PICA: fixing reading of gzipped files #251

pkiraly · Apr 4, 2023 · 3808b1e · 3808b1e
1 parent b0e8d73
commit 3808b1e
Show file tree

Hide file tree

Showing 4 changed files with 5 additions and 6 deletions.
diff --git a/src/main/java/de/gwdg/metadataqa/marc/utils/QAMarcReaderFactory.java b/src/main/java/de/gwdg/metadataqa/marc/utils/QAMarcReaderFactory.java
@@ -94,6 +94,7 @@ private MarcReader getPicaNormalizedStreamReader(InputStream stream, CommonParam
     String encoding = (parameters != null && StringUtils.isNotBlank(parameters.getDefaultEncoding())) ? parameters.getDefaultEncoding() : "UTF-8";
     PicaReader reader = new PicaNormalizedReader(stream, encoding);
     configurePicaReader(reader, parameters);
+    reader.parseIdField();
     return reader;
   }
 

diff --git a/src/main/java/de/gwdg/metadataqa/marc/utils/pica/reader/PicaNormalizedReader.java b/src/main/java/de/gwdg/metadataqa/marc/utils/pica/reader/PicaNormalizedReader.java
@@ -37,7 +37,7 @@ public PicaNormalizedReader(String fileName) {
   public PicaNormalizedReader(InputStream stream, String encoding) {
     try {
       bufferedReader = new BufferedReader(new InputStreamReader(stream, encoding));
-      parseIdField();
+      // parseIdField();
     } catch (IOException e) {
       logger.log(Level.WARNING, "error in PicaReader()", e);
     }

diff --git a/src/main/java/de/gwdg/metadataqa/marc/utils/pica/reader/PicaReader.java b/src/main/java/de/gwdg/metadataqa/marc/utils/pica/reader/PicaReader.java
@@ -11,7 +11,7 @@
 public abstract class PicaReader implements MarcReader {
 
   protected String idField = "003@$0";
-  protected String subfieldSeparator = null; // "$";
+  protected String subfieldSeparator = "$";
   protected String idTag = "003@";
   protected String idCode = "0";
   protected boolean parsed = false;
@@ -24,7 +24,7 @@ public abstract class PicaReader implements MarcReader {
   @Override
   public abstract Record next();
 
-  protected void parseIdField() {
+  public void parseIdField() {
     String[] parts = idField.split(Pattern.quote(subfieldSeparator));
     idTag = parts[0];
     idCode = parts[1];

diff --git a/src/test/java/de/gwdg/metadataqa/marc/utils/pica/reader/PicaReaderTest.java b/src/test/java/de/gwdg/metadataqa/marc/utils/pica/reader/PicaReaderTest.java
@@ -1,8 +1,6 @@
 package de.gwdg.metadataqa.marc.utils.pica.reader;
 
 import de.gwdg.metadataqa.marc.cli.CliTestUtils;
-import de.gwdg.metadataqa.marc.utils.pica.reader.PicaReader;
-import de.gwdg.metadataqa.marc.utils.pica.reader.PicaXmlReader;
 import org.junit.Before;
 import org.junit.Test;
 
@@ -53,7 +51,7 @@ public void getIdField() {
 
   @Test
   public void getSubfieldSeparator() {
-    assertNull(reader.getSubfieldSeparator());
+    assertNotNull(reader.getSubfieldSeparator());
     reader.setSubfieldSeparator("$");
     assertEquals("$", reader.getSubfieldSeparator());
   }