Attachment #116405 for bug #251748

Some Eclipse Foundation services are deprecated, or will be soon. Please ensure you've read this important communication.

View | Details | Raw Unified | Return to bug 251748
Collapse All | Expand All




	private static final QualifiedName[] SUPPORTED_OPTIONS = new QualifiedName[] {IContentDescription.CHARSET, IContentDescription.BYTE_ORDER_MARK};
	private static final String ENCODING = "encoding="; //$NON-NLS-1$
	private static final String XML_PREFIX = "<?xml "; //$NON-NLS-1$
	private static final String XML_DECL_END = "?>"; //$NON-NLS-1$

	public int describe(InputStream input, IContentDescription description) throws IOException {
		byte[] bom = getByteOrderMark(input);

		input.reset();
		if (bom != null) {
			if (bom == IContentDescription.BOM_UTF_16BE)
				xmlDeclEncoding = "UTF-16";//BE"; //$NON-NLS-1$
			else if (bom == IContentDescription.BOM_UTF_16LE)
				xmlDeclEncoding = "UTF-16";//LE"; //$NON-NLS-1$
			// skip BOM to make comparison simpler
			input.skip(bom.length);
			// set the BOM in the description if requested
			if (description != null && description.isRequested(IContentDescription.BYTE_ORDER_MARK))
				description.setProperty(IContentDescription.BYTE_ORDER_MARK, bom);
		}
		return internalDescribe(new InputStreamReader(input, xmlDeclEncoding), description);
		byte[] prefix = new byte[xmlPrefixBytes.length];
		if (input.read(prefix) < prefix.length)
			// there is not enough info to say anything
			return INDETERMINATE;
		for (int i = 0; i < prefix.length; i++)
			if (prefix[i] != xmlPrefixBytes[i])
				// we don't have a XMLDecl... there is not enough info to say anything
				return INDETERMINATE;
		if (description == null)
			return VALID;
		// describe charset if requested
		if (description.isRequested(IContentDescription.CHARSET)) {
			String fullXMLDecl = readFullXMLDecl(input, xmlDeclEncoding);
			if (fullXMLDecl != null) {
				String charset = getCharset(fullXMLDecl);
				if (charset != null && !isCharsetValid(charset))
					return INVALID;
				if (charset != null && !charset.matches("[uU][tT][fF](-)?8")) //$NON-NLS-1$
					// only set property if value is not default (avoid using a non-default content description)
					description.setProperty(IContentDescription.CHARSET, charset);
				if (charset == null || !charset.matches("[uU][tT][fF](-)?(8|16)"))
					description.setProperty(IContentDescription.BYTE_ORDER_MARK, null);
			}
		}
		return VALID;
	}
	
	public int describe(Reader input, IContentDescription description) throws IOException {	
		return internalDescribe(input, description);
	}

	private String readFullXMLDecl(InputStream input, String unicodeEncoding) throws IOException {
		byte[] xmlDecl = new byte[100];
		int c = 0;
		// looks for XMLDecl ending char (?)
		int read = 0;
		while (read < xmlDecl.length && (c = input.read()) != -1 && c !='\r' && c !='\n')
			xmlDecl[read++] = (byte) c;
		return new String(xmlDecl, 0, read, unicodeEncoding);
	}
	
	private int internalDescribe(Reader input, IContentDescription description) throws IOException {	
		String line = readXMLDecl(input);
		
		// end of stream
		if (line == null)
			return INDETERMINATE;

				description.setProperty(IContentDescription.CHARSET, charset);
		}
		return VALID;
	} 
	
	private boolean isFullXMLDecl(String xmlDecl) {
		return xmlDecl.endsWith(XML_DECL_END);
	}

	private String readXMLDecl(Reader input) throws IOException {
		BufferedReader reader = new BufferedReader(input);
		StringBuffer buffer = new StringBuffer();
		String line = null;

		while (buffer.length() < 100 && ((line = reader.readLine()) != null)) {
			buffer.append(line);
			if (line.indexOf(XML_DECL_END) != -1) {
				return buffer.substring(0, buffer.indexOf(XML_DECL_END) + XML_DECL_END.length());
			}
		}
		return buffer.toString();
	}

