<a name="110" href="#110">110</a>     <em>/**<em>*</em></em><a name="111" href="#111">111</a> <em>     * Create the single object (within which is one BDB database)</em><a name="112" href="#112">112</a> <em>     * inside which all the other queues live. </em><a name="113" href="#113">113</a> <em>     * </em><a name="114" href="#114">114</a> <em>     * @return the created BdbMultipleWorkQueues</em><a name="115" href="#115">115</a> <em>     * @throws DatabaseException</em><a name="116" href="#116">116</a> <em>     */</em><a name="117" href="#117">117</a>     <strong>private</strong> <a href="../../../../org/archive/crawler/frontier/BdbMultipleWorkQueues.html">BdbMultipleWorkQueues</a> createMultipleWorkQueues()<a name="118" href="#118">118</a>     throws DatabaseException {<a name="119" href="#119">119</a>         <strong>return</strong> <strong>new</strong> <a href="../../../../org/archive/crawler/frontier/BdbMultipleWorkQueues.html">BdbMultipleWorkQueues</a>(<strong>this</strong>.controller.getBdbEnvironment(),<a name="120" href="#120">120</a>             <strong>this</strong>.controller.getBdbEnvironment().getClassCatalog(),<a name="121" href="#121">121</a>             <strong>this</strong>.controller.isCheckpointRecover());<a name="122" href="#122">122</a>     }<a name="123" href="#123">123</a> <a name="124" href="#124">124</a>     <em>/**<em>*</em></em><a name="125" href="#125">125</a> <em>     * Create a UriUniqFilter that will serve as record </em><a name="126" href="#126">126</a> <em>     * of already seen URIs.</em><a name="127" href="#127">127</a> <em>     *</em><a name="128" href="#128">128</a> <em>     * @return A UURISet that will serve as a record of already seen URIs</em><a name="129" href="#129">129</a> <em>     * @throws IOException</em><a name="130" href="#130">130</a> <em>     */</em><a name="131" href="#131">131</a>     <strong>protected</strong> <a href="../../../../org/archive/crawler/datamodel/UriUniqFilter.html">UriUniqFilter</a> createAlreadyIncluded() throws IOException {<a name="132" href="#132">132</a>         <a href="../../../../org/archive/crawler/datamodel/UriUniqFilter.html">UriUniqFilter</a> uuf;<a name="133" href="#133">133</a>         String c = <strong>null</strong>;<a name="134" href="#134">134</a>         <strong>try</strong> {<a name="135" href="#135">135</a>             c = (String)getAttribute(<strong>null</strong>, ATTR_INCLUDED);<a name="136" href="#136">136</a>         } <strong>catch</strong> (AttributeNotFoundException e) {<a name="137" href="#137">137</a>             <em class="comment">// Do default action if attribute not in order.</em><a name="138" href="#138">138</a>         }<a name="139" href="#139">139</a>         <em class="comment">// TODO: avoid all this special-casing; enable some common</em><a name="140" href="#140">140</a>         <em class="comment">// constructor interface usable for all alt implemenations</em><a name="141" href="#141">141</a>         <strong>if</strong> (c != <strong>null</strong> &amp;&amp; c.equals(BloomUriUniqFilter.<strong>class</strong>.getName())) {<a name="142" href="#142">142</a>             uuf = <strong>this</strong>.controller.isCheckpointRecover()?<a name="143" href="#143">143</a>                     deserializeAlreadySeen(BloomUriUniqFilter.<strong>class</strong>,<a name="144" href="#144">144</a>                         <strong>this</strong>.controller.getCheckpointRecover().getDirectory()):<a name="145" href="#145">145</a>                     <strong>new</strong> <a href="../../../../org/archive/crawler/util/BloomUriUniqFilter.html">BloomUriUniqFilter</a>();<a name="146" href="#146">146</a>         } <strong>else</strong> <strong>if</strong> (c!=<strong>null</strong> &amp;&amp; c.equals(MemFPMergeUriUniqFilter.