Révision 2649
| tmp/org.txm.core/src/java/org/txm/xml/LocalNamesHookActivator.java (revision 2649) | ||
|---|---|---|
| 1 |
package org.txm.xml; |
|
| 2 |
|
|
| 3 |
import java.util.Arrays; |
|
| 4 |
import java.util.HashSet; |
|
| 5 |
import java.util.List; |
|
| 6 |
|
|
| 7 |
/** |
|
| 8 |
* Simple Activator activated when the localname matches the localname constructor parameter |
|
| 9 |
* |
|
| 10 |
* @author mdecorde |
|
| 11 |
* |
|
| 12 |
*/ |
|
| 13 |
public class LocalNamesHookActivator extends HookActivator {
|
|
| 14 |
|
|
| 15 |
HashSet<String> localnames; |
|
| 16 |
|
|
| 17 |
/** |
|
| 18 |
* @param localnames the element localnames to match |
|
| 19 |
*/ |
|
| 20 |
public LocalNamesHookActivator(Hook hook, List<String> localnames) {
|
|
| 21 |
super(); |
|
| 22 |
this.localnames = new HashSet<>(); |
|
| 23 |
this.localnames.addAll(localnames); |
|
| 24 |
} |
|
| 25 |
|
|
| 26 |
/** |
|
| 27 |
* @param localnames the element localnames to match |
|
| 28 |
*/ |
|
| 29 |
public LocalNamesHookActivator(Hook hook, String... localnames) {
|
|
| 30 |
this(hook, Arrays.asList(localnames)); |
|
| 31 |
} |
|
| 32 |
|
|
| 33 |
@Override |
|
| 34 |
public boolean mustActivate() {
|
|
| 35 |
if (localnames == null) return false; |
|
| 36 |
|
|
| 37 |
return localnames.contains(hook.getLocalName()); |
|
| 38 |
} |
|
| 39 |
|
|
| 40 |
@Override |
|
| 41 |
public void hookWasActivated() {
|
|
| 42 |
// nothing to do |
|
| 43 |
} |
|
| 44 |
} |
|
| 0 | 45 | |
| tmp/org.txm.core/src/java/org/txm/xml/XMLParser.java (revision 2649) | ||
|---|---|---|
| 1 |
package org.txm.xml; |
|
| 2 |
|
|
| 3 |
import java.io.BufferedOutputStream; |
|
| 4 |
import java.io.File; |
|
| 5 |
import java.io.FileOutputStream; |
|
| 6 |
import java.io.IOException; |
|
| 7 |
import java.io.InputStream; |
|
| 8 |
import java.net.URL; |
|
| 9 |
import java.util.Arrays; |
|
| 10 |
import java.util.HashMap; |
|
| 11 |
import java.util.LinkedHashMap; |
|
| 12 |
import java.util.Stack; |
|
| 13 |
|
|
| 14 |
import javax.xml.stream.XMLInputFactory; |
|
| 15 |
import javax.xml.stream.XMLOutputFactory; |
|
| 16 |
import javax.xml.stream.XMLStreamConstants; |
|
| 17 |
import javax.xml.stream.XMLStreamException; |
|
| 18 |
import javax.xml.stream.XMLStreamReader; |
|
| 19 |
import javax.xml.stream.XMLStreamWriter; |
|
| 20 |
|
|
| 21 |
import org.txm.importer.PersonalNamespaceContext; |
|
| 22 |
|
|
| 23 |
/** |
|
| 24 |
* parse a XML files using Stax and allow to set somes hooks activated with their HookActivator |
|
| 25 |
* |
|
| 26 |
* @author mdecorde |
|
| 27 |
* |
|
| 28 |
*/ |
|
| 29 |
public class XMLParser {
|
|
| 30 |
|
|
| 31 |
/** The input */ |
|
| 32 |
protected URL inputurl; |
|
| 33 |
|
|
| 34 |
protected InputStream inputData; |
|
| 35 |
|
|
| 36 |
protected XMLInputFactory factory; |
|
| 37 |
|
|
| 38 |
protected XMLStreamReader parser; |
|
| 39 |
|
|
| 40 |
public static String TXMNS = "http://textometrie.org/1.0"; |
|
| 41 |
|
|
| 42 |
public static String TXM = "txm"; |
|
| 43 |
|
|
| 44 |
public static String TEINS = "http://www.tei-c.org/ns/1.0"; |
|
| 45 |
|
|
| 46 |
public static String TEI = "tei"; |
|
| 47 |
|
|
| 48 |
protected static PersonalNamespaceContext Nscontext = new PersonalNamespaceContext(); |
|
| 49 |
|
|
| 50 |
protected StringBuilder currentXPath = new StringBuilder("");
|
|
| 51 |
|
|
| 52 |
protected String localname; |
|
| 53 |
|
|
| 54 |
int processingXInclude = 0; |
|
| 55 |
|
|
| 56 |
LinkedHashMap<String, Hook> hooks = new LinkedHashMap<>(); |
|
| 57 |
|
|
| 58 |
private Hook activeHook = null; |
|
| 59 |
|
|
| 60 |
public XMLParser(File infile) throws IOException, XMLStreamException {
|
|
| 61 |
this(infile.toURI().toURL()); |
|
| 62 |
} |
|
| 63 |
|
|
| 64 |
public XMLParser(URL inputurl) throws IOException, XMLStreamException {
|
|
| 65 |
this.inputurl = inputurl; |
|
| 66 |
this.inputData = inputurl.openStream(); |
|
| 67 |
this.factory = XMLInputFactory.newInstance(); |
|
| 68 |
this.parser = factory.createXMLStreamReader(inputData); |
|
| 69 |
} |
|
| 70 |
|
|
| 71 |
/** |
|
| 72 |
* Helper method to get an attribute value by its name |
|
| 73 |
* |
|
| 74 |
* @param name the attribute name |
|
| 75 |
* @return the value if any |
|
| 76 |
*/ |
|
| 77 |
public String getParserAttributeValue(String name) {
|
|
| 78 |
if (name == null) return null; |
|
| 79 |
|
|
| 80 |
int c = parser.getAttributeCount(); |
|
| 81 |
for (int i = 0; i < c; i++) {
|
|
| 82 |
if (name.equals(parser.getAttributeLocalName(i))) {
|
|
| 83 |
return parser.getAttributeValue(i); |
|
| 84 |
} |
|
| 85 |
} |
|
| 86 |
|
|
| 87 |
return null; |
|
| 88 |
} |
|
| 89 |
|
|
| 90 |
protected void before() {
|
|
| 91 |
|
|
| 92 |
} |
|
| 93 |
|
|
| 94 |
protected void after() throws XMLStreamException, IOException {
|
|
| 95 |
if (factory != null) {
|
|
| 96 |
factory = null; |
|
| 97 |
|
|
| 98 |
if (parser != null) {
|
|
| 99 |
parser.close(); |
|
| 100 |
} |
|
| 101 |
if (inputData != null) {
|
|
| 102 |
inputData.close(); |
|
| 103 |
} |
|
| 104 |
parser = null; |
|
| 105 |
|
|
| 106 |
if (parser != null) {
|
|
| 107 |
try {
|
|
| 108 |
parser.close(); |
|
| 109 |
} |
|
| 110 |
catch (Exception e) {
|
|
| 111 |
System.out.println("parser excep: " + e);
|
|
| 112 |
} |
|
| 113 |
} |
|
| 114 |
|
|
| 115 |
if (inputData != null) {
|
|
| 116 |
try {
|
|
| 117 |
inputData.close(); |
|
| 118 |
} |
|
| 119 |
catch (Exception e) {
|
|
| 120 |
System.out.println("inputData excep: " + e);
|
|
| 121 |
} |
|
| 122 |
} |
|
| 123 |
|
|
| 124 |
|
|
| 125 |
} |
|
| 126 |
} |
|
| 127 |
|
|
| 128 |
public final static String SLASH = "/"; |
|
| 129 |
|
|
| 130 |
public final boolean process() throws XMLStreamException, IOException {
|
|
| 131 |
|
|
| 132 |
// if (processingXInclude == 0) {
|
|
| 133 |
before(); // if you need to do something before reading the xml |
|
| 134 |
// } |
|
| 135 |
try {
|
|
| 136 |