	private String getCharset(String firstLine) {

		if (firstQuote == -1 || firstLine.length() == firstQuote - 1)
			return null;
		int secondQuote = firstLine.indexOf(quoteChar, firstQuote + 1);
		if (secondQuote == -1 && !isFullXMLDecl(firstLine))
			return null;
		else if (secondQuote == -1 && isFullXMLDecl(firstLine))
			return firstLine.substring(firstQuote + 1);
		return firstLine.substring(firstQuote + 1, secondQuote);
	}

	private boolean isCharsetValid(String charset) {
		return charset.matches("[A-Za-z]([A-Za-z0-9._\\-])*");
	}

	public QualifiedName[] getSupportedOptions() {
		return SUPPORTED_OPTIONS;
	}

Lines 45-66 Link Here

(-)src/org/eclipse/core/tests/runtime/content/XMLContentDescriberTest.java (-11 / +41 lines)
45		45
46	checkEncodedContents(ENCODING_NOTSUPPORTED, ENCODING_NOTSUPPORTED, ENCODING_UTF8);	46	checkEncodedContents(ENCODING_NOTSUPPORTED, ENCODING_NOTSUPPORTED, ENCODING_UTF8);
47	checkEncodedContents(ENCODING_NOTSUPPORTED, ENCODING_NOTSUPPORTED, ENCODING_UTF16);	47	checkEncodedContents(ENCODING_NOTSUPPORTED, ENCODING_NOTSUPPORTED, ENCODING_UTF16);
		48
		49	checkEncodedContents(ENCODING_UTF16, ENCODING_UTF16, ENCODING_UTF16, true);
		50	checkEncodedContents(ENCODING_UTF8, ENCODING_UTF8, ENCODING_UTF8, true);
		51
		52	checkEncodedContents(ENCODING_UTF16, ENCODING_UTF16, ENCODING_UTF8, true);
		53	checkEncodedContents(ENCODING_UTF8, ENCODING_UTF8, ENCODING_UTF16, true);
		54
		55	checkEncodedContents(ENCODING_NOTSUPPORTED, ENCODING_NOTSUPPORTED, ENCODING_UTF8, true);
		56	checkEncodedContents(ENCODING_NOTSUPPORTED, ENCODING_NOTSUPPORTED, ENCODING_UTF16, true);
48	}	57	}
49		58
50	public void testEncodedContents2() throws Exception {	59	public void testEncodedContents2() throws Exception {
51	checkEncodedContents2(ENCODING_UTF16, ENCODING_UTF16);	60	checkEncodedContents2(ENCODING_UTF16, ENCODING_UTF16);
52	checkEncodedContents2(ENCODING_UTF8, ENCODING_UTF8);	61	checkEncodedContents2(ENCODING_UTF8, ENCODING_UTF8);
53	checkEncodedContents2(ENCODING_NOTSUPPORTED, ENCODING_NOTSUPPORTED);	62	checkEncodedContents2(ENCODING_NOTSUPPORTED, ENCODING_NOTSUPPORTED);
		63
		64	checkEncodedContents2(ENCODING_UTF16, ENCODING_UTF16, true);
		65	checkEncodedContents2(ENCODING_UTF8, ENCODING_UTF8, true);
		66	checkEncodedContents2(ENCODING_NOTSUPPORTED, ENCODING_NOTSUPPORTED, true);
54	}	67	}
55		68
56	public void testEncodedContents3() throws Exception {	69	public void testEncodedContents3() throws Exception {
57	IContentDescription description = Platform.getContentTypeManager().getDescriptionFor(getInputStream(ENCODING_INCORRECT, ENCODING_UTF8), "fake.xml", new QualifiedName[] {IContentDescription.CHARSET});	70	IContentDescription description = Platform.getContentTypeManager().getDescriptionFor(getInputStream(ENCODING_INCORRECT, ENCODING_UTF8, false), "fake.xml", new QualifiedName[] {IContentDescription.CHARSET});
		71	assertNull("1.0", description);
		72
		73	description = Platform.getContentTypeManager().getDescriptionFor(getInputStream(ENCODING_INCORRECT, ENCODING_UTF16, false), "fake.xml", new QualifiedName[] {IContentDescription.CHARSET});
		74	assertNull("2.0", description);
		75
		76	description = Platform.getContentTypeManager().getDescriptionFor(getReader(ENCODING_INCORRECT, false), "fake.xml", new QualifiedName[] {IContentDescription.CHARSET});
		77	assertNull("3.0", description);
		78
		79	description = Platform.getContentTypeManager().getDescriptionFor(getInputStream(ENCODING_INCORRECT, ENCODING_UTF8, true), "fake.xml", new QualifiedName[] {IContentDescription.CHARSET});
58	assertNull("1.0", description);	80	assertNull("1.0", description);
59		81
60	description = Platform.getContentTypeManager().getDescriptionFor(getInputStream(ENCODING_INCORRECT, ENCODING_UTF16), "fake.xml", new QualifiedName[] {IContentDescription.CHARSET});	82	description = Platform.getContentTypeManager().getDescriptionFor(getInputStream(ENCODING_INCORRECT, ENCODING_UTF16, true), "fake.xml", new QualifiedName[] {IContentDescription.CHARSET});
61	assertNull("2.0", description);	83	assertNull("2.0", description);
62		84
63	description = Platform.getContentTypeManager().getDescriptionFor(getReader(ENCODING_INCORRECT), "fake.xml", new QualifiedName[] {IContentDescription.CHARSET});	85	description = Platform.getContentTypeManager().getDescriptionFor(getReader(ENCODING_INCORRECT, true), "fake.xml", new QualifiedName[] {IContentDescription.CHARSET});
