Macintosh-Tools/wrp
1
0
Fork 0
mirror of https://github.com/tenox7/wrp.git synced 2025-01-10 07:30:19 +00:00
Code Issues Projects Releases Wiki Activity
wrp/wrp.go

289 lines
8.8 KiB
Go
Raw Normal View History

add copyright
2019-05-29 18:53:05 -07:00
//
// WRP - Web Rendering Proxy
//
// Copyright (c) 2013-2018 Antoni Sawicki
// Copyright (c) 2019 Google LLC
//
gocdp initial checkin
2019-05-29 01:29:01 -07:00
package main
import (
log decode error instead of bailing out
2019-05-29 23:49:39 -07:00
"bytes"
add flags
2019-05-29 01:52:28 -07:00
"context"
"flag"
"fmt"
log decode error instead of bailing out
2019-05-29 23:49:39 -07:00
"image/gif"
"image/png"
add flags
2019-05-29 01:52:28 -07:00
"log"
randomized image url
2019-05-30 02:03:17 -07:00
"math/rand"
add flags
2019-05-29 01:52:28 -07:00
"net/http"
log decode error instead of bailing out
2019-05-29 23:49:39 -07:00
"net/url"
add flags
2019-05-29 01:52:28 -07:00
"strconv"
simple google search if url doesnt start with http
2019-05-31 00:36:53 -07:00
"strings"
add flags
2019-05-29 01:52:28 -07:00
"time"
gocdp initial checkin
2019-05-29 01:29:01 -07:00
add browser size setting
2019-05-29 02:39:06 -07:00
"github.com/chromedp/cdproto/emulation"
add simple page scrolling
2019-05-31 00:19:10 -07:00
"github.com/chromedp/cdproto/runtime"
add browser size setting
2019-05-29 02:39:06 -07:00
add flags
2019-05-29 01:52:28 -07:00
"github.com/chromedp/cdproto/cdp"
"github.com/chromedp/cdproto/dom"
"github.com/chromedp/chromedp"
added real color quantizer, number of colors works now
2019-06-06 01:06:36 -07:00
"github.com/ericpauley/go-quantize/quantize"
gocdp initial checkin
2019-05-29 01:29:01 -07:00
)
added more ismap cruft
2019-06-02 02:25:18 -07:00
// Ismap for server side processing
type Ismap struct {
xmin int64
ymin int64
xmax int64
ymax int64
url string
}
gocdp initial checkin
2019-05-29 01:29:01 -07:00
var (
version string
2019-06-02 16:19:08 -07:00
version = "3.0"
restored context
2019-06-04 01:23:46 -07:00
ctx context.Context
cancel context.CancelFunc
version string
2019-06-02 16:19:08 -07:00
gifmap = make(map[string]bytes.Buffer)
ismap = make(map[string][]Ismap)
gocdp initial checkin
2019-05-29 01:29:01 -07:00
)
better logging
2019-06-02 16:05:36 -07:00
func pageServer(out http.ResponseWriter, req *http.Request) {
req.ParseForm()
u := req.FormValue("url")
ISMAP checkbox
2019-05-31 01:56:34 -07:00
var istr string
version string
2019-06-02 16:19:08 -07:00
var ion string
ISMAP checkbox
2019-05-31 01:56:34 -07:00
var i bool
better logging
2019-06-02 16:05:36 -07:00
if req.FormValue("i") == "on" {
ISMAP checkbox
2019-05-31 01:56:34 -07:00
istr = "CHECKED"
i = true
version string
2019-06-02 16:19:08 -07:00
ion = "&i=on"
ISMAP checkbox
2019-05-31 01:56:34 -07:00
} else {
istr = ""
i = false
}
better logging
2019-06-02 16:05:36 -07:00
p, _ := strconv.ParseInt(req.FormValue("p"), 10, 64)
single line toolbar
2019-06-04 01:20:41 -07:00
if req.FormValue("pg") == "Dn" {
paginated scrolling
2019-05-31 16:41:25 -07:00
p++
single line toolbar
2019-06-04 01:20:41 -07:00
} else if req.FormValue("pg") == "Up" {
paginated scrolling
2019-05-31 16:41:25 -07:00
p--
default to 0 page when go button is pressed
2019-05-31 17:49:14 -07:00
} else {
p = 0
paginated scrolling
2019-05-31 16:41:25 -07:00
