📄 checkpointselftest.html
字号:
<a name="90" href="#90">90</a> <em class="comment">// Get order file from checkpoint dir.</em><a name="91" href="#91">91</a> File order = getFile(cpdir, <span class="string">"order.xml"</span>);<a name="92" href="#92">92</a> <a href="../../../../org/archive/crawler/settings/XMLSettingsHandler.html">XMLSettingsHandler</a> handler =<a name="93" href="#93">93</a> <strong>new</strong> <a href="../../../../org/archive/crawler/settings/XMLSettingsHandler.html">XMLSettingsHandler</a>(order);<a name="94" href="#94">94</a> handler.initialize();<a name="95" href="#95">95</a> <em class="comment">// Set recover-path to be this checkpoint dir.</em><a name="96" href="#96">96</a> handler.getOrder().setAttribute(<a name="97" href="#97">97</a> <strong>new</strong> Attribute(CrawlOrder.ATTR_RECOVER_PATH, cpdir.toString()));<a name="98" href="#98">98</a> <a href="../../../../org/archive/crawler/datamodel/Checkpoint.html">Checkpoint</a> cp =<a name="99" href="#99">99</a> CrawlController.getCheckpointRecover(handler.getOrder());<a name="100" href="#100">100</a> <strong>if</strong> (cp == <strong>null</strong>) {<a name="101" href="#101">101</a> <strong>throw</strong> <strong>new</strong> NullPointerException(<span class="string">"Failed read of checkpoint object"</span>);<a name="102" href="#102">102</a> }<a name="103" href="#103">103</a> <a href="../../../../org/archive/crawler/framework/CrawlController.html">CrawlController</a> c = (MBeanCrawlController)CheckpointUtils.<a name="104" href="#104">104</a> readObjectFromFile(MBeanCrawlController.<strong>class</strong>, cpdir);<a name="105" href="#105">105</a> c.initialize(handler);<a name="106" href="#106">106</a> c.addCrawlStatusListener(<strong>this</strong>);<a name="107" href="#107">107</a> c.addCrawlURIDispositionListener(<strong>this</strong>);<a name="108" href="#108">108</a> c.requestCrawlStart();<a name="109" href="#109">109</a> LOG.info(<span class="string">"Recover from selftest crawl started using "</span> +<a name="110" href="#110">110</a> order.toString() + <span class="string">"."</span>);<a name="111" href="#111">111</a> <em class="comment">// Wait here a while till its up and running?</em><a name="112" href="#112">112</a> <strong>while</strong>(!<strong>this</strong>.crawlEnded) {<a name="113" href="#113">113</a> LOG.info(<span class="string">"Waiting on recovered crawl to finish"</span>);<a name="114" href="#114">114</a> Thread.sleep(1000);<a name="115" href="#115">115</a> }<a name="116" href="#116">116</a> }<a name="117" href="#117">117</a> <a name="118" href="#118">118</a> <strong>private</strong> File getFile(<strong>final</strong> File parent, <strong>final</strong> String name)<a name="119" href="#119">119</a> throws IOException {<a name="120" href="#120">120</a> File f = <strong>new</strong> File(parent, name);<a name="121" href="#121">121</a> <strong>if</strong> (!f.exists()) {<a name="122" href="#122">122</a> <strong>throw</strong> <strong>new</strong> FileNotFoundException(f.getAbsolutePath());<a name="123" href="#123">123</a> }<a name="124" href="#124">124</a> <strong>if</strong> (!f.canRead()) {<a name="125" href="#125">125</a> <strong>throw</strong> <strong>new</strong> IOException(<span class="string">"Can't read "</span> + f.getAbsolutePath());<a name="126" href="#126">126</a> }<a name="127" href="#127">127</a> <strong>return</strong> f;<a name="128" href="#128">128</a> }<a name="129" href="#129">129</a> <a name="130" href="#130">130</a> <strong>public</strong> <strong>void</strong> crawlCheckpoint(File checkpointDir) throws Exception {<a name="131" href="#131">131</a> <em class="comment">// TODO Auto-generated method stub</em><a name="132" href="#132">132</a> <a name="133" href="#133">133</a> }<a name="134" href="#134">134</a> <a name="135" href="#135">135</a> <strong>public</strong> <strong>void</strong> crawlEnded(String sExitMessage) {<a name="136" href="#136">136</a> <strong>this</strong>.crawlEnded = <strong>true</strong>;<a name="137" href="#137">137</a> }<a name="138" href="#138">138</a> <a name="139" href="#139">139</a> <strong>public</strong> <strong>void</strong> crawlEnding(String sExitMessage) {<a name="140" href="#140">140</a> <em class="comment">// TODO Auto-generated method stub</em><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> crawlPaused(String statusMessage) {<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> <a name="149" href="#149">149</a> <strong>public</strong> <strong>void</strong> crawlPausing(String statusMessage) {<a name="150" href="#150">150</a> <em class="comment">// TODO Auto-generated method stub</em><a name="151" href="#151">151</a> <a name="152" href="#152">152</a> }<a name="153" href="#153">153</a> <a name="154" href="#154">154</a> <strong>public</strong> <strong>void</strong> crawlResuming(String statusMessage) {<a name="155" href="#155">155</a> <em class="comment">// TODO Auto-generated method stub</em><a name="156" href="#156">156</a> <a name="157" href="#157">157</a> }<a name="158" href="#158">158</a> <a name="159" href="#159">159</a> <strong>public</strong> <strong>void</strong> crawlStarted(String message) {<a name="160" href="#160">160</a> <em class="comment">// TODO Auto-generated method stub</em><a name="161" href="#161">161</a> <a name="162" href="#162">162</a> }<a name="163" href="#163">163</a> <a name="164" href="#164">164</a> <strong>public</strong> <strong>void</strong> crawledURIDisregard(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {<a name="165" href="#165">165</a> <em class="comment">// TODO Auto-generated method stub</em><a name="166" href="#166">166</a> <a name="167" href="#167">167</a> }<a name="168" href="#168">168</a> <a name="169" href="#169">169</a> <strong>public</strong> <strong>void</strong> crawledURIFailure(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {<a name="170" href="#170">170</a> <em class="comment">// TODO Auto-generated method stub</em><a name="171" href="#171">171</a> <a name="172" href="#172">172</a> }<a name="173" href="#173">173</a> <a name="174" href="#174">174</a> <strong>public</strong> <strong>void</strong> crawledURINeedRetry(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {<a name="175" href="#175">175</a> <em class="comment">// TODO Auto-generated method stub</em><a name="176" href="#176">176</a> <a name="177" href="#177">177</a> }<a name="178" href="#178">178</a> <a name="179" href="#179">179</a> <strong>public</strong> <strong>void</strong> crawledURISuccessful(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {<a name="180" href="#180">180</a> LOG.info(curi.toString());<a name="181" href="#181">181</a> }<a name="182" href="#182">182</a> }</pre><hr/><div id="footer">This page was automatically generated by <a href="http://maven.apache.org/">Maven</a></div></body></html>
⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -