changed id for arendt specific directory walk

Author: jbiermann

pom.xml

@@ -3,7 +3,7 @@
 	<modelVersion>4.0.0</modelVersion>
 	<groupId>de.sub.goettingen.arendt</groupId>
 	<artifactId>ocr2tei</artifactId>
-	<version>1.1-SNAPSHOT</version>
+	<version>1.1.1-SNAPSHOT</version>
 	<name>OCR2XML</name>
 	<dependencies>
 		<dependency>

src/main/java/de/sub/goettingen/arendt/ocrmapping/Cli.java

@@ -173,7 +173,7 @@ public FileVisitResult visitFile(Path file, BasicFileAttributes attrs) throws IO
 						// with file IV-NUMBER_TEI_Originalfilename.xml
 						if (Cli.commandDir.specificWalk) {
 							String newTEIDirname = file.getParent().getFileName().toString().replace("_xml", "_TEI");
-							documentId = "Volume-"+newTEIDirname;
+							documentId = newTEIDirname.substring(0, newTEIDirname.length()-4);
 							Path newTEIDir= file.getParent().resolveSibling(newTEIDirname);
 
 							String teiFilename = newTEIDirname +"_"+ file.getFileName().toString().replace("ocr_", "");

src/main/java/de/sub/goettingen/arendt/ocrmapping/ReadXML.java

@@ -59,16 +59,7 @@ public static void main(String args[]) throws Exception {
 		//TEI tei = convert2Tei("/tmp/ocr_1504081850380.xml", true);
 		TEI tei = convert2Tei("/Users/Johannes/Documents/Arendt/oc2tei/cloud/Hannah-Arendt/Scans/Band-XIV-NL/XIV-NL-02/ocr-03/Arendt-XIV-NL-02-03_xml/ocr_1504604413610.xml", 
 				true, "Lokaltest", true);
-		// distinct-values(//formatting/@lang)
-		//https://stackoverflow.com/questions/8607464/how-to-count-each-instance-of-distinct-values-in-xquery
-		/*
-		 * 
-for $v in distinct-values(//formatting/@lang)
-	return
-	($v, count(index-of(//formatting/@lang, $v)))
 
-http://www.baeldung.com/java-xpath
-		 */
 		OutputXML.writeXMLOutputToStdout(tei);
 	}
 
@@ -116,7 +107,7 @@ public static TEI convert2Tei (String inputFilePath,
 			// this should exactly represent that file. We can't get the original filename from the XML
 			// also if we have a document Id (CLI) option we add this as prefix, too.
 			if (documentId != null)
-				oPb.setN(documentId + String.format("%06d", pageNumber));
+				oPb.setN(documentId + "-" + String.format("%08d", pageNumber));
 			else
 				oPb.setN(String.format("%06d", pageNumber));