issue #137: allowableRecords

pkiraly · Jun 21, 2022 · 0bcf11b · 0bcf11b
1 parent d294a29
commit 0bcf11b
Show file tree

Hide file tree

Showing 7 changed files with 91 additions and 3 deletions.
diff --git a/src/main/java/de/gwdg/metadataqa/marc/cli/parameters/CommonParameters.java b/src/main/java/de/gwdg/metadataqa/marc/cli/parameters/CommonParameters.java
@@ -118,6 +118,7 @@ public CommonParameters(String[] arguments)  throws ParseException {
     trimId = cmd.hasOption("trimId");
     readIgnorableFields();
     readIgnorableRecords();
+    readAllowableRecords();
     readDefaultEncoding();
     readAlephseqLineType();
     readPicaIdField();
@@ -165,6 +166,7 @@ private void readIgnorableRecords() {
   }
 
   private void readAllowableRecords() {
+
     String allowableRecords = cmd.hasOption("allowableRecords") ? cmd.getOptionValue("allowableRecords") : "";
     setRecordFilter(allowableRecords);
   }
@@ -448,6 +450,9 @@ public void setRecordIgnorator(String ignorableRecords) {
     this.recordIgnorator = RecordIgnoratorFactory.create(schemaType, ignorableRecords.trim());
   }
 
+  public RecordFilter getRecordFilter() {
+    return recordFilter;
+  }
 
   public void setRecordFilter(String allowableRecords) {
     this.recordFilter = RecordFilterFactory.create(schemaType, allowableRecords.trim());

diff --git a/src/main/java/de/gwdg/metadataqa/marc/cli/utils/ignorablerecords/CriteriumPica.java b/src/main/java/de/gwdg/metadataqa/marc/cli/utils/ignorablerecords/CriteriumPica.java
@@ -127,4 +127,12 @@ private boolean subfieldEndsWith(List<MarcSubfield> instances) {
     return false;
   }
 
+  @Override
+  public String toString() {
+    return "CriteriumPica{" +
+      "path=" + path.getPath() +
+      ", operator=" + operator +
+      ", value='" + value + '\'' +
+      '}';
+  }
 }
diff --git a/src/main/java/de/gwdg/metadataqa/marc/cli/utils/ignorablerecords/Marc21Filter.java b/src/main/java/de/gwdg/metadataqa/marc/cli/utils/ignorablerecords/Marc21Filter.java
@@ -45,9 +45,6 @@ protected DataField parseField(String field) {
   }
 
   protected boolean met(MarcRecord marcRecord) {
-    if (isEmpty())
-      return false;
-
     for (DataField condition : conditions) {
       List<DataField> recordFields = marcRecord.getDatafield(condition.getTag());
       if (recordFields == null || recordFields.isEmpty())

diff --git a/src/main/java/de/gwdg/metadataqa/marc/cli/utils/ignorablerecords/RecordFilterMarc21.java b/src/main/java/de/gwdg/metadataqa/marc/cli/utils/ignorablerecords/RecordFilterMarc21.java
@@ -18,6 +18,14 @@ public boolean isEmpty() {
 
   @Override
   public boolean isAllowable(MarcRecord marcRecord) {
+    if (isEmpty())
+      return true;
+
     return met(marcRecord);
   }
+
+  @Override
+  public String toString() {
+    return isEmpty() ? "" : conditions.toString();
+  }
 }
diff --git a/src/main/java/de/gwdg/metadataqa/marc/cli/utils/ignorablerecords/RecordFilterPica.java b/src/main/java/de/gwdg/metadataqa/marc/cli/utils/ignorablerecords/RecordFilterPica.java
@@ -18,6 +18,9 @@ public boolean isEmpty() {
 
   @Override
   public boolean isAllowable(MarcRecord marcRecord) {
+    if (isEmpty())
+      return true;
+
     for (CriteriumPica criterium : criteria) {
       boolean passed = criterium.met(marcRecord);
       if (passed)

diff --git a/src/main/java/de/gwdg/metadataqa/marc/cli/utils/ignorablerecords/RecordIgnoratorMarc21.java b/src/main/java/de/gwdg/metadataqa/marc/cli/utils/ignorablerecords/RecordIgnoratorMarc21.java
@@ -19,6 +19,9 @@ public boolean isEmpty() {
 
