读书人

利用WebBrowser读取网页中表格的数据(

发布时间: 2012-06-15 19:37:05 作者: rapoo

利用WebBrowser读取网页中表格的数据(已有大部分代码,但读取失败)
比如这样一个网页:

HTML code
<html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=gb2312" /><title>无标题文档</title></head><body>    <table class="zhangdan_table" width="100%">      <tbody>        <tr class="table_top">          <td width="16%">编号</td>          <td width="19%">班别</td>          <td width="10%">年龄 </td>          <td width="25%">姓名 </td>          <td width="30%">出生年月 </td>        </tr>        <tr class="table_li">          <td><b>001</b> </td>          <td>一(1)</td>          <td><span>¥20</span> </td>          <td>小明</td>          <td>2000-08-01 </td>        </tr>      </tbody>    </table></body></html>





我想读取里面的数据,利用下面一段代码,但一直提示“找不到表格”
附代码的出处:http://topic.csdn.net/t/20060920/09/5033996.html



Delphi(Pascal) code
unit Unit1; 

interface

uses
Windows, Messages, SysUtils, Variants, Classes, Graphics, Controls, Forms,
Dialogs, MSHTML, StdCtrls, OleCtrls, SHDocVw;

type
TForm1 = class(TForm)
Button1: TButton;
WebBrowser1: TWebBrowser;
procedure FormCreate(Sender: TObject);
procedure Button1Click(Sender: TObject);
private
{ Private declarations }
public
{ Public declarations }
end;

var
Form1: TForm1;

implementation

{$R *.dfm}

function GetHtmlTableCell(aTable: IHTMLTable; aRow, aCol: Integer): IHTMLElement;
var
Row: IHTMLTableRow;

begin
Result := nil;
if aTable = nil then Exit;
if aTable.rows = nil then Exit;
Row := aTable.rows.item(aRow, aRow) as IHTMLTableRow;
if Row = nil then Exit;
Result := Row.cells.item(aCol, aCol) as IHTMLElement;
end;

function GetHtmlTable(aDoc: IHTMLDocument2; aIndex: Integer): IHTMLTable;
var
list: IHTMLElementCollection;
begin
Result := nil;
if aDoc = nil then Exit;
if aDoc.all = nil then Exit;
list := aDoc.all.tags('table ') as IHTMLElementCollection;
if list = nil then Exit;
Result := list.item(aIndex, aIndex) as IHTMLTable;
end;

function GetWebBrowserHtmlTableCellText(const AWebBrowser: TWebBrowser;
const TableIndex, RowIndex, ColIndex: Integer;
var ResValue: string): Boolean;
var
Docintf: IHTMLDocument2;
tblintf: IHTMLTable;
node: IHTMLElement;
begin
ResValue := ' ';
docintf := AWebBrowser.Document as IHTMLDocument2;
tblintf := GetHtmlTable(docintf, TableIndex);
node := GetHtmlTableCell(tblintf, RowIndex, ColIndex);
Result := node <> nil;
if Result then
ResValue := Trim(node.innerText);
end;

function GetHtmlTableRowHtml(aTable: IHTMLTable; aRow: Integer): IHTMLElement;
var
Row: IHTMLTableRow;
begin
Result := nil;
if aTable = nil then Exit;
if aTable.rows = nil then Exit;
Row := aTable.rows.item(aRow, aRow) as IHTMLTableRow;


if Row = nil then Exit;
Result := Row as IHTMLElement;
end;

function GetWebBrowserHtmlTableCellHtml(const AWebBrowser: TWebBrowser;
const TableIndex, RowIndex, ColIndex: Integer;
var ResValue: string): Boolean;
var
Docintf: IHTMLDocument2;
tblintf: IHTMLTable;
node: IHTMLElement;
begin
ResValue := ' ';
docintf := AWebBrowser.Document as IHTMLDocument2;
tblintf := GetHtmlTable(docintf, TableIndex);
node := GetHtmlTableCell(tblintf, RowIndex, ColIndex);
Result := node <> nil;
if Result then
ResValue := Trim(node.innerHTML);
end;


function GeHtmlTableHtml(aTable: IHTMLTable; aRow: Integer): IHTMLElement;
var
Row: IHTMLTableRow;
begin
Result := nil;
if aTable = nil then Exit;
if aTable.rows = nil then Exit;
Row := aTable.rows.item(aRow, aRow) as IHTMLTableRow;
if Row = nil then Exit;
Result := Row as IHTMLElement;
end;

function GetWebBrowserHtmlTableHtml(const AWebBrowser: TWebBrowser;
const TableIndex, RowIndex: Integer;
var ResValue: string): Boolean;
var
Docintf: IHTMLDocument2;
tblintf: IHTMLTable;
node: IHTMLElement;
begin
ResValue := ' ';
docintf := AWebBrowser.Document as IHTMLDocument2;
tblintf := GetHtmlTable(docintf, TableIndex);
node := GeHtmlTableHtml(tblintf, RowIndex);
Result := node <> nil;
if Result then
ResValue := node.innerHtml;
end;

procedure TForm1.FormCreate(Sender: TObject);
begin
WebBrowser1.Navigate('H:\ReadTable\00.html');
end;

procedure TForm1.Button1Click(Sender: TObject);
var
ResValue: string;
begin
if GetWebBrowserHtmlTableCellText(WebBrowser1, 0, 0, 0, ResValue) then
begin
ShowMessage(ResValue);
end else
begin
ShowMessage('指定的表格不存在 ');
end;
end;

end.




[解决办法]
list := aDoc.all.tags('table ') as IHTMLElementCollection;
改为
list := aDoc.all.tags('table') as IHTMLElementCollection;
去掉table后的空格

读书人网 >.NET

热点推荐