ref: 94389da8017a2bcca3cb29a4f954c928d577a31a
dir: /browser/website.go/
package browser
import (
"github.com/mjl-/duit"
"golang.org/x/net/html"
"golang.org/x/text/encoding"
"io/ioutil"
"net/url"
"github.com/psilva261/opossum"
"github.com/psilva261/opossum/domino"
"github.com/psilva261/opossum/nodes"
"github.com/psilva261/opossum/style"
"strings"
"sync"
)
const (
InitialLayout = iota
ClickRelayout
)
type Website struct {
duit.UI
opossum.ContentType
d *domino.Domino
mu sync.Mutex
html string
js []string
}
func (w *Website) Html() string {
w.mu.Lock()
defer w.mu.Unlock()
return w.html
}
func (w *Website) Js() []string {
w.mu.Lock()
defer w.mu.Unlock()
return append([]string{}, w.js...)
}
func (w *Website) layout(f opossum.Fetcher, htm string, layouting int) {
defer func() {
browser.statusBarMsg("", false)
}()
pass := func(htm string, csss ...string) (*html.Node, map[*html.Node]style.Map) {
if debugPrintHtml {
log.Printf("%v\n", htm)
}
var doc *html.Node
var err error
doc, err = html.ParseWithOptions(
strings.NewReader(htm),
html.ParseOptionEnableScripting(*ExperimentalJsInsecure),
)
if err != nil {
panic(err.Error())
}
log.Printf("Retrieving CSS Rules...")
var cssSize int
nodeMap := make(map[*html.Node]style.Map)
for i, css := range csss {
log.Printf("CSS size %v kB", cssSize/1024)
nm, err := style.FetchNodeMap(doc, css, 1280)
if err == nil {
log.Printf("[%v/%v] Fetch CSS Rules successful!", i+1, len(csss))
if debugPrintHtml {
log.Printf("%v", nm)
}
style.MergeNodeMaps(nodeMap, nm)
} else {
log.Errorf("Fetch CSS Rules failed: %v", err)
if *DebugDumpCSS {
ioutil.WriteFile("info.css", []byte(css), 0644)
}
}
}
return doc, nodeMap
}
log.Printf("1st pass")
doc, _ := pass(htm)
log.Printf("2nd pass")
log.Printf("Download style...")
csss := cssSrcs(f, doc)
doc, nodeMap := pass(htm, csss...)
// 3rd pass is only needed initially to load the scripts and set the goja VM
// state. During subsequent calls from click handlers that state is kept.
if *ExperimentalJsInsecure && layouting != ClickRelayout {
log.Printf("3rd pass")
nt := nodes.NewNodeTree(doc, style.Map{}, nodeMap, nil)
jsSrcs := domino.Srcs(nt)
downloads := make(map[string]string)
for _, src := range jsSrcs {
url, err := f.LinkedUrl(src)
if err != nil {
log.Printf("error parsing %v", src)
continue
}
log.Printf("Download %v", url)
buf, _, err := f.Get(url)
if err != nil {
log.Printf("error downloading %v", url)
continue
}
downloads[src] = string(buf)
}
codes := domino.Scripts(nt, downloads)
log.Infof("JS pipeline start")
if w.d != nil {
log.Infof("Stop existing JS instance")
w.d.Stop()
}
w.d = domino.NewDomino(htm, browser, nt)
w.d.Start()
jsProcessed, changed, err := processJS2(w.d, codes)
if changed && err == nil {
htm = jsProcessed
if debugPrintHtml {
log.Printf("%v\n", jsProcessed)
}
doc, nodeMap = pass(htm, csss...)
