From eb72b380a5ece6d93ce4b65b206b65dcbd6d27c9 Mon Sep 17 00:00:00 2001
From: lkrsnik <krsnik.luka92@gmail.com>
Date: Thu, 20 Jun 2024 12:13:27 +0200
Subject: [PATCH] Fixed taxonomy processing for KOST + Added ignoring of
 certain files in KOST

---
 build_instructions.md                 |  1 +
 src/main/java/alg/XML_processing.java | 11 ++++++++++-
 2 files changed, 11 insertions(+), 1 deletion(-)

diff --git a/build_instructions.md b/build_instructions.md
index 930ec08..405942e 100644
--- a/build_instructions.md
+++ b/build_instructions.md
@@ -3,6 +3,7 @@
 ```shell
 mvn package
 ```
+- results are in shade folder
 
 # Build executable using Launch4j
 - Install Java on Windows
diff --git a/src/main/java/alg/XML_processing.java b/src/main/java/alg/XML_processing.java
index 26ed8e5..a6e9e8e 100755
--- a/src/main/java/alg/XML_processing.java
+++ b/src/main/java/alg/XML_processing.java
@@ -501,6 +501,10 @@ public class XML_processing {
                         // this toggle is true when we're inside a header (next block of code executes)
                         // and false when we're not (skip reading unnecessary attributes)
                         insideHeader = true;
+                    } else if (corpusType == CorpusType.KOST && elementName.equals("standOff") ||
+                            corpusType == CorpusType.KOST && elementName.equals("TEI")
+                    ) {
+                        return resultTaxonomy;
                     }
 
                     if (insideHeader) {
@@ -849,6 +853,10 @@ public class XML_processing {
 
                         } else if (qName.equals("text")){
                             taxonomyMatch = true;
+                        } else if (stats.getCorpus().getCorpusType() == CorpusType.KOST && qName.equals("standOff") ||
+                                stats.getCorpus().getCorpusType() == CorpusType.KOST && qName.equals("TEI")
+                        ) {
+                            return true;
                         }
                         break;
 
@@ -911,7 +919,8 @@ public class XML_processing {
                         }
                         // fallback
                         else if (endElement.getName().getLocalPart().equalsIgnoreCase("div") &&
-                                stats.getCorpus().getCorpusType() == CorpusType.SSJ500K) {
+                                (stats.getCorpus().getCorpusType() == CorpusType.SSJ500K ||
+                                        stats.getCorpus().getCorpusType() == CorpusType.KOST)) {
                             // join corpus and stats
                             fj(corpus, stats);
                             corpus.clear();