Révision 3216
| tmp/org.txm.groovy.core/src/groovy/org/txm/macro/table/Table2XMLMacro.groovy (revision 3216) | ||
|---|---|---|
| 50 | 50 |
|
| 51 | 51 |
@Field @Option(name="structureContentColumnList", usage="structure textual content columns", widget="String", required=false, def="Titre propre,Résumé,Descripteurs (Aff. Lig.),Générique (Aff. Lig.) ,Séquences") |
| 52 | 52 |
def structureContentColumnList; |
| 53 |
|
|
| 54 |
@Field @Option(name="structureTitleColumnList", usage="structure textual content columns", widget="String", required=false, def="Titre propre,Date de diffusion") |
|
| 55 |
def structureTitleColumnList; |
|
| 53 | 56 |
|
| 54 | 57 |
@Field @Option(name="typesSeparator", usage="Columns types", widget="Separator", required=false, def="Columns types") |
| 55 | 58 |
def typesSeparator |
| ... | ... | |
| 78 | 81 |
structureMetadataColumnList.remove("")
|
| 79 | 82 |
structureContentColumnList = structureContentColumnList.split(",") as List
|
| 80 | 83 |
structureContentColumnList.remove("")
|
| 84 |
structureTitleColumnList = structureTitleColumnList.split(",") as List
|
|
| 85 |
structureTitleColumnList.remove("")
|
|
| 86 |
structureTitleColumnList |
|
| 81 | 87 |
dateColumnTypeList = dateColumnTypeList.split(",") as List
|
| 82 | 88 |
dateColumnTypeList.remove("")
|
| 83 | 89 |
prefixesColumnTypeList = prefixesColumnTypeList.split(",") as List
|
| ... | ... | |
| 99 | 105 |
println "textIDColumn, textJoinColumn=$textIDColumn, $textJoinColumn" |
| 100 | 106 |
println "textMetadataColumnList columns: $textMetadataColumnList" |
| 101 | 107 |
println "textContentColumnList columns: $textContentColumnList" |
| 108 |
println "structureTitleColumnList columns: $structureTitleColumnList" |
|
| 102 | 109 |
println "structureMetadataColumnList columns: $structureMetadataColumnList" |
| 103 | 110 |
println "structureContentColumnList columns: $structureContentColumnList" |
| 104 | 111 |
|
| 112 |
|
|
| 105 | 113 |
println "text selector="+textSelector |
| 106 | 114 |
println "structure selector="+structureSelector |
| 107 | 115 |
println "structureTag="+structureTag |
| ... | ... | |
| 125 | 133 |
def ok = true |
| 126 | 134 |
def hash = ["selection":[textIDColumn, textJoinColumn], "textMetadataColumnList":textMetadataColumnList, |
| 127 | 135 |
"textContentColumnList":textContentColumnList, "structureMetadataColumnList":structureMetadataColumnList, |
| 128 |
"structureContentColumnList":structureContentColumnList] |
|
| 136 |
"structureContentColumnList":structureContentColumnList, "structureTitleColumnList":structureTitleColumnList]
|
|
| 129 | 137 |
for (def key : hash.keySet()) {
|
| 130 | 138 |
for (def m : hash[key]) {
|
| 131 | 139 |
|
| ... | ... | |
| 197 | 205 |
} |
| 198 | 206 |
outputDirectory.mkdir() |
| 199 | 207 |
|
| 208 |
def metadata = new File(outputDirectory, "metadata.csv") |
|
| 209 |
metadata.delete() |
|
| 210 |
metadata << "id text-order\n" |
|
| 200 | 211 |
ConsoleProgressBar cpb = new ConsoleProgressBar(texts.size()) |
| 201 | 212 |
def nText = 0 |
| 202 | 213 |
for (def id : texts.keySet()) {
|
| ... | ... | |
| 231 | 242 |
|
| 232 | 243 |
textOrderWritten = false |
| 233 | 244 |
if (textOrderColumn != null && textOrderColumn.length() > 0 && textMetadataColumnList.contains(textOrderColumn)) {
|
| 234 |
writer.writeAttribute("text-order", text.get(textOrderColumn))
|
|
| 235 |
textOrderWritten = true |
|
| 245 |
textOrderWritten = text.get(textOrderColumn) |
|
| 236 | 246 |
} |
| 237 | 247 |
|
| 238 | 248 |
