351 lines
10 KiB
Go
351 lines
10 KiB
Go
//proxy/rewriter.go
|
|
package proxy
|
|
|
|
import (
|
|
"bytes"
|
|
"golang.org/x/net/html"
|
|
"net/url"
|
|
"strings"
|
|
)
|
|
|
|
type ContentRewriter struct {
|
|
baseURL *url.URL
|
|
token string
|
|
}
|
|
|
|
func NewContentRewriter(baseURL, token string) (*ContentRewriter, error) {
|
|
u, err := url.Parse(baseURL)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
|
|
return &ContentRewriter{
|
|
baseURL: u,
|
|
token: token,
|
|
}, nil
|
|
}
|
|
|
|
func (r *ContentRewriter) RewriteHTML(body []byte) ([]byte, error) {
|
|
doc, err := html.Parse(bytes.NewReader(body))
|
|
if err != nil {
|
|
return r.simpleRewriteHTML(body), nil
|
|
}
|
|
|
|
r.rewriteNode(doc)
|
|
|
|
var buf bytes.Buffer
|
|
if err := html.Render(&buf, doc); err != nil {
|
|
return r.simpleRewriteHTML(body), nil
|
|
}
|
|
|
|
return buf.Bytes(), nil
|
|
}
|
|
|
|
func (r *ContentRewriter) rewriteNode(n *html.Node) {
|
|
if n.Type == html.ElementNode {
|
|
|
|
// 在 head 标签中注入请求拦截脚本
|
|
if n.Data == "head" && n.FirstChild != nil {
|
|
script := &html.Node{
|
|
Type: html.ElementNode,
|
|
Data: "script",
|
|
}
|
|
script.AppendChild(&html.Node{
|
|
Type: html.TextNode,
|
|
Data: `(function(){var t="/p/` + r.token + `";var b="` + r.baseURL.String() + `";function r(u){if(!u||typeof u!=="string")return u;if(u.startsWith(t))return u;if(u.startsWith("http://")||u.startsWith("https://")||u.startsWith("data:")||u.startsWith("blob:")||u.startsWith("javascript:"))return u;if(u.startsWith("/")){return t+u}try{var a=new URL(u,b);if(a.origin===new URL(b).origin){return t+a.pathname+a.search+a.hash}}catch(e){}return u}var o=XMLHttpRequest.prototype.open;XMLHttpRequest.prototype.open=function(m,u){arguments[1]=r(u);return o.apply(this,arguments)};var f=window.fetch;window.fetch=function(u,opt){return f.call(this,r(u),opt)};var oa=Element.prototype.setAttribute;Element.prototype.setAttribute=function(n,v){if((n==="href"||n==="src")&&typeof v==="string"){v=r(v)}return oa.call(this,n,v)};var oi=Image;window.Image=function(){var i=new oi();var d=Object.getOwnPropertyDescriptor(HTMLImageElement.prototype,"src");if(d&&d.set){var os=d.set;Object.defineProperty(i,"src",{set:function(v){os.call(this,r(v))},get:d.get})}return i};document.addEventListener("submit",function(e){var a=e.target.action;if(a){e.target.action=r(a)}},true)})();`,
|
|
})
|
|
// 插入到 head 最前面
|
|
if n.FirstChild != nil {
|
|
script.NextSibling = n.FirstChild
|
|
n.FirstChild.PrevSibling = script
|
|
script.Parent = n
|
|
n.FirstChild = script
|
|
} else {
|
|
n.AppendChild(script)
|
|
}
|
|
}
|
|
|
|
attrs := map[string]bool{"href": true, "src": true, "action": true, "data": true}
|
|
|
|
for i, attr := range n.Attr {
|
|
if attrs[attr.Key] {
|
|
if rewritten := r.rewriteURL(attr.Val); rewritten != attr.Val {
|
|
n.Attr[i].Val = rewritten
|
|
}
|
|
}
|
|
|
|
if attr.Key == "srcset" {
|
|
n.Attr[i].Val = r.rewriteSrcset(attr.Val)
|
|
}
|
|
|
|
if attr.Key == "style" {
|
|
n.Attr[i].Val = r.rewriteInlineCSS(attr.Val)
|
|
}
|
|
}
|
|
|
|
if n.Data == "form" {