} else if err != nil {
log.Errorf("JS error: %v", err)
}
log.Infof("JS pipeline end")
}
var countHtmlNodes func(*html.Node) int
countHtmlNodes = func(n *html.Node) (num int) {
num++
for c := n.FirstChild; c != nil; c = c.NextSibling {
num += countHtmlNodes(c)
}
return
}
log.Printf("%v html nodes found...", countHtmlNodes(doc))
body := grep(doc, "body")
log.Printf("Layout website...")
scroller = NewScroll(
NodeToBox(0, browser, nodes.NewNodeTree(body, style.Map{}, nodeMap, &nodes.Node{})),
)
numElements := 0
TraverseTree(scroller, func(ui duit.UI) {
numElements++
})
w.UI = scroller
log.Printf("Layouting done (%v elements created)", numElements)
if numElements < 10 {
log.Errorf("Less than 10 elements layouted, seems css processing failed. Will layout without css")
scroller = NewScroll(
NodeToBox(0, browser, nodes.NewNodeTree(body, style.Map{}, make(map[*html.Node]style.Map), nil)),
)
w.UI = scroller
}
w.mu.Lock()
w.html = htm
w.mu.Unlock()
log.Flush()
}
func cssSrcs(f opossum.Fetcher, doc *html.Node) (csss []string) {
cssHrefs := style.Hrefs(doc)
inlines := make([]string, 0, 3)
ntAll := nodes.NewNodeTree(doc, style.Map{}, make(map[*html.Node]style.Map), nil)
inls := ntAll.FindAll("style")
for _, inl := range inls {
inlines = append(inlines, inl.ContentString(true))
}
csss = make([]string, 0, len(inlines)+len(cssHrefs))
csss = append(csss, style.AddOnCSS)
csss = append(csss, inlines...)
for _, href := range cssHrefs {
url, err := f.LinkedUrl(href)
if err != nil {
log.Printf("error parsing %v", href)
continue
}
log.Printf("Download %v", url)
buf, contentType, err := f.Get(url)
if err != nil {
log.Printf("error downloading %v", url)
continue
}
if contentType.IsCSS() {
csss = append(csss, string(buf))
} else {
log.Printf("css: unexpected %v", contentType)
}
}
return
}
func formData(n, submitBtn *html.Node) (data url.Values) {
data = make(url.Values)
nm := attr(*n, "name")
switch n.Data {
case "input", "select":
if attr(*n, "type") == "submit" && n != submitBtn {
return
}
if nm != "" {
data.Set(nm, attr(*n, "value"))
}
case "textarea":
nn := nodes.NewNodeTree(n, style.Map{}, make(map[*html.Node]style.Map), nil)
if nm != "" {
data.Set(nm, nn.ContentString(false))
}
}
for c := n.FirstChild; c != nil; c = c.NextSibling {
for k, vs := range formData(c, submitBtn) {
for _, v := range vs {
data.Add(k, v)
}
}
}
return
}
func escapeValues(ct opossum.ContentType, q url.Values) (qe url.Values) {
qe = make(url.Values)
enc := encoding.HTMLEscapeUnsupported(ct.Encoding().NewEncoder())
for k, vs := range q {
ke, err := enc.String(k)
if err != nil {
log.Errorf("string: %v", err)
ke = k
}
for _, v := range vs {
ve, err := enc.String(v)
if err != nil {
log.Errorf("string: %v", err)
ve = v
}
qe.Add(ke, ve)
}
}
return
}
func (b *Browser) submit(form *html.Node, submitBtn *html.Node) {
var err error
var buf []byte
var contentType opossum.ContentType
method := "GET" // TODO
if m := attr(*form, "method"); m != "" {
method = strings.ToUpper(m)
}
uri := b.URL()
if action := attr(*form, "action"); action != "" {
uri, err = b.LinkedUrl(action)
if err != nil {
log.Printf("error parsing %v", action)
return
}
}
if method == "GET" {
q := uri.Query()
for k, vs := range formData(form, submitBtn) {
q.Set(k, vs[0]) // TODO: what is with the rest?
}
uri.RawQuery = escapeValues(b.Website.ContentType, q).Encode()
buf, contentType, err = b.get(uri, true)
} else {
buf, contentType, err = b.PostForm(uri, formData(form, submitBtn))
}
if err != nil {
log.Errorf("submit form: %v", err)
return
}
if !contentType.IsHTML() {
log.Errorf("post: unexpected %v", contentType)
return
}
b.render(contentType, buf)
}