<strong>class</strong>.getName())) {<a name="147" href="#147">147</a>             <em class="comment">// TODO: add checkpointing for MemFPMergeUriUniqFilter</em><a name="148" href="#148">148</a>             uuf = <strong>new</strong> <a href="../../../../org/archive/crawler/util/MemFPMergeUriUniqFilter.html">MemFPMergeUriUniqFilter</a>();<a name="149" href="#149">149</a>         } <strong>else</strong> <strong>if</strong> (c!=<strong>null</strong> &amp;&amp; c.equals(DiskFPMergeUriUniqFilter.<strong>class</strong>.getName())) {<a name="150" href="#150">150</a>             <em class="comment">// TODO: add checkpointing for DiskFPMergeUriUniqFilter</em><a name="151" href="#151">151</a>             uuf = <strong>new</strong> <a href="../../../../org/archive/crawler/util/DiskFPMergeUriUniqFilter.html">DiskFPMergeUriUniqFilter</a>(controller.getScratchDisk());<a name="152" href="#152">152</a>         } <strong>else</strong> {<a name="153" href="#153">153</a>             <em class="comment">// Assume its BdbUriUniqFilter.</em><a name="154" href="#154">154</a>             uuf = <strong>this</strong>.controller.isCheckpointRecover()?<a name="155" href="#155">155</a>                 deserializeAlreadySeen(BdbUriUniqFilter.<strong>class</strong>,<a name="156" href="#156">156</a>                     <strong>this</strong>.controller.getCheckpointRecover().getDirectory()):<a name="157" href="#157">157</a>                 <strong>new</strong> <a href="../../../../org/archive/crawler/util/BdbUriUniqFilter.html">BdbUriUniqFilter</a>(<strong>this</strong>.controller.getBdbEnvironment());<a name="158" href="#158">158</a>             <strong>if</strong> (<strong>this</strong>.controller.isCheckpointRecover()) {<a name="159" href="#159">159</a>                 <em class="comment">// If recover, need to call reopen of the db.</em><a name="160" href="#160">160</a>                 <strong>try</strong> {<a name="161" href="#161">161</a>                     ((<a href="../../../../org/archive/crawler/util/BdbUriUniqFilter.html">BdbUriUniqFilter</a>)uuf).<a name="162" href="#162">162</a>                         reopen(<strong>this</strong>.controller.getBdbEnvironment());<a name="163" href="#163">163</a>                 } <strong>catch</strong> (DatabaseException e) {<a name="164" href="#164">164</a>                     <strong>throw</strong> <strong>new</strong> IOException(e.getMessage());<a name="165" href="#165">165</a>                 }<a name="166" href="#166">166</a>             }   <a name="167" href="#167">167</a>         }<a name="168" href="#168">168</a>         uuf.setDestination(<strong>this</strong>);<a name="169" href="#169">169</a>         <strong>return</strong> uuf;<a name="170" href="#170">170</a>     }<a name="171" href="#171">171</a>     <a name="172" href="#172">172</a>     <strong>protected</strong> <a href="../../../../org/archive/crawler/datamodel/UriUniqFilter.html">UriUniqFilter</a> deserializeAlreadySeen(<a name="173" href="#173">173</a>             <strong>final</strong> Class&lt;? <strong>extends</strong> UriUniqFilter> cls,<a name="174" href="#174">174</a>             <strong>final</strong> File dir)<a name="175" href="#175">175</a>     throws FileNotFoundException, IOException {<a name="176" href="#176">176</a>         <a href="../../../../org/archive/crawler/datamodel/UriUniqFilter.html">UriUniqFilter</a> uuf = <strong>null</strong>;<a name="177" href="#177">177</a>         <strong>try</strong> {<a name="178" href="#178">178</a>             logger.fine(<span class="string">"Started deserializing "</span> + cls.getName() +<a name="179" href="#179">179</a>                 <span class="string">" of checkpoint recover."</span>);<a name="180" href="#180">180</a>             uuf = CheckpointUtils.readObjectFromFile(cls, dir);<a name="181" href="#181">181</a>             logger.fine(<span class="string">"Finished deserializing bdbje as part "</span> +<a name="182" href="#182">182</a>                 <span class="string">"of checkpoint recover."