<a name="87" href="#87">87</a>      <strong>public</strong> <strong>static</strong> <strong>final</strong> String DEFAULT_MAP_SOURCE = <span class="string">""</span>;<a name="88" href="#88">88</a>      <a name="89" href="#89">89</a>      <em>/**<em>*</em></em><a name="90" href="#90">90</a>  <em>     * Mapping of classKey ranges (as represented by their start) to </em><a name="91" href="#91">91</a>  <em>     * crawlers (by abstract name/filename)</em><a name="92" href="#92">92</a>  <em>     */</em><a name="93" href="#93">93</a>      TreeMap&lt;String, String> map = <strong>new</strong> TreeMap&lt;String, String>();<a name="94" href="#94">94</a>  <a name="95" href="#95">95</a>      <em>/**<em>*</em></em><a name="96" href="#96">96</a>  <em>     * Constructor.</em><a name="97" href="#97">97</a>  <em>     * @param name Name of this processor.</em><a name="98" href="#98">98</a>  <em>     */</em><a name="99" href="#99">99</a>      <strong>public</strong> <a href="../../../../org/archive/crawler/processor/LexicalCrawlMapper.html">LexicalCrawlMapper</a>(String name) {<a name="100" href="#100">100</a>         <strong>super</strong>(name, <span class="string">"LexicalCrawlMapper. Maps URIs to a named "</span> +<a name="101" href="#101">101</a>                 <span class="string">"crawler by a lexical comparison of the URI's "</span> +<a name="102" href="#102">102</a>                 <span class="string">"classKey to a supplied ranges map."</span>);<a name="103" href="#103">103</a>         addElementToDefinition(<strong>new</strong> <a href="../../../../org/archive/crawler/settings/SimpleType.html">SimpleType</a>(ATTR_MAP_SOURCE,<a name="104" href="#104">104</a>             <span class="string">"Path (or HTTP URL) to map specification file. Each line "</span> +<a name="105" href="#105">105</a>             <span class="string">"should include 2 whitespace-separated tokens: the first a "</span> +<a name="106" href="#106">106</a>             <span class="string">"key indicating the end of a range, the second the crawler "</span> +<a name="107" href="#107">107</a>             <span class="string">"node to which URIs in the key range should be mapped."</span>,<a name="108" href="#108">108</a>             DEFAULT_MAP_SOURCE));<a name="109" href="#109">109</a>     }<a name="110" href="#110">110</a> <a name="111" href="#111">111</a>     <em>/**<em>*</em></em><a name="112" href="#112">112</a> <em>     * Look up the crawler node name to which the given CandidateURI </em><a name="113" href="#113">113</a> <em>     * should be mapped. </em><a name="114" href="#114">114</a> <em>     * </em><a name="115" href="#115">115</a> <em>     * @param cauri CandidateURI to consider</em><a name="116" href="#116">116</a> <em>     * @return String node name which should handle URI</em><a name="117" href="#117">117</a> <em>     */</em><a name="118" href="#118">118</a>     <strong>protected</strong> String map(<a href="../../../../org/archive/crawler/datamodel/CandidateURI.html">CandidateURI</a> cauri) {<a name="119" href="#119">119</a>         <em class="comment">// get classKey, via frontier to generate if necessary</em><a name="120" href="#120">120</a>         String classKey = getController().getFrontier().getClassKey(cauri);<a name="121" href="#121">121</a>         SortedMap tail = map.tailMap(classKey);<a name="122" href="#122">122</a>         <strong>if</strong>(tail.isEmpty()) {<a name="123" href="#123">123</a>             <em class="comment">// wraparound</em><a name="124" href="#124">124</a>             tail = map;<a name="125" href="#125">125</a>         }<a name="126" href="#126">126</a>         <em class="comment">// target node is value of nearest subsequent key</em><a name="127" href="#127">127</a>         <strong>return</strong> (String) tail.get(tail.firstKey());<a name="128" href="#128">128</a>     }<a name="129" href="#129">129</a> <a name="130" href="#130">130</a>     <strong>protected</strong> <strong>void</strong> initialTasks() {<a name="131" href="#131">131</a>         <strong>super</strong>.initialTasks();<a name="132" href="#132">132</a>         <strong>try</strong> {<a name="133" href="#133">133</a>             loadMap();<a name="134" href="#134">134</a>         } <strong>catch</strong> (IOException e) {<a name="135" href="#135">135</a>             e.printStackTrace();<a name="136" href="#136">136</a>             <strong>throw</strong> <strong>new</strong> RuntimeException(e);<a name="137" href="#137">137</a>         }<a name="138" href="#138">138</a>     }<a name="139" href="#139">139</a> <a name="140" href="#140">140</a>     <em>/**<em>*</em></em><a name="141" href="#141">141</a> <em>     * Retrieve and parse the mapping specification from a local path or</em><a name="142" href="#142">142</a> <em>     * HTTP URL. </em><a name="143" href="#143">143</a> <em>     * </em><a name="144" href="#144">144</a> <em>     * @throws IOException</em><a name="145" href="#145">145</a> <em>     */</em><a name="146" href="#146">146</a>     <strong>protected</strong> <strong>void</strong> loadMap() throws IOException {<a name="147" href="#147">147</a>         map.clear();<a name="148" href="#148">148</a>         String mapSource = (String) getUncheckedAttribute(<strong>null</strong>,ATTR_MAP_SOURCE);<a name="149" href="#149">149</a>         Reader reader = <strong>null</strong>;<a name="150" href="#150">150</a>         <strong>if</strong>(!mapSource.startsWith(<span class="string">"http://"</span>)) {<a name="151" href="#151">151</a>             <em class="comment">// file-based source</em><a name="152" href="#152">152</a>             File source = <strong>new</strong> File(mapSource);<a name="153" href="#153">153</a>             <strong>if</strong> (!source.isAbsolute()) {<a name="154" href="#154">154</a>                 source = <strong>new</strong> File(getSettingsHandler().getOrder()<a name="155" href="#155">155</a>                         .getController().getDisk(), mapSource);<a name="156" href="#156">156</a>             }<a name="157" href="#157">157</a>             reader = <strong>new</strong> FileReader(source);<a name="158" href="#158">158</a>         } <strong>else</strong> {<a name="159" href="#159">159</a>             URLConnection conn = (<strong>new</strong> URL(mapSource)).openConnection();<a name="160" href="#160">160</a>             reader = <strong>new</strong> InputStreamReader(conn.getInputStream());<a name="161" href="#161">161</a>         }<a name="162" href="#162">162</a>         reader = <strong>new</strong> BufferedReader(reader);<a name="163" href="#163">163</a>         Iterator iter = <a name="164" href="#164">164</a>             <strong>new</strong> <a href="../../../../org/archive/util/iterator/RegexpLineIterator.html">RegexpLineIterator</a>(<a name="165" href="#165">165</a>                     <strong>new</strong> <a href="../../../../org/archive/util/iterator/LineReadingIterator.html">LineReadingIterator</a>((BufferedReader) reader),<a name="166" href="#166">166</a>                     RegexpLineIterator.COMMENT_LINE,<a name="167" href="#167">167</a>                     RegexpLineIterator.TRIMMED_ENTRY_TRAILING_COMMENT,<a name="168" href="#168">168</a>                     RegexpLineIterator.ENTRY);<a name="169" href="#169">169</a>         <strong>while</strong> (iter.hasNext()) {<a name="170" href="#170">170</a>             String[] entry = ((String) iter.next()).split(<span class="string">"&#47;&#47;s+"</span>);<a name="171" href="#171">171</a>             map.put(entry[0],entry[1]);<a name="172" href="#172">172</a>         }<a name="173" href="#173">173</a>         reader.close();<a name="174" href="#174">174</a>     }<a name="175" href="#175">175</a> }</pre><hr/><div id="footer">This page was automatically generated by <a href="http://maven.apache.org/">Maven</a></div></body></html>