root / tmp / org.txm.analec.rcp / src / org / txm / macro / urs / democrat / Metadata2TeiHeaderMacro.groovy @ 2027
History | View | Annotate | Download (7 kB)
| 1 |
// STANDARD DECLARATIONS
|
|---|---|
| 2 |
package org.txm.macro.urs.democrat
|
| 3 |
|
| 4 |
import java.nio.charset.Charset |
| 5 |
|
| 6 |
import org.kohsuke.args4j.* |
| 7 |
import groovy.transform.Field |
| 8 |
|
| 9 |
import org.txm.importer.StaxIdentityParser |
| 10 |
import org.txm.objects.* |
| 11 |
import org.txm.rcp.swt.widget.parameters.* |
| 12 |
import org.txm.searchengine.cqp.corpus.* |
| 13 |
import org.txm.utils.CsvReader |
| 14 |
import org.txm.utils.io.IOUtils |
| 15 |
import org.txm.scripts.importer.XPathResult |
| 16 |
import javax.xml.xpath.XPathConstants |
| 17 |
|
| 18 |
class Metadata2TEiHeaderMacro { |
| 19 |
|
| 20 |
int debug = 0; |
| 21 |
public Metadata2TEiHeaderMacro(int debug) { |
| 22 |
this.debug = debug
|
| 23 |
} |
| 24 |
//@Field @Option(name="teiHeaderTemplateFile", usage="the default teiHeader of texts", widget="FileOpen", required=true, def="teiHeaderTemplateFile.xml")
|
| 25 |
//def teiHeaderTemplateFile
|
| 26 |
//
|
| 27 |
//@Field @Option(name="xpathFile", usage="properties file to redirect metadata column to the teiHeader locations", widget="FileOpen", required=true, def="xpathFile.properties")
|
| 28 |
//def xpathFile
|
| 29 |
//
|
| 30 |
//@Field @Option(name="metadataFile", usage="the TSV file containing the metadata values per text", widget="FileOpen", required=true, def="metadataFile.tsv")
|
| 31 |
//def metadataFile
|
| 32 |
//
|
| 33 |
//@Field @Option(name="debug", usage="Show internal variable content", widget="StringArray", metaVar="OFF ON ALL REALLY ALL", required=true, def="OFF")
|
| 34 |
//debug
|
| 35 |
//
|
| 36 |
//if (!ParametersDialog.open(this)) return;
|
| 37 |
//if (debug == "OFF") debug = 0; else if (debug == "ON") debug = 1; else if (debug == "ALL") debug = 2 else if (debug == "REALLY ALL") debug = 3
|
| 38 |
//
|
| 39 |
//
|
| 40 |
//
|
| 41 |
//def xpathProperties = new Properties()
|
| 42 |
//xpathProperties.load(IOUtils.getReader(xpathFile))
|
| 43 |
//println xpathProperties
|
| 44 |
//
|
| 45 |
//def csvReader = new CsvReader(metadataFile.getAbsolutePath(), "\t".charAt(0), Charset.forName("UTF-8"))
|
| 46 |
//println csvReader
|
| 47 |
//
|
| 48 |
//csvReader.readHeaders();
|
| 49 |
//def header = csvReader.getHeaders()
|
| 50 |
//if (!header.contains("id")) {
|
| 51 |
// println "** 'id' column not found in $metadataFile header=$header"
|
| 52 |
// return;
|
| 53 |
//}
|
| 54 |
//if (!header.contains("corpus_id")) {
|
| 55 |
// println "** 'corpus_id' column not found in $metadataFile header=$header"
|
| 56 |
// return;
|
| 57 |
//}
|
| 58 |
//
|
| 59 |
//def corpora = [:]
|
| 60 |
//for (Project project : Workspace.getInstance().getProjects()) {
|
| 61 |
// for (MainCorpus corpus : project.getChildren(MainCorpus.class)) {
|
| 62 |
// corpora[corpus.getID()] = corpus;
|
| 63 |
// }
