[12395] | 1 | package org.nzdl.gsdl;
|
---|
| 2 |
|
---|
| 3 | import java.io.*;
|
---|
| 4 |
|
---|
| 5 | import javax.xml.transform.Transformer;
|
---|
| 6 | import javax.xml.transform.TransformerConfigurationException;
|
---|
| 7 | import javax.xml.transform.TransformerException;
|
---|
| 8 | import javax.xml.transform.TransformerFactory;
|
---|
| 9 | import javax.xml.transform.stream.StreamResult;
|
---|
| 10 | import javax.xml.transform.stream.StreamSource;
|
---|
| 11 |
|
---|
[12594] | 12 | import javax.xml.parsers.*;
|
---|
| 13 | import javax.xml.transform.dom.*;
|
---|
| 14 | import org.w3c.dom.*;
|
---|
| 15 |
|
---|
| 16 |
|
---|
| 17 |
|
---|
[12395] | 18 | /**
|
---|
| 19 | * Use the TraX interface to perform a transformation in the simplest manner possible
|
---|
| 20 | * (3 statements).
|
---|
| 21 | */
|
---|
| 22 | public class ApplyXSLT
|
---|
| 23 | {
|
---|
| 24 |
|
---|
[20875] | 25 | public static final String DOC_START = new String ("<?DocStart?>");
|
---|
| 26 | public static final String DOC_END = new String ("<?DocEnd?>");
|
---|
| 27 | public static final String INPUT_END = new String ("<?Done?>");
|
---|
[12395] | 28 |
|
---|
[20875] | 29 | private static final String RECORD_ELEMENT = "record";
|
---|
| 30 | private static final String CONTROLFIELD_ELEMENT = "controlfield";
|
---|
| 31 | private static final String SUBFIELD_ELEMENT = "subfield";
|
---|
| 32 | private static final String LEADER_ELEMENT = "leader";
|
---|
[12594] | 33 |
|
---|
[20875] | 34 | private final int BEFORE_READING = 0;
|
---|
| 35 | private final int IS_READING = 1;
|
---|
| 36 | private String xsl_file;
|
---|
| 37 | private String mapping_file;
|
---|
[12395] | 38 |
|
---|
[20875] | 39 | public ApplyXSLT(){}
|
---|
[13470] | 40 |
|
---|
[20875] | 41 | public ApplyXSLT(String xsl_file)
|
---|
| 42 | {
|
---|
| 43 | this.xsl_file = xsl_file;
|
---|
| 44 | }
|
---|
[12395] | 45 |
|
---|
[20875] | 46 | public ApplyXSLT(String xsl_file,String mapping_file){
|
---|
| 47 | this.xsl_file = xsl_file;
|
---|
| 48 | this.mapping_file = mapping_file;
|
---|
| 49 | }
|
---|
[12594] | 50 |
|
---|
[20875] | 51 | private boolean process()
|
---|
| 52 | {
|
---|
| 53 | try{
|
---|
[13223] | 54 |
|
---|
[20875] | 55 | // Use System InputStream to receive piped data from the perl program
|
---|
| 56 | InputStreamReader ir = new InputStreamReader(System.in, "UTF8");
|
---|
| 57 | BufferedReader br = new BufferedReader(ir);
|
---|
[12395] | 58 |
|
---|
[20875] | 59 | int system_status = BEFORE_READING;
|
---|
| 60 | String a_doc = new String();
|
---|
| 61 | String output_file = new String();
|
---|
[12395] | 62 |
|
---|
[13223] | 63 |
|
---|
[20875] | 64 | while (br.ready()) {
|
---|
[12594] | 65 |
|
---|
[20875] | 66 | String this_line = br.readLine();
|
---|
| 67 | if(system_status == BEFORE_READING){
|
---|
| 68 | if(this_line.compareTo(DOC_START) == 0){
|
---|
| 69 | output_file = br.readLine(); // read the next line as the output file name
|
---|
| 70 | system_status = IS_READING;
|
---|
| 71 | a_doc = "";
|
---|
| 72 | }
|
---|
| 73 | else if(this_line.compareTo(INPUT_END) == 0){
|
---|
| 74 | return true;
|
---|
| 75 | }
|
---|
| 76 | else{
|
---|
