wrp/wrp.go

286 lines
8.7 KiB
Go
Raw Normal View History

2019-05-30 01:53:05 +00:00
//
// WRP - Web Rendering Proxy
//
// Copyright (c) 2013-2018 Antoni Sawicki
// Copyright (c) 2019 Google LLC
//
2019-05-29 08:29:01 +00:00
package main
import (
"bytes"
2019-05-29 08:52:28 +00:00
"context"
"flag"
"fmt"
_ "image"
"image/gif"
"image/png"
2019-05-29 08:52:28 +00:00
"log"
2019-05-30 09:03:17 +00:00
"math/rand"
2019-05-29 08:52:28 +00:00
"net/http"
"net/url"
2019-05-29 08:52:28 +00:00
"strconv"
"strings"
2019-05-29 08:52:28 +00:00
"time"
2019-05-29 08:29:01 +00:00
2019-05-29 09:39:06 +00:00
"github.com/chromedp/cdproto/emulation"
2019-05-31 07:19:10 +00:00
"github.com/chromedp/cdproto/runtime"
2019-05-29 09:39:06 +00:00
2019-05-29 08:52:28 +00:00
"github.com/chromedp/cdproto/cdp"
"github.com/chromedp/cdproto/dom"
"github.com/chromedp/chromedp"
2019-05-29 08:29:01 +00:00
)
2019-06-02 09:25:18 +00:00
// Ismap for server side processing
type Ismap struct {
xmin int64
ymin int64
xmax int64
ymax int64
url string
}
2019-05-29 08:29:01 +00:00
var (
2019-06-02 23:19:08 +00:00
version = "3.0"
gifmap = make(map[string]bytes.Buffer)
ismap = make(map[string][]Ismap)
2019-05-29 08:29:01 +00:00
)
2019-06-02 23:05:36 +00:00
func pageServer(out http.ResponseWriter, req *http.Request) {
req.ParseForm()
u := req.FormValue("url")
2019-05-31 08:56:34 +00:00
var istr string
2019-06-02 23:19:08 +00:00
var ion string
2019-05-31 08:56:34 +00:00
var i bool
2019-06-02 23:05:36 +00:00
if req.FormValue("i") == "on" {
2019-05-31 08:56:34 +00:00
istr = "CHECKED"
i = true
2019-06-02 23:19:08 +00:00
ion = "&i=on"
2019-05-31 08:56:34 +00:00
} else {
istr = ""
i = false
}
2019-06-02 23:05:36 +00:00
p, _ := strconv.ParseInt(req.FormValue("p"), 10, 64)
if req.FormValue("pg") == "Next" {
2019-05-31 23:41:25 +00:00
p++
2019-06-02 23:05:36 +00:00
} else if req.FormValue("pg") == "Prev" {
2019-05-31 23:41:25 +00:00
p--
} else {
p = 0
2019-05-31 23:41:25 +00:00
}
2019-06-02 23:05:36 +00:00
w, _ := strconv.ParseInt(req.FormValue("w"), 10, 64)
2019-05-31 01:08:48 +00:00
if w < 10 {
w = 1024
}
2019-06-02 23:05:36 +00:00
h, _ := strconv.ParseInt(req.FormValue("h"), 10, 64)
2019-05-31 01:08:48 +00:00
if h < 10 {
h = 768
2019-05-29 08:52:28 +00:00
}
2019-06-02 23:05:36 +00:00
s, _ := strconv.ParseFloat(req.FormValue("s"), 64)
2019-05-31 01:08:48 +00:00
if s < 0.1 {
s = 1.0
}
2019-06-03 00:06:41 +00:00
c, _ := strconv.ParseInt(req.FormValue("c"), 10, 64)
if c < 2 || c > 256 {
c = 256
}
2019-06-02 23:05:36 +00:00
log.Printf("%s Page Reqest for url=\"%s\" [%s]\n", req.RemoteAddr, u, req.URL.Path)
2019-05-29 08:52:28 +00:00
out.Header().Set("Content-Type", "text/html")
2019-06-02 23:24:46 +00:00
fmt.Fprintf(out, "<!-- Web Rendering Proxy Version %s -->\n", version)
fmt.Fprintf(out, "<HTML>\n<HEAD><TITLE>WRP %s</TITLE></HEAD>\n<BODY BGCOLOR=\"#F0F0F0\">\n", u)
2019-06-03 00:06:41 +00:00
fmt.Fprintf(out, "<FORM ACTION=\"/\">URL/Search: <INPUT TYPE=\"TEXT\" NAME=\"url\" VALUE=\"%s\" SIZE=\"20\">", u)
fmt.Fprintf(out, "<INPUT TYPE=\"SUBMIT\" VALUE=\"Go\"> \n")
fmt.Fprintf(out, "Page:<INPUT TYPE=\"SUBMIT\" NAME=\"pg\" VALUE=\"Prev\"> \n")