|
| 64 |
//}
|
| 65 |
//
|
| 66 |
//while (csvReader.readRecord()) {
|
| 67 |
// String text_id = csvReader.get("id")
|
| 68 |
// if (text_id == null || text_id.length() == 0) {
|
| 69 |
// //println "** @id not found for record="+csvReader.getRawRecord()
|
| 70 |
// continue;
|
| 71 |
// }
|
| 72 |
//
|
| 73 |
// String corpus_id = csvReader.get("corpus_id")
|
| 74 |
// if (corpus_id == null || corpus_id.length() == 0) {
|
| 75 |
// println "** @corpus_id not found for record="+csvReader.getRawRecord()
|
| 76 |
// continue;
|
| 77 |
// }
|
| 78 |
//
|
| 79 |
// MainCorpus corpus = corpora[corpus_id]
|
| 80 |
// if (corpus == null) {
|
| 81 |
// println "** no corpus found for ID=$corpus_id"
|
| 82 |
// continue;
|
| 83 |
// }
|
| 84 |
// def project = corpus.getProject()
|
| 85 |
//
|
| 86 |
// def txmDir = new File(project.getProjectDirectory(), "txm/"+corpus.getID())
|
| 87 |
//
|
| 88 |
// if (!txmDir.exists()) {
|
| 89 |
// println "** the selected corpus has no XML-TXM files. Aborting."
|
| 90 |
// }
|
| 91 |
//
|
| 92 |
//
|
| 93 |
// File txmFile = new File(txmDir, text_id+".xml")
|
| 94 |
// if (txmFile.exists()) {
|
| 95 |
//
|
| 96 |
// def data = [:]
|
| 97 |
//
|
| 98 |
// for (def h : header) data[h] = csvReader.get(h)
|
| 99 |
//
|
| 100 |
// String xmlteiHeaderContent = getCustomizedTEiHeader(teiHeaderTemplateFile, data, xpathProperties);
|
| 101 |
// if (xmlteiHeaderContent != null && xmlteiHeaderContent.length() > 0) {
|
| 102 |
// injecting(txmFile, xmlteiHeaderContent)
|
| 103 |
// } else {
|
| 104 |
// println "** Text header not updated: $txmFile"
|
| 105 |
// }
|
| 106 |
// } else {
|
| 107 |
// println "** Text not found: $txmFile"
|
| 108 |
// }
|
| 109 |
//}
|
| 110 |
|
| 111 |
def getCustomizedTEiHeader(File teiHeaderTemplateFile, def data, Properties xpathProperties) { |
| 112 |
XPathResult xpathProcessor = new XPathResult(teiHeaderTemplateFile);
|
| 113 |
for (String info : data.keySet()) { |
| 114 |
String xpath = xpathProperties[info];
|
| 115 |
String value = data[info]
|
| 116 |
if (xpath == null) { |
| 117 |
continue; // not a data to inject |
| 118 |
} |
| 119 |
if (debug > 1) println " injecting '$info'='$value' in '$xpath'" |
| 120 |
|
| 121 |
def expr = xpathProcessor.xpath.compile(xpath);
|
| 122 |
def first = expr.evaluate(xpathProcessor.doc.getDocumentElement(), XPathConstants.NODE); |
| 123 |
if (first != null) { |
| 124 |
switch (first.getNodeType()) {
|
| 125 |
case 1: // element |
| 126 |
if (debug > 1) println " patching: $info with "+value |
| 127 |
def newChild = xpathProcessor.doc.createElement(first.getTagName())
|
| 128 |
newChild.appendChild(xpathProcessor.doc.createTextNode(value)) |
| 129 |
first.getParentNode().replaceChild(newChild, first) |
| 130 |
break;
|
| 131 |
case 2: // attribute |
| 132 |
if (debug > 1) println " patching attribute: $info with "+value |
| 133 |
first.setNodeValue(value) |
| 134 |
break;
|
| 135 |
case 3: // text |
| 136 |