| 77 | System.err.println("Undefined process status:" + this_line);
|
---|
| 78 | system_status = BEFORE_READING;
|
---|
| 79 | }
|
---|
[12594] | 80 |
|
---|
[20875] | 81 | }
|
---|
| 82 | else if(system_status == IS_READING){
|
---|
| 83 | if(this_line.compareTo(DOC_END) == 0){
|
---|
| 84 | boolean result = false;
|
---|
| 85 | if (mapping_file !=null && !mapping_file.equals("")){
|
---|
| 86 | result = translateXMLWithMapping(a_doc,output_file);
|
---|
| 87 | }
|
---|
| 88 | else{
|
---|
| 89 | result = translateXML(a_doc,output_file);
|
---|
| 90 | }
|
---|
[12594] | 91 |
|
---|
[20875] | 92 | if (!result){
|
---|
| 93 | System.err.println("Translation Failed!!");
|
---|
| 94 | return false;
|
---|
| 95 | }
|
---|
[12594] | 96 |
|
---|
[20875] | 97 | system_status = BEFORE_READING;
|
---|
[12594] | 98 |
|
---|
[20875] | 99 | }
|
---|
| 100 | else{
|
---|
| 101 | a_doc = a_doc + this_line + "\n";
|
---|
[12594] | 102 |
|
---|
[20875] | 103 | }
|
---|
| 104 | }
|
---|
| 105 | else{
|
---|
| 106 | System.err.println ("Undefined system status in ApplyXSLT.java main().");
|
---|
| 107 | System.exit(-1);
|
---|
| 108 | }
|
---|
[12594] | 109 |
|
---|
[20875] | 110 | }
|
---|
| 111 | }catch (Exception e)
|
---|
| 112 | {
|
---|
| 113 | System.err.println("Receiving piped data error!" + e.toString());
|
---|
| 114 | }
|
---|
[12395] | 115 |
|
---|
[20875] | 116 | return false;
|
---|
| 117 | }
|
---|
[12594] | 118 |
|
---|
| 119 |
|
---|
[20875] | 120 | private boolean translateXML(String full_doc, String output_file)
|
---|
| 121 | throws IOException,TransformerException, TransformerConfigurationException, FileNotFoundException
|
---|
| 122 | {
|
---|
[12594] | 123 |
|
---|
[20875] | 124 | StringReader str = new StringReader(full_doc) ;
|
---|
[12395] | 125 |
|
---|
[20875] | 126 | TransformerFactory tFactory = TransformerFactory.newInstance();
|
---|
| 127 | Transformer transformer = tFactory.newTransformer(new StreamSource(xsl_file));
|
---|
| 128 | transformer.transform(new StreamSource(str), new StreamResult(new FileOutputStream(output_file)));
|
---|
| 129 | return true;
|
---|
| 130 | }
|
---|
[12395] | 131 |
|
---|
[20875] | 132 | private boolean translateXMLWithMapping(String full_doc, String output_file)
|
---|
| 133 | throws IOException,TransformerException, TransformerConfigurationException, FileNotFoundException
|
---|
| 134 | {
|
---|
| 135 | StringReader str = new StringReader(full_doc) ;
|
---|
[12594] | 136 |
|
---|
[20875] | 137 | try{
|
---|
| 138 | TransformerFactory tFactory = TransformerFactory.newInstance();
|
---|
| 139 | Transformer transformer = tFactory.newTransformer(new StreamSource(xsl_file));
|
---|
[12594] | 140 |
|
---|
[20875] | 141 | Document mapping_doc = DocumentBuilderFactory.newInstance().newDocumentBuilder().parse(mapping_file);
|
---|
| 142 | Element mapping =mapping_doc.getDocumentElement();
|
---|
[12594] | 143 |
|
---|
[20875] | 144 | transformer.setParameter("mapping",mapping);
|
---|
[12594] | 145 |
|
---|
[20875] | 146 | Document output_doc = DocumentBuilderFactory.newInstance().newDocumentBuilder().newDocument();
|
---|
[12594] | 147 |
|
---|
[20875] | 148 | transformer.transform(new StreamSource(str), new DOMResult(output_doc));
|
---|
[12594] | 149 |