fmt.Fprintf(out, "<INPUT TYPE=\"TEXT\" NAME=\"p\" VALUE=\"%d\" SIZE=\"2\"> \n", p)
fmt.Fprintf(out, "<INPUT TYPE=\"SUBMIT\" NAME=\"pg\" VALUE=\"Next\"> <P>\n")
2019-06-01 01:20:55 +00:00
fmt.Fprintf(out, "ISMAP:<INPUT TYPE=\"CHECKBOX\" NAME=\"i\" %s> \n", istr)
2019-05-31 07:19:10 +00:00
fmt.Fprintf(out, "Width:<INPUT TYPE=\"TEXT\" NAME=\"w\" VALUE=\"%d\" SIZE=\"4\"> \n", w)
fmt.Fprintf(out, "Height:<INPUT TYPE=\"TEXT\" NAME=\"h\" VALUE=\"%d\" SIZE=\"4\"> \n", h)
fmt.Fprintf(out, "Scale:<INPUT TYPE=\"TEXT\" NAME=\"s\" VALUE=\"%1.2f\" SIZE=\"3\"> \n", s)
2019-06-03 00:06:41 +00:00
fmt.Fprintf(out, "Colors:<INPUT TYPE=\"TEXT\" NAME=\"c\" VALUE=\"%d\" SIZE=\"3\"> \n", c)
2019-06-01 01:20:55 +00:00
fmt.Fprintf(out, "</FORM><P>\n")
if len(u) > 1 {
if strings.HasPrefix(u, "http") {
2019-06-03 00:06:41 +00:00
capture(u, w, h, s, int(c), p, i, req.RemoteAddr, out)
} else {
2019-06-03 00:06:41 +00:00
capture(fmt.Sprintf("http://www.google.com/search?q=%s", url.QueryEscape(u)), w, h, s, int(c), p, i, req.RemoteAddr, out)
}
2019-05-31 01:08:48 +00:00
} else {
fmt.Fprintf(out, "No URL or search query specified")
2019-05-29 08:52:28 +00:00
}
2019-06-03 00:06:41 +00:00
fmt.Fprintf(out, "\n<P><A HREF=\"/?url=https://github.com/tenox7/wrp/&w=%d&h=%d&s=%1.2f&c=%d%s\">Web Rendering Proxy Version %s</A></BODY>\n</HTML>\n", w, h, s, c, ion, version)
2019-05-29 08:29:01 +00:00
}
func imgServer(out http.ResponseWriter, req *http.Request) {
2019-06-01 01:20:55 +00:00
log.Printf("%s IMG Request for %s\n", req.RemoteAddr, req.URL.Path)
2019-06-03 05:23:41 +00:00
gifbuf, ok := gifmap[req.URL.Path]
if !ok || gifbuf.Bytes() == nil {
fmt.Fprintf(out, "Unable to find image %s\n", req.URL.Path)
log.Printf("Unable to find image %s\n", req.URL.Path)
return
}
2019-05-30 09:03:17 +00:00
defer delete(gifmap, req.URL.Path)
2019-05-30 01:02:29 +00:00
out.Header().Set("Content-Type", "image/gif")
out.Header().Set("Content-Length", strconv.Itoa(len(gifbuf.Bytes())))
out.Write(gifbuf.Bytes())
2019-05-30 09:03:17 +00:00
out.(http.Flusher).Flush()
2019-05-29 08:29:01 +00:00
}
2019-06-01 01:20:55 +00:00
func mapServer(out http.ResponseWriter, req *http.Request) {
2019-06-02 22:55:58 +00:00
log.Printf("%s ISMAP Request for %s [%+v]\n", req.RemoteAddr, req.URL.Path, req.URL.RawQuery)
var loc string
var x, y int64
n, err := fmt.Sscanf(req.URL.RawQuery, "%d,%d", &x, &y)
if err != nil || n != 2 {
fmt.Fprintf(out, "n=%d, err=%s\n", n, err)
2019-06-02 23:05:36 +00:00
log.Printf("%s ISMAP n=%d, err=%s\n", req.RemoteAddr, n, err)
2019-06-02 22:55:58 +00:00
return
}
2019-06-03 05:23:41 +00:00
is, ok := ismap[req.URL.Path]
if !ok || is == nil {
fmt.Fprintf(out, "Unable to find map %s\n", req.URL.Path)
log.Printf("Unable to find map %s\n", req.URL.Path)
return
}
2019-06-02 22:55:58 +00:00
defer delete(ismap, req.URL.Path)
for _, i := range is {
if x >= i.xmin && x <= i.xmax && y >= i.ymin && y <= i.ymax {
loc = i.url
}
}
if len(loc) < 1 {
loc = is[0].url
}
log.Printf("%s ISMAP Redirect to: http://%s%s\n", req.RemoteAddr, req.Context().Value(http.LocalAddrContextKey), loc)
