xmlreader.html
来自「rsslibj-1_0RC2,RSS阅读」· HTML 代码 · 共 679 行 · 第 1/4 页
HTML
679 行
<a name="502" href="#502">502</a> <a name="503" href="#503">503</a> <strong>private</strong> <strong>static</strong> <strong>final</strong> Pattern CHARSET_PATTERN = Pattern.compile(<span class="string">"charset=([.[^; ]]*)"</span>);<a name="504" href="#504">504</a> <a name="505" href="#505">505</a> <em class="comment">// returns charset parameter value, NULL if not present, NULL if httpContentType is NULL</em><a name="506" href="#506">506</a> <strong>private</strong> <strong>static</strong> String getContentTypeEncoding(String httpContentType) {<a name="507" href="#507">507</a> String encoding = <strong>null</strong>;<a name="508" href="#508">508</a> <strong>if</strong> (httpContentType!=<strong>null</strong>) {<a name="509" href="#509">509</a> <strong>int</strong> i = httpContentType.indexOf(<span class="string">";"</span>);<a name="510" href="#510">510</a> <strong>if</strong> (i>-1) {<a name="511" href="#511">511</a> String postMime = httpContentType.substring(i+1);<a name="512" href="#512">512</a> Matcher m = CHARSET_PATTERN.matcher(postMime);<a name="513" href="#513">513</a> encoding = (m.find()) ? m.group(1) : <strong>null</strong>;<a name="514" href="#514">514</a> encoding = (encoding!=<strong>null</strong>) ? encoding.toUpperCase() : <strong>null</strong>;<a name="515" href="#515">515</a> }<a name="516" href="#516">516</a> <strong>if</strong> (encoding != <strong>null</strong> &&<a name="517" href="#517">517</a> ((encoding.startsWith(<span class="string">"\""</span>) && encoding.endsWith(<span class="string">"\""</span>)) ||<a name="518" href="#518">518</a> (encoding.startsWith(<span class="string">"'"</span>) && encoding.endsWith(<span class="string">"'"</span>))<a name="519" href="#519">519</a> )) {<a name="520" href="#520">520</a> encoding = encoding.substring(1, encoding.length() - 1);<a name="521" href="#521">521</a> }<a name="522" href="#522">522</a> }<a name="523" href="#523">523</a> <strong>return</strong> encoding;<a name="524" href="#524">524</a> }<a name="525" href="#525">525</a> <a name="526" href="#526">526</a> <em class="comment">// returns the BOM in the stream, NULL if not present,</em><a name="527" href="#527">527</a> <em class="comment">// if there was BOM the in the stream it is consumed</em><a name="528" href="#528">528</a> <strong>private</strong> <strong>static</strong> String getBOMEncoding(BufferedInputStream is) throws IOException {<a name="529" href="#529">529</a> String encoding = <strong>null</strong>;<a name="530" href="#530">530</a> <strong>int</strong>[] bytes = <strong>new</strong> <strong>int</strong>[3];<a name="531" href="#531">531</a> is.mark(3);<a name="532" href="#532">532</a> bytes[0] = is.read();<a name="533" href="#533">533</a> bytes[1] = is.read();<a name="534" href="#534">534</a> bytes[2] = is.read();<a name="535" href="#535">535</a> <a name="536" href="#536">536</a> <strong>if</strong> (bytes[0] == 0xFE && bytes[1] == 0xFF) {<a name="537" href="#537">537</a> encoding = UTF_16BE;<a name="538" href="#538">538</a> is.reset();<a name="539" href="#539">539</a> is.read();<a name="540" href="#540">540</a> is.read();<a name="541" href="#541">541</a> }<a name="542" href="#542">542</a> <strong>else</strong><a name="543" href="#543">543</a> <strong>if</strong> (bytes[0] == 0xFF && bytes[1] == 0xFE) {<a name="544" href="#544">544</a> encoding = UTF_16LE;<a name="545" href="#545">545</a> is.reset();<a name="546" href="#546">546</a> is.read();<a name="547" href="#547">547</a> is.read();<a name="548" href="#548">548</a> }<a name="549" href="#549">549</a> <strong>else</strong><a name="550" href="#550">550</a> <strong>if</strong> (bytes[0] == 0xEF && bytes[1] == 0xBB && bytes[2] == 0xBF) {<a name="551" href="#551">551</a> encoding = UTF_8;<a name="552" href="#552">552</a> }<a name="553" href="#553">553</a> <strong>else</strong> {<a name="554" href="#554">554</a> is.reset();<a name="555" href="#555">555</a> }<a name="556" href="#556">556</a> <strong>return</strong> encoding;<a name="557" href="#557">557</a> }<a name="558" href="#558">558</a> <a name="559" href="#559">559</a> <em class="comment">// returns the best guess for the encoding by looking the first bytes of the stream, '<?'