|
---|
[20875] | 150 | calculateRecordsLength(output_doc);
|
---|
[12594] | 151 |
|
---|
[20875] | 152 | transformer = tFactory.newTransformer();
|
---|
[12594] | 153 |
|
---|
[20875] | 154 | transformer.transform(new DOMSource(output_doc), new StreamResult(new FileOutputStream(output_file)));
|
---|
[12594] | 155 |
|
---|
| 156 | }
|
---|
[20875] | 157 | catch(Exception e){
|
---|
| 158 | e.printStackTrace();
|
---|
| 159 | return false;
|
---|
| 160 | }
|
---|
[12594] | 161 |
|
---|
[20875] | 162 | return true;
|
---|
| 163 | }
|
---|
[12594] | 164 |
|
---|
[20875] | 165 | private void calculateRecordsLength(Document output_doc){
|
---|
| 166 | NodeList records = output_doc.getDocumentElement().getElementsByTagName(RECORD_ELEMENT);
|
---|
| 167 |
|
---|
| 168 | for(int i=0;i<records.getLength();i++){
|
---|
| 169 | Element record = (Element)records.item(i);
|
---|
| 170 | calculateRecordLength(record);
|
---|
[12594] | 171 | }
|
---|
[20875] | 172 | }
|
---|
[12594] | 173 |
|
---|
[20875] | 174 | private void calculateRecordLength(Element record){
|
---|
| 175 | int total_length =0;
|
---|
| 176 | NodeList controlfileds = record.getElementsByTagName(CONTROLFIELD_ELEMENT);
|
---|
| 177 | for(int i=0;i<controlfileds.getLength();i++){
|
---|
| 178 | Element controlfiled = (Element)controlfileds.item(i);
|
---|
| 179 | total_length +=getElementTextValue(controlfiled).length();
|
---|
| 180 | }
|
---|
[12594] | 181 |
|
---|
[20875] | 182 | NodeList subfileds = record.getElementsByTagName(SUBFIELD_ELEMENT);
|
---|
| 183 | for(int i=0;i<subfileds.getLength();i++){
|
---|
| 184 | Element subfiled = (Element)subfileds.item(i);
|
---|
| 185 | total_length +=getElementTextValue(subfiled).length();
|
---|
| 186 | }
|
---|
[12594] | 187 |
|
---|
[20875] | 188 | String record_length = total_length+"";
|
---|
| 189 | //fill in a extra digit as record length needs to be five characters long
|
---|
| 190 | if (total_length < 10000){
|
---|
| 191 | record_length = "0"+record_length;
|
---|
| 192 | if (total_length < 1000){
|
---|
| 193 | record_length = "0"+record_length;
|
---|
| 194 | }
|
---|
| 195 | if (total_length < 100){
|
---|
| 196 | record_length = "0"+record_length;
|
---|
| 197 | }
|
---|
| 198 | if (total_length < 10){
|
---|
| 199 | record_length = "0"+record_length;
|
---|
| 200 | }
|
---|
[12594] | 201 |
|
---|
[20875] | 202 | }
|
---|
[12594] | 203 |
|
---|
[20875] | 204 | NodeList leaders = record.getElementsByTagName(LEADER_ELEMENT);
|
---|
[12594] | 205 |
|
---|
[20875] | 206 | //only one leader element
|
---|
| 207 | if (leaders.getLength() >0){
|
---|
| 208 | Element leader_element = (Element)leaders.item(0);
|
---|
| 209 | removeFirstTextNode(leader_element);
|
---|
| 210 | leader_element.insertBefore(leader_element.getOwnerDocument().createTextNode(record_length),leader_element.getFirstChild());
|
---|
[12594] | 211 | }
|
---|
| 212 |
|
---|
[20875] | 213 | }
|
---|
[12594] | 214 |
|
---|
[20875] | 215 | private void removeFirstTextNode(Element element){
|
---|
| 216 | //remove the first text node
|
---|
| 217 | NodeList children_nodelist = element.getChildNodes();
|
---|
| 218 | for (int i = 0; i < children_nodelist.getLength(); i++) {
|
---|