http.Redirect(out, req, fmt.Sprintf("http://%s%s", req.Context().Value(http.LocalAddrContextKey), loc), 301)
2019-06-01 01:20:55 +00:00
}
2019-06-03 00:06:41 +00:00
func capture(gourl string, w int64, h int64, s float64, co int, p int64, i bool, c string, out http.ResponseWriter) {
2019-05-29 08:52:28 +00:00
var nodes []*cdp.Node
ctxx := chromedp.FromContext(ctx)
2019-05-30 09:03:17 +00:00
var pngbuf []byte
var gifbuf bytes.Buffer
2019-05-30 01:48:07 +00:00
var loc string
2019-05-31 07:19:10 +00:00
var res *runtime.RemoteObject
2019-06-02 09:25:18 +00:00
is := make([]Ismap, 0)
2019-06-02 23:18:53 +00:00
var ion string
2019-05-29 08:29:01 +00:00
2019-06-02 23:05:36 +00:00
log.Printf("%s Processing Caputure Request for %s\n", c, gourl)
2019-05-29 09:39:06 +00:00
2019-05-30 09:03:17 +00:00
// Run ChromeDP Magic
2019-05-31 07:41:46 +00:00
err := chromedp.Run(ctx,
2019-05-31 01:08:48 +00:00
emulation.SetDeviceMetricsOverride(w, h, s, false),
2019-05-30 01:48:07 +00:00
chromedp.Navigate(gourl),
2019-05-31 23:41:25 +00:00
chromedp.Evaluate(fmt.Sprintf("window.scrollTo(0, %d);", p*int64(float64(h)*float64(0.9))), &res),
2019-05-31 07:19:10 +00:00
chromedp.Sleep(time.Second*1),
2019-05-30 09:03:17 +00:00
chromedp.CaptureScreenshot(&pngbuf),
2019-05-30 01:48:07 +00:00
chromedp.Location(&loc),
2019-05-29 08:52:28 +00:00
chromedp.Nodes("a", &nodes, chromedp.ByQueryAll))
2019-05-29 08:29:01 +00:00
2019-05-31 07:41:46 +00:00
if err != nil {
2019-06-02 23:05:36 +00:00
log.Printf("%s %s", c, err)
2019-05-31 07:41:46 +00:00
fmt.Fprintf(out, "<BR>%s<BR>", err)
return
}
2019-06-02 23:05:36 +00:00
log.Printf("%s Landed on: %s, Nodes: %d\n", c, loc, len(nodes))
2019-05-30 01:48:07 +00:00
2019-05-30 09:03:17 +00:00
// Process Screenshot Image
2019-05-30 09:15:52 +00:00
bytes.NewReader(pngbuf).Seek(0, 0)
2019-05-30 09:03:17 +00:00
img, err := png.Decode(bytes.NewReader(pngbuf))
2019-05-30 01:48:07 +00:00
if err != nil {
2019-06-02 23:05:36 +00:00
log.Printf("%s Failed to decode screenshot: %s\n", c, err)
fmt.Fprintf(out, "<BR>Unable to decode page screenshot:<BR>%s<BR>\n", err)
return
2019-05-30 01:48:07 +00:00
}
gifbuf.Reset()
2019-06-03 00:06:41 +00:00
err = gif.Encode(&gifbuf, img, &gif.Options{NumColors: co})
2019-05-30 07:53:59 +00:00
if err != nil {
2019-06-02 23:05:36 +00:00
log.Printf("%s Failed to encode GIF: %s\n", c, err)
2019-05-30 07:53:59 +00:00
fmt.Fprintf(out, "<BR>Unable to encode GIF:<BR>%s<BR>\n", err)
return
}
2019-06-02 09:25:18 +00:00
seq := rand.Intn(9999)
imgpath := fmt.Sprintf("/img/%04d.gif", seq)
mappath := fmt.Sprintf("/map/%04d.map", seq)
2019-06-02 23:05:36 +00:00
log.Printf("%s Encoded GIF image: %s, Size: %dKB\n", c, imgpath, len(gifbuf.Bytes())/1024)
2019-05-30 09:03:17 +00:00
gifmap[imgpath] = gifbuf
2019-05-30 01:02:29 +00:00
2019-05-30 09:03:17 +00:00
// Process Nodes
2019-05-30 01:48:07 +00:00
base, _ := url.Parse(loc)
2019-06-02 09:25:18 +00:00
if i {
2019-06-03 00:07:31 +00:00
fmt.Fprintf(out, "<A HREF=\"%s\"><IMG SRC=\"%s\" ALT=\"wrp\" BORDER=\"0\" ISMAP></A>", mappath, imgpath)
2019-06-03 00:06:41 +00:00
is = append(is, Ismap{xmin: -1, xmax: -1, ymin: -1, ymax: -1, url: fmt.Sprintf("/?url=%s&w=%d&h=%d&s=%1.2f&c=%d&i=on", loc, w, h, s, co)})
