1 | package clarin.cmdi.schema.cmd; |
---|
2 | |
---|
3 | import clarin.cmdi.xml.Saxon; |
---|
4 | import java.io.File; |
---|
5 | import java.io.IOException; |
---|
6 | import java.io.InputStream; |
---|
7 | import java.net.MalformedURLException; |
---|
8 | import java.net.URL; |
---|
9 | import java.util.List; |
---|
10 | import javax.xml.XMLConstants; |
---|
11 | import javax.xml.parsers.DocumentBuilderFactory; |
---|
12 | import javax.xml.parsers.ParserConfigurationException; |
---|
13 | import javax.xml.transform.Source; |
---|
14 | import javax.xml.transform.dom.DOMSource; |
---|
15 | import javax.xml.transform.stream.StreamSource; |
---|
16 | import javax.xml.validation.Schema; |
---|
17 | import javax.xml.validation.SchemaFactory; |
---|
18 | import net.sf.saxon.s9api.DOMDestination; |
---|
19 | import net.sf.saxon.s9api.QName; |
---|
20 | import net.sf.saxon.s9api.SaxonApiException; |
---|
21 | import net.sf.saxon.s9api.XdmDestination; |
---|
22 | import net.sf.saxon.s9api.XdmItem; |
---|
23 | import net.sf.saxon.s9api.XdmNode; |
---|
24 | import net.sf.saxon.s9api.XsltExecutable; |
---|
25 | import net.sf.saxon.s9api.XsltTransformer; |
---|
26 | import org.w3c.dom.Document; |
---|
27 | import org.xml.sax.SAXException; |
---|
28 | |
---|
29 | /** |
---|
30 | * The Validator class handles the XSD and Schematron validation of CMD profile and component specifications. |
---|
31 | * |
---|
32 | * The class caches thread safe versions of the XSD Schema and the Schematron XSLT. |
---|
33 | * So multiple instances of this Validator class can be used in parallel and use the same cached schema and transformer. |
---|
34 | * Although a single instance of the Validator class can't be accessed in parallel it can be used to validate multiple CMD |
---|
35 | * profiles/components in sequence. |
---|
36 | * |
---|
37 | * @author menwin |
---|
38 | * @author twagoo |
---|
39 | */ |
---|
40 | public class Validator { |
---|
41 | |
---|
42 | /** |
---|
43 | * Default location of the CMD schema |
---|
44 | */ |
---|
45 | static final String CMD_SCHEMA_URL = "http://www.clarin.eu/cmd/general-component-schema.xsd"; |
---|
46 | /** |
---|
47 | * The immutable location of the CMD schema that is used in this instance |
---|
48 | */ |
---|
49 | private final URL cmdSchemaUri; |
---|
50 | /** |
---|
51 | * The "immutable, and therefore thread-safe," "compiled form of [the Schematron] stylesheet". |
---|
52 | */ |
---|
53 | private XsltExecutable cmdSchematron = null; |
---|
54 | /** |
---|
55 | * The "immutable in-memory representation of [the XSD] grammar". |
---|
56 | */ |
---|
57 | private Schema cmdSchema = null; |
---|
58 | /** |
---|
59 | * The list of validation messages compiled a the last run of the validator. |
---|
60 | */ |
---|
61 | private List<Message> msgList = null; |
---|
62 | /** |
---|
63 | * The Schematron SVRL validation report |
---|
64 | */ |
---|
65 | private XdmNode validationReport = null; |
---|
66 | |
---|
67 | /** |
---|
68 | * Creates a Validator that uses a specific schema specified by its URL |
---|
69 | * |
---|
70 | * @param cmdSchemaUri Schema URI to use |
---|
71 | * @see #CMD_SCHEMA_URL |
---|
72 | */ |
---|
73 | public Validator(URL cmdSchemaUri) { |
---|
74 | this.cmdSchemaUri = cmdSchemaUri; |
---|
75 | } |
---|
76 | |
---|
77 | /** |
---|
78 | * Convenience method to build a XSLT transformer from a resource. |
---|
79 | * |
---|
80 | * @param uri The location of the resource |
---|
81 | * @return An executable XSLT |
---|
82 | * @throws Exception |
---|
83 | */ |
---|