64	assertNull("3.0", description);	86	assertNull("3.0", description);
65	}	87	}
66		88
Lines 68-94 Link Here
68	return new TestSuite(XMLContentDescriberTest.class);	90	return new TestSuite(XMLContentDescriberTest.class);
69	}	91	}
70		92
71	private void checkEncodedContents(String resultEncoding, String encodingInContent, String encoding) throws Exception {	93	private void checkEncodedContents(String resultEncoding, String encodingInContent, String encoding, boolean encInNewLine) throws Exception {
72	IContentDescription description = Platform.getContentTypeManager().getDescriptionFor(getInputStream(encodingInContent, encoding), "fake.xml", new QualifiedName[] {IContentDescription.CHARSET});	94	IContentDescription description = Platform.getContentTypeManager().getDescriptionFor(getInputStream(encodingInContent, encoding, encInNewLine), "fake.xml", new QualifiedName[] {IContentDescription.CHARSET});
73	assertNotNull("1.0", description);	95	assertNotNull("1.0", description);
74	assertEquals("1.1", Platform.PI_RUNTIME + ".xml", description.getContentType().getId());	96	assertEquals("1.1", Platform.PI_RUNTIME + ".xml", description.getContentType().getId());
75	assertEquals("1.2", resultEncoding, description.getProperty(IContentDescription.CHARSET));	97	assertEquals("1.2", resultEncoding, description.getProperty(IContentDescription.CHARSET));
76	}	98	}
77		99
78	private void checkEncodedContents2(String resultEncoding, String encodingInContent) throws Exception {	100	private void checkEncodedContents(String resultEncoding, String encodingInContent, String encoding) throws Exception {
79	IContentDescription description = Platform.getContentTypeManager().getDescriptionFor(getReader(encodingInContent), "fake.xml", new QualifiedName[] {IContentDescription.CHARSET});	101	checkEncodedContents(resultEncoding, encodingInContent, encoding, false);
		102	}
		103
		104	private void checkEncodedContents2(String resultEncoding, String encodingInContent, boolean encInNewLine) throws Exception {
		105	IContentDescription description = Platform.getContentTypeManager().getDescriptionFor(getReader(encodingInContent, encInNewLine), "fake.xml", new QualifiedName[] {IContentDescription.CHARSET});
80	assertNotNull("2.0", description);	106	assertNotNull("2.0", description);
81	assertEquals("2.1", Platform.PI_RUNTIME + ".xml", description.getContentType().getId());	107	assertEquals("2.1", Platform.PI_RUNTIME + ".xml", description.getContentType().getId());
82	assertEquals("2.2", resultEncoding, description.getProperty(IContentDescription.CHARSET));	108	assertEquals("2.2", resultEncoding, description.getProperty(IContentDescription.CHARSET));
83	}	109	}
84		110
85	private InputStream getInputStream(String encodingInContent, String encoding) throws UnsupportedEncodingException {	111	private void checkEncodedContents2(String resultEncoding, String encodingInContent) throws Exception {
86	String content = "<?xml version=\"1.0\" encoding=\"" + encodingInContent + "\"?><root attribute=\"" + ENCODED_TEXT + "\">";	112	checkEncodedContents2(resultEncoding, encodingInContent, false);
		113	}
		114
		115	private InputStream getInputStream(String encodingInContent, String encoding, boolean encInNewLine) throws UnsupportedEncodingException {
		116	String content = "<?xml version=\"1.0\"" + (encInNewLine ? "\n" : "") + "encoding=\"" + encodingInContent + "\"?><root attribute=\"" + ENCODED_TEXT + "\">";
87	return new ByteArrayInputStream(content.getBytes(encoding));	117	return new ByteArrayInputStream(content.getBytes(encoding));
88	}	118	}
89		119
90	private Reader getReader(String encodingInContent) {	120	private Reader getReader(String encodingInContent, boolean encInNewLine) {
91	String content = "<?xml version=\"1.0\" encoding=\"" + encodingInContent + "\"?><root attribute=\"" + ENCODED_TEXT + "\">";	121	String content = "<?xml version=\"1.0\"" + (encInNewLine ? "\n" : "") + " encoding=\"" + encodingInContent + "\"?><root attribute=\"" + ENCODED_TEXT + "\">";
92	return new InputStreamReader(new ByteArrayInputStream(content.getBytes()));	122	return new InputStreamReader(new ByteArrayInputStream(content.getBytes()));
93	}	123	}
94	}	124	}