2019-06-02 23:18:53 +00:00
ion = "&i=on"
2019-06-01 01:20:55 +00:00
} else {
2019-06-03 00:07:31 +00:00
fmt.Fprintf(out, "<IMG SRC=\"%s\" ALT=\"wrp\" BORDER=\"0\" USEMAP=\"#map\">\n<MAP NAME=\"map\">\n", imgpath)
2019-06-01 01:20:55 +00:00
}
2019-05-29 08:29:01 +00:00
2019-05-29 08:52:28 +00:00
for _, n := range nodes {
b, err := dom.GetBoxModel().WithNodeID(n.NodeID).Do(cdp.WithExecutor(ctx, ctxx.Target))
2019-05-30 01:48:07 +00:00
if err != nil {
continue
}
tgt, err := base.Parse(n.AttributeValue("href"))
if err != nil {
continue
2019-05-29 08:52:28 +00:00
}
2019-06-03 00:06:41 +00:00
target := fmt.Sprintf("/?url=%s&w=%d&h=%d&s=%1.2f&c=%d%s", tgt, w, h, s, co, ion) // no page# here
2019-05-29 08:29:01 +00:00
2019-05-30 01:48:07 +00:00
if len(b.Content) > 6 && len(target) > 7 {
2019-06-02 09:25:18 +00:00
if i {
is = append(is, Ismap{
xmin: int64(b.Content[0] * s), ymin: int64(b.Content[1] * s),
xmax: int64(b.Content[4] * s), ymax: int64(b.Content[5] * s),
url: target})
2019-06-01 01:20:55 +00:00
} else {
2019-06-01 08:17:50 +00:00
fmt.Fprintf(out, "<AREA SHAPE=\"RECT\" COORDS=\"%.f,%.f,%.f,%.f\" ALT=\"%s\" TITLE=\"%s\" HREF=\"%s\">\n",
b.Content[0]*s, b.Content[1]*s, b.Content[4]*s, b.Content[5]*s, n.AttributeValue("href"), n.AttributeValue("href"), target)
2019-06-01 01:20:55 +00:00
}
2019-05-29 08:52:28 +00:00
}
}
2019-05-29 08:29:01 +00:00
2019-06-02 23:05:36 +00:00
if i {
log.Printf("%s Encoded ISMAP %s\n", c, mappath)
} else {
2019-06-01 01:20:55 +00:00
fmt.Fprintf(out, "</MAP>\n")
}
2019-05-30 09:03:17 +00:00
out.(http.Flusher).Flush()
2019-06-02 23:05:36 +00:00
log.Printf("%s Done with caputure for %s\n", c, gourl)
2019-06-02 09:25:18 +00:00
ismap[mappath] = is
2019-05-29 08:29:01 +00:00
}
func main() {
2019-05-29 08:52:28 +00:00
var addr string
var head, headless bool
var debug bool
2019-05-29 08:52:28 +00:00
flag.StringVar(&addr, "l", ":8080", "Listen address:port, default :8080")
flag.BoolVar(&head, "h", false, "Headed mode - display browser window")
flag.BoolVar(&debug, "d", false, "Debug ChromeDP")
2019-05-29 08:52:28 +00:00
flag.Parse()
if head {
headless = false
} else {
headless = true
}
opts := append(chromedp.DefaultExecAllocatorOptions[:],
chromedp.Flag("headless", headless),
)
actx, cancel := chromedp.NewExecAllocator(context.Background(), opts...)
defer cancel()
if debug {
2019-06-04 08:20:19 +00:00
ctx, cancel := chromedp.NewContext(actx, chromedp.WithDebugf(log.Printf))
} else {
2019-06-04 08:20:19 +00:00
ctx, cancel := chromedp.NewContext(actx)
}
defer cancel()
2019-05-30 09:03:17 +00:00
rand.Seed(time.Now().UnixNano())
2019-05-29 08:52:28 +00:00
http.HandleFunc("/", pageServer)
2019-05-30 09:03:17 +00:00
http.HandleFunc("/img/", imgServer)
2019-06-01 01:20:55 +00:00
http.HandleFunc("/map/", mapServer)
2019-05-30 01:47:03 +00:00
http.HandleFunc("/favicon.ico", http.NotFound)
2019-06-02 23:24:46 +00:00
log.Printf("Web Rendering Proxy Version %s\n", version)
2019-05-31 06:40:43 +00:00
log.Printf("Starting WRP http server on %s\n", addr)
2019-05-29 08:52:28 +00:00
http.ListenAndServe(addr, nil)
2019-05-29 08:29:01 +00:00
}