wrp/wrp.go

236 lines
7.2 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/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"
2019-06-26 00:07:43 +00:00
"encoding/json"
"encoding/base64"
"net"
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/chromedp"
"github.com/ericpauley/go-quantize/quantize"
2019-05-29 08:29:01 +00:00
)
var (
2019-06-02 23:19:08 +00:00
version = "3.0"
2019-06-18 06:53:22 +00:00
srv http.Server
2019-06-04 08:23:46 +00:00
ctx context.Context
cancel context.CancelFunc
2019-06-02 23:19:08 +00:00
gifmap = make(map[string]bytes.Buffer)
2019-05-29 08:29:01 +00:00
)
2019-06-26 00:07:43 +00:00
// Params - Page Configuration Parameters
type Params struct {
U string // url
P int64 // page
W int64 // width
H int64 // height
S float64 // scale
C int64 // #colors
r string // remote addr
l string // local addr
}
2019-06-24 07:40:34 +00:00
2019-06-26 00:07:43 +00:00
func (p *Params) parseParams(req *http.Request) {
req.ParseForm()
p.r = string(req.RemoteAddr)
p.l = req.Context().Value(http.LocalAddrContextKey).(*net.TCPAddr).IP.String()
p.U = req.FormValue("url")
var x,y int64
fmt.Sscanf(req.URL.RawQuery, "%d,%d", &x, &y)
log.Printf("%s Page Request for %s url=\"%s\" [%+v]\n", req.RemoteAddr, req.URL.Path, p.U, req.URL.RawQuery)
if len(p.U) > 1 && !strings.HasPrefix(p.U, "http") {
p.U = fmt.Sprintf("http://www.google.com/search?q=%s", url.QueryEscape(p.U))
}
p.P, _ = strconv.ParseInt(req.FormValue("p"), 10, 64)
2019-06-04 08:20:41 +00:00
if req.FormValue("pg") == "Dn" {
2019-06-26 00:07:43 +00:00
p.P++
2019-06-04 08:20:41 +00:00
} else if req.FormValue("pg") == "Up" {
2019-06-26 00:07:43 +00:00
p.P--
} else {
2019-06-26 00:07:43 +00:00
p.P = 0
2019-05-31 23:41:25 +00:00
}
2019-06-26 00:07:43 +00:00
p.W, _ = strconv.ParseInt(req.FormValue("w"), 10, 64)
if p.P < 10 {
p.P = 1024
2019-05-31 01:08:48 +00:00
}
2019-06-26 00:07:43 +00:00
p.H, _ = strconv.ParseInt(req.FormValue("h"), 10, 64)
if p.H < 10 {
p.H = 768
2019-05-29 08:52:28 +00:00
}
2019-06-26 00:07:43 +00:00
p.S, _ = strconv.ParseFloat(req.FormValue("s"), 64)
if p.S < 0.1 {
p.S = 1.0
2019-05-31 01:08:48 +00:00
}
2019-06-26 00:07:43 +00:00
p.C, _ = strconv.ParseInt(req.FormValue("c"), 10, 64)
if p.C < 2 || p.C > 256 {
p.C = 256
2019-06-03 00:06:41 +00:00
}
2019-06-26 00:07:43 +00:00
m, _ := json.Marshal(*p)
log.Printf("DEBUG: Struct: %+v Json: %s Base64: %s\n", p, string(m), base64.StdEncoding.EncodeToString([]byte(m)))
}
func pageServer(out http.ResponseWriter, req *http.Request) {
var p Params
p.parseParams(req)
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)
2019-06-26 00:07:43 +00:00
fmt.Fprintf(out, "<HTML>\n<HEAD><TITLE>WRP %s</TITLE></HEAD>\n<BODY BGCOLOR=\"#F0F0F0\">\n", p.U)
fmt.Fprintf(out, "<FORM ACTION=\"/\"><INPUT TYPE=\"TEXT\" NAME=\"url\" VALUE=\"%s\" SIZE=\"20\">", p.U)
2019-06-03 00:06:41 +00:00
fmt.Fprintf(out, "<INPUT TYPE=\"SUBMIT\" VALUE=\"Go\"> \n")
2019-06-04 08:20:41 +00:00
fmt.Fprintf(out, "<INPUT TYPE=\"SUBMIT\" NAME=\"pg\" VALUE=\"Up\"> \n")
2019-06-26 00:07:43 +00:00
fmt.Fprintf(out, "<INPUT TYPE=\"TEXT\" NAME=\"p\" VALUE=\"%d\" SIZE=\"2\"> \n", p.P)
2019-06-04 08:20:41 +00:00
fmt.Fprintf(out, "<INPUT TYPE=\"SUBMIT\" NAME=\"pg\" VALUE=\"Dn\"> \n")
2019-06-26 00:07:43 +00:00
fmt.Fprintf(out, "W <INPUT TYPE=\"TEXT\" NAME=\"w\" VALUE=\"%d\" SIZE=\"4\"> \n", p.W)
fmt.Fprintf(out, "H <INPUT TYPE=\"TEXT\" NAME=\"h\" VALUE=\"%d\" SIZE=\"4\"> \n", p.H)
fmt.Fprintf(out, "S <INPUT TYPE=\"TEXT\" NAME=\"s\" VALUE=\"%1.2f\" SIZE=\"3\"> \n", p.S)