84 | static XsltExecutable buildTransformer(File file) throws ValidatorException { |
---|
85 | try { |
---|
86 | XdmNode xslDoc = Saxon.buildDocument(new javax.xml.transform.stream.StreamSource(file)); |
---|
87 | return Saxon.buildTransformer(xslDoc); |
---|
88 | } catch (SaxonApiException ex) { |
---|
89 | throw new ValidatorException(ex); |
---|
90 | } |
---|
91 | } |
---|
92 | |
---|
93 | /** |
---|
94 | * Convenience method to build a XSLT transformer from a resource. |
---|
95 | * |
---|
96 | * @param uri The location of the resource |
---|
97 | * @return An executable XSLT |
---|
98 | * @throws Exception |
---|
99 | */ |
---|
100 | static XsltExecutable buildTransformer(URL url) throws ValidatorException { |
---|
101 | try { |
---|
102 | XdmNode xslDoc = Saxon.buildDocument(new javax.xml.transform.stream.StreamSource(url.toExternalForm())); |
---|
103 | return Saxon.buildTransformer(xslDoc); |
---|
104 | } catch (SaxonApiException ex) { |
---|
105 | throw new ValidatorException(ex); |
---|
106 | } |
---|
107 | |
---|
108 | } |
---|
109 | |
---|
110 | /** |
---|
111 | * Convenience method to build a XSLT transformer from a resource. |
---|
112 | * |
---|
113 | * @param uri The location of the resource |
---|
114 | * @return An executable XSLT |
---|
115 | * @throws Exception |
---|
116 | */ |
---|
117 | static XsltExecutable buildTransformer(InputStream stream) throws ValidatorException { |
---|
118 | try { |
---|
119 | XdmNode xslDoc = Saxon.buildDocument(new javax.xml.transform.stream.StreamSource(stream)); |
---|
120 | return Saxon.buildTransformer(xslDoc); |
---|
121 | } catch (SaxonApiException ex) { |
---|
122 | throw new ValidatorException(ex); |
---|
123 | } |
---|
124 | |
---|
125 | } |
---|
126 | |
---|
127 | /** |
---|
128 | * Returns the CMD XSD schema, and loads it just-in-time. |
---|
129 | * |
---|
130 | * @return An in-memory representation of the grammar |
---|
131 | * @throws Exception |
---|
132 | */ |
---|
133 | private synchronized Schema getSchema() throws ValidatorException, IOException { |
---|
134 | if (cmdSchema == null) { |
---|
135 | SchemaFactory factory = SchemaFactory.newInstance(XMLConstants.W3C_XML_SCHEMA_NS_URI); |
---|
136 | |
---|
137 | // Load the CMD XSD. |
---|
138 | Source schemaFile = new StreamSource(cmdSchemaUri.openStream()); |
---|
139 | try { |
---|
140 | cmdSchema = factory.newSchema(schemaFile); |
---|
141 | } catch (SAXException ex) { |
---|
142 | throw new ValidatorException(ex); |
---|
143 | } |
---|
144 | } |
---|
145 | return cmdSchema; |
---|
146 | } |
---|
147 | |
---|
148 | /** |
---|
149 | * Validation of a loaded CMD profile/component against the XSD schema. |
---|
150 | * |
---|
151 | * Unfortunately we can't use the Saxon XSD validator as that is limited to a commercial version of Saxon. |
---|
152 | * |
---|
153 | * @param src The loaded CMD profile/component |
---|
154 | * @return Is the CMD profile/component valid or not? |
---|
155 | * @throws Exception |
---|
156 | */ |
---|
157 | public boolean validateXSD(XdmNode src) throws ValidatorException, IOException { |
---|
158 | try { |
---|
159 | Document doc = DocumentBuilderFactory.newInstance().newDocumentBuilder().newDocument(); |
---|
160 | DOMDestination dst = new DOMDestination(doc); |
---|
161 | Saxon.getProcessor().writeXdmValue(src, dst); |
---|
162 | |
---|
163 | try { |
---|
164 | // Create a Validator object, which can be used to validate |
---|
165 | // an instance document. |
---|
166 | javax.xml.validation.Validator validator = getSchema().newValidator(); |
---|
