seedcachingscopetest.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="100" href="#100">100</a>    <em class="comment">/*<em class="comment"> (non-Javadoc)</em></em>
<a name="101" href="#101">101</a> <em class="comment">    * @see org.archive.util.TmpDirTestCase#setUp()</em>
<a name="102" href="#102">102</a> <em class="comment">    */</em>
<a name="103" href="#103">103</a>    <strong>protected</strong> <strong>void</strong> setUp() throws Exception {
<a name="104" href="#104">104</a>        <strong>super</strong>.setUp();
<a name="105" href="#105">105</a> 
<a name="106" href="#106">106</a>        <em class="comment">// First create array of seeds and add to treeset.</em>
<a name="107" href="#107">107</a>        SeedCachingScopeTest.seeds = <strong>new</strong> TreeSet(SeedCachingScopeTest.CMP);
<a name="108" href="#108">108</a>        String [] uris = {<span class="string">"mailto:www.google.com"</span>,
<a name="109" href="#109">109</a>            <span class="string">"http://www.port.com:80/etc/motd2"</span>,
<a name="110" href="#110">110</a>            <span class="string">"http://a:[e-mail]/etc/motd2"</span>,
<a name="111" href="#111">111</a>            <span class="string">"news:www.google.com"</span>,
<a name="112" href="#112">112</a>            <span class="string">"http://www.google.com"</span>,
<a name="113" href="#113">113</a>            <span class="string">"https://www.google.com"</span>,
<a name="114" href="#114">114</a>            <span class="string">"gopher://www.google.com"</span>,
<a name="115" href="#115">115</a>            <span class="string">"news://www.google.com"</span>,
<a name="116" href="#116">116</a>            <span class="string">"rss://www.google.com"</span>,
<a name="117" href="#117">117</a>            <span class="string">"telnet://www.google.com"</span>,
<a name="118" href="#118">118</a>            <span class="string">"ftp://[e-mail]/etc/motd"</span>,
<a name="119" href="#119">119</a>            <span class="string">"ftp://example.com/etc/motd2"</span>
<a name="120" href="#120">120</a>        };
<a name="121" href="#121">121</a>        <strong>for</strong> (<strong>int</strong> i = 0; i &lt; uris.length; i++) {
<a name="122" href="#122">122</a>            SeedCachingScopeTest.seeds.add(UURIFactory.getInstance(uris[i]));
<a name="123" href="#123">123</a>        }
<a name="124" href="#124">124</a> 
<a name="125" href="#125">125</a>        <em class="comment">// Write a seeds file w/ our list of seeds.</em>
<a name="126" href="#126">126</a>        <strong>this</strong>.seedsfile = <strong>new</strong> File(getTmpDir(),
<a name="127" href="#127">127</a>                SeedCachingScopeTest.<strong>class</strong>.getName() + <span class="string">".seedfile"</span>);
<a name="128" href="#128">128</a>        PrintWriter writer = <strong>new</strong> PrintWriter(<strong>new</strong> FileWriter(<strong>this</strong>.seedsfile));
<a name="129" href="#129">129</a>        <strong>for</strong> (<strong>int</strong> i = 0; i &lt; uris.length; i++) {
<a name="130" href="#130">130</a>            writer.println(uris[i]);
<a name="131" href="#131">131</a>        }
<a name="132" href="#132">132</a>        writer.close();
<a name="133" href="#133">133</a>    }
<a name="134" href="#134">134</a> 
<a name="135" href="#135">135</a> 
<a name="136" href="#136">136</a>    <em class="comment">/*<em class="comment"> (non-Javadoc)</em></em>
<a name="137" href="#137">137</a> <em class="comment">    * @see org.archive.util.TmpDirTestCase#tearDown()</em>
<a name="138" href="#138">138</a> <em class="comment">    */</em>
<a name="139" href="#139">139</a>    <strong>protected</strong> <strong>void</strong> tearDown() throws Exception {
<a name="140" href="#140">140</a>        <strong>super</strong>.tearDown();
<a name="141" href="#141">141</a>        <strong>if</strong> (<strong>this</strong>.seedsfile.exists()) {
<a name="142" href="#142">142</a>             <strong>this</strong>.seedsfile.delete();
<a name="143" href="#143">143</a>        }
<a name="144" href="#144">144</a>    }
<a name="145" href="#145">145</a> 
<a name="146" href="#146">146</a>    <strong>public</strong> <strong>void</strong> testGeneral() throws URIException {
<a name="147" href="#147">147</a>        <em class="comment">// First make sure that I can get the seed set from seed file.</em>
<a name="148" href="#148">148</a>        <a href="../../../../org/archive/crawler/scope/SeedCachingScope.html">SeedCachingScope</a> sl = checkContent(SeedCachingScopeTest.seeds);
<a name="149" href="#149">149</a>        <em class="comment">// Now do add and see if get set matches seed file content.</em>