   @Override
   public boolean isIgnorable(MarcRecord marcRecord) {
+    if (isEmpty())
+      return false;
+
     return met(marcRecord);
   }
 

diff --git a/src/test/java/de/gwdg/metadataqa/marc/cli/parameters/CommonParametersTest.java b/src/test/java/de/gwdg/metadataqa/marc/cli/parameters/CommonParametersTest.java
@@ -1,5 +1,9 @@
 package de.gwdg.metadataqa.marc.cli.parameters;
 
+import de.gwdg.metadataqa.marc.cli.utils.ignorablerecords.RecordFilterMarc21;
+import de.gwdg.metadataqa.marc.cli.utils.ignorablerecords.RecordFilterPica;
+import de.gwdg.metadataqa.marc.cli.utils.ignorablerecords.RecordIgnoratorMarc21;
+import de.gwdg.metadataqa.marc.cli.utils.ignorablerecords.RecordIgnoratorPica;
 import de.gwdg.metadataqa.marc.definition.DataSource;
 import de.gwdg.metadataqa.marc.definition.MarcFormat;
 import de.gwdg.metadataqa.marc.definition.bibliographic.SchemaType;
@@ -317,4 +321,64 @@ public void getPicaRecordType_set() {
     assertEquals("003$d", parameters.getPicaRecordTypeField());
   }
 
+  @Test
+  public void getRecordIgnorator_pica() {
+    String[] arguments = new String[]{"--schemaType", "PICA", "--ignorableRecords", "[email protected] !~ '^L'"};
+    CommonParameters parameters = null;
+    try {
+      parameters = new CommonParameters(arguments);
+    } catch (ParseException e) {
+      logger.log(Level.WARNING, "error in schemaType()", e);
+    }
+    assertEquals("RecordIgnoratorPica", parameters.getRecordIgnorator().getClass().getSimpleName());
+    assertEquals(1, ((RecordIgnoratorPica)parameters.getRecordIgnorator()).getCriteria().size());
+    assertEquals("CriteriumPica{[email protected], operator=NOT_MATCH, value='^L'}",
+      ((RecordIgnoratorPica)parameters.getRecordIgnorator()).getCriteria().get(0).toString());
+  }
+
+  @Test
+  public void getRecordIgnorator_marc21() {
+    String[] arguments = new String[]{"--schemaType", "MARC21", "--ignorableRecords", "STA$a=SUPPRESSED"};
+    CommonParameters parameters = null;
+    try {
+      parameters = new CommonParameters(arguments);
+    } catch (ParseException e) {
+      logger.log(Level.WARNING, "error in schemaType()", e);
+    }
+    assertEquals("RecordIgnoratorMarc21", parameters.getRecordIgnorator().getClass().getSimpleName());
+    assertEquals(
+      "[DataField{STA, ind1=' ', ind2=' ', subfields=[MarcSubfield{code='a', value='SUPPRESSED'}]}]",
+      ((RecordIgnoratorMarc21)parameters.getRecordIgnorator()).toString());
+  }
+
+  @Test
+  public void getRecordFilter_pica() {
+    String[] arguments = new String[]{"--schemaType", "PICA", "--allowableRecords", "[email protected] !~ '^L'"};
+    CommonParameters parameters = null;
+    try {
+      parameters = new CommonParameters(arguments);
+    } catch (ParseException e) {
+      logger.log(Level.WARNING, "error in schemaType()", e);
+    }
+    assertEquals("RecordFilterPica", parameters.getRecordFilter().getClass().getSimpleName());
+    assertEquals(1, ((RecordFilterPica)parameters.getRecordFilter()).getCriteria().size());
+    assertEquals("CriteriumPica{[email protected], operator=NOT_MATCH, value='^L'}",
+      ((RecordFilterPica)parameters.getRecordFilter()).getCriteria().get(0).toString());
+  }
+
+  @Test
+  public void getRecordFilter_marc21() {
+    String[] arguments = new String[]{"--schemaType", "MARC21", "--allowableRecords", "STA$a=SUPPRESSED"};
+    CommonParameters parameters = null;
+    try {
+      parameters = new CommonParameters(arguments);
+    } catch (ParseException e) {
+      logger.log(Level.WARNING, "error in schemaType()", e);
+    }
+    assertEquals("RecordFilterMarc21", parameters.getRecordFilter().getClass().getSimpleName());
+    assertEquals(
+      "[DataField{STA, ind1=' ', ind2=' ', subfields=[MarcSubfield{code='a', value='SUPPRESSED'}]}]",
+      ((RecordFilterMarc21)parameters.getRecordFilter()).toString());
+  }
+
 }