}
better logging
2019-06-02 16:05:36 -07:00
w, _ := strconv.ParseInt(req.FormValue("w"), 10, 64)
add width height and scale input boxes
2019-05-30 18:08:48 -07:00
if w < 10 {
w = 1024
}
better logging
2019-06-02 16:05:36 -07:00
h, _ := strconv.ParseInt(req.FormValue("h"), 10, 64)
add width height and scale input boxes
2019-05-30 18:08:48 -07:00
if h < 10 {
h = 768
add flags
2019-05-29 01:52:28 -07:00
}
better logging
2019-06-02 16:05:36 -07:00
s, _ := strconv.ParseFloat(req.FormValue("s"), 64)
add width height and scale input boxes
2019-05-30 18:08:48 -07:00
if s < 0.1 {
s = 1.0
}
added number of colors
2019-06-02 17:06:41 -07:00
c, _ := strconv.ParseInt(req.FormValue("c"), 10, 64)
if c < 2 || c > 256 {
c = 256
}
wrp.go: Fixed typos
2019-06-06 04:01:19 -04:00
log.Printf("%s Page Request for url=\"%s\" [%s]\n", req.RemoteAddr, u, req.URL.Path)
add flags
2019-05-29 01:52:28 -07:00
out.Header().Set("Content-Type", "text/html")
version branding etc
2019-06-02 16:24:46 -07:00
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)
single line toolbar
2019-06-04 01:20:41 -07:00
fmt.Fprintf(out, "<FORM ACTION=\"/\"><INPUT TYPE=\"TEXT\" NAME=\"url\" VALUE=\"%s\" SIZE=\"20\">", u)
added number of colors
2019-06-02 17:06:41 -07:00
fmt.Fprintf(out, "<INPUT TYPE=\"SUBMIT\" VALUE=\"Go\"> \n")
single line toolbar
2019-06-04 01:20:41 -07:00
fmt.Fprintf(out, "<INPUT TYPE=\"SUBMIT\" NAME=\"pg\" VALUE=\"Up\"> \n")
added number of colors
2019-06-02 17:06:41 -07:00
fmt.Fprintf(out, "<INPUT TYPE=\"TEXT\" NAME=\"p\" VALUE=\"%d\" SIZE=\"2\"> \n", p)
single line toolbar
2019-06-04 01:20:41 -07:00
fmt.Fprintf(out, "<INPUT TYPE=\"SUBMIT\" NAME=\"pg\" VALUE=\"Dn\"> \n")
fmt.Fprintf(out, "I <INPUT TYPE=\"CHECKBOX\" NAME=\"i\" %s> \n", istr)
fmt.Fprintf(out, "W <INPUT TYPE=\"TEXT\" NAME=\"w\" VALUE=\"%d\" SIZE=\"4\"> \n", w)
fmt.Fprintf(out, "H <INPUT TYPE=\"TEXT\" NAME=\"h\" VALUE=\"%d\" SIZE=\"4\"> \n", h)
fmt.Fprintf(out, "S <INPUT TYPE=\"TEXT\" NAME=\"s\" VALUE=\"%1.2f\" SIZE=\"3\"> \n", s)
added real color quantizer, number of colors works now
2019-06-06 01:06:36 -07:00
fmt.Fprintf(out, "C <INPUT TYPE=\"TEXT\" NAME=\"c\" VALUE=\"%d\" SIZE=\"3\"> \n", c)
single line toolbar
2019-06-04 01:20:41 -07:00
fmt.Fprintf(out, "</FORM><BR>\n")
obtain server address from context, -l no longer needed
2019-06-04 00:58:45 -07:00
if len(u) > 1 {
simple google search if url doesnt start with http
2019-05-31 00:36:53 -07:00
if strings.HasPrefix(u, "http") {
added number of colors
2019-06-02 17:06:41 -07:00
capture(u, w, h, s, int(c), p, i, req.RemoteAddr, out)
simple google search if url doesnt start with http
2019-05-31 00:36:53 -07:00
} else {
added number of colors
2019-06-02 17:06:41 -07:00
capture(fmt.Sprintf("http://www.google.com/search?q=%s", url.QueryEscape(u)), w, h, s, int(c), p, i, req.RemoteAddr, out)
simple google search if url doesnt start with http
2019-05-31 00:36:53 -07:00
}
add width height and scale input boxes
2019-05-30 18:08:48 -07:00
} else {
simple google search if url doesnt start with http
2019-05-31 00:36:53 -07:00
fmt.Fprintf(out, "No URL or search query specified")
add flags
2019-05-29 01:52:28 -07:00
}
added number of colors
2019-06-02 17:06:41 -07:00
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)
gocdp initial checkin
2019-05-29 01:29:01 -07:00
}
func imgServer(out http.ResponseWriter, req *http.Request) {
more ismap stuff
2019-05-31 18:20:55 -07:00
log.Printf("%s IMG Request for %s\n", req.RemoteAddr, req.URL.Path)
dont crash on missing image or map
2019-06-02 22:23:41 -07:00
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
}
randomized image url
2019-05-30 02:03:17 -07:00
defer delete(gifmap, req.URL.Path)
encode image as gif
2019-05-29 18:02:29 -07:00
out.Header().Set("Content-Type", "image/gif")
out.Header().Set("Content-Length", strconv.Itoa(len(gifbuf.Bytes())))
out.Write(gifbuf.Bytes())
randomized image url
2019-05-30 02:03:17 -07:00
out.(http.Flusher).Flush()
gocdp initial checkin
2019-05-29 01:29:01 -07:00
}
more ismap stuff
2019-05-31 18:20:55 -07:00
func mapServer(out http.ResponseWriter, req *http.Request) {
ismap server cruft
2019-06-02 15:55:58 -07:00
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)
better logging
2019-06-02 16:05:36 -07:00
log.Printf("%s ISMAP n=%d, err=%s\n", req.RemoteAddr, n, err)
ismap server cruft
2019-06-02 15:55:58 -07:00
return
}
dont crash on missing image or map
2019-06-02 22:23:41 -07:00
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
}
ismap server cruft
2019-06-02 15:55:58 -07:00
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
}
obtain server address from context, -l no longer needed
2019-06-04 00:58:45 -07:00
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)
more ismap stuff
2019-05-31 18:20:55 -07:00
}
added number of colors
2019-06-02 17:06:41 -07:00
func capture(gourl string, w int64, h int64, s float64, co int, p int64, i bool, c string, out http.ResponseWriter) {
add flags
2019-05-29 01:52:28 -07:00
var nodes []*cdp.Node
ctxx := chromedp.FromContext(ctx)
randomized image url
2019-05-30 02:03:17 -07:00
var pngbuf []byte
var gifbuf bytes.Buffer
resolve relative a href urls
2019-05-29 18:48:07 -07:00
var loc string
add simple page scrolling
2019-05-31 00:19:10 -07:00
var res *runtime.RemoteObject
added more ismap cruft
2019-06-02 02:25:18 -07:00
is := make([]Ismap, 0)
better handling of ismap variable
2019-06-02 16:18:53 -07:00
var ion string
gocdp initial checkin
2019-05-29 01:29:01 -07:00
wrp.go: Fixed typos
2019-06-06 04:01:19 -04:00
log.Printf("%s Processing Capture Request for %s\n", c, gourl)
add browser size setting
2019-05-29 02:39:06 -07:00
randomized image url
2019-05-30 02:03:17 -07:00
// Run ChromeDP Magic
chromedp.run error handling
2019-05-31 00:41:46 -07:00
err := chromedp.Run(ctx,
better scaling
2019-06-04 01:30:00 -07:00
emulation.SetDeviceMetricsOverride(int64(float64(w)/s), int64(float64(h)/s), s, false),
resolve relative a href urls
2019-05-29 18:48:07 -07:00
chromedp.Navigate(gourl),
paginated scrolling
2019-05-31 16:41:25 -07:00
chromedp.Evaluate(fmt.Sprintf("window.scrollTo(0, %d);", p*int64(float64(h)*float64(0.9))), &res),
add simple page scrolling
2019-05-31 00:19:10 -07:00
chromedp.Sleep(time.Second*1),
randomized image url
2019-05-30 02:03:17 -07:00
chromedp.CaptureScreenshot(&pngbuf),
resolve relative a href urls
2019-05-29 18:48:07 -07:00
chromedp.Location(&loc),
add flags
2019-05-29 01:52:28 -07:00
chromedp.Nodes("a", &nodes, chromedp.ByQueryAll))
gocdp initial checkin
2019-05-29 01:29:01 -07:00
