Fix escaping by using RFC compliant parser (#496)

By configuring the CSVReader with an RFC-compliant parser the escaping is fixed. - update opencsv dependency to version 5.9 - add test
metafacture · Oct 18, 2024 · 6f39fcc · 6f39fcc
1 parent 71c367c
commit 6f39fcc
Show file tree

Hide file tree

Showing 3 changed files with 40 additions and 6 deletions.
diff --git a/metafacture-csv/build.gradle b/metafacture-csv/build.gradle
@@ -19,7 +19,7 @@ description = 'Modules for processing comma-separated values'
 
 dependencies {
   api project(':metafacture-framework')
-  implementation 'com.opencsv:opencsv:3.10'
+  implementation 'com.opencsv:opencsv:5.9'
   testImplementation "junit:junit:${versions.junit}"
   testImplementation "org.mockito:mockito-core:${versions.mockito}"
 }
diff --git a/metafacture-csv/src/main/java/org/metafacture/csv/CsvDecoder.java b/metafacture-csv/src/main/java/org/metafacture/csv/CsvDecoder.java
@@ -1,5 +1,5 @@
 /*
- * Copyright 2013, 2014 Deutsche Nationalbibliothek
+ * Copyright 2013-2024 Deutsche Nationalbibliothek and hbz
  *
  * Licensed under the Apache License, Version 2.0 the "License";
  * you may not use this file except in compliance with the License.
@@ -24,6 +24,10 @@
 import org.metafacture.framework.helpers.DefaultObjectPipe;
 
 import com.opencsv.CSVReader;
+import com.opencsv.CSVReaderBuilder;
+import com.opencsv.RFC4180Parser;
+import com.opencsv.RFC4180ParserBuilder;
+import com.opencsv.exceptions.CsvException;
 
 import java.io.IOException;
 import java.io.StringReader;
@@ -48,6 +52,7 @@ public final class CsvDecoder extends DefaultObjectPipe<String, StreamReceiver>
     private String[] header = new String[0];
     private int count;
     private boolean hasHeader;
+    private RFC4180Parser parser;
 
     /**
      * Creates an instance of {@link CsvDecoder} with a given separator.
@@ -56,6 +61,7 @@ public final class CsvDecoder extends DefaultObjectPipe<String, StreamReceiver>
      */
     public CsvDecoder(final String separator) {
         this.separator = separator.charAt(0);
+        initializeCsvParser();
     }
 
     /**
@@ -65,13 +71,21 @@ public CsvDecoder(final String separator) {
      */
     public CsvDecoder(final char separator) {
         this.separator = separator;
+        initializeCsvParser();
     }
 
     /**
      * Creates an instance of {@link CsvDecoder}. The default separator is
      * {@value #DEFAULT_SEP}.
      */
     public CsvDecoder() {
+        initializeCsvParser();
+    }
+
+    private void initializeCsvParser() {
+        this.parser = new RFC4180ParserBuilder()
+            .withSeparator(separator)
+            .build();
     }
 
     @Override
@@ -105,18 +119,19 @@ else if (parts.length == header.length) {
         }
     }
 
-    private String[] parseCsv(final String string) {
+    private String[] parseCsv(final String csv) {
         String[] parts = new String[0];
         try {
-            final CSVReader reader = new CSVReader(new StringReader(string),
-                    separator);
+            final CSVReader reader = new CSVReaderBuilder(new StringReader(csv))
+                .withCSVParser(parser)
+                .build();
             final List<String[]> lines = reader.readAll();
             if (lines.size() > 0) {
                 parts = lines.get(0);
             }
             reader.close();
         }
-        catch (final IOException e) {
+        catch (final IOException | CsvException e) {
             e.printStackTrace();
         }
         return parts;
@@ -139,5 +154,6 @@ public void setHasHeader(final boolean hasHeader) {
      */
     public void setSeparator(final String separator) {
         this.separator = separator.charAt(0);
+        initializeCsvParser();
     }
 }
diff --git a/metafacture-csv/src/test/java/org/metafacture/csv/CsvDecoderTest.java b/metafacture-csv/src/test/java/org/metafacture/csv/CsvDecoderTest.java
@@ -89,4 +89,22 @@ public void testTabSeparated() {
         ordered.verify(receiver).endRecord();
     }
 
+    /**
+     * In: "a","b\t","c\\t","\","\cd\"
+     * Out: a, b	, c\\t, \, \cd\
+     */
+    @Test
+    public void issue496_escaping() {
+        decoder.setHasHeader(false);
+        decoder.process("\"a\",\"b\t\",\"c\\t\",\"\\\",\"\\cd\\\"");
+        final InOrder ordered = inOrder(receiver);
+        ordered.verify(receiver).startRecord("1");
+        ordered.verify(receiver).literal("0", "a");
+        ordered.verify(receiver).literal("1", "b\t");
+        ordered.verify(receiver).literal("2", "c\\t");
+        ordered.verify(receiver).literal("3", "\\");
+        ordered.verify(receiver).literal("4", "\\cd\\");
+        ordered.verify(receiver).endRecord();
+    }
+
 }