for (int event = parser.next(); event != XMLStreamConstants.END_DOCUMENT; event = parser.next()) {
|
|
| 137 |
|
|
| 138 |
// preparing process |
|
| 139 |
if (event == XMLStreamConstants.START_ELEMENT) {
|
|
| 140 |
localname = parser.getLocalName(); |
|
| 141 |
currentXPath.append(SLASH); |
|
| 142 |
currentXPath.append(localname); |
|
| 143 |
buildCurrentAttributes(); // for easier access later |
|
| 144 |
} |
|
| 145 |
else if (event == XMLStreamConstants.END_ELEMENT) {
|
|
| 146 |
localname = parser.getLocalName(); |
|
| 147 |
} |
|
| 148 |
|
|
| 149 |
if (activeHook != null) {
|
|
| 150 |
if (activeHook.mustDeactivate()) {
|
|
| 151 |
activeHook.deactivate(); |
|
| 152 |
activeHook = null; |
|
| 153 |
} |
|
| 154 |
else {
|
|
| 155 |
activeHook.processParserEvent(event); |
|
| 156 |
} |
|
| 157 |
} |
|
| 158 |
|
|
| 159 |
if (activeHook == null && event == XMLStreamConstants.START_ELEMENT) { // only activate when starting an element (for now, thus we cannot process only text nodes)
|
|
| 160 |
for (Hook hook : hooks.values()) {
|
|
| 161 |
if (hook.mustActivate()) {
|
|
| 162 |
if (hook.activate()) {
|
|
| 163 |
activeHook = hook; |
|
| 164 |
|
|
| 165 |
hook.processParserEvent(event); |
|
| 166 |
break; |
|
| 167 |
} |
|
| 168 |
} |
|
| 169 |
} |
|
| 170 |
} |
|
| 171 |
|
|
| 172 |
if (activeHook == null) {
|
|
| 173 |
this.processParserEvent(event); |
|
| 174 |
} |
|
| 175 |
|
|
| 176 |
// post-preparing process |
|
| 177 |
if (event == XMLStreamConstants.END_ELEMENT) {
|
|
| 178 |
currentXPath.setLength(currentXPath.length() - localname.length() - 1); |
|
| 179 |
popCurrentAttributes(); |
|
| 180 |
} |
|
| 181 |
} |
|
| 182 |
} |
|
| 183 |
catch (Exception e) {
|
|
| 184 |
System.out.println("Unexpected error while parsing file " + inputurl + " : " + e);
|
|
| 185 |
System.out.println("Location line: " + parser.getLocation().getLineNumber() + " character: " + parser.getLocation().getColumnNumber());
|
|
| 186 |
org.txm.utils.logger.Log.printStackTrace(e); |
|
| 187 |
// e.printStackTrace(); |
|
| 188 |
return false; |
|
| 189 |
} |
|
| 190 |
finally {
|
|
| 191 |
after(); // if you need to do something before closing the parser(); |
|
| 192 |
} |
|
| 193 |
|
|
| 194 |
return true; |
|
| 195 |
} |
|
| 196 |
|
|
| 197 |
protected final void processParserEvent(int event) throws XMLStreamException, IOException {
|
|
| 198 |
switch (event) {
|
|
| 199 |
case XMLStreamConstants.NAMESPACE: |
|
| 200 |
processNamespace(); |
|
| 201 |
break; |
|
| 202 |
case XMLStreamConstants.START_ELEMENT: |
|
| 203 |
processStartElement(); |
|
| 204 |
break; |
|
| 205 |
case XMLStreamConstants.CHARACTERS: |
|
| 206 |
processCharacters(); |
|
| 207 |
break; |
|
| 208 |
case XMLStreamConstants.END_ELEMENT: |
|
| 209 |
processEndElement(); |
|
| 210 |
break; |
|
| 211 |
case XMLStreamConstants.PROCESSING_INSTRUCTION: |
|
| 212 |
processProcessingInstruction(); |
|
| 213 |
break; |
|
| 214 |
case XMLStreamConstants.DTD: |
|
| 215 |
processDTD(); |
|
| 216 |
break; |
|
| 217 |
case XMLStreamConstants.CDATA: |
|
| 218 |
processCDATA(); |
|
| 219 |
break; |
|
| 220 |
case XMLStreamConstants.COMMENT: |
|
| 221 |
processComment(); |
|
| 222 |
break; |
|
| 223 |
case XMLStreamConstants.END_DOCUMENT: |
|
| 224 |
processEndDocument(); |
|
| 225 |
break; |
|
| 226 |
case XMLStreamConstants.ENTITY_REFERENCE: |
|
| 227 |
processEntityReference(); |
|
| 228 |
break; |
|
| 229 |
} |
|
| 230 |
} |
|
| 231 |
|
|
| 232 |
/** |
|
| 233 |
* The start element has already been written |
|
| 234 |
* |
|
| 235 |
* @param tagname |
|
| 236 |
* @throws XMLStreamException |
|
| 237 |
* @throws IOException |
|
| 238 |
*/ |
|
| 239 |
public void goToEnd(String tagname) throws XMLStreamException, IOException {
|
|
| 240 |
int elements = 1; |
|
| 241 |
for (int event = parser.next(); event != XMLStreamConstants.END_DOCUMENT; event = parser.next()) {
|
|
| 242 |
// System.out.println("event "+event
|
|
| 243 |
switch (event) {
|
|
| 244 |
case XMLStreamConstants.NAMESPACE: |
|
| 245 |
processNamespace(); |
|
| 246 |
break; |
|
| 247 |
case XMLStreamConstants.START_ELEMENT: |
|
| 248 |
elements++; |
|
| 249 |
localname = parser.getLocalName(); |
|
| 250 |
currentXPath.append(SLASH); |
|
| 251 |
currentXPath.append(localname); |
|
| 252 |
buildCurrentAttributes(); |
|
| 253 |
|
|
| 254 |
processStartElement(); |
|
| 255 |
break; |
|
| 256 |
case XMLStreamConstants.CHARACTERS: |
|
| 257 |
processCharacters(); |
|
| 258 |
break; |
|
| 259 |
case XMLStreamConstants.PROCESSING_INSTRUCTION: |
|
| 260 |
processProcessingInstruction(); |
|
| 261 |
break; |
|
| 262 |
case XMLStreamConstants.DTD: |
|
| 263 |
processDTD(); |
|
| 264 |
break; |
|
| 265 |
case XMLStreamConstants.CDATA: |
|
| 266 |
processCDATA(); |
|
| 267 |
break; |
|
| 268 |
case XMLStreamConstants.COMMENT: |
|
| 269 |
processComment(); |
|
| 270 |
break; |
|
| 271 |
case XMLStreamConstants.END_ELEMENT: |
|
| 272 |
elements--; |
|
| 273 |
localname = parser.getLocalName(); |
|
| 274 |
|
|
| 275 |
currentXPath.setLength(currentXPath.length() - localname.length() - 1); |
|
| 276 |
|
|
| 277 |
if (elements == 0 && localname == tagname) |
|
| 278 |
return; |
|
| 279 |
break; |
|
| 280 |
case XMLStreamConstants.END_DOCUMENT: |
|
| 281 |
processEndDocument(); |
|
| 282 |
break; |
|
| 283 |
case XMLStreamConstants.ENTITY_REFERENCE: |
|
| 284 |
processEntityReference(); |
|
| 285 |
break; |
|
| 286 |
} |
|
| 287 |
} |
|
| 288 |
|
|
| 289 |
} |
|
| 290 |
|
|
| 291 |
public final String getLocation() {
|
|
| 292 |
if (parser != null) {
|
|
| 293 |
return "Line: " + parser.getLocation().getLineNumber() + " Col: " + parser.getLocation().getColumnNumber(); |
|
| 294 |
} |
|
| 295 |
return null; |
|
| 296 |
} |
|
| 297 |
|
|
| 298 |
private Stack<LinkedHashMap<String, String>> attributesStack = new Stack<>(); |
|
| 299 |
|
|
| 300 |
private final HashMap<String, String> popCurrentAttributes() {