Return to bug 251748

Lines 30-35 Link Here

(-)src/org/eclipse/core/internal/content/XMLContentDescriber.java (-45 / +35 lines)
30	private static final QualifiedName[] SUPPORTED_OPTIONS = new QualifiedName[] {IContentDescription.CHARSET, IContentDescription.BYTE_ORDER_MARK};	30	private static final QualifiedName[] SUPPORTED_OPTIONS = new QualifiedName[] {IContentDescription.CHARSET, IContentDescription.BYTE_ORDER_MARK};
31	private static final String ENCODING = "encoding="; //$NON-NLS-1$	31	private static final String ENCODING = "encoding="; //$NON-NLS-1$
32	private static final String XML_PREFIX = "<?xml "; //$NON-NLS-1$	32	private static final String XML_PREFIX = "<?xml "; //$NON-NLS-1$
		33	private static final String XML_DECL_END = "?>"; //$NON-NLS-1$
33		34
34	public int describe(InputStream input, IContentDescription description) throws IOException {	35	public int describe(InputStream input, IContentDescription description) throws IOException {
35	byte[] bom = getByteOrderMark(input);	36	byte[] bom = getByteOrderMark(input);
Lines 37-96 Link Here
37	input.reset();	38	input.reset();
38	if (bom != null) {	39	if (bom != null) {
39	if (bom == IContentDescription.BOM_UTF_16BE)	40	if (bom == IContentDescription.BOM_UTF_16BE)
40	xmlDeclEncoding = "UTF-16BE"; //$NON-NLS-1$	41	xmlDeclEncoding = "UTF-16";//BE"; //$NON-NLS-1$
41	else if (bom == IContentDescription.BOM_UTF_16LE)	42	else if (bom == IContentDescription.BOM_UTF_16LE)
42	xmlDeclEncoding = "UTF-16LE"; //$NON-NLS-1$	43	xmlDeclEncoding = "UTF-16";//LE"; //$NON-NLS-1$
43	// skip BOM to make comparison simpler	44	// skip BOM to make comparison simpler
44	input.skip(bom.length);	45	input.skip(bom.length);
45	// set the BOM in the description if requested	46	// set the BOM in the description if requested
46	if (description != null && description.isRequested(IContentDescription.BYTE_ORDER_MARK))	47	if (description != null && description.isRequested(IContentDescription.BYTE_ORDER_MARK))
47	description.setProperty(IContentDescription.BYTE_ORDER_MARK, bom);	48	description.setProperty(IContentDescription.BYTE_ORDER_MARK, bom);
48	}	49	}
49	byte[] xmlPrefixBytes = XML_PREFIX.getBytes(xmlDeclEncoding);	50	return internalDescribe(new InputStreamReader(input, xmlDeclEncoding), description);
50	byte[] prefix = new byte[xmlPrefixBytes.length];
51	if (input.read(prefix) < prefix.length)
52	// there is not enough info to say anything
53	return INDETERMINATE;
54	for (int i = 0; i < prefix.length; i++)
55	if (prefix[i] != xmlPrefixBytes[i])
56	// we don't have a XMLDecl... there is not enough info to say anything
57	return INDETERMINATE;
58	if (description == null)
59	return VALID;
60	// describe charset if requested
61	if (description.isRequested(IContentDescription.CHARSET)) {
62	String fullXMLDecl = readFullXMLDecl(input, xmlDeclEncoding);
63	if (fullXMLDecl != null) {
64	String charset = getCharset(fullXMLDecl);
65	if (charset != null && !isCharsetValid(charset))
66	return INVALID;
67	if (charset != null && !charset.matches("[uU][tT][fF](-)?8")) //$NON-NLS-1$
68	// only set property if value is not default (avoid using a non-default content description)
69	description.setProperty(IContentDescription.CHARSET, charset);
