用 WebBrowser 访问百度

用 WebBrowser 访问百度

#light

open System
open System.Windows.Forms
open System.IO
open System.Text

let (html:HtmlDocument ref) = ref null

let getPage (url:string) =
// let (html:HtmlDocument ref) = ref null

let handler (sender:obj) (e: WebBrowserDocumentCompletedEventArgs) =
// let handler (sender:obj) (e: _) =
let wb = sender :?> (WebBrowser)
html := (wb.Document)

use wb = new WebBrowser()
wb.Visible<-true
wb.DocumentCompleted.Add(handler wb)
wb.Navigate(url)

while wb.ReadyState <> WebBrowserReadyState.Complete do
Application.DoEvents()
html

let writeToFile filename obj =
let fp = __SOURCE_DIRECTORY__ + "\\" + filename
File.AppendAllText( fp, obj, Encoding.Default )

getPage "http://www.baidu.com"

html.Value.All.Item("wd").SetAttribute("value","XXXXXX")
html.Value.All.Item("su").InvokeMember("click")
let url = (html.Value.All.Item("wd")).OuterHtml

getPage ("http://www.baidu.com" + url )

writeToFile "baidu-Search.txt"
([ for i in html.Value.All do yield i.OuterText] |> Seq.toList |> Seq.reduce (fun a b -> a + b))

后面的一句 getPage 非常重要。没有这一句，运行会没有输出。但是，单步运行是可以的。