Laboratoire ICAR » Plateforme TXM

import org.txm.Toolbox

import org.txm.utils.ConsoleProgressBar

import org.txm.utils.io.IOUtils

def groovyUserDirectory = new File(Toolbox.getTxmHomePath() + "/scripts/groovy/user/org/txm/macro/");

if (!groovyUserDirectory.exists()) {

	groovyUserDirectory.mkdirs();

}

if (!groovyUserDirectory.exists()) {

	monitorShowError("Le répertoire $groovyUserDirectory n'a pas pu être créé. Abandon.")

	return;

}

@Field @Option(name="targets", usage="Path to the macro to update", widget="String", required=true, def="")

def targets

@Field @Option(name="svn", usage="TXM SVN repository", widget="String", required=true, def="http://forge.cbp.ens-lyon.fr/svn/txm/tmp/org.txm.groovy.core/src/groovy/org/txm/macro/")

def svn

@Field @Option(name="debug", usage="Show devug messages", widget="Boolean", required=false, def="false")

def debug

if (!ParametersDialog.open(this)) return false;

if (!svn.endsWith("/")) svn += "/"

def macros = targets.split(",").collect(){it.trim()}

println "Copyping ${macros.size()} files: "

def errors = []

for (def m : macros) {

	def f = new File(groovyUserDirectory, m+"Macro.groovy")

	f.getParentFile().mkdirs()

	String remoteUrl = svn+m+"Macro.groovy"

	try {

		if (debug) print "\n"+remoteUrl

		boolean replaced = f.exists()

		def file = new FileOutputStream(f)

		def out = new BufferedOutputStream(file)

		out << new URL(remoteUrl).openStream()

		out.close()

		if (replaced) print "*"

		else print "."

	} catch(Exception e ) {

		print "x"

		errors << "\nError while fetching $f from $remoteUrl: $e"

	}

}

if (errors.size() > 0) {

	println errors.join("\n")

}

println "\n$targets updated."

def monitorShowError(String message) {

	monitor.syncExec(new Runnable() {

		public void run() {

			org.eclipse.jface.dialogs.MessageDialog.openError(org.eclipse.swt.widgets.Display.getCurrent().getActiveShell(), "Error", message)

		}

	});

}

textMetadataColumnList = textMetadataColumnList.split(",") as List

textMetadataColumnList.remove("")

textContentColumnList = textContentColumnList.split(",") as List

textContentColumnList.remove("")

structureMetadataColumnList = structureMetadataColumnList.split(",") as List

structureMetadataColumnList.remove("")

structureContentColumnList = structureContentColumnList.split(",") as List

structureContentColumnList.remove("")

dateColumnTypeList = dateColumnTypeList.split(",") as List

dateColumnTypeList.remove("")

prefixesColumnTypeList = prefixesColumnTypeList.split(",") as List

prefixesColumnTypeList.remove("")

listColumnTypeList = listColumnTypeList.split(",") as List

listColumnTypeList.remove("")

	textSelector = [textSelector.substring(0, textSelector.indexOf("=")), textSelector.substring(textSelector.indexOf("=")+1)]

	structureSelector = [structureSelector.substring(0, structureSelector.indexOf("=")), structureSelector.substring(structureSelector.indexOf("=")+1)]

	}

	String id = reader.get(textIDColumn)

	String join = reader.get(textJoinColumn)

	String textSelectorValue = reader.get(textSelector[0])

	String structureSelectorValue = reader.get(structureSelector[0])

println "N records to join: "+nRecordToWrite

ConsoleProgressBar cpb = new ConsoleProgressBar(texts.size())

	cpb.tick()

	def toWrite = texts[id]

	def text = toWrite[0]

	String textSelectorValue = text.get(textSelector[0])

	if (textSelectorValue == null || (textSelectorValue != null && textSelectorValue.matches(textSelector[1]))) {

		if (debug) println "Processing text: $id"

		File outputfile = new File(outputDirectory, id+".xml")

		XMLOutputFactory factory = XMLOutputFactory.newInstance()

		FileOutputStream output = new FileOutputStream(outputfile)

		XMLStreamWriter writer = factory.createXMLStreamWriter(output, "UTF-8")

		writer.writeStartDocument("UTF-8","1.0")

		writer.writeCharacters("\n") // simple XML formating

		writer.writeStartElement("TEI")

		writer.writeCharacters("\n") // simple XML formating

		writer.writeStartElement("teiHeader")

		writer.writeEndElement() // teiHeader

		writer.writeCharacters("\n") // simple XML formating

		writer.writeStartElement("text")

		writer.writeAttribute("id", id)

		for (String att : textMetadataColumnList) {

				writeMetadataDate(text, att, writer)

				writer.writeAttribute(AsciiUtils.buildAttributeId(att), text.get(att)) // struct

		writer.writeCharacters("\n") // simple XML formating

		textMetadataColumnList.each { att ->

			writer.writeCharacters(att+" : "+text.get(att).replaceAll("\n", ";"))

		for (String att : textContentColumnList) {

				writeTextContent(text, att, writer)

		int pb_n = 1;

		for (int i = 1 ; i < toWrite.size() ; i++) {

			def record = toWrite[i]

			writer.writeEmptyElement("pb") // <pb/>

			writer.writeAttribute("n", ""+pb_n++)

			writer.writeAttribute("type", "record")

			writer.writeStartElement(structureTag)

			for (String att : structureMetadataColumnList) {

				if (att in dateColumnTypeList) {

					writeMetadataDate(record, att, writer)

