doctest.html
来自「网络爬虫开源代码」· HTML 代码 · 共 208 行 · 第 1/2 页
HTML
208 行
<a name="96" href="#96">96</a> c.close();<a name="97" href="#97">97</a> } <strong>catch</strong> (IOException e) {<a name="98" href="#98">98</a> e.printStackTrace();<a name="99" href="#99">99</a> }<a name="100" href="#100">100</a> }<a name="101" href="#101">101</a> <a name="102" href="#102">102</a> <a name="103" href="#103">103</a> <strong>private</strong> <strong>int</strong> runFiles(File doc, File expected) <a name="104" href="#104">104</a> throws IOException {<a name="105" href="#105">105</a> FileInputStream expectedIn = <strong>new</strong> FileInputStream(expected);<a name="106" href="#106">106</a> Reader expectedReader = <strong>new</strong> InputStreamReader(expectedIn, <span class="string">"UTF-16BE"</span>);<a name="107" href="#107">107</a> Reader docReader = Doc.getText(doc);<a name="108" href="#108">108</a> <strong>try</strong> {<a name="109" href="#109">109</a> <strong>return</strong> runReaders(docReader, expectedReader);<a name="110" href="#110">110</a> } <strong>finally</strong> {<a name="111" href="#111">111</a> close(docReader);<a name="112" href="#112">112</a> close(expectedReader);<a name="113" href="#113">113</a> }<a name="114" href="#114">114</a> }<a name="115" href="#115">115</a> <a name="116" href="#116">116</a> <a name="117" href="#117">117</a> <strong>private</strong> <strong>int</strong> runReaders(Reader doc, Reader expected) <a name="118" href="#118">118</a> throws IOException {<a name="119" href="#119">119</a> <strong>int</strong> count = 0;<a name="120" href="#120">120</a> <strong>int</strong> errors = 0;<a name="121" href="#121">121</a> <strong>boolean</strong> go = <strong>true</strong>;<a name="122" href="#122">122</a> <strong>while</strong> (go) {<a name="123" href="#123">123</a> <strong>int</strong> ch = doc.read();<a name="124" href="#124">124</a> <strong>int</strong> expectedCh = correctPOI(expected.read());<a name="125" href="#125">125</a> <strong>if</strong> ((ch < 0) || (expectedCh < 0)) {<a name="126" href="#126">126</a> go = false;<a name="127" href="#127">127</a> <strong>if</strong> ((ch >= 0) || (expectedCh >= 0)) {<a name="128" href="#128">128</a> errors++;<a name="129" href="#129">129</a> System.out.println(<span class="string">"File lengths differ."</span>);<a name="130" href="#130">130</a> }<a name="131" href="#131">131</a> }<a name="132" href="#132">132</a> <strong>if</strong> (ch != expectedCh) {<a name="133" href="#133">133</a> errors += 1;<a name="134" href="#134">134</a> report(count, expectedCh, ch);<a name="135" href="#135">135</a> }<a name="136" href="#136">136</a> count++;<a name="137" href="#137">137</a> }<a name="138" href="#138">138</a> <strong>return</strong> errors;<a name="139" href="#139">139</a> }<a name="140" href="#140">140</a> <a name="141" href="#141">141</a> <a name="142" href="#142">142</a> <strong>private</strong> <strong>void</strong> report(<strong>int</strong> count, <strong>int</strong> expected, <strong>int</strong> actual) {<a name="143" href="#143">143</a> StringBuilder msg = <strong>new</strong> StringBuilder(<span class="string">"#"</span>).append(count);<a name="144" href="#144">144</a> msg.append(<span class="string">": Expected "</span>);<a name="145" href="#145">145</a> msg.append(expected).append(<span class="string">" ("</span>).append(toChar(expected));<a name="146" href="#146">146</a> msg.append(<span class="string">") but got "</span>).append(actual).append(<span class="string">" ("</span>);<a name="147" href="#147">147</a> msg.append(toChar(actual)).append(<span class="string">")."</span>);<a name="148" href="#148">148</a> System.out.println(msg);<a name="149" href="#149">149</a> }<a name="150" href="#150">150</a> <a name="151" href="#151">151</a> <a name="152" href="#152">152</a> <strong>private</strong> <strong>static</strong> String toChar(<strong>int</strong> ch) {<a name="153" href="#153">153</a> <strong>if</strong> (ch < 0) {<a name="154" href="#154">154</a> <strong>return</strong> <span class="string">"EOF"</span>;<a name="155" href="#155">155</a> } <strong>else</strong> {<a name="156" href="#156">156</a> <strong>return</strong> Character.toString((<strong>char</strong>)ch);<a name="157" href="#157">157</a> }<a name="158" href="#158">158</a> }<a name="159" href="#159">159</a> <a name="160" href="#160">160</a> <em>/**<em>*</em></em><a name="161" href="#161">161</a> <em> * Corrects POI's Cp1252 output. There's a bug somewhere in POI that</em><a name="162" href="#162">162</a> <em> * makes it produce incorrect characters. Not sure where and don't have</em><a name="163" href="#163">163</a> <em> * time to track it down. But I have visually checked the input </em><a name="164" href="#164">164</a> <em> * documents to verify that Doc is producing the right character, and</em><a name="165" href="#165">165</a> <em> * that POI is not.</em><a name="166" href="#166">166</a> <em> * </em><a name="167" href="#167">167</a> <em> * @param ch the POI-produced character to check</em><a name="168" href="#168">168</a> <em> * @return the corrected character</em><a name="169" href="#169">169</a> <em> */</em><a name="170" href="#170">170</a> <strong>private</strong> <strong>static</strong> <strong>int</strong> correctPOI(<strong>int</strong> ch) {<a name="171" href="#171">171</a> <strong>switch</strong> (ch) {<a name="172" href="#172">172</a> <strong>case</strong> 8734:<a name="173" href="#173">173</a> <em class="comment">// POI produced the infinity sign when it should have </em><a name="174" href="#174">174</a> <em class="comment">// produced the degrees sign.</em><a name="175" href="#175">175</a> <strong>return</strong> 176;<a name="176" href="#176">176</a> <strong>case</strong> 214:<a name="177" href="#177">177</a> <em class="comment">// POI produced an umat O instead of an ellipses mark.</em><a name="178" href="#178">178</a> <strong>return</strong> 8230;<a name="179" href="#179">179</a> <strong>case</strong> 237:<a name="180" href="#180">180</a> <em class="comment">// POI produced an acute i instead of a fancy single quote</em><a name="181" href="#181">181</a> <strong>return</strong> 8217;<a name="182" href="#182">182</a> <strong>case</strong> 236:<a name="183" href="#183">183</a> <em class="comment">// POI produced a reverse acute i instead of fancy double quote</em><a name="184" href="#184">184</a> <strong>return</strong> 8220;<a name="185" href="#185">185</a> <strong>case</strong> 238:<a name="186" href="#186">186</a> <em class="comment">// POI produced a caret i instead of fancy double quote</em><a name="187" href="#187">187</a> <strong>return</strong> 8221;<a name="188" href="#188">188</a> <strong>default</strong>:<a name="189" href="#189">189</a> <strong>return</strong> ch;<a name="190" href="#190">190</a> }<a name="191" href="#191">191</a> }<a name="192" href="#192">192</a> <a name="193" href="#193">193</a> <a name="194" href="#194">194</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
增大字号Ctrl + =
减小字号Ctrl + -
显示快捷键?