testxmlreader.html

解析RSS工具源码

<a name="89" href="#89">89</a>          }
<a name="90" href="#90">90</a>          <strong>else</strong> {
<a name="91" href="#91">91</a>              assertEquals(xmlReader.getEncoding().substring(0,encoding.length()),encoding);
<a name="92" href="#92">92</a>          }
<a name="93" href="#93">93</a>      }
<a name="94" href="#94">94</a>  
<a name="95" href="#95">95</a>      <strong>protected</strong> <strong>void</strong> _testRawBomInvalid(String bomEnc,String streamEnc,String prologEnc) throws Exception {
<a name="96" href="#96">96</a>          InputStream is = getXmlStream(bomEnc,XML3,streamEnc,prologEnc);
<a name="97" href="#97">97</a>          <strong>try</strong> {
<a name="98" href="#98">98</a>              XmlReader xmlReader = <strong>new</strong> XmlReader(is,false);
<a name="99" href="#99">99</a>              fail(<span class="string">"It should have failed for BOM "</span>+bomEnc+<span class="string">", streamEnc "</span>+streamEnc+<span class="string">" and prologEnc "</span>+prologEnc);
<a name="100" href="#100">100</a>         }
<a name="101" href="#101">101</a>         <strong>catch</strong> (IOException ex) {
<a name="102" href="#102">102</a>             assertTrue(ex.getMessage().indexOf(<span class="string">"Invalid encoding,"</span>)>-1);
<a name="103" href="#103">103</a>         }
<a name="104" href="#104">104</a>      }
<a name="105" href="#105">105</a> 
<a name="106" href="#106">106</a>     <strong>public</strong> <strong>void</strong> testRawBom() throws Exception {
<a name="107" href="#107">107</a>         _testRawBomValid(<span class="string">"UTF-8"</span>);
<a name="108" href="#108">108</a>         _testRawBomValid(<span class="string">"UTF-16BE"</span>);
<a name="109" href="#109">109</a>         _testRawBomValid(<span class="string">"UTF-16LE"</span>);
<a name="110" href="#110">110</a>         _testRawBomValid(<span class="string">"UTF-16"</span>);
<a name="111" href="#111">111</a> 
<a name="112" href="#112">112</a>         _testRawBomInvalid(<span class="string">"UTF-8-bom"</span>,<span class="string">"US-ASCII"</span>,<span class="string">"US-ASCII"</span>);
<a name="113" href="#113">113</a>         _testRawBomInvalid(<span class="string">"UTF-8-bom"</span>,<span class="string">"ISO-8859-1"</span>,<span class="string">"ISO-8859-1"</span>);
<a name="114" href="#114">114</a>         _testRawBomInvalid(<span class="string">"UTF-8-bom"</span>,<span class="string">"UTF-8"</span>,<span class="string">"UTF-16"</span>);
<a name="115" href="#115">115</a>         _testRawBomInvalid(<span class="string">"UTF-8-bom"</span>,<span class="string">"UTF-8"</span>,<span class="string">"UTF-16BE"</span>);
<a name="116" href="#116">116</a>         _testRawBomInvalid(<span class="string">"UTF-8-bom"</span>,<span class="string">"UTF-8"</span>,<span class="string">"UTF-16LE"</span>);
<a name="117" href="#117">117</a>         _testRawBomInvalid(<span class="string">"UTF-16BE-bom"</span>,<span class="string">"UTF-16BE"</span>,<span class="string">"UTF-16LE"</span>);
<a name="118" href="#118">118</a>         _testRawBomInvalid(<span class="string">"UTF-16LE-bom"</span>,<span class="string">"UTF-16LE"</span>,<span class="string">"UTF-16BE"</span>);
<a name="119" href="#119">119</a>         _testRawBomInvalid(<span class="string">"UTF-16LE-bom"</span>,<span class="string">"UTF-16LE"</span>,<span class="string">"UTF-8"</span>);
<a name="120" href="#120">120</a>     }
<a name="121" href="#121">121</a> 
<a name="122" href="#122">122</a>     <strong>public</strong> <strong>void</strong> testHttp() throws Exception {
<a name="123" href="#123">123</a>         _testHttpValid(<span class="string">"application/xml"</span>,<span class="string">"no-bom"</span>,<span class="string">"US-ASCII"</span>,<strong>null</strong>);
