issue #145: parsing PICA path

pkiraly · Jun 21, 2022 · f1b7f3f · f1b7f3f
1 parent 1108a26
commit f1b7f3f
Show file tree

Hide file tree

Showing 4 changed files with 75 additions and 34 deletions.
diff --git a/src/main/java/de/gwdg/metadataqa/marc/utils/pica/path/PicaPath.java b/src/main/java/de/gwdg/metadataqa/marc/utils/pica/path/PicaPath.java
@@ -2,27 +2,20 @@
 
 public class PicaPath {
 
-  public enum SubfieldType {
-    SINGLE,
-    MULTI,
-    ALL
-  }
 
 
   private String path;
   private String tag = null;
   private String xtag = null;
   private Occurrence occurrence = null;
-  private String subfields = null;
-  private SubfieldType subfieldType = null;
+  private Subfields subfields = null;
 
-  public PicaPath(String path, String tag, String xtag, Occurrence occurrence, String subfields, SubfieldType subfieldType) {
+  public PicaPath(String path, String tag, String xtag, Occurrence occurrence, Subfields subfields) {
     this.path = path;
     this.tag = tag;
     this.xtag = xtag;
     this.occurrence = occurrence;
     this.subfields = subfields;
-    this.subfieldType = subfieldType;
   }
 
   public String getPath() {
@@ -41,11 +34,7 @@ public Occurrence getOccurrence() {
     return occurrence;
   }
 
-  public String getSubfields() {
+  public Subfields getSubfields() {
     return subfields;
   }
-
-  public SubfieldType getSubfieldType() {
-    return subfieldType;
-  }
 }
diff --git a/src/main/java/de/gwdg/metadataqa/marc/utils/pica/path/PicaPathParser.java b/src/main/java/de/gwdg/metadataqa/marc/utils/pica/path/PicaPathParser.java
@@ -9,14 +9,12 @@ public class PicaPathParser {
   private static final Pattern XTAG = Pattern.compile("^(2[0-9\\.][0-9\\.][A-Z@\\.]x\\d+)");
   private static final Pattern OCCURENCE = Pattern.compile("^/((\\d+)-(\\d+)|(\\d{1,3})|(\\*))");
   private static final Pattern SUBFIELDS = Pattern.compile("^[\\$.]?(([A-Za-z0-9]+)|(\\*))");
-
   public static PicaPath parse(String input) {
     String path = input;
     String tag = null;
     String xtag = null;
     Occurrence occurrence = null;
-    String subfields = null;
-    PicaPath.SubfieldType subfieldType = null;
+    Subfields subfields = null;
 
     String remainder = null;
     Matcher m = null;
@@ -56,18 +54,23 @@ public static PicaPath parse(String input) {
     if (remainder != null) {
       m = SUBFIELDS.matcher(remainder);
       if (m.find()) {
-        subfields = m.group(1);
+        String subfieldsRaw = m.group(1);
+        Subfields.Type subfieldType = null;
         if (m.group(2) != null) {
-          subfieldType = m.group(2).length() == 1 ? PicaPath.SubfieldType.SINGLE : PicaPath.SubfieldType.MULTI;
-        } else if (m.group(3) != null)
-          subfieldType = PicaPath.SubfieldType.ALL;
+          subfieldType = m.group(2).length() == 1 ? Subfields.Type.SINGLE : Subfields.Type.MULTI;
+        } else if (m.group(3) != null) {
+          subfieldType = Subfields.Type.ALL;
+        }
+        if (subfieldType == null)
+          throw new IllegalArgumentException("The input does not fit to rules: " + input);
+        subfields = new Subfields(subfieldType, subfieldsRaw);
         remainder = m.end() < remainder.length() ? remainder.substring(m.end()) : null;
       }
     }
 
     if (remainder != null)
       throw new IllegalArgumentException("The input does not fit to rules: " + input);
 
-    return new PicaPath(path, tag, xtag, occurrence, subfields, subfieldType);
+    return new PicaPath(path, tag, xtag, occurrence, subfields);
   }
 }
diff --git a/src/main/java/de/gwdg/metadataqa/marc/utils/pica/path/Subfields.java b/src/main/java/de/gwdg/metadataqa/marc/utils/pica/path/Subfields.java
@@ -0,0 +1,34 @@
+package de.gwdg.metadataqa.marc.utils.pica.path;
+
+import java.util.Arrays;
+import java.util.List;
+
+public class Subfields {
+  public enum Type {
+    SINGLE,
+    MULTI,
+    ALL
+  }
+
+  private Type type;
+  private String input;
+  private List<String> codes;
+
+  public Subfields(Type type, String input) {
+    this.type = type;
+    this.input = input;
+    codes = Arrays.asList(input.split(""));
+  }
+
+  public Type getType() {
+    return type;
+  }
+
+  public String getInput() {
+    return input;
+  }
+
+  public List<String> getCodes() {
+    return codes;
+  }
+}
diff --git a/src/test/java/de/gwdg/metadataqa/marc/utils/pica/path/PicaPathParserTest.java b/src/test/java/de/gwdg/metadataqa/marc/utils/pica/path/PicaPathParserTest.java
@@ -2,7 +2,10 @@
 
 import org.junit.Test;
 
+import java.util.Arrays;
+
 import static org.junit.Assert.assertEquals;
+import static org.junit.Assert.assertNull;
 
 public class PicaPathParserTest {
 
@@ -12,6 +15,7 @@ public void tag() {
     assertEquals("003@", path.getPath());
     assertEquals("003@", path.getTag());
     assertEquals(null, path.getXtag());
+    assertNull(path.getSubfields());
   }
 