| 219 | Node child_node = children_nodelist.item(i);
|
---|
| 220 | if (child_node.getNodeType() == Node.TEXT_NODE) {
|
---|
| 221 | element.removeChild(child_node);
|
---|
| 222 | return;
|
---|
| 223 | }
|
---|
[12594] | 224 | }
|
---|
| 225 |
|
---|
[20875] | 226 | }
|
---|
[12594] | 227 |
|
---|
[20875] | 228 | private String getElementTextValue(Element element)
|
---|
| 229 | {
|
---|
| 230 | String text ="";
|
---|
[12594] | 231 |
|
---|
[20875] | 232 | // Find the node child
|
---|
| 233 | NodeList children_nodelist = element.getChildNodes();
|
---|
| 234 | for (int i = 0; i < children_nodelist.getLength(); i++) {
|
---|
| 235 | Node child_node = children_nodelist.item(i);
|
---|
| 236 | if (child_node.getNodeType() == Node.TEXT_NODE) {
|
---|
| 237 | text +=child_node.getNodeValue();
|
---|
| 238 | }
|
---|
[12594] | 239 | }
|
---|
[20875] | 240 |
|
---|
| 241 | return text;
|
---|
| 242 | }
|
---|
[12594] | 243 |
|
---|
| 244 |
|
---|
[20875] | 245 | private void setMappingVariable(Document style_doc){
|
---|
| 246 | Node child = style_doc.getDocumentElement().getFirstChild();
|
---|
| 247 | while(child != null) {
|
---|
| 248 | String name = child.getNodeName();
|
---|
| 249 | if (name.equals("xsl:variable")) {
|
---|
| 250 | Element variable_element = (Element)child;
|
---|
| 251 | if ( variable_element.getAttribute("name").trim().equals("mapping")){
|
---|
| 252 | variable_element.setAttribute("select","document('"+mapping_file+"')/Mapping");
|
---|
[12594] | 253 | }
|
---|
[20875] | 254 | }
|
---|
| 255 | child = child.getNextSibling();
|
---|
| 256 | }
|
---|
[12594] | 257 |
|
---|
[20875] | 258 | }
|
---|
[12594] | 259 |
|
---|
| 260 |
|
---|
[20875] | 261 | private void translate(String xml_file, String xsl_file, String output_file)throws IOException,TransformerException, TransformerConfigurationException, FileNotFoundException, IOException{
|
---|
[13473] | 262 |
|
---|
[20875] | 263 | TransformerFactory tFactory = TransformerFactory.newInstance();
|
---|
| 264 | Transformer transformer = tFactory.newTransformer(new StreamSource(xsl_file));
|
---|
[13473] | 265 |
|
---|
[20875] | 266 | OutputStreamWriter output = null;
|
---|
| 267 | if (output_file.equals("")) {
|
---|
| 268 | output = new OutputStreamWriter(System.out, "UTF-8");
|
---|
| 269 | }
|
---|
| 270 | else{
|
---|
| 271 | output = new OutputStreamWriter(new FileOutputStream(output_file), "UTF-8");
|
---|
| 272 | }
|
---|
[13473] | 273 |
|
---|
[20875] | 274 | transformer.transform(new StreamSource(new File(xml_file)),new StreamResult(output));
|
---|
[13473] | 275 |
|
---|
[20875] | 276 | }
|
---|
[13470] | 277 |
|
---|
[20875] | 278 | static public String replaceAll(String source_string, String match_regexp, String replace_string)
|
---|
| 279 | {
|
---|
| 280 | return source_string.replaceAll(match_regexp, replace_string);
|
---|
| 281 | }
|
---|
[13473] | 282 |
|
---|
| 283 |
|
---|
[20875] | 284 | public static void main(String[] args)
|
---|
| 285 | {
|
---|
[13470] | 286 |
|
---|
[20875] | 287 | String xml_file="";
|
---|
| 288 | String xsl_file="";
|
---|
| 289 | String mapping_file="";
|
---|
| 290 | String output_file="";
|
---|
[13470] | 291 |
|
---|
[20875] | 292 | // Checking Arguments