for (String att : textMetadataColumnList) {
|
| 239 | 249 |
if (att in dateColumnTypeList) {
|
| 240 | 250 |
String tri = writeMetadataDate(text, att, writer) |
| 241 | 251 |
if (!textOrderWritten) {
|
| 242 |
writer.writeAttribute("text-order", tri)
|
|
| 243 |
textOrderWritten = true; |
|
| 252 |
textOrderWritten = tri; |
|
| 244 | 253 |
} |
| 245 | 254 |
} else {
|
| 246 | 255 |
writer.writeAttribute(AsciiUtils.buildAttributeId(att), text.get(att)) // struct |
| ... | ... | |
| 248 | 257 |
} |
| 249 | 258 |
|
| 250 | 259 |
if (!textOrderWritten) {
|
| 251 |
writer.writeAttribute("text-order", String.sprintf("%08d", nText))
|
|
| 252 |
textOrderWritten = true; |
|
| 260 |
textOrderWritten = String.sprintf("%08d", nText);
|
|
| 253 | 261 |
} |
| 254 | 262 |
|
| 263 |
metadata << "$id $textOrderWritten\n" |
|
| 264 |
|
|
| 255 | 265 |
writer.writeCharacters("\n") // simple XML formating
|
| 256 |
|
|
| 257 |
writer.writeStartElement("metadata")
|
|
| 258 |
writer.writeStartElement("list")
|
|
| 259 |
writer.writeAttribute("type", "unordered")
|
|
| 260 |
writer.writeCharacters("\n")
|
|
| 261 |
textMetadataColumnList.each { att ->
|
|
| 262 |
writer.writeStartElement("item")
|
|
| 263 |
writer.writeCharacters(att+" : "+text.get(att).replaceAll("\n", ";"))
|
|
| 264 |
writer.writeEndElement() // item |
|
| 265 |
writer.writeCharacters("\n")
|
|
| 266 |
} |
|
| 267 |
writer.writeEndElement() // list |
|
| 268 |
writer.writeEndElement() // metadata |
|
| 269 |
writer.writeCharacters("\n")
|
|
| 270 |
|
|
| 266 |
|
|
| 271 | 267 |
for (String att : textContentColumnList) {
|
| 272 | 268 |
|
| 273 | 269 |
if (att in prefixesColumnTypeList) {
|
| ... | ... | |
| 289 | 285 |
writer.writeEmptyElement("pb") // <pb/>
|
| 290 | 286 |
writer.writeAttribute("n", ""+pb_n++)
|
| 291 | 287 |
writer.writeAttribute("type", "record")
|
| 288 |
writer.writeCharacters("\n")
|
|
| 292 | 289 |
|
| 293 | 290 |
writer.writeStartElement(structureTag) |
| 294 | 291 |
|
| ... | ... | |
| 301 | 298 |
} |
| 302 | 299 |
writer.writeCharacters("\n")
|
| 303 | 300 |
|
| 301 |
for (String att : structureTitleColumnList) {
|
|
| 302 |
writer.writeStartElement("head")
|
|
| 303 |
writer.writeCharacters(record.get(att).trim()) |
|
| 304 |
writer.writeEndElement() // head |
|
| 305 |
writer.writeCharacters("\n")
|
|
| 306 |
} |
|
| 307 |
|
|
| 304 | 308 |
writer.writeStartElement("metadata")
|
| 305 | 309 |
writer.writeStartElement("list")
|
| 306 | 310 |
writer.writeAttribute("type", "unordered")
|
| ... | ... | |
| 350 | 354 |
writer.writeCharacters(att) |
| 351 | 355 |
writer.writeEndElement() // hi |
| 352 | 356 |
writer.writeEndElement() // head |
| 357 |
writer.writeCharacters("\n")
|
|
| 353 | 358 |
|
| 354 | 359 |
def value = record.get(att).replaceAll(" ++", " ")
|
| 355 | 360 |
if (value.length() <= 1) return |
| ... | ... | |
| 389 | 394 |
writer.writeCharacters(att) |
| 390 | 395 |
writer.writeEndElement() // hi |
| 391 | 396 |
writer.writeEndElement() // head |
| 397 |
writer.writeCharacters("\n")
|
|
| 392 | 398 |
|
| 393 | 399 |
def value = record.get(att).replaceAll(" ++", " ")
|
| 394 | 400 |
if (value.length() == 0) return |
| ... | ... | |
| 412 | 418 |
writer.writeCharacters(att) |
| 413 | 419 |
writer.writeEndElement() // hi |
| 414 | 420 |
writer.writeEndElement() // head |
| 421 |
writer.writeCharacters("\n")
|
|
| 415 | 422 |
|
| 416 | 423 |
def value = record.get(att).replaceAll(" ++", " ")
|
| 417 | 424 |
if (value.length() <= 1) return |
Formats disponibles : Unified diff