167 | |
---|
168 | // Validate the DOM tree. |
---|
169 | validator.validate(new DOMSource(doc)); |
---|
170 | |
---|
171 | } catch (SAXException e) { |
---|
172 | Message msg = new Message(); |
---|
173 | msg.error = true; |
---|
174 | msg.text = e.getMessage(); |
---|
175 | msgList.add(msg); |
---|
176 | return false; |
---|
177 | } |
---|
178 | } catch (SaxonApiException ex) { |
---|
179 | throw new ValidatorException(ex); |
---|
180 | } catch (ParserConfigurationException ex) { |
---|
181 | throw new ValidatorException(ex); |
---|
182 | } |
---|
183 | |
---|
184 | return true; |
---|
185 | } |
---|
186 | |
---|
187 | /** |
---|
188 | * Returns the CMD Schematron XSLT, and loads it just-in-time. |
---|
189 | * |
---|
190 | * @return The compiled Schematron XSLT |
---|
191 | * @throws Exception |
---|
192 | */ |
---|
193 | private synchronized XsltExecutable getSchematron() throws ValidatorException, IOException { |
---|
194 | if (cmdSchematron == null) { |
---|
195 | try { |
---|
196 | // Load the schema |
---|
197 | XdmNode schema = Saxon.buildDocument(new javax.xml.transform.stream.StreamSource(cmdSchemaUri.openStream())); |
---|
198 | // Load the Schematron XSL to extract the Schematron rules; |
---|
199 | XsltTransformer extractSchXsl = buildTransformer(Validator.class.getResource("/schematron/ExtractSchFromXSD-2.xsl")).load(); |
---|
200 | // Load the Schematron XSLs to 'compile' Schematron rules; |
---|
201 | XsltTransformer includeSchXsl = buildTransformer(Validator.class.getResource("/schematron/iso_dsdl_include.xsl")).load(); |
---|
202 | XsltTransformer expandSchXsl = buildTransformer(Validator.class.getResource("/schematron/iso_abstract_expand.xsl")).load(); |
---|
203 | XsltTransformer compileSchXsl = buildTransformer(Validator.class.getResource("/schematron/iso_svrl_for_xslt2.xsl")).load(); |
---|
204 | // Setup the pipeline |
---|
205 | XdmDestination destination = new XdmDestination(); |
---|
206 | extractSchXsl.setSource(schema.asSource()); |
---|
207 | extractSchXsl.setDestination(includeSchXsl); |
---|
208 | includeSchXsl.setDestination(expandSchXsl); |
---|
209 | expandSchXsl.setDestination(compileSchXsl); |
---|
210 | compileSchXsl.setDestination(destination); |
---|
211 | // Extract the Schematron rules from the schema |
---|
212 | extractSchXsl.transform(); |
---|
213 | // Compile the Schematron rules XSL |
---|
214 | cmdSchematron = Saxon.buildTransformer(destination.getXdmNode()); |
---|
215 | } catch (SaxonApiException ex) { |
---|
216 | throw new ValidatorException(ex); |
---|
217 | } |
---|
218 | } |
---|
219 | return cmdSchematron; |
---|
220 | } |
---|
221 | |
---|
222 | /** |
---|
223 | * Validation of a loaded CMD profile/component against the Schematron XSLT |
---|
224 | * |
---|
225 | * @param src The loaded CMD profile/component |
---|
226 | * @return Is the CMD profile/component valid or not? |
---|
227 | * @throws Exception |
---|
228 | */ |
---|
229 | public boolean validateSchematron(XdmNode src) throws ValidatorException, IOException { |
---|
230 | try { |
---|
231 | XsltTransformer schematronXsl = getSchematron().load(); |
---|
232 | schematronXsl.setSource(src.asSource()); |
---|
233 | XdmDestination destination = new XdmDestination(); |
---|
234 | schematronXsl.setDestination(destination); |
---|
235 | schematronXsl.transform(); |
---|
236 | |
---|
237 | validationReport = destination.getXdmNode(); |
---|
238 | |
---|
239 | Saxon.declareXPathNamespace("svrl", "http://purl.oclc.org/dsdl/svrl"); |
---|