fmt.Fprintf(out, "C <INPUT TYPE=\"TEXT\" NAME=\"c\" VALUE=\"%d\" SIZE=\"3\"> \n", p.C)
2019-06-04 08:20:41 +00:00
fmt.Fprintf(out, "</FORM><BR>\n")
2019-06-26 00:07:43 +00:00
fmt.Fprintf(out, "\n<P><A HREF=\"/?url=https://github.com/tenox7/wrp/&w=%d&h=%d&s=%1.2f&c=%d\">Web Rendering Proxy Version %s</A> | <A HREF=\"/shutdown/\">Shutdown WRP</A></BODY>\n</HTML>\n", p.W, p.H, p.S, p.C, 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
}
2019-06-26 00:07:43 +00:00
func (p Params) capture(c string, out http.ResponseWriter) {
2019-05-29 08:52:28 +00:00
var nodes []*cdp.Node
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
2019-05-29 08:29:01 +00:00
2019-06-26 00:07:43 +00:00
log.Printf("%s Processing Capture Request for %s\n", c, p.U)
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,
2019-06-26 00:07:43 +00:00
emulation.SetDeviceMetricsOverride(int64(float64(p.W)/p.S), int64(float64(p.H)/p.S), p.S, false),
chromedp.Navigate(p.U),
chromedp.Evaluate(fmt.Sprintf("window.scrollTo(0, %d);", p.P*int64(float64(p.H)*float64(0.9))), &res),
2019-05-31 07:19:10 +00:00
chromedp.Sleep(time.Second*1),
chromedp.Location(&loc))
2019-05-29 08:29:01 +00:00
2019-05-31 07:41:46 +00:00
if err != nil {
if err.Error() == "context canceled" {
log.Printf("%s Contex cancelled, try again", c)
fmt.Fprintf(out, "<BR>%s<BR> -- restarting, try again", err)
ctx, cancel = chromedp.NewContext(context.Background())
} else {
log.Printf("%s %s", c, err)
fmt.Fprintf(out, "<BR>%s<BR>", err)
}
2019-06-17 07:21:32 +00:00
return
2019-05-31 07:41:46 +00:00
}
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
2019-06-24 07:40:34 +00:00
err = chromedp.Run(ctx, chromedp.CaptureScreenshot(&pngbuf))
if err != nil {
log.Printf("%s Failed to capture screenshot: %s\n", c, err)
fmt.Fprintf(out, "<BR>Unable to capture screenshot:<BR>%s<BR>\n", err)
return
}
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()
2019-06-26 00:07:43 +00:00
err = gif.Encode(&gifbuf, img, &gif.Options{NumColors: int(p.C), Quantizer: quantize.MedianCutQuantizer{}})
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
}
2019-06-26 00:07:43 +00:00
imgpath := fmt.Sprintf("/img/%04d.gif", rand.Intn(9999))
log.Printf("%s Encoded GIF image: %s, Size: %dKB, Colors: %d\n", c, imgpath, len(gifbuf.Bytes())/1024, p.C)
2019-05-30 09:03:17 +00:00
gifmap[imgpath] = gifbuf
2019-05-30 01:02:29 +00:00
2019-06-24 07:40:34 +00:00
// Gif location
fmt.Fprintf(out, "<A HREF=\"/w=100/h=100/s=1.0/pg=3/url=%s\"><IMG SRC=\"%s\" ALT=\"wrp\" BORDER=\"0\" ISMAP></A>", loc, imgpath)
2019-05-29 08:29:01 +00:00
2019-05-30 09:03:17 +00:00
out.(http.Flusher).Flush()
2019-06-26 00:07:43 +00:00
log.Printf("%s Done with caputure for %s\n", c, p.U)
2019-05-29 08:29:01 +00:00
}
2019-06-18 06:53:22 +00:00
func haltServer(out http.ResponseWriter, req *http.Request) {
log.Printf("%s Shutdown Request for %s\n", req.RemoteAddr, req.URL.Path)
defer cancel()
srv.Shutdown(context.Background())
}
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 08:23:46 +00:00
ctx, cancel = chromedp.NewContext(actx, chromedp.WithDebugf(log.Printf))
} else {
2019-06-04 08:23:46 +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)
2019-06-18 06:53:22 +00:00
http.HandleFunc("/shutdown/", haltServer)
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)
2019-06-18 06:53:22 +00:00
srv.Addr = addr
err := srv.ListenAndServe()
if err != nil {
log.Fatal(err)
}
2019-05-29 08:29:01 +00:00
}