70	if (charset == null \|\| !charset.matches("[uU][tT][fF](-)?(8\|16)"))
71	description.setProperty(IContentDescription.BYTE_ORDER_MARK, null);
72	}
73	}
74	return VALID;
75	}	51	}
76		52
77	private boolean isCharsetValid(String charset){	53	public int describe(Reader input, IContentDescription description) throws IOException {
78	return charset.matches("[A-Za-z]([A-Za-z0-9._\\-])*");	54	return internalDescribe(input, description);
79	}
80
81	private String readFullXMLDecl(InputStream input, String unicodeEncoding) throws IOException {
82	byte[] xmlDecl = new byte[100];
83	int c = 0;
84	// looks for XMLDecl ending char (?)
85	int read = 0;
86	while (read < xmlDecl.length && (c = input.read()) != -1 && c !='\r' && c !='\n')
87	xmlDecl[read++] = (byte) c;
88	return new String(xmlDecl, 0, read, unicodeEncoding);
89	}	55	}
90		56
91	public int describe(Reader input, IContentDescription description) throws IOException {	57	private int internalDescribe(Reader input, IContentDescription description) throws IOException {
92	BufferedReader reader = new BufferedReader(input);	58	String line = readXMLDecl(input);
93	String line = reader.readLine();	59
94	// end of stream	60	// end of stream
95	if (line == null)	61	if (line == null)
96	return INDETERMINATE;	62	return INDETERMINATE;
Lines 109-114 Link Here
109	description.setProperty(IContentDescription.CHARSET, charset);	75	description.setProperty(IContentDescription.CHARSET, charset);
110	}	76	}
111	return VALID;	77	return VALID;
		78	}
		79
		80	private boolean isFullXMLDecl(String xmlDecl) {
		81	return xmlDecl.endsWith(XML_DECL_END);
		82	}
		83
		84	private String readXMLDecl(Reader input) throws IOException {
		85	BufferedReader reader = new BufferedReader(input);
		86	StringBuffer buffer = new StringBuffer();
		87	String line = null;
		88
		89	while (buffer.length() < 100 && ((line = reader.readLine()) != null)) {
		90	buffer.append(line);
		91	if (line.indexOf(XML_DECL_END) != -1) {
		92	return buffer.substring(0, buffer.indexOf(XML_DECL_END) + XML_DECL_END.length());
		93	}
		94	}
		95	return buffer.toString();
112	}	96	}
113		97
114	private String getCharset(String firstLine) {	98	private String getCharset(String firstLine) {
Lines 124-134 Link Here
124	if (firstQuote == -1 \|\| firstLine.length() == firstQuote - 1)	108	if (firstQuote == -1 \|\| firstLine.length() == firstQuote - 1)
125	return null;	109	return null;
126	int secondQuote = firstLine.indexOf(quoteChar, firstQuote + 1);	110	int secondQuote = firstLine.indexOf(quoteChar, firstQuote + 1);
127	if (secondQuote == -1)	111	if (secondQuote == -1 && !isFullXMLDecl(firstLine))
128	return null;	112	return null;
		113	else if (secondQuote == -1 && isFullXMLDecl(firstLine))
		114	return firstLine.substring(firstQuote + 1);
129	return firstLine.substring(firstQuote + 1, secondQuote);	115	return firstLine.substring(firstQuote + 1, secondQuote);
130	}	116	}
131		117
		118	private boolean isCharsetValid(String charset) {
		119	return charset.matches("[A-Za-z]([A-Za-z0-9._\\-])*");
		120	}
		121
132	public QualifiedName[] getSupportedOptions() {	122	public QualifiedName[] getSupportedOptions() {
133	return SUPPORTED_OPTIONS;	123	return SUPPORTED_OPTIONS;
134	}	124	}