<a name="150" href="#150">150</a>        <strong>final</strong> <a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a> curi = <strong>new</strong> <a href="../../../../org/archive/crawler/datamodel/CrawlURI.html">CrawlURI</a>(UURIFactory.getInstance(<span class="string">"http://one.two.three"</span>));
<a name="151" href="#151">151</a>        sl.addSeed(curi);
<a name="152" href="#152">152</a>        Set set = <strong>new</strong> TreeSet(SeedCachingScopeTest.CMP);
<a name="153" href="#153">153</a>        set.addAll(SeedCachingScopeTest.seeds);
<a name="154" href="#154">154</a>        set.add(curi.getUURI());
<a name="155" href="#155">155</a>        checkContent(sl, set);
<a name="156" href="#156">156</a>    }
<a name="157" href="#157">157</a> 
<a name="158" href="#158">158</a>    <strong>public</strong> <strong>void</strong> testNoScheme() throws IOException {
<a name="159" href="#159">159</a>        <strong>final</strong> String NOSCHEME = <span class="string">"x.y.z"</span>;
<a name="160" href="#160">160</a>        FileWriter fw = <strong>new</strong> FileWriter(<strong>this</strong>.seedsfile, <strong>true</strong>);
<a name="161" href="#161">161</a>        <em class="comment">// Write to new (last) line the URL.</em>
<a name="162" href="#162">162</a>        fw.write(<span class="string">"\n"</span>);
<a name="163" href="#163">163</a>        fw.write(NOSCHEME);
<a name="164" href="#164">164</a>        fw.flush();
<a name="165" href="#165">165</a>        fw.close();
<a name="166" href="#166">166</a>        <strong>boolean</strong> found = false;
<a name="167" href="#167">167</a>        <a href="../../../../org/archive/crawler/scope/SeedCachingScope.html">SeedCachingScope</a> sl = <strong>new</strong> UnitTestSeedCachingScope(seedsfile);
<a name="168" href="#168">168</a>        <strong>for</strong> (Iterator i = sl.seedsIterator(); i.hasNext();) {
<a name="169" href="#169">169</a>            <a href="../../../../org/archive/net/UURI.html">UURI</a> uuri = (UURI)i.next();
<a name="170" href="#170">170</a>            <strong>if</strong> (uuri.getHost() == <strong>null</strong>) {
<a name="171" href="#171">171</a>                <strong>continue</strong>;
<a name="172" href="#172">172</a>            }
<a name="173" href="#173">173</a>            <strong>if</strong> (uuri.getHost().equals(NOSCHEME)) {
<a name="174" href="#174">174</a>                found = <strong>true</strong>;
<a name="175" href="#175">175</a>                <strong>break</strong>;
<a name="176" href="#176">176</a>            }
<a name="177" href="#177">177</a>        }
<a name="178" href="#178">178</a>        assertTrue(<span class="string">"Did not find "</span> + NOSCHEME, found);
<a name="179" href="#179">179</a>    }
<a name="180" href="#180">180</a> 
<a name="181" href="#181">181</a>    <strong>private</strong> <a href="../../../../org/archive/crawler/scope/SeedCachingScope.html">SeedCachingScope</a> checkContent(Set seedSet) {
<a name="182" href="#182">182</a>        <strong>return</strong> checkContent(<strong>null</strong>, seedSet);
<a name="183" href="#183">183</a>    }
<a name="184" href="#184">184</a> 
<a name="185" href="#185">185</a>    <strong>private</strong> <a href="../../../../org/archive/crawler/scope/SeedCachingScope.html">SeedCachingScope</a> checkContent(<a href="../../../../org/archive/crawler/scope/SeedCachingScope.html">SeedCachingScope</a> sl, Set seedSet) {
<a name="186" href="#186">186</a>        <strong>if</strong> (sl == <strong>null</strong>) {
<a name="187" href="#187">187</a>            sl = <strong>new</strong> UnitTestSeedCachingScope(<strong>this</strong>.seedsfile);
<a name="188" href="#188">188</a>        }
<a name="189" href="#189">189</a>        <strong>int</strong> count = 0;
<a name="190" href="#190">190</a>        <strong>for</strong> (Iterator i = sl.seedsIterator(); i.hasNext();) {
<a name="191" href="#191">191</a>            count++;
<a name="192" href="#192">192</a>            <a href="../../../../org/archive/net/UURI.html">UURI</a> uuri = (UURI)i.next();
<a name="193" href="#193">193</a>            assertTrue(<span class="string">"Does not contain: "</span> + uuri.toString(),
<a name="194" href="#194">194</a>                seedSet.contains(uuri));
<a name="195" href="#195">195</a>        }
<a name="196" href="#196">196</a>        assertTrue(<span class="string">"Different sizes: "</span> + count + <span class="string">", "</span> + seedSet.size(),
<a name="197" href="#197">197</a>            count == seedSet.size());
<a name="198" href="#198">198</a>        <strong>return</strong> sl;
<a name="199" href="#199">199</a>    }
<a name="200" href="#200">200</a> }
<a name="201" href="#201">201</a> 
</pre>
<hr/><div id="footer">This page was automatically generated by <a href="http://maven.apache.org/">Maven</a></div></body>
</html>