</span>);<a name="183" href="#183">183</a>         } <strong>catch</strong> (ClassNotFoundException e) {<a name="184" href="#184">184</a>             <strong>throw</strong> <strong>new</strong> IOException(<span class="string">"Failed to deserialize "</span>  +<a name="185" href="#185">185</a>                 cls.getName() + <span class="string">": "</span> + e.getMessage());<a name="186" href="#186">186</a>         }<a name="187" href="#187">187</a>         <strong>return</strong> uuf;<a name="188" href="#188">188</a>     }<a name="189" href="#189">189</a> <a name="190" href="#190">190</a>     <em>/**<em>*</em></em><a name="191" href="#191">191</a> <em>     * Return the work queue for the given CrawlURI's classKey. URIs</em><a name="192" href="#192">192</a> <em>     * are ordered and politeness-delayed within their 'class'.</em><a name="193" href="#193">193</a> <em>     * </em><a name="194" href="#194">194</a> <em>     * @param curi CrawlURI to base queue on</em><a name="195" href="#195">195</a> <em>     * @return the found or created BdbWorkQueue</em><a name="196" href="#196">196</a> <em>     */</em><a name="197" href="#197">197</a>     <strong>protected</strong> <a href="../../../../org/archive/crawler/frontier/WorkQueue.html">WorkQueue</a> getQueueFor(<a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi) {<a name="198" href="#198">198</a>         <a href="../../../../org/archive/crawler/frontier/WorkQueue.html">WorkQueue</a> wq;<a name="199" href="#199">199</a>         String classKey = curi.getClassKey();<a name="200" href="#200">200</a>         <strong>synchronized</strong> (allQueues) {<a name="201" href="#201">201</a>             wq = (WorkQueue)allQueues.get(classKey);<a name="202" href="#202">202</a>             <strong>if</strong> (wq == <strong>null</strong>) {<a name="203" href="#203">203</a>                 wq = <strong>new</strong> <a href="../../../../org/archive/crawler/frontier/BdbWorkQueue.html">BdbWorkQueue</a>(classKey, <strong>this</strong>);<a name="204" href="#204">204</a>                 wq.setTotalBudget(((Long)getUncheckedAttribute(<a name="205" href="#205">205</a>                     curi,ATTR_QUEUE_TOTAL_BUDGET)).longValue());<a name="206" href="#206">206</a>                 allQueues.put(classKey, wq);<a name="207" href="#207">207</a>             }<a name="208" href="#208">208</a>         }<a name="209" href="#209">209</a>         <strong>return</strong> wq;<a name="210" href="#210">210</a>     }<a name="211" href="#211">211</a>     <a name="212" href="#212">212</a>     <em>/**<em>*</em></em><a name="213" href="#213">213</a> <em>     * Return the work queue for the given classKey, or null</em><a name="214" href="#214">214</a> <em>     * if no such queue exists.</em><a name="215" href="#215">215</a> <em>     * </em><a name="216" href="#216">216</a> <em>     * @param classKey key to look for</em><a name="217" href="#217">217</a> <em>     * @return the found WorkQueue</em><a name="218" href="#218">218</a> <em>     */</em><a name="219" href="#219">219</a>     <strong>protected</strong> <a href="../../../../org/archive/crawler/frontier/WorkQueue.html">WorkQueue</a> getQueueFor(String classKey) {<a name="220" href="#220">220</a>         <a href="../../../../org/archive/crawler/frontier/WorkQueue.html">WorkQueue</a> wq; <a name="221" href="#221">221</a>         <strong>synchronized</strong> (allQueues) {<a name="222" href="#222">222</a>             wq = (WorkQueue)allQueues.get(classKey);<a name="223" href="#223">223</a>         }<a name="224" href="#224">224</a>         <strong>return</strong> wq;<a name="225" href="#225">225</a>     }<a name="226" href="#226">226</a> <a name="227" href="#227">227</a>     <strong>public</strong> <a href="../../../../org/archive/crawler/framework/FrontierMarker.html">FrontierMarker</a> getInitialMarker(String regexpr,