|
|
| 301 |
return attributesStack.pop(); |
|
| 302 |
} |
|
| 303 |
|
|
| 304 |
protected final void buildCurrentAttributes() throws XMLStreamException {
|
|
| 305 |
LinkedHashMap<String, String> currentAttributes = new LinkedHashMap<>(); // keep order |
|
| 306 |
for (int i = 0; i < parser.getAttributeCount(); i++) {
|
|
| 307 |
currentAttributes.put(parser.getAttributeLocalName(i), parser.getAttributeValue(i)); |
|
| 308 |
} |
|
| 309 |
attributesStack.push(currentAttributes); |
|
| 310 |
} |
|
| 311 |
|
|
| 312 |
public final LinkedHashMap<String, String> getCurrentAttributes() {
|
|
| 313 |
return attributesStack.lastElement(); |
|
| 314 |
} |
|
| 315 |
|
|
| 316 |
protected void processNamespace() throws XMLStreamException {}
|
|
| 317 |
|
|
| 318 |
protected void processStartElement() throws XMLStreamException, IOException {}
|
|
| 319 |
|
|
| 320 |
protected void processCharacters() throws XMLStreamException {}
|
|
| 321 |
|
|
| 322 |
protected void processProcessingInstruction() throws XMLStreamException {}
|
|
| 323 |
|
|
| 324 |
protected void processDTD() throws XMLStreamException {}
|
|
| 325 |
|
|
| 326 |
protected void processCDATA() throws XMLStreamException {}
|
|
| 327 |
|
|
| 328 |
protected void processComment() throws XMLStreamException {}
|
|
| 329 |
|
|
| 330 |
protected void processEndElement() throws XMLStreamException {}
|
|
| 331 |
|
|
| 332 |
protected void processEndDocument() throws XMLStreamException {}
|
|
| 333 |
|
|
| 334 |
protected void processEntityReference() throws XMLStreamException {}
|
|
| 335 |
|
|
| 336 |
public String getCurrentPath() {
|
|
| 337 |
return currentXPath.toString(); |
|
| 338 |
} |
|
| 339 |
|
|
| 340 |
public static void main(String[] args) {
|
|
| 341 |
try {
|
|
| 342 |
File input = new File(System.getProperty("user.home"), "xml/tdm80j/tdm80j.xml");
|
|
| 343 |
if (!(input.exists() && input.canRead())) {
|
|
| 344 |
System.out.println("cannot found $input");
|
|
| 345 |
return; |
|
| 346 |
} |
|
| 347 |
XMLParser builder; |
|
| 348 |
|
|
| 349 |
builder = new XMLParser(input.toURI().toURL()); |
|
| 350 |
|
|
| 351 |
// IdentityHook hook1 = new IdentityHook("hook1", new LocalNameHookActivator(null, "p"), builder) {
|
|
| 352 |
// |
|
| 353 |
// @Override |
|
| 354 |
// public boolean _activate() {
|
|
| 355 |
// System.out.println("ACTIVATING " + name + " AT " + getLocation());
|
|
| 356 |
// return true; |
|
| 357 |
// } |
|
| 358 |
// |
|
| 359 |
// @Override |
|
| 360 |
// public boolean deactivate() {
|
|
| 361 |
// System.out.println("DEACTIVATING " + name + " AT " + getLocation());
|
|
| 362 |
// return true; |
|
| 363 |
// } |
|
| 364 |
// |
|
| 365 |
// @Override |
|
| 366 |
// public void processStartElement() throws XMLStreamException, IOException {
|
|
| 367 |
// super.processStartElement(); |
|
| 368 |
// parentParser.writer.writeAttribute("hook", name);
|
|
| 369 |
// } |
|
| 370 |
// }; |
|
| 371 |
XPathHookActivator xpathActivator = new XPathHookActivator(null, "//p"); |
|
| 372 |
XPathsHookActivator xpathsActivator = new XPathsHookActivator(null, Arrays.asList("//p", "//p"));
|
|
| 373 |
// IdentityHook hook2 = new IdentityHook("hook2", xpathsActivator, builder) {
|
|
| 374 |
// |
|
| 375 |
// @Override |
|
| 376 |
// public boolean _activate() {
|
|
| 377 |
// System.out.println("ACTIVATING " + name + " AT " + getLocation());
|
|
| 378 |
// return true; |
|
| 379 |
// } |
|
| 380 |
// |
|
| 381 |
// @Override |
|
| 382 |
// public boolean deactivate() {
|
|
| 383 |
// System.out.println("DEACTIVATING " + name + " AT " + getLocation());
|
|
| 384 |
// return true; |
|
| 385 |
// } |
|
| 386 |
// |
|
| 387 |
// @Override |
|
| 388 |
// public void processStartElement() throws XMLStreamException, IOException {
|
|
| 389 |
// super.processStartElement(); |
|
| 390 |
// parentParser.writer.writeAttribute("hook", name);
|
|
| 391 |
// } |
|
| 392 |
// }; |
|
| 393 |
|
|
| 394 |
long time = System.currentTimeMillis(); |
|
| 395 |
if (builder.process()) {
|
|
| 396 |
System.out.println("Time=" + (System.currentTimeMillis() - time));
|
|
| 397 |
System.out.println("XPaths activator has done working ? " + xpathsActivator.hasAllXpathsBeenProcessed());
|
|
| 398 |
} |
|
| 399 |
else {
|
|
| 400 |
System.out.println("failure !");
|
|
| 401 |
} |
|
| 402 |
} |
|
| 403 |
catch (Exception e) {
|
|
| 404 |
e.printStackTrace(); |
|
| 405 |
} |
|
| 406 |
} |
|
| 407 |
|
|
| 408 |
/** |
|
| 409 |
* |
|
| 410 |
* @return the actual hooks hash |
|
| 411 |
*/ |
|
| 412 |
public LinkedHashMap<String, Hook> getHooks() {
|
|
| 413 |
return hooks; |
|
| 414 |
} |
|
| 415 |
|
|
| 416 |
public void addHook(String name, Hook hook) {
|
|
| 417 |
hooks.put(name, hook); |
|
| 418 |
} |
|
| 419 |
} |
|
| 0 | 420 | |
| tmp/org.txm.core/src/java/org/txm/xml/Hook.java (revision 2649) | ||
|---|---|---|
| 4 | 4 |
|
| 5 | 5 |
import javax.xml.stream.XMLStreamConstants; |
| 6 | 6 |
import javax.xml.stream.XMLStreamException; |
| 7 |
import javax.xml.stream.XMLStreamReader; |
|
| 7 | 8 |
|
| 8 | 9 |
import org.txm.utils.logger.Log; |
| 9 | 10 |
|
| 10 |
public abstract class Hook {
|
|
| 11 |
public abstract class Hook<T extends XMLParser> {
|
|
| 11 | 12 |
|
| 12 |
protected XMLProcessor parentParser = null;
|
|
| 13 |
protected T parentParser = null;
|
|
| 13 | 14 |
|
| 14 | 15 |
protected HookActivator activator = null; |
| 15 | 16 |
|
| ... | ... | |
| 17 | 18 |
|
| 18 | 19 |
private int elements = 0; |
| 19 | 20 |
|
| 21 |
protected XMLStreamReader parser; |
|
| 22 |
|
|
| 20 | 23 |
/** |
| 21 | 24 |
* create a hook parser. When a hook parser is activated, the processXYZ methods are used instead of the parent processXTZ methods. |
| 22 | 25 |
* |
| ... | ... | |
| 28 | 31 |
* @throws IOException |
| 29 | 32 |
* @throws XMLStreamException |
| 30 | 33 |
*/ |
| 31 |
public Hook(String name, HookActivator activator, XMLProcessor parentParser) throws IOException, XMLStreamException {
