selftestcrawljobhandler.html

用JAVA编写的,在做实验的时候留下来的,本来想删的,但是传上来,大家分享吧

<a name="80" href="#80">80</a>      }
<a name="81" href="#81">81</a>      
<a name="82" href="#82">82</a>      @Override
<a name="83" href="#83">83</a>      <strong>public</strong> <strong>void</strong> crawlStarted(String message) {
<a name="84" href="#84">84</a>      	<strong>super</strong>.crawlStarted(message);
<a name="85" href="#85">85</a>      	<strong>this</strong>.getCurrentJob().getController().
<a name="86" href="#86">86</a>      		addCrawlURIDispositionListener(<strong>this</strong>);
<a name="87" href="#87">87</a>      }
<a name="88" href="#88">88</a>  
<a name="89" href="#89">89</a>      <strong>public</strong> <strong>void</strong> crawlEnded(String sExitMessage)  {
<a name="90" href="#90">90</a>          TestResult result = <strong>null</strong>;
<a name="91" href="#91">91</a>          <strong>try</strong> {
<a name="92" href="#92">92</a>              <strong>super</strong>.crawlEnded(sExitMessage);
<a name="93" href="#93">93</a>  
<a name="94" href="#94">94</a>              <em class="comment">// At crawlEnded time, there is no current job.  Get the selftest</em>
<a name="95" href="#95">95</a>              <em class="comment">// job by pulling from the completedCrawlJobs queue.</em>
<a name="96" href="#96">96</a>              List completedCrawlJobs = getCompletedJobs();
<a name="97" href="#97">97</a>              <strong>if</strong> (completedCrawlJobs == <strong>null</strong> || completedCrawlJobs.size() &lt;= 0) {
<a name="98" href="#98">98</a>                  logger.severe(<span class="string">"Selftest job did not complete."</span>);
<a name="99" href="#99">99</a>              } <strong>else</strong> {
<a name="100" href="#100">100</a>                 <a href="../../../../org/archive/crawler/admin/CrawlJob.html">CrawlJob</a> job = (CrawlJob)completedCrawlJobs.
<a name="101" href="#101">101</a>                     get(completedCrawlJobs.size()-1);
<a name="102" href="#102">102</a>                 Test test = <strong>null</strong>;
<a name="103" href="#103">103</a>                 <strong>if</strong> (<strong>this</strong>.selfTestName != <strong>null</strong> &amp;&amp;
<a name="104" href="#104">104</a>                         <strong>this</strong>.selfTestName.length() > 0) {
<a name="105" href="#105">105</a>                     <em class="comment">// Run single selftest only.</em>
<a name="106" href="#106">106</a>                     <em class="comment">// Get class for the passed single selftest.</em>
<a name="107" href="#107">107</a>                     <em class="comment">// Assume test to run is in this package.</em>
<a name="108" href="#108">108</a>                     String <strong>this</strong>ClassName = <strong>this</strong>.getClass().getName();
<a name="109" href="#109">109</a>                     String pkg = thisClassName.
<a name="110" href="#110">110</a>                         substring(0, thisClassName.lastIndexOf('.'));
<a name="111" href="#111">111</a>                     <em class="comment">// All selftests end in 'SelfTest'.</em>
<a name="112" href="#112">112</a>                     String selftestClass = pkg + '.' + <strong>this</strong>.selfTestName +
<a name="113" href="#113">113</a>                         <span class="string">"SelfTest"</span>;
<a name="114" href="#114">114</a>                     <em class="comment">// Need to make a list.  Make an array first.</em>
<a name="115" href="#115">115</a>                     List classList = <strong>new</strong> ArrayList();
<a name="116" href="#116">116</a>                     classList.add(Class.forName(selftestClass));
<a name="117" href="#117">117</a>                     test = AllSelfTestCases.suite(<strong>this</strong>.selfTestUrl,
<a name="118" href="#118">118</a>                         job, job.getDirectory(), Heritrix.getHttpServer().