|
|
hasAction := false
|
|
for i, attr := range n.Attr {
|
|
if attr.Key == "action" {
|
|
hasAction = true
|
|
if attr.Val == "" {
|
|
n.Attr[i].Val = r.rewriteURL(r.baseURL.String())
|
|
}
|
|
break
|
|
}
|
|
}
|
|
if !hasAction {
|
|
n.Attr = append(n.Attr, html.Attribute{
|
|
Key: "action",
|
|
Val: r.rewriteURL(r.baseURL.String()),
|
|
})
|
|
}
|
|
}
|
|
|
|
if n.Data == "base" {
|
|
for i, attr := range n.Attr {
|
|
if attr.Key == "href" {
|
|
n.Attr[i].Val = r.baseURL.String()
|
|
}
|
|
}
|
|
}
|
|
|
|
if n.Data == "style" && n.FirstChild != nil {
|
|
if n.FirstChild.Type == html.TextNode {
|
|
n.FirstChild.Data = r.rewriteInlineCSS(n.FirstChild.Data)
|
|
}
|
|
}
|
|
|
|
if n.Data == "script" {
|
|
for _, attr := range n.Attr {
|
|
if attr.Key == "src" {
|
|
if r.isTrackingScript(attr.Val) {
|
|
if n.Parent != nil {
|
|
n.Parent.RemoveChild(n)
|
|
return
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
for c := n.FirstChild; c != nil; c = c.NextSibling {
|
|
r.rewriteNode(c)
|
|
}
|
|
}
|
|
|
|
func (r *ContentRewriter) rewriteURL(urlStr string) string {
|
|
urlStr = strings.TrimSpace(urlStr)
|
|
|
|
if strings.HasPrefix(urlStr, "javascript:") ||
|
|
strings.HasPrefix(urlStr, "data:") ||
|
|
strings.HasPrefix(urlStr, "mailto:") ||
|
|
strings.HasPrefix(urlStr, "tel:") ||
|
|
strings.HasPrefix(urlStr, "#") ||
|
|
urlStr == "" {
|
|
return urlStr
|
|
}
|
|
|
|
// 防止重复添加 token
|
|
if strings.HasPrefix(urlStr, "/p/"+r.token) {
|
|
return urlStr
|
|
}
|
|
|
|
if strings.HasPrefix(urlStr, "//") {
|
|
urlStr = r.baseURL.Scheme + ":" + urlStr
|
|
}
|
|
|
|
if strings.HasPrefix(urlStr, "/") && !strings.HasPrefix(urlStr, "//") {
|
|
return "/p/" + r.token + urlStr
|
|
}
|
|
|
|
u, err := url.Parse(urlStr)
|
|
if err != nil {
|
|
return urlStr
|
|
}
|
|
|
|
if !u.IsAbs() {
|
|
resolved := r.baseURL.ResolveReference(u)
|
|
if resolved.Host == r.baseURL.Host {
|
|
proxyPath := resolved.Path
|
|
if resolved.RawQuery != "" {
|
|
proxyPath += "?" + resolved.RawQuery
|
|
}
|
|
if resolved.Fragment != "" {
|
|
proxyPath += "#" + resolved.Fragment
|
|
}
|
|
return "/p/" + r.token + proxyPath
|
|
}
|
|
return "/p/" + r.token + "/" + resolved.String()
|
|
}
|
|
|
|
if u.Host == r.baseURL.Host {
|
|
proxyPath := u.Path
|
|
if u.RawQuery != "" {
|
|
proxyPath += "?" + u.RawQuery
|
|
}
|
|
if u.Fragment != "" {
|
|
proxyPath += "#" + u.Fragment
|
|
}
|
|
return "/p/" + r.token + proxyPath
|
|
}
|
|
|
|
return "/p/" + r.token + "/" + u.String()
|
|
}
|
|
|
|
func (r *ContentRewriter) rewriteSrcset(srcset string) string {
|
|
if srcset == "" {
|
|
return srcset
|
|
}
|
|
|
|
parts := strings.Split(srcset, ",")
|
|
var rewritten []string
|
|
|
|
for _, part := range parts {
|
|
part = strings.TrimSpace(part)
|
|
fields := strings.Fields(part)
|
|
|
|
if len(fields) > 0 {
|
|
fields[0] = r.rewriteURL(fields[0])
|
|
rewritten = append(rewritten, strings.Join(fields, " "))
|
|
}
|
|
}
|
|
|
|
return strings.Join(rewritten, ", ")
|
|
}
|
|
|
|
func (r *ContentRewriter) RewriteCSS(body []byte) []byte {
|
|
content := string(body)
|
|