				} else {

					writer.writeAttribute(AsciiUtils.buildAttributeId(att), record.get(att)) // struct

				}

			}

			writer.writeCharacters("\n")

			writer.writeStartElement("metadata")

			writer.writeStartElement("list")

			writer.writeAttribute("type", "unordered")

			writer.writeCharacters("\n")

			structureMetadataColumnList.each { att ->

				writer.writeStartElement("item")

				writer.writeCharacters(att+" : "+record.get(att).replaceAll("\n", ";"))

				writer.writeEndElement() // item

				writer.writeCharacters("\n")

			}

			writer.writeEndElement() // list

			writer.writeEndElement() // metadata

			writer.writeCharacters("\n")

			for (String att : structureContentColumnList) {

				if (att in prefixesColumnTypeList) {

					writePrefixTextContent(record, att, writer)

				} else if (att in listColumnTypeList) {

					writeListTextContent(record, att, writer)

				} else {

					writeTextContent(record, att, writer)

				}

			}

			writer.writeEndElement() // struct

			writer.writeCharacters("\n") // simple XML formating

		}

		writer.writeEndElement() // text

		writer.writeEndElement() // TEI

		writer.close()

		output.close()

		reader.close()

	} else {

		// error

		println "ERROR: '$id' text group with  no text line"

cpb.done()

	def value = record.get(att).replaceAll(" ++", " ")

	if (value.length() <= 1) return

	found = false

	for (String s : value.split("\n")) {

		s = s.trim()

		if (s.length() == 0) continue;

		if (s.startsWith("-")) s = s.substring(1).trim();

		if (s.length() == 0) continue;

		found = true

		writer.writeCharacters("\t")

		writer.writeStartElement("p")

		writer.writeAttribute("rend", "list")

		writer.writeCharacters(s)

		writer.writeEndElement() // p

		writer.writeCharacters("\n")

	}

	if (!found) {

		println "Warning: '$att' list not found in '$value'"

		writer.writeCharacters("\t")

		writer.writeStartElement("p")

		writer.writeAttribute("rend", "no-list")

		writer.writeCharacters(value)

		writer.writeEndElement() // p

		writer.writeCharacters("\n")

	}

def writeTextContent(def record, def att, def writer) {

	def value = record.get(att).replaceAll(" ++", " ")

	if (value.length() == 0) return

	writer.writeStartElement("head")

	writer.writeStartElement("hi")

	writer.writeCharacters(att)

	writer.writeEndElement() // hi

	writer.writeEndElement() // head

	writer.writeStartElement(AsciiUtils.buildAttributeId(att));

	writer.writeAttribute("type", att.trim())

	if (record.get(textIDColumn) != null) {

		writer.writeAttribute("id", record.get(textIDColumn))

	}

	writer.writeCharacters("\n") // simple XML formating

	writer.writeCharacters(value) // get textColumnList content

	writer.writeCharacters("\n") // simple XML formating)

	def value = record.get(att).replaceAll(" ++", " ")

	if (value.length() <= 1) return

	writer.writeStartElement("list")

	writer.writeAttribute("rend", att.trim())

	writer.writeAttribute("type", "unordered")

	writer.writeCharacters("\n")

	found = false

	value.findAll( /(?s)[A-Z]{3}:? *([^;\n]+?) +[;\n]/ ).each { desc ->

		found = true

		writer.writeCharacters("\t")

		writer.writeStartElement("item")

		matches = (desc =~ /(?s)([A-Z]{3}):? *([^;\n]+?) +[;\n]/)

		writer.writeAttribute("type", matches[0][1])

		writer.writeStartElement("span")

		writer.writeCharacters(matches[0][1]+" ")

		writer.writeEndElement() // span

		writer.writeCharacters(matches[0][2])

		writer.writeEndElement() // item

		writer.writeCharacters("\n")

	}

	if (!found) {

		println "Warning: '$att' prefixes not found in: '$value'"

		writer.writeCharacters(value)

	}

	writer.writeEndElement() // list

	writer.writeCharacters("\n")

	String value = record.get(att)

	if (value.length() == 0) return

	String att_normalized = AsciiUtils.buildAttributeId(att)

	matches = (value =~ /([0-9]{2})\/([0-9]{2})\/([0-9]{4})/)

	writer.writeAttribute(att_normalized+"-jour", matches[0][1])

	writer.writeAttribute(att_normalized+"-joursemaine", new java.text.SimpleDateFormat('EEEE').format(Date.parse("dd/MM/yyyy", value)))

	writer.writeAttribute(att_normalized+"-mois", matches[0][2])

	writer.writeAttribute(att_normalized+"-annee", matches[0][3])

	writer.writeAttribute(att_normalized+"-tri", matches[0][3]+"-"+matches[0][2]+"-"+matches[0][1])

//			println "NEW HTML: "+outfile

//			println "TAGS: "+tags