<a name="119" href="#119">119</a>                         getWebappPath(SELFTEST_WEBAPP), classList);
<a name="120" href="#120">120</a>                 } <strong>else</strong> {
<a name="121" href="#121">121</a>                     <em class="comment">// Run all tests.</em>
<a name="122" href="#122">122</a>                     test = AllSelfTestCases.suite(<strong>this</strong>.selfTestUrl,
<a name="123" href="#123">123</a>                         job, job.getDirectory(), Heritrix.getHttpServer().
<a name="124" href="#124">124</a>                         getWebappPath(SELFTEST_WEBAPP));
<a name="125" href="#125">125</a>                 }
<a name="126" href="#126">126</a>                 result = junit.textui.TestRunner.run(test);
<a name="127" href="#127">127</a>             }
<a name="128" href="#128">128</a>         } <strong>catch</strong> (Exception e) {
<a name="129" href="#129">129</a>             logger.info(<span class="string">"Failed running selftest analysis: "</span> + e.getMessage());
<a name="130" href="#130">130</a>         } <strong>finally</strong>  {
<a name="131" href="#131">131</a>             <em class="comment">// TODO: This technique where I'm calling shutdown directly means</em>
<a name="132" href="#132">132</a>             <em class="comment">// we bypass the running of other crawlended handlers.  Means</em>
<a name="133" href="#133">133</a>             <em class="comment">// that such as the statistics tracker have no chance to run so</em>
<a name="134" href="#134">134</a>             <em class="comment">// reports are never generated.  Fix -- but preserve 0 or 1 exit</em>
<a name="135" href="#135">135</a>             <em class="comment">// code.</em>
<a name="136" href="#136">136</a>             logger.info((<strong>new</strong> Date()).toString() + <span class="string">" Selftest "</span> +
<a name="137" href="#137">137</a>                 (result != <strong>null</strong> &amp;&amp; result.wasSuccessful()? <span class="string">"PASSED"</span>: <span class="string">"FAILED"</span>));
<a name="138" href="#138">138</a>             stop();
<a name="139" href="#139">139</a>             Heritrix.shutdown(((result !=  <strong>null</strong>) &amp;&amp; result.wasSuccessful())?
<a name="140" href="#140">140</a>                 0: 1);
<a name="141" href="#141">141</a>         }
<a name="142" href="#142">142</a>     }
<a name="143" href="#143">143</a> 
<a name="144" href="#144">144</a> 	<strong>public</strong> <strong>void</strong> crawledURIDisregard(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {
<a name="145" href="#145">145</a> 		<em class="comment">// TODO Auto-generated method stub</em>
<a name="146" href="#146">146</a> 	}
<a name="147" href="#147">147</a> 
<a name="148" href="#148">148</a> 	<strong>public</strong> <strong>void</strong> crawledURIFailure(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {
<a name="149" href="#149">149</a> 		<em class="comment">// TODO Auto-generated method stub</em>
<a name="150" href="#150">150</a> 	}
<a name="151" href="#151">151</a> 
<a name="152" href="#152">152</a> 	<strong>public</strong> <strong>void</strong> crawledURINeedRetry(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {
<a name="153" href="#153">153</a> 		<em class="comment">// TODO Auto-generated method stub</em>
<a name="154" href="#154">154</a> 	}
<a name="155" href="#155">155</a> 
<a name="156" href="#156">156</a> 	<strong>public</strong> <strong>void</strong> crawledURISuccessful(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {
<a name="157" href="#157">157</a> 		<em class="comment">// If curi ends in 'Checkpoint/index.html', then run a Checkpoint.</em>
<a name="158" href="#158">158</a> 		<strong>if</strong> (curi.toString().endsWith(<span class="string">"/Checkpoint/"</span>)) {
<a name="159" href="#159">159</a> 			<strong>this</strong>.getCurrentJob().getController().requestCrawlCheckpoint();
<a name="160" href="#160">160</a> 		}
<a name="161" href="#161">161</a> 	}
<a name="162" href="#162">162</a> }
</pre>
<hr/><div id="footer">This page was automatically generated by <a href="http://maven.apache.org/">Maven</a></div></body>
</html>