Resolve hostname for chromium RDP requests (#2174)

This commit is contained in:
bnkai 2022-01-04 06:47:39 +02:00 committed by GitHub
parent 1714efc92f
commit be5dc7e545
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
2 changed files with 31 additions and 2 deletions

View file

@ -5,6 +5,7 @@ import (
"context"
"fmt"
"io"
"net"
"net/http"
"net/url"
"os"
@ -87,7 +88,7 @@ func loadURL(ctx context.Context, loadURL string, client *http.Client, scraperCo
// func urlFromCDP uses chrome cdp and DOM to load and process the url
// if remote is set as true in the scraperConfig it will try to use localhost:9222
// else it will look for google-chrome in path
func urlFromCDP(ctx context.Context, url string, driverOptions scraperDriverOptions, globalConfig GlobalConfig) (io.Reader, error) {
func urlFromCDP(ctx context.Context, urlCDP string, driverOptions scraperDriverOptions, globalConfig GlobalConfig) (io.Reader, error) {
if !driverOptions.UseCDP {
return nil, fmt.Errorf("url shouldn't be fetched through CDP")
@ -107,6 +108,33 @@ func urlFromCDP(ctx context.Context, url string, driverOptions scraperDriverOpti
if isCDPPathHTTP(globalConfig) || isCDPPathWS(globalConfig) {
remote := cdpPath
// -------------------------------------------------------------------
// #1023
// when chromium is listening over RDP it only accepts requests
// with host headers that are either IPs or `localhost`
cdpURL, err := url.Parse(remote)
if err != nil {
return nil, fmt.Errorf("failed to parse CDP Path: %v", err)
}
hostname := cdpURL.Hostname()
if hostname != "localhost" {
if net.ParseIP(hostname) == nil { // not an IP
addr, err := net.LookupIP(hostname)
if err != nil || len(addr) == 0 { // can not resolve to IP
return nil, fmt.Errorf("CDP: hostname <%s> can not be resolved", hostname)
}
if len(addr[0]) == 0 { // nil IP
return nil, fmt.Errorf("CDP: hostname <%s> resolved to nil", hostname)
}
// addr is a valid IP
// replace the host part of the cdpURL with the IP
cdpURL.Host = strings.Replace(cdpURL.Host, hostname, addr[0].String(), 1)
// use that for remote
remote = cdpURL.String()
}
}
// --------------------------------------------------------------------
// if CDPPath is http(s) then we need to get the websocket URL
if isCDPPathHTTP(globalConfig) {
var err error
@ -150,7 +178,7 @@ func urlFromCDP(ctx context.Context, url string, driverOptions scraperDriverOpti
setCDPCookies(driverOptions),
printCDPCookies(driverOptions, "Cookies found"),
network.SetExtraHTTPHeaders(network.Headers(headers)),
chromedp.Navigate(url),
chromedp.Navigate(urlCDP),
chromedp.Sleep(sleepDuration),
setCDPClicks(driverOptions),
chromedp.OuterHTML("html", &res, chromedp.ByQuery),

View file

@ -4,6 +4,7 @@
* Show counts on list tabs in Performer, Studio and Tag pages. ([#2169](https://github.com/stashapp/stash/pull/2169))
### 🐛 Bug fixes
* Resolve CDP hostname if necessary. ([#2174](https://github.com/stashapp/stash/pull/2174))
* Generate sprites for short video files. ([#2167](https://github.com/stashapp/stash/pull/2167))
* Fix stash-box scraping including underscores in ethnicity. ([#2191](https://github.com/stashapp/stash/pull/2191))
* Fix stash-box batch performer task not setting birthdate. ([#2189](https://github.com/stashapp/stash/pull/2189))