|
|
| 34 |
public Hook(String name, HookActivator activator, T parentParser) throws IOException, XMLStreamException {
|
|
| 32 | 35 |
this.parentParser = parentParser; |
| 33 | 36 |
this.name = name; |
| 34 | 37 |
setActivator(activator); // set the activator |
| 35 | 38 |
parentParser.addHook(name, this); // register the hook to the XMLProcessor |
| 39 |
this.parser = parentParser.parser; |
|
| 36 | 40 |
} |
| 37 | 41 |
|
| 38 | 42 |
public void setActivator(HookActivator activator) {
|
| ... | ... | |
| 43 | 47 |
} |
| 44 | 48 |
|
| 45 | 49 |
/** |
| 50 |
* |
|
| 51 |
* @return the hook's name |
|
| 52 |
*/ |
|
| 53 |
public String getName() {
|
|
| 54 |
return name; |
|
| 55 |
} |
|
| 56 |
|
|
| 57 |
/** |
|
| 46 | 58 |
* called when the parser is activated by its parent |
| 47 | 59 |
* |
| 48 | 60 |
* @return true if correctly activated |
| tmp/org.txm.core/src/java/org/txm/xml/UDPipeAnnotate.java (revision 2649) | ||
|---|---|---|
| 1 |
package org.txm.xml; |
|
| 2 |
|
|
| 3 |
|
|
| 4 |
public class UDPipeAnnotate {
|
|
| 5 |
|
|
| 6 |
} |
|
| 0 | 7 | |
| tmp/org.txm.core/src/java/org/txm/xml/XMLProcessor.java (revision 2649) | ||
|---|---|---|
| 4 | 4 |
import java.io.File; |
| 5 | 5 |
import java.io.FileOutputStream; |
| 6 | 6 |
import java.io.IOException; |
| 7 |
import java.io.InputStream; |
|
| 8 | 7 |
import java.net.URL; |
| 9 | 8 |
import java.util.Arrays; |
| 10 |
import java.util.HashMap; |
|
| 11 | 9 |
import java.util.LinkedHashMap; |
| 12 |
import java.util.Stack; |
|
| 13 | 10 |
|
| 14 |
import javax.xml.stream.XMLInputFactory; |
|
| 15 | 11 |
import javax.xml.stream.XMLOutputFactory; |
| 16 | 12 |
import javax.xml.stream.XMLStreamConstants; |
| 17 | 13 |
import javax.xml.stream.XMLStreamException; |
| 18 |
import javax.xml.stream.XMLStreamReader; |
|
| 19 | 14 |
import javax.xml.stream.XMLStreamWriter; |
| 20 | 15 |
|
| 21 |
import org.txm.importer.PersonalNamespaceContext; |
|
| 22 |
|
|
| 23 |
public class XMLProcessor {
|
|
| 16 |
/** |
|
| 17 |
* Parse and write a XML file. If no hook specified the file is not transformed |
|
| 18 |
* |
|
| 19 |
* @author mdecorde |
|
| 20 |
* |
|
| 21 |
*/ |
|
| 22 |
public class XMLProcessor extends XMLParser {
|
|
| 24 | 23 |
|
| 25 |
/** The input */ |
|
| 26 |
protected URL inputurl; |
|
| 27 |
|
|
| 28 |
protected InputStream inputData; |
|
| 29 |
|
|
| 30 |
protected XMLInputFactory factory; |
|
| 31 |
|
|
| 32 |
protected XMLStreamReader parser; |
|
| 33 |
|
|
| 34 | 24 |
/** The output. */ |
| 35 | 25 |
protected XMLOutputFactory outfactory = XMLOutputFactory.newInstance(); |
| 36 | 26 |
|
| ... | ... | |
| 38 | 28 |
|
| 39 | 29 |
protected XMLStreamWriter writer; |
| 40 | 30 |
|
| 41 |
public static String TXMNS = "http://textometrie.org/1.0"; |
|
| 42 |
|
|
| 43 |
public static String TXM = "txm"; |
|
| 44 |
|
|
| 45 |
public static String TEINS = "http://www.tei-c.org/ns/1.0"; |
|
| 46 |
|
|
| 47 |
public static String TEI = "tei"; |
|
| 48 |
|
|
| 49 |
protected static PersonalNamespaceContext Nscontext = new PersonalNamespaceContext(); |
|
| 50 |
|
|
| 51 |
protected StringBuilder currentXPath = new StringBuilder("");
|
|
| 52 |
|
|
| 53 |
protected String localname; |
|
| 54 |
|
|
| 55 |
int processingXInclude = 0; |
|
| 56 |
|
|
| 57 |
LinkedHashMap<String, Hook> hooks = new LinkedHashMap<>(); |
|
| 58 |
|
|
| 59 |
private Hook activeHook = null; |
|
| 60 |
|
|
| 61 | 31 |
public XMLProcessor(File infile) throws IOException, XMLStreamException {
|
| 62 | 32 |
this(infile.toURI().toURL()); |
| 63 | 33 |
} |
| 64 | 34 |
|
| 65 | 35 |
public XMLProcessor(URL inputurl) throws IOException, XMLStreamException {
|
| 66 |
this.inputurl = inputurl; |
|
| 67 |
this.inputData = inputurl.openStream(); |
|
| 68 |
this.factory = XMLInputFactory.newInstance(); |
|
| 69 |
this.parser = factory.createXMLStreamReader(inputData); |
|
| 36 |
super(inputurl); |
|
| 70 | 37 |
} |
| 71 | 38 |
|
| 72 | 39 |
/** |
| 73 |
* called when the parser is activated by its parent |
|
| 74 |
* |
|
| 75 |
* @return true if correctly activated |
|
| 76 |
*/ |
|
| 77 |
public boolean activate() {
|
|
| 78 |
return false; |
|
| 79 |
} |
|
| 80 |
|
|
| 81 |
/** |
|
| 82 |
* called when the parser is deactivated by its parent |
|
| 83 |
* |
|
| 84 |
* @return true if correctly activated |
|
| 85 |
*/ |
|
| 86 |
public boolean deactivate() {
|
|
| 87 |
return false; |
|
| 88 |
} |
|
| 89 |
|
|
| 90 |
/** |
|
| 91 |
* |
|
| 92 |
* @return true if this hook parser must be activated |
|
| 93 |
*/ |
|
| 94 |
public boolean mustActivate() {
|
|
| 95 |
return false; |
|
| 96 |
} |
|
| 97 |
|
|
| 98 |
/** |
|
| 99 |
* |
|
| 100 |
* @return true if this hook parser must be desactivated |
|
| 101 |
*/ |
|
| 102 |
public boolean mustDeactivate() {
|
|
| 103 |
return true; |
|
| 104 |
} |
|
| 105 |
|
|
| 106 |
/** |
|
| 107 | 40 |
* Helper method to get an attribute value by its name |
| 108 | 41 |
* |
| 109 | 42 |
* @param name the attribute name |
| 110 | 43 |
* @return the value if any |
| 111 | 44 |
*/ |
| 45 |
@Override |
|
| 112 | 46 |
public String getParserAttributeValue(String name) {
|
| 113 | 47 |
if (name == null) return null; |
| 114 | 48 |
|
| ... | ... | |
| 122 | 56 |
return null; |
| 123 | 57 |
} |
| 124 | 58 |
|
| 59 |
@Override |
|
| 125 | 60 |
protected void before() {
|
| 126 | 61 |
|
| 62 |
super.before(); |
|
| 127 | 63 |
} |
| 128 | 64 |
|
| 65 |
@Override |
|
| 129 | 66 |
protected void after() throws XMLStreamException, IOException {
|
| 67 |
|
|
| 68 |
super.after(); |
|
| 69 |
|
|
| 130 | 70 |
if (factory != null) {
|
| 131 | 71 |
factory = null; |
| 132 | 72 |
if (parser != null) {
|
| ... | ... | |
| 142 | 82 |
writer = null; |
| 143 | 83 |
parser = null; |
| 144 | 84 |
} |
| 85 |
|
|
| 86 |
if (writer != null) writer.close(); |
|
| 87 |
|
|
| 88 |
if (output != null) output.close(); |