<a name="124" href="#124">124</a>         _testHttpValid(<span class="string">"application/xml"</span>,<span class="string">"UTF-8-bom"</span>,<span class="string">"US-ASCII"</span>,<strong>null</strong>);
<a name="125" href="#125">125</a>         _testHttpValid(<span class="string">"application/xml"</span>,<span class="string">"UTF-8-bom"</span>,<span class="string">"UTF-8"</span>,<strong>null</strong>);
<a name="126" href="#126">126</a>         _testHttpValid(<span class="string">"application/xml"</span>,<span class="string">"UTF-8-bom"</span>,<span class="string">"UTF-8"</span>,<span class="string">"UTF-8"</span>);
<a name="127" href="#127">127</a>         _testHttpValid(<span class="string">"application/xml;charset=UTF-8"</span>,<span class="string">"UTF-8-bom"</span>,<span class="string">"UTF-8"</span>,<strong>null</strong>);
<a name="128" href="#128">128</a>         _testHttpValid(<span class="string">"application/xml;charset=UTF-8"</span>,<span class="string">"UTF-8-bom"</span>,<span class="string">"UTF-8"</span>,<span class="string">"UTF-8"</span>);
<a name="129" href="#129">129</a>         _testHttpValid(<span class="string">"application/xml;charset=UTF-16"</span>,<span class="string">"UTF-16BE-bom"</span>,<span class="string">"UTF-16BE"</span>,<strong>null</strong>);
<a name="130" href="#130">130</a>         _testHttpValid(<span class="string">"application/xml;charset=UTF-16"</span>,<span class="string">"UTF-16BE-bom"</span>,<span class="string">"UTF-16BE"</span>,<span class="string">"UTF-16"</span>);
<a name="131" href="#131">131</a>         _testHttpValid(<span class="string">"application/xml;charset=UTF-16"</span>,<span class="string">"UTF-16BE-bom"</span>,<span class="string">"UTF-16BE"</span>,<span class="string">"UTF-16BE"</span>);
<a name="132" href="#132">132</a> 
<a name="133" href="#133">133</a>         _testHttpInvalid(<span class="string">"application/xml;charset=UTF-16BE"</span>,<span class="string">"UTF-16BE-bom"</span>,<span class="string">"UTF-16BE"</span>,<strong>null</strong>);
<a name="134" href="#134">134</a>         _testHttpInvalid(<span class="string">"application/xml;charset=UTF-16BE"</span>,<span class="string">"UTF-16BE-bom"</span>,<span class="string">"UTF-16BE"</span>,<span class="string">"UTF-16"</span>);
<a name="135" href="#135">135</a>         _testHttpInvalid(<span class="string">"application/xml;charset=UTF-16BE"</span>,<span class="string">"UTF-16BE-bom"</span>,<span class="string">"UTF-16BE"</span>,<span class="string">"UTF-16BE"</span>);
<a name="136" href="#136">136</a>         _testHttpInvalid(<span class="string">"application/xml"</span>,<span class="string">"UTF-8-bom"</span>,<span class="string">"US-ASCII"</span>,<span class="string">"US-ASCII"</span>);
<a name="137" href="#137">137</a>         _testHttpInvalid(<span class="string">"application/xml;charset=UTF-16"</span>,<span class="string">"UTF-16LE"</span>,<span class="string">"UTF-8"</span>,<span class="string">"UTF-8"</span>);
<a name="138" href="#138">138</a>         _testHttpInvalid(<span class="string">"application/xml;charset=UTF-16"</span>,<span class="string">"no-bom"</span>,<span class="string">"UTF-16BE"</span>,<span class="string">"UTF-16BE"</span>);
<a name="139" href="#139">139</a> 
<a name="140" href="#140">140</a>         _testHttpValid(<span class="string">"text/xml"</span>,<span class="string">"no-bom"</span>,<span class="string">"US-ASCII"</span>,<strong>null</strong>);
<a name="141" href="#141">141</a>         _testHttpValid(<span class="string">"text/xml;charset=UTF-8"</span>,<span class="string">"UTF-8-bom"</span>,<span class="string">"UTF-8"</span>,<span class="string">"UTF-8"</span>);
<a name="142" href="#142">142</a>         _testHttpValid(<span class="string">"text/xml;charset=UTF-8"</span>,<span class="string">"UTF-8-bom"</span>,<span class="string">"UTF-8"</span>,<strong>null</strong>);