   @Test
@@ -20,6 +24,7 @@ public void xtag() {
     assertEquals("203@x1", path.getPath());
     assertEquals(null, path.getTag());
     assertEquals("203@x1", path.getXtag());
+    assertNull(path.getSubfields());
   }
 
   @Test(expected = IllegalArgumentException.class)
@@ -28,6 +33,7 @@ public void illegalTag() {
     assertEquals("203@x1", path.getPath());
     assertEquals(null, path.getTag());
     assertEquals("203@x1", path.getXtag());
+    assertNull(path.getSubfields());
   }
 
   @Test
@@ -39,6 +45,7 @@ public void occurenceNumber() {
     assertEquals(Occurrence.Type.SINGLE, path.getOccurrence().getType());
     assertEquals(2, path.getOccurrence().getStart().intValue());
     assertEquals(null, path.getOccurrence().getEnd());
+    assertNull(path.getSubfields());
   }
 
   @Test
@@ -50,6 +57,7 @@ public void occurenceRange() {
     assertEquals(Occurrence.Type.RANGE, path.getOccurrence().getType());
     assertEquals(2, path.getOccurrence().getStart().intValue());
     assertEquals(3, path.getOccurrence().getEnd().intValue());
+    assertNull(path.getSubfields());
   }
 
   @Test
@@ -59,6 +67,7 @@ public void occurenceAsteriks() {
     assertEquals("003@", path.getTag());
     assertEquals(null, path.getXtag());
     assertEquals(Occurrence.Type.ALL, path.getOccurrence().getType());
+    assertNull(path.getSubfields());
   }
 
   @Test
@@ -68,8 +77,9 @@ public void occurenceAsteriks_withsubfield() {
     assertEquals("003@", path.getTag());
     assertEquals(null, path.getXtag());
     assertEquals(Occurrence.Type.ALL, path.getOccurrence().getType());
-    assertEquals(PicaPath.SubfieldType.ALL, path.getSubfieldType());
-    assertEquals("*", path.getSubfields());
+    assertEquals(Subfields.Type.ALL, path.getSubfields().getType());
+    assertEquals("*", path.getSubfields().getInput());
+    assertEquals(Arrays.asList("*"), path.getSubfields().getCodes());
   }
 
   @Test
@@ -79,8 +89,9 @@ public void subfieldsTag() {
     assertEquals("003@", path.getTag());
     assertEquals(null, path.getXtag());
     assertEquals(null, path.getOccurrence());
-    assertEquals(PicaPath.SubfieldType.SINGLE, path.getSubfieldType());
-    assertEquals("a", path.getSubfields());
+    assertEquals(Subfields.Type.SINGLE, path.getSubfields().getType());
+    assertEquals("a", path.getSubfields().getInput());
+    assertEquals(Arrays.asList("a"), path.getSubfields().getCodes());
   }
 
   @Test
@@ -90,8 +101,9 @@ public void subfieldsTag_multiple() {
     assertEquals("003@", path.getTag());
     assertEquals(null, path.getXtag());
     assertEquals(null, path.getOccurrence());
-    assertEquals(PicaPath.SubfieldType.MULTI, path.getSubfieldType());
-    assertEquals("abc", path.getSubfields());
+    assertEquals(Subfields.Type.MULTI, path.getSubfields().getType());
+    assertEquals("abc", path.getSubfields().getInput());
+    assertEquals(Arrays.asList("a", "b", "c"), path.getSubfields().getCodes());
   }
 
   @Test
@@ -101,8 +113,9 @@ public void subfieldsTag_multiple_mixed() {
     assertEquals("003@", path.getTag());
     assertEquals(null, path.getXtag());
     assertEquals(null, path.getOccurrence());
-    assertEquals(PicaPath.SubfieldType.MULTI, path.getSubfieldType());
-    assertEquals("aBc", path.getSubfields());
+    assertEquals(Subfields.Type.MULTI, path.getSubfields().getType());
+    assertEquals("aBc", path.getSubfields().getInput());
+    assertEquals(Arrays.asList("a", "B", "c"), path.getSubfields().getCodes());
   }
 
   @Test
@@ -112,8 +125,9 @@ public void subfieldsTag_asteriks() {
     assertEquals("003@", path.getTag());
     assertEquals(null, path.getXtag());
     assertEquals(null, path.getOccurrence());
-    assertEquals(PicaPath.SubfieldType.ALL, path.getSubfieldType());
-    assertEquals("*", path.getSubfields());
+    assertEquals(Subfields.Type.ALL, path.getSubfields().getType());
+    assertEquals("*", path.getSubfields().getInput());
+    assertEquals(Arrays.asList("*"), path.getSubfields().getCodes());
   }
 
   @Test
@@ -123,7 +137,8 @@ public void subfieldsTag_asteriks_nodollar() {
     assertEquals("003@", path.getTag());
     assertEquals(null, path.getXtag());
     assertEquals(null, path.getOccurrence());
-    assertEquals(PicaPath.SubfieldType.ALL, path.getSubfieldType());
-    assertEquals("*", path.getSubfields());
+    assertEquals(Subfields.Type.ALL, path.getSubfields().getType());
+    assertEquals("*", path.getSubfields().getInput());
+    assertEquals(Arrays.asList("*"), path.getSubfields().getCodes());
   }
 }