|
| 145 | 89 |
} |
| 146 | 90 |
|
| 147 |
protected void closeForError() throws XMLStreamException, IOException {
|
|
| 148 |
if (factory != null) {
|
|
| 149 |
if (parser != null) {
|
|
| 150 |
parser.close(); |
|
| 151 |
} |
|
| 152 |
if (inputData != null) {
|
|
| 153 |
inputData.close(); |
|
| 154 |
} |
|
| 155 |
} |
|
| 156 |
} |
|
| 157 |
|
|
| 158 | 91 |
/** |
| 159 | 92 |
* Creates the output. |
| 160 | 93 |
* |
| ... | ... | |
| 182 | 115 |
} |
| 183 | 116 |
} |
| 184 | 117 |
|
| 185 |
public boolean process(File outfile) throws XMLStreamException, IOException {
|
|
| 186 |
if (factory == null) {
|
|
| 187 |
System.out.println("Error: this parser is a Hook parser to be used with another StaxIdentityParser as a Hook");
|
|
| 118 |
public boolean process(File outFile) throws XMLStreamException, IOException {
|
|
| 119 |
|
|
| 120 |
if (!createOutput(outFile)) {
|
|
| 188 | 121 |
return false; |
| 189 | 122 |
} |
| 190 |
if (!createOutput(outfile)) {
|
|
| 191 |
return false; |
|
| 192 |
} |
|
| 193 | 123 |
|
| 194 | 124 |
writer.writeStartDocument("UTF-8", "1.0");
|
| 195 | 125 |
writer.writeCharacters("\n");
|
| 126 |
|
|
| 196 | 127 |
boolean ret = process(writer); |
| 197 |
if (writer != null) {
|
|
| 198 |
writer.close(); |
|
| 199 |
} |
|
| 200 |
if (output != null) {
|
|
| 201 |
try {
|
|
| 202 |
output.close(); |
|
| 203 |
} |
|
| 204 |
catch (Exception e) {
|
|
| 205 |
System.out.println("output excep: " + e);
|
|
| 206 |
} |
|
| 207 |
} |
|
| 208 | 128 |
|
| 209 |
if (parser != null) {
|
|
| 210 |
try {
|
|
| 211 |
parser.close(); |
|
| 212 |
} |
|
| 213 |
catch (Exception e) {
|
|
| 214 |
System.out.println("parser excep: " + e);
|
|
| 215 |
} |
|
| 216 |
} |
|
| 217 |
|
|
| 218 |
if (inputData != null) {
|
|
| 219 |
try {
|
|
| 220 |
inputData.close(); |
|
| 221 |
} |
|
| 222 |
catch (Exception e) {
|
|
| 223 |
System.out.println("inputData excep: " + e);
|
|
| 224 |
} |
|
| 225 |
} |
|
| 226 |
|
|
| 227 | 129 |
return ret; |
| 228 | 130 |
} |
| 229 | 131 |
|
| 230 |
public final static String SLASH = "/"; |
|
| 231 |
|
|
| 232 | 132 |
public boolean process(XMLStreamWriter awriter) throws XMLStreamException, IOException {
|
| 233 | 133 |
|
| 234 | 134 |
if (factory == null) {
|
| ... | ... | |
| 237 | 137 |
} |
| 238 | 138 |
|
| 239 | 139 |
this.writer = awriter; |
| 240 |
// if (processingXInclude == 0) {
|
|
| 241 |
before(); // if you need to do something before reading the xml |
|
| 242 |
// } |
|
| 243 |
try {
|
|
| 244 |
for (int event = parser.next(); event != XMLStreamConstants.END_DOCUMENT; event = parser.next()) {
|
|
| 245 |
|
|
| 246 |
// preparing process |
|
| 247 |
if (event == XMLStreamConstants.START_ELEMENT) {
|
|
| 248 |
localname = parser.getLocalName(); |
|
| 249 |
currentXPath.append(SLASH); |
|
| 250 |
currentXPath.append(localname); |
|
| 251 |
buildCurrentAttributes(); // for easier access later |
|
| 252 |
} |
|
| 253 |
else if (event == XMLStreamConstants.END_ELEMENT) {
|
|
| 254 |
localname = parser.getLocalName(); |
|
| 255 |
} |
|
| 256 |
|
|
| 257 |
if (activeHook != null) {
|
|
| 258 |
if (activeHook.mustDeactivate()) {
|
|
| 259 |
activeHook.deactivate(); |
|
| 260 |
activeHook = null; |
|
| 261 |
} |
|
| 262 |
else {
|
|
| 263 |
activeHook.processParserEvent(event); |
|
| 264 |
} |
|
| 265 |
} |
|
| 266 |
|
|
| 267 |
if (activeHook == null && event == XMLStreamConstants.START_ELEMENT) { // only activate when starting an element (for now, thus we cannot process only text nodes)
|
|
| 268 |
for (Hook hook : hooks.values()) {
|
|
| 269 |
if (hook.mustActivate()) {
|
|
| 270 |
if (hook.activate()) {
|
|
| 271 |
activeHook = hook; |
|
| 272 |
|
|
| 273 |
hook.processParserEvent(event); |
|
| 274 |
break; |
|
| 275 |
} |
|
| 276 |
} |
|
| 277 |
} |
|
| 278 |
} |
|
| 279 |
|
|
| 280 |
if (activeHook == null) {
|
|
| 281 |
this.processParserEvent(event); |
|
| 282 |
} |
|
| 283 |
|
|
| 284 |
// post-preparing process |
|
| 285 |
if (event == XMLStreamConstants.END_ELEMENT) {
|
|
| 286 |
currentXPath.setLength(currentXPath.length() - localname.length() - 1); |
|
| 287 |
popCurrentAttributes(); |
|
| 288 |
} |
|
| 289 |
} |
|
| 290 |
} |
|
| 291 |
catch (Exception e) {
|
|
| 292 |
System.out.println("Unexpected error while parsing file " + inputurl + " : " + e);
|
|
| 293 |
System.out.println("Location line: " + parser.getLocation().getLineNumber() + " character: " + parser.getLocation().getColumnNumber());
|
|
| 294 |
org.txm.utils.logger.Log.printStackTrace(e); |
|
| 295 |
// e.printStackTrace(); |
|
| 296 |
if (writer != null) writer.close(); |
|
| 297 |
if (output != null) output.close(); |
|
| 298 |
parser.close(); |
|
| 299 |
inputData.close(); |
|
| 300 |
return false; |
|
| 301 |
} |
|
| 302 | 140 |
|
| 303 |
// if (processingXInclude == 0) {
|
|
| 304 |
after(); // if you need to do something before closing the parser(); |
|
| 305 |
// } |
|
| 306 |
return true; |
|
| 141 |
return process(); |
|
| 307 | 142 |
|
| 308 | 143 |
} |
| 309 | 144 |
|
| 310 |
protected void processParserEvent(int event) throws XMLStreamException, IOException {
|
|
| 311 |
switch (event) {
|
|
| 312 |
case XMLStreamConstants.NAMESPACE: |
|
| 313 |
processNamespace(); |
|
| 314 |
break; |
|
| 315 |
case XMLStreamConstants.START_ELEMENT: |
|
| 316 |
processStartElement(); |
|
| 317 |
break; |
|
| 318 |
case XMLStreamConstants.CHARACTERS: |
|
| 319 |
processCharacters(); |
|
| 320 |
break; |
|
| 321 |
case XMLStreamConstants.END_ELEMENT: |
|
| 322 |
processEndElement(); |
|
| 323 |
break; |
|
| 324 |
case XMLStreamConstants.PROCESSING_INSTRUCTION: |
|
| 325 |
processProcessingInstruction(); |
|
| 326 |
break; |
|
| 327 |
case XMLStreamConstants.DTD: |
|
| 328 |
processDTD(); |
|
| 329 |
break; |
|
| 330 |
case XMLStreamConstants.CDATA: |
|
| 331 |
processCDATA(); |
|
| 332 |
break; |
|
| 333 |
case XMLStreamConstants.COMMENT: |
|
| 334 |
processComment(); |