<a name="143" href="#143">143</a>         _testHttpValid(<span class="string">"text/xml;charset=UTF-16"</span>,<span class="string">"UTF-16BE-bom"</span>,<span class="string">"UTF-16BE"</span>,<strong>null</strong>);
<a name="144" href="#144">144</a>         _testHttpValid(<span class="string">"text/xml;charset=UTF-16"</span>,<span class="string">"UTF-16BE-bom"</span>,<span class="string">"UTF-16BE"</span>,<span class="string">"UTF-16"</span>);
<a name="145" href="#145">145</a>         _testHttpValid(<span class="string">"text/xml;charset=UTF-16"</span>,<span class="string">"UTF-16BE-bom"</span>,<span class="string">"UTF-16BE"</span>,<span class="string">"UTF-16BE"</span>);
<a name="146" href="#146">146</a>         _testHttpValid(<span class="string">"text/xml"</span>,<span class="string">"UTF-8-bom"</span>,<span class="string">"US-ASCII"</span>,<strong>null</strong>);
<a name="147" href="#147">147</a> 
<a name="148" href="#148">148</a>         _testHttpInvalid(<span class="string">"text/xml;charset=UTF-16BE"</span>,<span class="string">"UTF-16BE-bom"</span>,<span class="string">"UTF-16BE"</span>,<strong>null</strong>);
<a name="149" href="#149">149</a>         _testHttpInvalid(<span class="string">"text/xml;charset=UTF-16BE"</span>,<span class="string">"UTF-16BE-bom"</span>,<span class="string">"UTF-16BE"</span>,<span class="string">"UTF-16"</span>);
<a name="150" href="#150">150</a>         _testHttpInvalid(<span class="string">"text/xml;charset=UTF-16BE"</span>,<span class="string">"UTF-16BE-bom"</span>,<span class="string">"UTF-16BE"</span>,<span class="string">"UTF-16BE"</span>);
<a name="151" href="#151">151</a>         _testHttpInvalid(<span class="string">"text/xml;charset=UTF-16"</span>,<span class="string">"no-bom"</span>,<span class="string">"UTF-16BE"</span>,<span class="string">"UTF-16BE"</span>);
<a name="152" href="#152">152</a>         _testHttpInvalid(<span class="string">"text/xml;charset=UTF-16"</span>,<span class="string">"no-bom"</span>,<span class="string">"UTF-16BE"</span>,<strong>null</strong>);
<a name="153" href="#153">153</a> 
<a name="154" href="#154">154</a>         _testHttpLenient(<span class="string">"text/xml"</span>,<span class="string">"no-bom"</span>,<span class="string">"US-ASCII"</span>,<strong>null</strong>, <span class="string">"US-ASCII"</span>);
<a name="155" href="#155">155</a>         _testHttpLenient(<span class="string">"text/xml;charset=UTF-8"</span>,<span class="string">"UTF-8-bom"</span>,<span class="string">"UTF-8"</span>,<span class="string">"UTF-8"</span>, <span class="string">"UTF-8"</span>);
<a name="156" href="#156">156</a>         _testHttpLenient(<span class="string">"text/xml;charset=UTF-8"</span>,<span class="string">"UTF-8-bom"</span>,<span class="string">"UTF-8"</span>,<strong>null</strong>, <span class="string">"UTF-8"</span>);
<a name="157" href="#157">157</a>         _testHttpLenient(<span class="string">"text/xml;charset=UTF-16"</span>,<span class="string">"UTF-16BE-bom"</span>,<span class="string">"UTF-16BE"</span>,<strong>null</strong>, <span class="string">"UTF-16BE"</span>);
<a name="158" href="#158">158</a>         _testHttpLenient(<span class="string">"text/xml;charset=UTF-16"</span>,<span class="string">"UTF-16BE-bom"</span>,<span class="string">"UTF-16BE"</span>,<span class="string">"UTF-16"</span>, <span class="string">"UTF-16"</span>);
<a name="159" href="#159">159</a>         _testHttpLenient(<span class="string">"text/xml;charset=UTF-16"</span>,<span class="string">"UTF-16BE-bom"</span>,<span class="string">"UTF-16BE"</span>,<span class="string">"UTF-16BE"</span>, <span class="string">"UTF-16BE"</span>);