chromedp.run error handling
2019-05-31 00:41:46 -07:00
if err != nil {
better logging
2019-06-02 16:05:36 -07:00
log.Printf("%s %s", c, err)
chromedp.run error handling
2019-05-31 00:41:46 -07:00
fmt.Fprintf(out, "<BR>%s<BR>", err)
return
}
better logging
2019-06-02 16:05:36 -07:00
log.Printf("%s Landed on: %s, Nodes: %d\n", c, loc, len(nodes))
resolve relative a href urls
2019-05-29 18:48:07 -07:00
randomized image url
2019-05-30 02:03:17 -07:00
// Process Screenshot Image
unexpected eof on windows
2019-05-30 02:15:52 -07:00
bytes.NewReader(pngbuf).Seek(0, 0)
randomized image url
2019-05-30 02:03:17 -07:00
img, err := png.Decode(bytes.NewReader(pngbuf))
resolve relative a href urls
2019-05-29 18:48:07 -07:00
if err != nil {
better logging
2019-06-02 16:05:36 -07:00
log.Printf("%s Failed to decode screenshot: %s\n", c, err)
log decode error instead of bailing out
2019-05-29 23:49:39 -07:00
fmt.Fprintf(out, "<BR>Unable to decode page screenshot:<BR>%s<BR>\n", err)
return
resolve relative a href urls
2019-05-29 18:48:07 -07:00
}
gifbuf.Reset()
added real color quantizer, number of colors works now
2019-06-06 01:06:36 -07:00
err = gif.Encode(&gifbuf, img, &gif.Options{NumColors: co, Quantizer: quantize.MedianCutQuantizer{}})
log encode errors
2019-05-30 00:53:59 -07:00
if err != nil {
better logging
2019-06-02 16:05:36 -07:00
log.Printf("%s Failed to encode GIF: %s\n", c, err)
log encode errors
2019-05-30 00:53:59 -07:00
fmt.Fprintf(out, "<BR>Unable to encode GIF:<BR>%s<BR>\n", err)
return
}
added more ismap cruft
2019-06-02 02:25:18 -07:00
seq := rand.Intn(9999)
imgpath := fmt.Sprintf("/img/%04d.gif", seq)
mappath := fmt.Sprintf("/map/%04d.map", seq)
added real color quantizer, number of colors works now
2019-06-06 01:06:36 -07:00
log.Printf("%s Encoded GIF image: %s, Size: %dKB, Colors: %d\n", c, imgpath, len(gifbuf.Bytes())/1024, co)
randomized image url
2019-05-30 02:03:17 -07:00
gifmap[imgpath] = gifbuf
encode image as gif
2019-05-29 18:02:29 -07:00
randomized image url
2019-05-30 02:03:17 -07:00
// Process Nodes
resolve relative a href urls
2019-05-29 18:48:07 -07:00
base, _ := url.Parse(loc)
added more ismap cruft
2019-06-02 02:25:18 -07:00
if i {
img border=0
2019-06-02 17:07:31 -07:00
fmt.Fprintf(out, "<A HREF=\"%s\"><IMG SRC=\"%s\" ALT=\"wrp\" BORDER=\"0\" ISMAP></A>", mappath, imgpath)
added number of colors
2019-06-02 17:06:41 -07:00
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)})
better handling of ismap variable
2019-06-02 16:18:53 -07:00
ion = "&i=on"
more ismap stuff
2019-05-31 18:20:55 -07:00
} else {
img border=0
2019-06-02 17:07:31 -07:00
fmt.Fprintf(out, "<IMG SRC=\"%s\" ALT=\"wrp\" BORDER=\"0\" USEMAP=\"#map\">\n<MAP NAME=\"map\">\n", imgpath)
more ismap stuff
2019-05-31 18:20:55 -07:00
}
gocdp initial checkin
2019-05-29 01:29:01 -07:00
add flags
2019-05-29 01:52:28 -07:00
for _, n := range nodes {
b, err := dom.GetBoxModel().WithNodeID(n.NodeID).Do(cdp.WithExecutor(ctx, ctxx.Target))
resolve relative a href urls
2019-05-29 18:48:07 -07:00
if err != nil {
continue
}
tgt, err := base.Parse(n.AttributeValue("href"))
if err != nil {
continue
add flags
2019-05-29 01:52:28 -07:00
}
added number of colors
2019-06-02 17:06:41 -07:00
target := fmt.Sprintf("/?url=%s&w=%d&h=%d&s=%1.2f&c=%d%s", tgt, w, h, s, co, ion) // no page# here