240 | return ((net.sf.saxon.value.BooleanValue) Saxon.evaluateXPath(validationReport, "empty(//svrl:failed-assert[(preceding-sibling::svrl:fired-rule)[last()][empty(@role) or @role!='warning']])").evaluateSingle().getUnderlyingValue()).getBooleanValue(); |
---|
241 | } catch (SaxonApiException ex) { |
---|
242 | throw new ValidatorException(ex); |
---|
243 | } |
---|
244 | } |
---|
245 | |
---|
246 | /** |
---|
247 | * Validation of a loaded CMD profile/component against both the XSD and the Schematron XSLT |
---|
248 | * |
---|
249 | * After validation any messages can be accessed using the {@link getMessages()} method. |
---|
250 | * Notice that even if a CMD profile/component is valid there might be warning messages. |
---|
251 | * |
---|
252 | * @param prof The CMD profile/component |
---|
253 | * @return Is the CMD profile/component valid or not? |
---|
254 | * @throws Exception |
---|
255 | */ |
---|
256 | public boolean validateProfile(Source prof) throws ValidatorException, IOException { |
---|
257 | // Initalize |
---|
258 | msgList = new java.util.ArrayList<Message>(); |
---|
259 | validationReport = null; |
---|
260 | |
---|
261 | try { |
---|
262 | // load the document |
---|
263 | XdmNode doc = Saxon.buildDocument(prof); |
---|
264 | |
---|
265 | // step 1: validate against XML Schema |
---|
266 | if (!this.validateXSD(doc)) { |
---|
267 | return false; |
---|
268 | } |
---|
269 | |
---|
270 | // step 2: validate Schematron rules |
---|
271 | return validateSchematron(doc); |
---|
272 | } catch (SaxonApiException ex) { |
---|
273 | throw new ValidatorException(ex); |
---|
274 | } |
---|
275 | |
---|
276 | } |
---|
277 | |
---|
278 | /** |
---|
279 | * Get the list of messages accumulated in the last validation run. |
---|
280 | * |
---|
281 | * @return The list of messages |
---|
282 | * @throws Exception |
---|
283 | */ |
---|
284 | public List<Message> getMessages() throws ValidatorException { |
---|
285 | if (validationReport != null) { |
---|
286 | try { |
---|
287 | for (XdmItem assertion : Saxon.evaluateXPath(validationReport, "//svrl:failed-assert")) { |
---|
288 | Message msg = new Message(); |
---|
289 | msg.context = Saxon.evaluateXPath(assertion, "(preceding-sibling::svrl:fired-rule)[last()]/@context").evaluateSingle().getStringValue(); |
---|
290 | msg.test = ((XdmNode) assertion).getAttributeValue(new QName("test")); |
---|
291 | msg.location = ((XdmNode) assertion).getAttributeValue(new QName("location")); |
---|
292 | msg.error = !((net.sf.saxon.value.BooleanValue) Saxon.evaluateXPath(assertion, "(preceding-sibling::svrl:fired-rule)[last()]/@role='warning'").evaluateSingle().getUnderlyingValue()).getBooleanValue(); |
---|
293 | msg.text = assertion.getStringValue(); |
---|
294 | msgList.add(msg); |
---|
295 | } |
---|
296 | validationReport = null; |
---|
297 | } catch (SaxonApiException ex) { |
---|
298 | throw new ValidatorException(ex); |
---|
299 | } |
---|
300 | } |
---|
301 | return msgList; |
---|
302 | } |
---|
303 | |
---|
304 | /** |
---|
305 | * Print the list of messages accumulated in the last validation run. |
---|
306 | * |
---|
307 | * @param out |
---|
308 | * @throws Exception |
---|
309 | */ |
---|
310 | public void printMessages(java.io.PrintStream out) throws Exception { |
---|
311 | for (Message msg : getMessages()) { |
---|
312 | out.println("" + (msg.isError() ? "ERROR" : "WARNING") + (msg.getLocation() != null ? " at " + msg.getLocation() : "")); |
---|
313 | out.println(" " + msg.getText()); |
---|
314 | } |
---|
315 | } |
---|
316 | |
---|
317 | /** |
---|