<a name="160" href="#160">160</a>         _testHttpLenient(<span class="string">"text/xml"</span>,<span class="string">"UTF-8-bom"</span>,<span class="string">"US-ASCII"</span>,<strong>null</strong>, <span class="string">"US-ASCII"</span>);
<a name="161" href="#161">161</a> 
<a name="162" href="#162">162</a>         _testHttpLenient(<span class="string">"text/xml;charset=UTF-16BE"</span>,<span class="string">"UTF-16BE-bom"</span>,<span class="string">"UTF-16BE"</span>,<strong>null</strong>, <span class="string">"UTF-16BE"</span>);
<a name="163" href="#163">163</a>         _testHttpLenient(<span class="string">"text/xml;charset=UTF-16BE"</span>,<span class="string">"UTF-16BE-bom"</span>,<span class="string">"UTF-16BE"</span>,<span class="string">"UTF-16"</span>, <span class="string">"UTF-16"</span>);
<a name="164" href="#164">164</a>         _testHttpLenient(<span class="string">"text/xml;charset=UTF-16BE"</span>,<span class="string">"UTF-16BE-bom"</span>,<span class="string">"UTF-16BE"</span>,<span class="string">"UTF-16BE"</span>, <span class="string">"UTF-16BE"</span>);
<a name="165" href="#165">165</a>         _testHttpLenient(<span class="string">"text/xml;charset=UTF-16"</span>,<span class="string">"no-bom"</span>,<span class="string">"UTF-16BE"</span>,<span class="string">"UTF-16BE"</span>, <span class="string">"UTF-16BE"</span>);
<a name="166" href="#166">166</a>         _testHttpLenient(<span class="string">"text/xml;charset=UTF-16"</span>,<span class="string">"no-bom"</span>,<span class="string">"UTF-16BE"</span>,<strong>null</strong>, <span class="string">"UTF-16"</span>);
<a name="167" href="#167">167</a> 
<a name="168" href="#168">168</a>         _testHttpLenient(<span class="string">"text/html"</span>,<span class="string">"no-bom"</span>,<span class="string">"US-ASCII"</span>,<span class="string">"US-ASCII"</span>, <span class="string">"US-ASCII"</span>);
<a name="169" href="#169">169</a>         _testHttpLenient(<span class="string">"text/html"</span>,<span class="string">"no-bom"</span>,<span class="string">"US-ASCII"</span>,<strong>null</strong>, <span class="string">"US-ASCII"</span>);
<a name="170" href="#170">170</a>         _testHttpLenient(<span class="string">"text/html;charset=UTF-8"</span>,<span class="string">"no-bom"</span>,<span class="string">"US-ASCII"</span>,<span class="string">"UTF-8"</span>, <span class="string">"UTF-8"</span>);
<a name="171" href="#171">171</a>         _testHttpLenient(<span class="string">"text/html;charset=UTF-16BE"</span>,<span class="string">"no-bom"</span>,<span class="string">"US-ASCII"</span>,<span class="string">"UTF-8"</span>, <span class="string">"UTF-8"</span>);
<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> _testHttpValid(String cT,String bomEnc,String streamEnc,String prologEnc) throws Exception {
<a name="175" href="#175">175</a>         InputStream is = getXmlStream(bomEnc,(prologEnc==<strong>null</strong>)?XML1 :XML3,streamEnc,prologEnc);
<a name="176" href="#176">176</a>         XmlReader xmlReader = <strong>new</strong> XmlReader(is,cT,false);
<a name="177" href="#177">177</a>         <strong>if</strong> (!streamEnc.equals(<span class="string">"UTF-16"</span>)) {
<a name="178" href="#178">178</a>             <em class="comment">// we can not assert things here becuase UTF-8, US-ASCII and ISO-8859-1 look alike for the chars used for detection</em>
<a name="179" href="#179">179</a>         }
<a name="180" href="#180">180</a>         <strong>else</strong> {
<a name="181" href="#181">181</a>             assertEquals(xmlReader.getEncoding().substring(0,streamEnc.length()),streamEnc);
<a name="182" href="#182">182</a>         }
<a name="183" href="#183">183</a>     }
<a name="184" href="#184">184</a> 
<a name="185" href="#185">185</a>     <strong>protected</strong> <strong>void</strong> _testHttpInvalid(String cT,String bomEnc,String streamEnc,String prologEnc) throws Exception {