if (debug > 1) println " patching text: $info with "+value |
| 137 |
first.setNodeValue(value) |
| 138 |
break;
|
| 139 |
default:
|
| 140 |
break
|
| 141 |
} |
| 142 |
} else {
|
| 143 |
println "** not found $xpath"
|
| 144 |
} |
| 145 |
} |
| 146 |
|
| 147 |
// return xpathProcessor.doc.getElementsByTagName("teiHeader").item(0).getNodeValue();
|
| 148 |
//println "DOC="+xpathProcessor.doc.getDocumentElement().getElementsByTagName("teiHeader").item(0).toString()
|
| 149 |
String content = ""+xpathProcessor.doc.getDocumentElement().getElementsByTagName("teiHeader").item(0) |
| 150 |
content = content.substring('<?xml version="1.0" encoding="UTF-8"?>'.length())
|
| 151 |
} |
| 152 |
|
| 153 |
/**
|
| 154 |
* Replace the teiHeader
|
| 155 |
* @param txmFile
|
| 156 |
* @param data
|
| 157 |
* @param paths
|
| 158 |
* @param xmlteiHeaderContent
|
| 159 |
* @return
|
| 160 |
*/
|
| 161 |
def injecting(File txmFile, String xmlteiHeaderContent) { |
| 162 |
println "Editing: $txmFile..."
|
| 163 |
StaxIdentityParser sparser = new StaxIdentityParser(txmFile) {
|
| 164 |
boolean start = false; |
| 165 |
public void processStartElement() { |
| 166 |
|
| 167 |
if (localname == "teiHeader") { |
| 168 |
start = true;
|
| 169 |
if (debug > 2) println " replacing teiHeader" |
| 170 |
if (debug > 3) println " with $xmlteiHeaderContent" |
| 171 |
output.write(xmlteiHeaderContent.getBytes(Charset.forName("UTF-8"))); // REPLACE CONTENT ! |
| 172 |
} |
| 173 |
if (!start) {
|
| 174 |
super.processStartElement();
|
| 175 |
} |
| 176 |
} |
| 177 |
|
| 178 |
public void processEndElement() { |
| 179 |
if (!start) {
|
| 180 |
super.processEndElement();
|
| 181 |
} |
| 182 |
if (localname == "teiHeader") { |
| 183 |
start = false;
|
| 184 |
if (debug > 2) println " replace done" |
| 185 |
} |
| 186 |
} |
| 187 |
|
| 188 |
public void processCharacters() { |
| 189 |
if (!start) {
|
| 190 |
super.processCharacters();
|
| 191 |
} |
| 192 |
} |
| 193 |
|
| 194 |
public void processComment() { |
| 195 |
if (!start) {
|
| 196 |
super.processComment();
|
| 197 |
} |
| 198 |
} |
| 199 |
} |
| 200 |
|
| 201 |
File tmpDirectory = new File("/home/mdecorde/TEMP"); |
| 202 |
File outfile = new File(tmpDirectory, "tmp_"+txmFile.getName()) |
| 203 |
File copyFile = new File(tmpDirectory, "copy_"+txmFile.getName()) |
| 204 |
|
| 205 |
if (sparser.process(outfile)) { // replace inputFile |
| 206 |
if (txmFile.renameTo(copyFile)) {
|
| 207 |
if (outfile.renameTo(txmFile)) {
|
| 208 |
println " -> SUCCESS see $txmFile"
|
| 209 |
return true |
| 210 |
} else {
|
| 211 |
println " -> FAIL could not replace $txmFile"
|
| 212 |
println " -> see result in $outfile"
|
| 213 |
|
| 214 |
} |
| 215 |
} else {
|
| 216 |
println " -> FAIL could not make a copy of $txmFile in $copyFile"
|
| 217 |
println " -> see result in $outfile"
|
| 218 |
} |
| 219 |
|
| 220 |
} else {
|
| 221 |
println " -> FAIL see $outfile"
|
| 222 |
} |
| 223 |
|
| 224 |
return false; |
| 225 |
} |
| 226 |
} |