|
| 335 |
break; |
|
| 336 |
case XMLStreamConstants.END_DOCUMENT: |
|
| 337 |
processEndDocument(); |
|
| 338 |
break; |
|
| 339 |
case XMLStreamConstants.ENTITY_REFERENCE: |
|
| 340 |
processEntityReference(); |
|
| 341 |
break; |
|
| 342 |
} |
|
| 343 |
} |
|
| 344 |
|
|
| 345 | 145 |
/** |
| 346 |
* The start element has already been written |
|
| 146 |
* Go to the end of an element : |
|
| 147 |
* - The start element has already been written |
|
| 148 |
* - the content is skipped and not written |
|
| 347 | 149 |
* |
| 348 | 150 |
* @param tagname |
| 349 | 151 |
* @throws XMLStreamException |
| 350 | 152 |
* @throws IOException |
| 351 | 153 |
*/ |
| 154 |
@Override |
|
| 352 | 155 |
public void goToEnd(String tagname) throws XMLStreamException, IOException {
|
| 353 | 156 |
int elements = 1; |
| 354 | 157 |
try {
|
| ... | ... | |
| 365 | 168 |
currentXPath.append(localname); |
| 366 | 169 |
buildCurrentAttributes(); |
| 367 | 170 |
|
| 368 |
_processStartElement();
|
|
| 171 |
processStartElement(); |
|
| 369 | 172 |
break; |
| 370 | 173 |
case XMLStreamConstants.CHARACTERS: |
| 371 | 174 |
processCharacters(); |
| ... | ... | |
| 406 | 209 |
System.out.println("Error while parsing file " + inputurl);
|
| 407 | 210 |
System.out.println("Location " + parser.getLocation());
|
| 408 | 211 |
org.txm.utils.logger.Log.printStackTrace(e); |
| 409 |
output.close(); |
|
| 410 |
parser.close(); |
|
| 212 |
after(); |
|
| 411 | 213 |
return; |
| 412 | 214 |
} |
| 413 | 215 |
} |
| 414 | 216 |
|
| 415 |
public String getLocation() {
|
|
| 416 |
if (parser != null) {
|
|
| 417 |
return "Line: " + parser.getLocation().getLineNumber() + " Col: " + parser.getLocation().getColumnNumber(); |
|
| 418 |
} |
|
| 419 |
return null; |
|
| 420 |
} |
|
| 421 |
|
|
| 422 |
public static final String INCLUDE = "include"; |
|
| 423 |
|
|
| 424 |
public static final String XI = "xi"; |
|
| 425 |
|
|
| 426 |
|
|
| 217 |
@Override |
|
| 427 | 218 |
protected void processNamespace() throws XMLStreamException {
|
| 428 | 219 |
writer.writeNamespace(parser.getPrefix(), parser.getNamespaceURI()); |
| 429 | 220 |
} |
| 430 | 221 |
|
| 431 | 222 |
|
| 223 |
@Override |
|
| 432 | 224 |
protected void processStartElement() throws XMLStreamException, IOException {
|
| 433 | 225 |
String prefix = parser.getPrefix(); |
| 434 |
if (prefix != null && prefix.length() > 0) {
|
|
| 435 |
writer.writeStartElement(Nscontext.getNamespaceURI(prefix), localname); |
|
| 436 |
} |
|
| 437 |
else {
|
|
| 438 |
writer.writeStartElement(localname); |
|
| 439 |
} |
|
| 440 |
|
|
| 441 |
for (int i = 0; i < parser.getNamespaceCount(); i++) {
|
|
| 442 |
writer.writeNamespace(parser.getNamespacePrefix(i), parser.getNamespaceURI(i)); |
|
| 443 |
} |
|
| 444 |
|
|
| 445 |
writeAttributes(); |
|
| 446 |
} |
|
| 447 |
|
|
| 448 |
private void _processStartElement() throws XMLStreamException, IOException {
|
|
| 449 |
String prefix = parser.getPrefix(); |
|
| 450 |
if (prefix != null && prefix.length() > 0) {
|
|
| 451 |
writer.writeStartElement(Nscontext.getNamespaceURI(prefix), localname); |
|
| 452 |
} |
|
| 453 |
else {
|
|
| 454 |
writer.writeStartElement(localname); |
|
| 455 |
} |
|
| 456 |
for (int i = 0; i < parser.getNamespaceCount(); i++) {
|
|
| 457 |
writer.writeNamespace(parser.getNamespacePrefix(i), parser.getNamespaceURI(i)); |
|
| 458 |
} |
|
| 459 |
|
|
| 460 |
writeAttributes(); |
|
| 461 |
} |
|
| 462 |
|
|
| 463 |
private Stack<LinkedHashMap<String, String>> attributesStack = new Stack<>(); |
|
| 464 |
|
|
| 465 |
private HashMap<String, String> popCurrentAttributes() {
|
|
| 466 |
return attributesStack.pop(); |
|
| 467 |
} |
|
| 468 |
|
|
| 469 |
private void buildCurrentAttributes() throws XMLStreamException {
|
|
| 470 |
LinkedHashMap<String, String> currentAttributes = new LinkedHashMap<>(); // keep order |
|
| 471 |
for (int i = 0; i < parser.getAttributeCount(); i++) {
|
|
| 472 |
currentAttributes.put(parser.getAttributeLocalName(i), parser.getAttributeValue(i)); |
|
| 226 |
if (prefix != null && prefix.length() > 0) {
|
|
| 227 |
writer.writeStartElement(Nscontext.getNamespaceURI(prefix), localname); |
|
| 473 | 228 |
} |
| 474 |
attributesStack.push(currentAttributes); |
|
| 229 |
else {
|
|
| 230 |
writer.writeStartElement(localname); |
|
| 231 |
} |
|
| 232 |
|
|
| 233 |
for (int i = 0; i < parser.getNamespaceCount(); i++) {
|
|
| 234 |
writer.writeNamespace(parser.getNamespacePrefix(i), parser.getNamespaceURI(i)); |
|
| 235 |
} |
|
| 236 |
|
|
| 237 |
writeAttributes(); |
|
| 475 | 238 |
} |
| 476 | 239 |
|
| 477 |
public LinkedHashMap<String, String> getCurrentAttributes() {
|
|
| 478 |
return attributesStack.lastElement(); |
|
| 479 |
} |
|
| 480 |
|
|
| 481 | 240 |
protected void writeAttributes() throws XMLStreamException {
|
| 482 | 241 |
for (int i = 0; i < parser.getAttributeCount(); i++) {
|
| 483 | 242 |
writeAttribute(parser.getAttributePrefix(i), parser.getAttributeLocalName(i), parser.getAttributeValue(i)); |
| ... | ... | |
| 493 | 252 |
} |
| 494 | 253 |
} |
| 495 | 254 |
|
| 255 |
@Override |
|
| 496 | 256 |
protected void processCharacters() throws XMLStreamException {
|
| 497 | 257 |
writer.writeCharacters(parser.getText()); |
| 498 | 258 |
} |
| 499 | 259 |
|
| 260 |
@Override |
|
| 500 | 261 |
protected void processProcessingInstruction() throws XMLStreamException {
|
| 501 | 262 |
writer.writeProcessingInstruction(parser.getPITarget(), parser.getPIData()); |
| 502 | 263 |
} |
| 503 | 264 |
|
| 265 |
@Override |
|
| 504 | 266 |
protected void processDTD() throws XMLStreamException {
|
| 505 | 267 |
writer.writeDTD(parser.getText()); |
| 506 | 268 |
} |
| 507 | 269 |
|
| 270 |
@Override |
|
| 508 | 271 |
protected void processCDATA() throws XMLStreamException {
|
| 509 | 272 |
writer.writeCData(parser.getText()); |
| 510 | 273 |
} |
| 511 | 274 |
|
| 275 |
@Override |
|