</em><a name="560" href="#560">560</a> <strong>private</strong> <strong>static</strong> String getXMLGuessEncoding(BufferedInputStream is) throws IOException {<a name="561" href="#561">561</a> String encoding = <strong>null</strong>;<a name="562" href="#562">562</a> <strong>int</strong>[] bytes = <strong>new</strong> <strong>int</strong>[4];<a name="563" href="#563">563</a> is.mark(4);<a name="564" href="#564">564</a> bytes[0] = is.read();<a name="565" href="#565">565</a> bytes[1] = is.read();<a name="566" href="#566">566</a> bytes[2] = is.read();<a name="567" href="#567">567</a> bytes[3] = is.read();<a name="568" href="#568">568</a> is.reset();<a name="569" href="#569">569</a> <a name="570" href="#570">570</a> <strong>if</strong> (bytes[0] == 0x00 && bytes[1] == 0x3C && bytes[2] == 0x00 && bytes[3] == 0x3F) {<a name="571" href="#571">571</a> encoding = UTF_16BE;<a name="572" href="#572">572</a> }<a name="573" href="#573">573</a> <strong>else</strong><a name="574" href="#574">574</a> <strong>if</strong> (bytes[0] == 0x3C && bytes[1] == 0x00 && bytes[2] == 0x3F && bytes[3] == 0x00) {<a name="575" href="#575">575</a> encoding = UTF_16LE;<a name="576" href="#576">576</a> }<a name="577" href="#577">577</a> <strong>else</strong><a name="578" href="#578">578</a> <strong>if</strong> (bytes[0] == 0x3C && bytes[1] == 0x3F && bytes[2] == 0x78 && bytes[3] == 0x6D) {<a name="579" href="#579">579</a> encoding = UTF_8;<a name="580" href="#580">580</a> }<a name="581" href="#581">581</a> <strong>return</strong> encoding;<a name="582" href="#582">582</a> }<a name="583" href="#583">583</a> <a name="584" href="#584">584</a> <a name="585" href="#585">585</a> <strong>private</strong> <strong>static</strong> <strong>final</strong> Pattern ENCODING_PATTERN =<a name="586" href="#586">586</a> Pattern.compile(<span class="string">"<//?xml.*encoding[//s]*=[//s]*((?:\".[^\"]*\")|(?:'.[^']*'))"</span>, Pattern.MULTILINE);<a name="587" href="#587">587</a> <a name="588" href="#588">588</a> <em class="comment">// returns the encoding declared in the <?xml encoding=...?>, NULL if none</em><a name="589" href="#589">589</a> <strong>private</strong> <strong>static</strong> String getXmlProlog(BufferedInputStream is,String guessedEnc) throws IOException {<a name="590" href="#590">590</a> String encoding = <strong>null</strong>;<a name="591" href="#591">591</a> <strong>if</strong> (guessedEnc!=<strong>null</strong>) {<a name="592" href="#592">592</a> byte[] bytes = <strong>new</strong> byte[BUFFER_SIZE];<a name="593" href="#593">593</a> is.mark(BUFFER_SIZE);<a name="594" href="#594">594</a> <strong>int</strong> offset = 0;<a name="595" href="#595">595</a> <strong>int</strong> max = BUFFER_SIZE;<a name="596" href="#596">596</a> <strong>int</strong> c = is.read(bytes,offset,max);<a name="597" href="#597">597</a> <strong>int</strong> firstGT = -1;<a name="598" href="#598">598</a> <strong>while</strong> (c!=-1 && firstGT==-1 && offset< BUFFER_SIZE) {<a name="599" href="#599">599</a> offset += c;<a name="600" href="#600">600</a> max -= c;<a name="601" href="#601">601</a> c = is.read(bytes,offset,max);<a name="602" href="#602">602</a> firstGT = <strong>new</strong> String(bytes, 0, offset).indexOf(<span class="string">">"</span>);<a name="603" href="#603">603</a> }<a name="604" href="#604">604</a> <strong>if</strong> (firstGT == -1) {<a name="605" href="#605">605</a> <strong>if</strong> (c == -1) {<a name="606" href="#606">606</a> <strong>throw</strong> <strong>new</strong> IOException(<span class="string">"Unexpected end of XML stream"</span>);<a name="607" href="#607">607</a> }<a name="608" href="#608">608</a> <strong>else</strong> {<a name="609" href="#609">609</a> <strong>throw</strong> <strong>new</strong> IOException(<span class="string">"XML prolog or ROOT element not found on first "</span> + offset + <span class="string">" bytes"</span>);<a name="610" href="#610">610</a> }<a name="611" href="#611">611</a> }<a name="612" href="#612">612</a> <strong>int</strong> bytesRead = offset;<a name="613" href="#613">613</a> <strong>if</strong> (bytesRead>0) {<a name="614" href="#614">614</a> is.reset();<a name="615" href="#615">615</a> Reader reader = <strong>new</strong> InputStreamReader(<strong>new</strong> ByteArrayInputStream(bytes,0,firstGT + 1), guessedEnc);<a name="616" href="#616">616</a> BufferedReader bReader = <strong>new</strong> BufferedReader(reader);<a name="617" href="#617">617</a> StringBuffer prolog = <strong>new</strong> StringBuffer();<a