let postHeight = () => {window.parent.postMessage({frameHeight: document.documentElement.offsetHeight}, '*')}; window.addEventListener('load', postHeight); window.addEventListener('message', (event) => {if (event.source === window.parent && event.data.requestOffsetHeight) postHeight()});}</script>" ) )
}
2021-04-20 06:25:08 +08:00
_ = pw . Close ( )
wg . Wait ( )
return err
}
// ErrUnsupportedRenderType represents
type ErrUnsupportedRenderType struct {
Type string
}
func ( err ErrUnsupportedRenderType ) Error ( ) string {
return fmt . Sprintf ( "Unsupported render type: %s" , err . Type )
}
func renderByType ( ctx * RenderContext , input io . Reader , output io . Writer ) error {
if renderer , ok := renderers [ ctx . Type ] ; ok {
return render ( ctx , renderer , input , output )
}
return ErrUnsupportedRenderType { ctx . Type }
}
// ErrUnsupportedRenderExtension represents the error when extension doesn't supported to render
type ErrUnsupportedRenderExtension struct {
Extension string
}
2023-03-24 07:12:23 +01:00
func IsErrUnsupportedRenderExtension ( err error ) bool {
_ , ok := err . ( ErrUnsupportedRenderExtension )
return ok
}
2021-04-20 06:25:08 +08:00
func ( err ErrUnsupportedRenderExtension ) Error ( ) string {
return fmt . Sprintf ( "Unsupported render extension: %s" , err . Extension )
}
func renderFile ( ctx * RenderContext , input io . Reader , output io . Writer ) error {
2022-06-16 11:33:23 +08:00
extension := strings . ToLower ( filepath . Ext ( ctx . RelativePath ) )
2021-04-20 06:25:08 +08:00
if renderer , ok := extRenderers [ extension ] ; ok {
2022-06-16 11:33:23 +08:00
if r , ok := renderer . ( ExternalRenderer ) ; ok && r . DisplayInIFrame ( ) {
if ! ctx . InStandalonePage {
// for an external render, it could only output its content in a standalone page
// otherwise, a <iframe> should be outputted to embed the external rendered page
return renderIFrame ( ctx , output )
}
}
2021-04-20 06:25:08 +08:00
return render ( ctx , renderer , input , output )
}
return ErrUnsupportedRenderExtension { extension }
}
// Type returns if markup format via the filename
func Type ( filename string ) string {
if parser := GetRendererByFileName ( filename ) ; parser != nil {
return parser . Name ( )
}
return ""
}
// IsMarkupFile reports whether file is a markup type file
func IsMarkupFile ( name , markup string ) bool {
if parser := GetRendererByFileName ( name ) ; parser != nil {
return parser . Name ( ) == markup
}
return false
}
2023-03-24 07:12:23 +01:00
func PreviewableExtensions ( ) [ ] string {
extensions := make ( [ ] string , 0 , len ( extRenderers ) )
for extension := range extRenderers {
extensions = append ( extensions , extension )
}
return extensions
}