| 512 | 276 |
protected void processComment() throws XMLStreamException {
|
| 513 | 277 |
writer.writeComment(parser.getText()); |
| 514 | 278 |
} |
| 515 | 279 |
|
| 280 |
@Override |
|
| 516 | 281 |
protected void processEndElement() throws XMLStreamException {
|
| 517 |
if (localname == INCLUDE && parser.getPrefix() == XI) {
|
|
| 518 |
// nothing !! |
|
| 519 |
} |
|
| 520 |
else {
|
|
| 521 |
writer.writeEndElement(); |
|
| 522 |
} |
|
| 282 |
writer.writeEndElement(); |
|
| 523 | 283 |
} |
| 524 | 284 |
|
| 285 |
@Override |
|
| 525 | 286 |
protected void processEndDocument() throws XMLStreamException {
|
| 526 | 287 |
writer.writeEndDocument(); |
| 527 | 288 |
} |
| 528 | 289 |
|
| 290 |
@Override |
|
| 529 | 291 |
protected void processEntityReference() throws XMLStreamException {
|
| 530 | 292 |
writer.writeEntityRef(parser.getLocalName()); |
| 531 | 293 |
} |
| 532 | 294 |
|
| 295 |
@Override |
|
| 533 | 296 |
public String getCurrentPath() {
|
| 534 | 297 |
return currentXPath.toString(); |
| 535 | 298 |
} |
| ... | ... | |
| 568 | 331 |
// }; |
| 569 | 332 |
XPathHookActivator xpathActivator = new XPathHookActivator(null, "//p"); |
| 570 | 333 |
XPathsHookActivator xpathsActivator = new XPathsHookActivator(null, Arrays.asList("//p", "//p"));
|
| 571 |
// IdentityHook hook2 = new IdentityHook("hook2", xpathsActivator, builder) {
|
|
| 572 |
// |
|
| 573 |
// @Override |
|
| 574 |
// public boolean _activate() {
|
|
| 575 |
// System.out.println("ACTIVATING " + name + " AT " + getLocation());
|
|
| 576 |
// return true; |
|
| 577 |
// } |
|
| 578 |
// |
|
| 579 |
// @Override |
|
| 580 |
// public boolean deactivate() {
|
|
| 581 |
// System.out.println("DEACTIVATING " + name + " AT " + getLocation());
|
|
| 582 |
// return true; |
|
| 583 |
// } |
|
| 584 |
// |
|
| 585 |
// @Override |
|
| 586 |
// public void processStartElement() throws XMLStreamException, IOException {
|
|
| 587 |
// super.processStartElement(); |
|
| 588 |
// parentParser.writer.writeAttribute("hook", name);
|
|
| 589 |
// } |
|
| 590 |
// }; |
|
| 591 |
|
|
| 592 |
DOMIdentityHook hook3 = new DOMIdentityHook("hook3", xpathsActivator, builder) {
|
|
| 593 |
|
|
| 334 |
IdentityHook hook2 = new IdentityHook("hook2", xpathsActivator, builder) {
|
|
| 335 |
|
|
| 336 |
@Override |
|
| 337 |
public boolean _activate() {
|
|
| 338 |
System.out.println("ACTIVATING " + name + " AT " + getLocation());
|
|
| 339 |
return true; |
|
| 340 |
} |
|
| 341 |
|
|
| 342 |
@Override |
|
| 343 |
public boolean deactivate() {
|
|
| 344 |
System.out.println("DEACTIVATING " + name + " AT " + getLocation());
|
|
| 345 |
return true; |
|
| 346 |
} |
|
| 347 |
|
|
| 348 |
@Override |
|
| 349 |
public void processStartElement() throws XMLStreamException, IOException {
|
|
| 350 |
super.processStartElement(); |
|
| 351 |
parentParser.writer.writeAttribute("hook", name);
|
|
| 352 |
} |
|
| 594 | 353 |
}; |
| 595 | 354 |
|
| 355 |
// DOMIdentityHook hook3 = new DOMIdentityHook("hook3", xpathsActivator, builder) {
|
|
| 356 |
// |
|
| 357 |
// }; |
|
| 358 |
|
|
| 596 | 359 |
long time = System.currentTimeMillis(); |
| 597 | 360 |
if (builder.process(output)) {
|
| 598 | 361 |
System.out.println("Time=" + (System.currentTimeMillis() - time));
|
| 599 |
System.out.println("XPaths activator has done working ? "+xpathsActivator.hasAllXpathsBeenProcessed());
|
|
| 362 |
System.out.println("XPaths activator has done working ? " + xpathsActivator.hasAllXpathsBeenProcessed());
|
|
| 600 | 363 |
} |
| 601 | 364 |
else {
|
| 602 | 365 |
System.out.println("failure !");
|
| ... | ... | |
| 612 | 375 |
* |
| 613 | 376 |
* @return the actual hooks hash |
| 614 | 377 |
*/ |
| 378 |
@Override |
|
| 615 | 379 |
public LinkedHashMap<String, Hook> getHooks() {
|
| 616 | 380 |
return hooks; |
| 617 | 381 |
} |
| 618 | 382 |
|
| 383 |
@Override |
|
| 619 | 384 |
public void addHook(String name, Hook hook) {
|
| 620 | 385 |
hooks.put(name, hook); |
| 621 | 386 |
} |
| tmp/org.txm.core/src/java/org/txm/xml/DOMIdentityHook.java (revision 2649) | ||
|---|---|---|
| 4 | 4 |
|
| 5 | 5 |
import javax.xml.stream.XMLStreamException; |
| 6 | 6 |
|
| 7 |
import org.txm.utils.xml.DomUtils; |
|
| 8 | 7 |
import org.w3c.dom.Element; |
| 9 | 8 |
import org.w3c.dom.Node; |
| 10 | 9 |
import org.w3c.dom.NodeList; |
| 11 | 10 |
|
| 12 |
public class DOMIdentityHook extends Hook {
|
|
| 13 |
|
|
| 11 |
public class DOMIdentityHook extends Hook<XMLProcessor> {
|
|
| 12 |
|
|
| 14 | 13 |
Element dom; |
| 15 | 14 |
|
| 16 | 15 |
public DOMIdentityHook(String name, HookActivator activator, XMLProcessor processor) throws IOException, XMLStreamException {
|
| 17 | 16 |
super(name, activator, processor); |
| 18 | 17 |
} |
| 19 |
|
|
| 18 |
|
|
| 20 | 19 |
@Override |
| 21 | 20 |
public boolean _activate() {
|
| 22 | 21 |
StaxDomConstructor domConstructor = new StaxDomConstructor(parentParser.parser); |
| 23 | 22 |
try {
|
| 24 | 23 |
dom = domConstructor.getDom(); |
| 25 | 24 |
return dom != null; |
| 26 |
} catch (Exception e) {
|
|
| 25 |
} |
|
| 26 |
catch (Exception e) {
|
|
| 27 | 27 |
// TODO Auto-generated catch block |
| 28 | 28 |
e.printStackTrace(); |
| 29 | 29 |
return false; |
| ... | ... | |
| 35 | 35 |
boolean b = dom != null; // the DOM has been constructed and the XMLProcess can continue parsing |
| 36 | 36 |
if (b) {
|
| 37 | 37 |
try {
|
| 38 |
processDom(); |
|
| 38 | 39 |
writeDOM(); |
| 39 | 40 |
dom = null; |
| 40 |
} catch (XMLStreamException e) {
|
|
| 41 |
} |
|
| 42 |
catch (XMLStreamException e) {
|
|
| 41 | 43 |
e.printStackTrace(); |
| 42 | 44 |
} |
| 43 | 45 |
} |
| 44 | 46 |
return b; |
| 45 | 47 |
} |
| 46 |
|
|
| 48 |
|
|
| 49 |
/** |
|
| 50 |
* extends this method to process the DOM before it is written |
|
| 51 |
*/ |
|
| 52 |
public void processDom() {
|
|
| 53 |
// nothing since its the DomIdentityHook |
|
| 54 |
} |
|
| 55 |
|
|
| 47 | 56 |