318 | * @param args One or more CMD profile/component files to validate. |
---|
319 | */ |
---|
320 | public static void main(String[] args) { |
---|
321 | try { |
---|
322 | |
---|
323 | URL schemaURL = null; |
---|
324 | |
---|
325 | int startArg = 0; |
---|
326 | if (args.length > 0) { |
---|
327 | if ("-s".equals(args[0].trim())) { |
---|
328 | if (args.length > 1) { |
---|
329 | String schemaArg = args[1]; |
---|
330 | System.err.println("Using schema URL" + schemaArg); |
---|
331 | schemaURL = new URL(schemaArg); |
---|
332 | startArg = 2; |
---|
333 | } else { |
---|
334 | printUsage(args); |
---|
335 | return; |
---|
336 | } |
---|
337 | } |
---|
338 | } else { |
---|
339 | printUsage(args); |
---|
340 | return; |
---|
341 | } |
---|
342 | |
---|
343 | if (schemaURL == null) { |
---|
344 | schemaURL = new URL(CMD_SCHEMA_URL); |
---|
345 | } |
---|
346 | |
---|
347 | final Validator cmdValidator = new Validator(schemaURL); |
---|
348 | |
---|
349 | for (int i = startArg; i < args.length; i++) { |
---|
350 | String f = args[i]; |
---|
351 | System.out.print("CMD validate[" + f + "] "); |
---|
352 | try { |
---|
353 | Source src = new javax.xml.transform.stream.StreamSource(new java.io.File(f)); |
---|
354 | if (cmdValidator.validateProfile(src)) { |
---|
355 | System.out.println("valid"); |
---|
356 | } else { |
---|
357 | System.out.println("invalid"); |
---|
358 | } |
---|
359 | cmdValidator.printMessages(System.out); |
---|
360 | } catch (Exception e) { |
---|
361 | System.err.println("failed:"); |
---|
362 | e.printStackTrace(System.out); |
---|
363 | } |
---|
364 | } |
---|
365 | } catch (MalformedURLException e) { |
---|
366 | System.err.println("failed:"); |
---|
367 | e.printStackTrace(System.out); |
---|
368 | } |
---|
369 | } |
---|
370 | |
---|
371 | private static void printUsage(String[] args) { |
---|
372 | System.err.println("Arguments: [-s schemafileURL] files..."); |
---|
373 | } |
---|
374 | |
---|
375 | /** |
---|
376 | * Public inner class to represent validation messages. |
---|
377 | */ |
---|
378 | public final static class Message { |
---|
379 | |
---|
380 | /** |
---|
381 | * Is the message and error or an warning? |
---|
382 | */ |
---|
383 | boolean error = false; |
---|
384 | /** |
---|
385 | * The context of the message (might be null). |
---|
386 | */ |
---|
387 | String context = null; |
---|
388 | /** |
---|
389 | * The test that triggered the message (might be null). |
---|
390 | */ |
---|
391 | String test = null; |
---|
392 | /** |
---|
393 | * The location that triggered the test (might be null). |
---|
394 | */ |
---|
395 | String location = null; |
---|
396 | /** |
---|
397 | * The actual message. |
---|
398 | */ |
---|
399 | String text = null; |
---|
400 | |
---|
401 | /** |
---|
402 | * @return the error |
---|
403 | */ |
---|
404 | public boolean isError() { |
---|
405 | return error; |
---|
406 | } |
---|
407 | |
---|
408 | /** |
---|
409 | * @return the context |
---|
410 | */ |
---|
411 | public String getContext() { |
---|
412 | return context; |
---|
413 | } |
---|
414 | |
---|
415 | /** |
---|
416 | * @return the test |
---|
417 | */ |
---|
418 | public String getTest() { |
---|
419 | return test; |
---|
420 | } |
---|
421 | |
---|
422 | /** |
---|
423 | * @return the location |
---|
424 | */ |
---|
425 | public String getLocation() { |
---|
426 | return location; |
---|
427 | } |
---|
428 | |
---|
429 | /** |
---|
430 | * @return the text |
---|
431 | */ |
---|
432 | public String getText() { |
---|
433 | return text; |
---|
434 | } |
---|
435 | } |
---|
436 | } |
---|