gocdp initial checkin
2019-05-29 01:29:01 -07:00
resolve relative a href urls
2019-05-29 18:48:07 -07:00
if len(b.Content) > 6 && len(target) > 7 {
added more ismap cruft
2019-06-02 02:25:18 -07:00
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})
more ismap stuff
2019-05-31 18:20:55 -07:00
} else {
formatting
2019-06-01 01:17:50 -07:00
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)
more ismap stuff
2019-05-31 18:20:55 -07:00
}
add flags
2019-05-29 01:52:28 -07:00
}
}
gocdp initial checkin
2019-05-29 01:29:01 -07:00
better logging
2019-06-02 16:05:36 -07:00
if i {
log.Printf("%s Encoded ISMAP %s\n", c, mappath)
} else {
more ismap stuff
2019-05-31 18:20:55 -07:00
fmt.Fprintf(out, "</MAP>\n")
}
assign to map before flush
2019-06-06 00:15:57 -07:00
ismap[mappath] = is
randomized image url
2019-05-30 02:03:17 -07:00
out.(http.Flusher).Flush()
better logging
2019-06-02 16:05:36 -07:00
log.Printf("%s Done with caputure for %s\n", c, gourl)
gocdp initial checkin
2019-05-29 01:29:01 -07:00
}
func main() {
add flags
2019-05-29 01:52:28 -07:00
var addr string
obtain server address from context, -l no longer needed
2019-06-04 00:58:45 -07:00
var head, headless bool
added headed mode and chromedp debug output
2019-06-03 17:50:16 -07:00
var debug bool
add flags
2019-05-29 01:52:28 -07:00
flag.StringVar(&addr, "l", ":8080", "Listen address:port, default :8080")
added headed mode and chromedp debug output
2019-06-03 17:50:16 -07:00
flag.BoolVar(&head, "h", false, "Headed mode - display browser window")
flag.BoolVar(&debug, "d", false, "Debug ChromeDP")
add flags
2019-05-29 01:52:28 -07:00
flag.Parse()
added headed mode and chromedp debug output
2019-06-03 17:50:16 -07:00
if head {
headless = false
} else {
headless = true
}
opts := append(chromedp.DefaultExecAllocatorOptions[:],
obtain server address from context, -l no longer needed
2019-06-04 00:58:45 -07:00
chromedp.Flag("headless", headless),
added headed mode and chromedp debug output
2019-06-03 17:50:16 -07:00
)
actx, cancel := chromedp.NewExecAllocator(context.Background(), opts...)
defer cancel()
if debug {
restored context
2019-06-04 01:23:46 -07:00
ctx, cancel = chromedp.NewContext(actx, chromedp.WithDebugf(log.Printf))
added headed mode and chromedp debug output
2019-06-03 17:50:16 -07:00
} else {
restored context
2019-06-04 01:23:46 -07:00
ctx, cancel = chromedp.NewContext(actx)
added headed mode and chromedp debug output
2019-06-03 17:50:16 -07:00
}
defer cancel()
randomized image url
2019-05-30 02:03:17 -07:00
rand.Seed(time.Now().UnixNano())
add flags
2019-05-29 01:52:28 -07:00
http.HandleFunc("/", pageServer)
randomized image url
2019-05-30 02:03:17 -07:00
http.HandleFunc("/img/", imgServer)
more ismap stuff
2019-05-31 18:20:55 -07:00
http.HandleFunc("/map/", mapServer)
added favico handler
2019-05-29 18:47:03 -07:00
http.HandleFunc("/favicon.ico", http.NotFound)
version branding etc
2019-06-02 16:24:46 -07:00
log.Printf("Web Rendering Proxy Version %s\n", version)
scale area map too
2019-05-30 23:40:43 -07:00
log.Printf("Starting WRP http server on %s\n", addr)
add flags
2019-05-29 01:52:28 -07:00
http.ListenAndServe(addr, nil)
gocdp initial checkin
2019-05-29 01:29:01 -07:00
}
Reference in New Issue Copy Permalink