name="618" href="#618">618</a> String line = bReader.readLine();<a name="619" href="#619">619</a> <strong>while</strong> (line != <strong>null</strong>) {<a name="620" href="#620">620</a> prolog.append(line);<a name="621" href="#621">621</a> line = bReader.readLine();<a name="622" href="#622">622</a> }<a name="623" href="#623">623</a> Matcher m = ENCODING_PATTERN.matcher(prolog);<a name="624" href="#624">624</a> <strong>if</strong> (m.find()) {<a name="625" href="#625">625</a> encoding = m.group(1).toUpperCase();<a name="626" href="#626">626</a> encoding = encoding.substring(1,encoding.length()-1);<a name="627" href="#627">627</a> }<a name="628" href="#628">628</a> }<a name="629" href="#629">629</a> }<a name="630" href="#630">630</a> <strong>return</strong> encoding;<a name="631" href="#631">631</a> }<a name="632" href="#632">632</a> <a name="633" href="#633">633</a> <em class="comment">// indicates if the MIME type belongs to the APPLICATION XML family</em><a name="634" href="#634">634</a> <strong>private</strong> <strong>static</strong> <strong>boolean</strong> isAppXml(String mime) {<a name="635" href="#635">635</a> <strong>return</strong> mime!=<strong>null</strong> &&<a name="636" href="#636">636</a> (mime.equals(<span class="string">"application/xml"</span>) ||<a name="637" href="#637">637</a> mime.equals(<span class="string">"application/xml-dtd"</span>) ||<a name="638" href="#638">638</a> mime.equals(<span class="string">"application/xml-external-parsed-entity"</span>) ||<a name="639" href="#639">639</a> (mime.startsWith(<span class="string">"application/"</span>) && mime.endsWith(<span class="string">"+xml"</span>)));<a name="640" href="#640">640</a> }<a name="641" href="#641">641</a> <a name="642" href="#642">642</a> <em class="comment">// indicates if the MIME type belongs to the TEXT XML family</em><a name="643" href="#643">643</a> <strong>private</strong> <strong>static</strong> <strong>boolean</strong> isTextXml(String mime) {<a name="644" href="#644">644</a> <strong>return</strong> mime!=<strong>null</strong> &&<a name="645" href="#645">645</a> (mime.equals(<span class="string">"text/xml"</span>) ||<a name="646" href="#646">646</a> mime.equals(<span class="string">"text/xml-external-parsed-entity"</span>) ||<a name="647" href="#647">647</a> (mime.startsWith(<span class="string">"text/"</span>) && mime.endsWith(<span class="string">"+xml"</span>)));<a name="648" href="#648">648</a> }<a name="649" href="#649">649</a> <a name="650" href="#650">650</a> <strong>private</strong> <strong>static</strong> <strong>final</strong> MessageFormat RAW_EX_1 = <strong>new</strong> MessageFormat(<a name="651" href="#651">651</a> <span class="string">"Invalid encoding, BOM [{0}] XML guess [{1}] XML prolog [{2}] encoding mismatch"</span>);<a name="652" href="#652">652</a> <a name="653" href="#653">653</a> <strong>private</strong> <strong>static</strong> <strong>final</strong> MessageFormat RAW_EX_2 = <strong>new</strong> MessageFormat(<a name="654" href="#654">654</a> <span class="string">"Invalid encoding, BOM [{0}] XML guess [{1}] XML prolog [{2}] unknown BOM"</span>);<a name="655" href="#655">655</a> <a name="656" href="#656">656</a> <strong>private</strong> <strong>static</strong> <strong>final</strong> MessageFormat HTTP_EX_1 = <strong>new</strong> MessageFormat(<a name="657" href="#657">657</a> <span class="string">"Invalid encoding, CT-MIME [{0}] CT-Enc [{1}] BOM [{2}] XML guess [{3}] XML prolog [{4}], BOM must be NULL"</span>);<a name="658" href="#658">658</a> <a name="659" href="#659">659</a> <strong>private</strong> <strong>static</strong> <strong>final</strong> MessageFormat HTTP_EX_2 = <strong>new</strong> MessageFormat(<a name="660" href="#660">660</a> <span class="string">"Invalid encoding, CT-MIME [{0}] CT-Enc [{1}] BOM [{2}] XML guess [{3}] XML prolog [{4}], encoding mismatch"</span>);<a name="661" href="#661">661</a> <a name="662" href="#662">662</a> <strong>private</strong> <strong>static</strong> <strong>final</strong> MessageFormat HTTP_EX_3 = <strong>new</strong> MessageFormat(<a name="663" href="#663">663</a> <span class="string">"Invalid encoding, CT-MIME [{0}] CT-Enc [{1}] BOM [{2}] XML guess [{3}] XML prolog [{4}], Invalid MIME"</span>);<a name="664" href="#664">664</a> <a name="665" href="#665">665</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?