return []byte(r.rewriteInlineCSS(content))
|
|
}
|
|
|
|
func (r *ContentRewriter) rewriteInlineCSS(css string) string {
|
|
result := css
|
|
|
|
patterns := []struct {
|
|
prefix string
|
|
suffix string
|
|
}{
|
|
{`url("`, `")`},
|
|
{`url('`, `')`},
|
|
{`url(`, `)`},
|
|
}
|
|
|
|
for _, pattern := range patterns {
|
|
start := 0
|
|
for {
|
|
idx := strings.Index(result[start:], pattern.prefix)
|
|
if idx == -1 {
|
|
break
|
|
}
|
|
|
|
idx += start
|
|
urlStart := idx + len(pattern.prefix)
|
|
urlEnd := strings.Index(result[urlStart:], pattern.suffix)
|
|
|
|
if urlEnd == -1 {
|
|
break
|
|
}
|
|
|
|
urlEnd += urlStart
|
|
originalURL := result[urlStart:urlEnd]
|
|
rewrittenURL := r.rewriteURL(originalURL)
|
|
|
|
result = result[:urlStart] + rewrittenURL + result[urlEnd:]
|
|
start = urlStart + len(rewrittenURL)
|
|
}
|
|
}
|
|
|
|
result = r.rewriteImports(result)
|
|
|
|
return result
|
|
}
|
|
|
|
func (r *ContentRewriter) rewriteImports(css string) string {
|
|
result := css
|
|
|
|
patterns := []string{`@import "`, `@import '`, `@import url("`, `@import url('`}
|
|
|
|
for _, pattern := range patterns {
|
|
start := 0
|
|
for {
|
|
idx := strings.Index(result[start:], pattern)
|
|
if idx == -1 {
|
|
break
|
|
}
|
|
|
|
idx += start
|
|
urlStart := idx + len(pattern)
|
|
|
|
var endChar string
|
|
if strings.Contains(pattern, `"`) {
|
|
endChar = `"`
|
|
} else {
|
|
endChar = `'`
|
|
}
|
|
|
|
urlEnd := strings.Index(result[urlStart:], endChar)
|
|
if urlEnd == -1 {
|
|
break
|
|
}
|
|
|
|
urlEnd += urlStart
|
|
originalURL := result[urlStart:urlEnd]
|
|
rewrittenURL := r.rewriteURL(originalURL)
|
|
|
|
result = result[:urlStart] + rewrittenURL + result[urlEnd:]
|
|
start = urlStart + len(rewrittenURL)
|
|
}
|
|
}
|
|
|
|
return result
|
|
}
|
|
|
|
func (r *ContentRewriter) simpleRewriteHTML(body []byte) []byte {
|
|
content := string(body)
|
|
baseStr := r.baseURL.Scheme + "://" + r.baseURL.Host
|
|
|
|
replacements := []struct {
|
|
old string
|
|
new string
|
|
}{
|
|
{`href="` + baseStr, `href="/p/` + r.token},
|
|
{`src="` + baseStr, `src="/p/` + r.token},
|
|
{`action="` + baseStr, `action="/p/` + r.token},
|
|
{`href='` + baseStr, `href='/p/` + r.token},
|
|
{`src='` + baseStr, `src='/p/` + r.token},
|
|
{`href="/`, `href="/p/` + r.token + `/`},
|
|
{`src="/`, `src="/p/` + r.token + `/`},
|
|
{`action="/`, `action="/p/` + r.token + `/`},
|
|
{`href='/`, `href='/p/` + r.token + `/`},
|
|
{`src='/`, `src='/p/` + r.token + `/`},
|
|
}
|
|
|
|
for _, rep := range replacements {
|
|
content = strings.ReplaceAll(content, rep.old, rep.new)
|
|
}
|
|
|
|
return []byte(content)
|
|
}
|
|
|
|
func (r *ContentRewriter) isTrackingScript(src string) bool {
|
|
trackingDomains := []string{
|
|
"google-analytics.com",
|
|
"googletagmanager.com",
|
|
"facebook.net",
|
|
"doubleclick.net",
|
|
"analytics.js",
|
|
"ga.js",
|
|
"gtag.js",
|
|
}
|
|
|
|
srcLower := strings.ToLower(src)
|
|
for _, domain := range trackingDomains {
|
|
if strings.Contains(srcLower, domain) {
|
|
return true
|
|
}
|
|
}
|
|
|
|
return false
|
|
} |