求正则表达式把一段字符串形式的静态html的table表里的每个td或者th里的文字值取出来

w091b 2013-11-25 04:47:58

要求如题目所言



<thead>

  <tr>

      <td class="emptyRow" colspan="5"></td>

  </tr>

  <tr>

      <th style="width: 40% ;text-align:left" axis="length" colspan="2" scope="col">星期一, 十月 28, 2013 </th>

      <th style="width: 30% ;text-align:left " axis="length" scope="col">位置</th>

      <th style="width: 9%;text-align:left" axis="length" scope="col">时间</th>

      <th style="width: 25% ;text-align:left" class="lastChild" axis="length" scope="col">件</th>

			

   </tr>

</thead>

<tbody>

   <tr>

	<td style="width: 5% ;text-align:left" class="">16</td>

	<td style="text-align:left" class="">已派送并签收: PUSHPA</td>

	<td style="text-align:left" class=""> PORT OUIS </td>

	<td class="">09:08</td>

	<td class="lastChild "><!--start contentteaser -->

	   <div class="dhl">

		<div>

	          <!--optional: start Expandable links -->

		   <div class="article_list">

		    <div class="dijitArticleTitlePane dijitContentPane" id="dijit_ArticleTitlePane_1" title="" widgetid="dijit_ArticleTitlePane_1">

	<div dojoattachpoint="focusNode" class="dijitArticleTitlePaneTitle dijitOpen" wairole="button" tabindex="0" dojoattachevent="onclick:toggle,onkeypress: _onTitleKey,onfocus:_handleFocus,onblur:_handleFocus" role="button" aria-haspopup="true">

		   <div class="dijitInline dijitArrowNode" dojoattachpoint="arrowNode"><span class="dijitArrowNodeInner" dojoattachpoint="arrowNodeInner">-</span></div>

		   <div class="dijitArticleTitlePaneTextNode" dojoattachpoint="titleNode" style="-moz-user-select: none;">1 件</div>

	      </div>

	<div dojoattachpoint="hideNode" class="dijitArticleTitlePaneContentOuter" style="">

		<div dojoattachpoint="wipeNode" class="dijitReset" style="overflow: hidden; height: auto;">

			<div tabindex="-1" wairole="region" dojoattachpoint="containerNode" class="dijitArticleTitlePaneContentInner" role="region" aria-labelledby="">

				<!-- nested divs because wipeIn()/wipeOut() doesn't work right on node w/padding etc.  Put padding on inner div. -->

			

			<div class="ArticleTitleContent">

			<div>JD014557964330001499</div>			

			</div>			

			</div>

		</div>

	</div>

</div>

			</div>

			<!--optional: end Expandable links --> 

			<div class="clearAll"> </div>

			</div>

			</div>

			<!--end contentteaser --></td>

		</tr>

	</tbody>