@Override |
| 48 | 57 |
public boolean deactivate() {
|
| 49 | 58 |
return dom == null; // dom should be null |
| 50 | 59 |
} |
| 51 |
|
|
| 60 |
|
|
| 52 | 61 |
private void writeDOM() throws XMLStreamException {
|
| 53 | 62 |
writeDOM(dom); |
| 54 | 63 |
} |
| 55 | 64 |
|
| 56 | 65 |
private void writeDOM(Node node) throws XMLStreamException {
|
| 57 | 66 |
if (node.getNodeType() == Node.ELEMENT_NODE) {
|
| 58 |
Element e = (Element)node; |
|
| 67 |
Element e = (Element) node;
|
|
| 59 | 68 |
NodeList children = e.getChildNodes(); |
| 60 | 69 |
if (children.getLength() > 0) {
|
| 61 | 70 |
parentParser.writer.writeStartElement(e.getNamespaceURI(), e.getLocalName()); |
| 62 |
} else {
|
|
| 63 |
parentParser.writer.writeEmptyElement(e.getNamespaceURI(),e.getLocalName()); |
|
| 64 | 71 |
} |
| 72 |
else {
|
|
| 73 |
parentParser.writer.writeEmptyElement(e.getNamespaceURI(), e.getLocalName()); |
|
| 74 |
} |
|
| 65 | 75 |
|
| 66 |
for (int i = 0 ; i < e.getAttributes().getLength() ; i++) {
|
|
| 76 |
for (int i = 0; i < e.getAttributes().getLength(); i++) {
|
|
| 67 | 77 |
Node att = e.getAttributes().item(i); |
| 68 | 78 |
parentParser.writer.writeAttribute(att.getNamespaceURI(), att.getNodeName(), att.getNodeValue()); |
| 69 | 79 |
} |
| 70 | 80 |
|
| 71 |
for (int c = 0 ; c < children.getLength() ; c++) {
|
|
| 81 |
for (int c = 0; c < children.getLength(); c++) {
|
|
| 72 | 82 |
writeDOM(children.item(c)); |
| 73 | 83 |
} |
| 74 | 84 |
if (children.getLength() > 0) {
|
| 75 | 85 |
parentParser.writer.writeEndElement(); |
| 76 | 86 |
} |
| 77 |
} else if (node.getNodeType() == Node.TEXT_NODE) {
|
|
| 87 |
} |
|
| 88 |
else if (node.getNodeType() == Node.TEXT_NODE) {
|
|
| 78 | 89 |
parentParser.writer.writeCharacters(node.getTextContent()); |
| 79 | 90 |
} |
| 80 | 91 |
} |
| 81 |
|
|
| 92 |
|
|
| 82 | 93 |
@Override |
| 83 |
protected void processNamespace() throws XMLStreamException { }
|
|
| 84 |
|
|
| 94 |
protected void processNamespace() throws XMLStreamException {}
|
|
| 95 |
|
|
| 85 | 96 |
@Override |
| 86 |
protected void processStartElement() throws XMLStreamException, IOException { }
|
|
| 87 |
|
|
| 97 |
protected void processStartElement() throws XMLStreamException, IOException {}
|
|
| 98 |
|
|
| 88 | 99 |
@Override |
| 89 |
protected void processCharacters() throws XMLStreamException { }
|
|
| 90 |
|
|
| 100 |
protected void processCharacters() throws XMLStreamException {}
|
|
| 101 |
|
|
| 91 | 102 |
@Override |
| 92 |
protected void processProcessingInstruction() throws XMLStreamException { }
|
|
| 93 |
|
|
| 103 |
protected void processProcessingInstruction() throws XMLStreamException {}
|
|
| 104 |
|
|
| 94 | 105 |
@Override |
| 95 |
protected void processDTD() throws XMLStreamException { }
|
|
| 96 |
|
|
| 106 |
protected void processDTD() throws XMLStreamException {}
|
|
| 107 |
|
|
| 97 | 108 |
@Override |
| 98 |
protected void processCDATA() throws XMLStreamException { }
|
|
| 99 |
|
|
| 109 |
protected void processCDATA() throws XMLStreamException {}
|
|
| 110 |
|
|
| 100 | 111 |
@Override |
| 101 |
protected void processComment() throws XMLStreamException { }
|
|
| 102 |
|
|
| 112 |
protected void processComment() throws XMLStreamException {}
|
|
| 113 |
|
|
| 103 | 114 |
@Override |
| 104 |
protected void processEndElement() throws XMLStreamException { }
|
|
| 105 |
|
|
| 115 |
protected void processEndElement() throws XMLStreamException {}
|
|
| 116 |
|
|
| 106 | 117 |
@Override |
| 107 |
protected void processEndDocument() throws XMLStreamException { }
|
|
| 108 |
|
|
| 118 |
protected void processEndDocument() throws XMLStreamException {}
|
|
| 119 |
|
|
| 109 | 120 |
@Override |
| 110 |
protected void processEntityReference() throws XMLStreamException { }
|
|
| 121 |
protected void processEntityReference() throws XMLStreamException {}
|
|
| 111 | 122 |
} |
| tmp/org.txm.core/src/java/org/txm/xml/IdentityHook.java (revision 2649) | ||
|---|---|---|
| 4 | 4 |
|
| 5 | 5 |
import javax.xml.stream.XMLStreamException; |
| 6 | 6 |
|
| 7 |
public abstract class IdentityHook extends Hook {
|
|
| 7 |
public abstract class IdentityHook extends Hook<XMLProcessor> {
|
|
| 8 | 8 |
|
| 9 | 9 |
public IdentityHook(String name, HookActivator activator, XMLProcessor parentParser) throws IOException, XMLStreamException {
|
| 10 | 10 |
super(name, activator, parentParser); |
| tmp/org.txm.core/src/java/org/txm/xml/SimpleHook.java (revision 2649) | ||
|---|---|---|
| 1 |
package org.txm.xml; |
|
| 2 |
|
|
| 3 |
import java.io.IOException; |
|
| 4 |
|
|
| 5 |
import javax.xml.stream.XMLStreamException; |
|
| 6 |
|
|
| 7 |
/** |
|
| 8 |
* Simple wrapper to process only the start element, end element and characters Stax events |
|
| 9 |
* |
|
| 10 |
* @author mdecorde |
|
| 11 |
* |
|
| 12 |
*/ |
|
| 13 |
public abstract class SimpleHook<T extends XMLParser> extends Hook<T> {
|
|
| 14 |
|
|
| 15 |
public SimpleHook(String name, HookActivator activator, T parentParser) throws IOException, XMLStreamException {
|
|
| 16 |
super(name, activator, parentParser); |
|
| 17 |
} |
|
| 18 |
|
|
| 19 |
@Override |
|
| 20 |
protected void processNamespace() throws XMLStreamException {}
|
|
| 21 |
|
|
| 22 |
@Override |
|
| 23 |
protected void processProcessingInstruction() throws XMLStreamException {}
|
|
| 24 |
|
|
| 25 |
@Override |
|
| 26 |
protected void processDTD() throws XMLStreamException {}
|
|
| 27 |
|
|
| 28 |
@Override |
|
| 29 |
protected void processCDATA() throws XMLStreamException {}
|
|
| 30 |
|
|
| 31 |
@Override |
|
| 32 |
protected void processComment() throws XMLStreamException {}
|
|
| 33 |
|
|
| 34 |
@Override |
|
| 35 |
protected void processEndDocument() throws XMLStreamException {}
|
|
| 36 |
|
|
| 37 |
@Override |
|
| 38 |
protected void processEntityReference() throws XMLStreamException {}
|
|
| 39 |
} |
|
| 0 | 40 | |
Formats disponibles : Unified diff