|
---|
| 293 | if(args.length < 1)
|
---|
| 294 | {
|
---|
| 295 | printUsage();
|
---|
| 296 | }
|
---|
[13470] | 297 |
|
---|
[20875] | 298 | for (int i=0;i<args.length;i++){
|
---|
| 299 | if (args[i].equals("-m") && i+1 < args.length && !args[i+1].startsWith("-")){
|
---|
| 300 | mapping_file = args[++i];
|
---|
| 301 | checkFile(mapping_file.replaceAll("file:///",""));
|
---|
| 302 | }
|
---|
| 303 | else if (args[i].equals("-x") && i+1 < args.length && !args[i+1].startsWith("-")){
|
---|
| 304 | xml_file = args[++i];
|
---|
| 305 | checkFile(xml_file.replaceAll("file:///",""));
|
---|
| 306 | }
|
---|
| 307 | else if(args[i].equals("-t") && i+1 < args.length && !args[i+1].startsWith("-")){
|
---|
| 308 | xsl_file = args[++i];
|
---|
| 309 | checkFile( xsl_file.replaceAll("file:///",""));
|
---|
| 310 | }
|
---|
| 311 | else if(args[i].equals("-o") && i+1 < args.length && !args[i+1].startsWith("-")){
|
---|
| 312 | output_file = args[++i];
|
---|
[13470] | 313 |
|
---|
[20875] | 314 | }
|
---|
| 315 | else if(args[i].equals("-h")){
|
---|
| 316 | printUsage();
|
---|
| 317 | }
|
---|
| 318 | else{
|
---|
| 319 | printUsage();
|
---|
| 320 | }
|
---|
[13470] | 321 |
|
---|
[20875] | 322 | }
|
---|
[12594] | 323 |
|
---|
[13470] | 324 |
|
---|
[20875] | 325 | ApplyXSLT core = null;
|
---|
[13470] | 326 |
|
---|
[20875] | 327 | if (xml_file.equals("") && !xsl_file.equals("")){//read from pipe line
|
---|
| 328 | if (mapping_file.equals("")){
|
---|
| 329 | core = new ApplyXSLT(xsl_file);
|
---|
| 330 | }
|
---|
| 331 | else{
|
---|
| 332 | core = new ApplyXSLT(xsl_file,mapping_file);
|
---|
| 333 | }
|
---|
[13470] | 334 |
|
---|
[20875] | 335 | if (core != null){
|
---|
| 336 | core.process();
|
---|
| 337 | }
|
---|
| 338 | else{
|
---|
| 339 | printUsage();
|
---|
| 340 | }
|
---|
| 341 | }
|
---|
| 342 | else if(!xml_file.equals("") && !xsl_file.equals("")){
|
---|
| 343 | core = new ApplyXSLT();
|
---|
| 344 | try {
|
---|
| 345 | core.translate(xml_file,xsl_file,output_file);
|
---|
| 346 | }
|
---|
| 347 | catch(Exception e){e.printStackTrace();}
|
---|
| 348 | }
|
---|
| 349 | else{
|
---|
| 350 | printUsage();
|
---|
| 351 | }
|
---|
[13470] | 352 |
|
---|
[20875] | 353 | }
|
---|
[13470] | 354 |
|
---|
[20875] | 355 | private static void checkFile(String filename){
|
---|
| 356 | File file = new File(filename);
|
---|
| 357 | if (!file.exists()){
|
---|
| 358 | System.out.println("Error: "+filename+" doesn't exist!");
|
---|
| 359 | System.exit(-1);
|
---|
[13470] | 360 | }
|
---|
[20875] | 361 | }
|
---|
[13470] | 362 |
|
---|
[20875] | 363 | private static void printUsage(){
|
---|
| 364 | System.out.println("Usage:ApplyXSLT -x File -t File [-m File] [-o File]");
|
---|
| 365 | System.out.println("\t-x spacifies the xml file (Note: optional for piped xml data)");
|
---|
| 366 | System.out.println("\t-t spacifies the xsl file");
|
---|
| 367 | System.out.println("\t-m specifies the mapping file (for MARCXMLPlugout.pm only)");
|
---|
| 368 | System.out.println("\t-o specifies the output file name (output to screen if this option is absent)");
|
---|
| 369 | System.exit(-1);
|
---|
| 370 | }
|
---|
[12395] | 371 | }
|
---|
| 372 |
|
---|
| 373 |
|
---|