Merge pull request #100 from melissalinkert/readfile-encoding

Add DataTools.readFile signature that accepts an encoding name

Author: sbesson

src/main/java/loci/common/DataTools.java

@@ -74,11 +74,12 @@ private DataTools() { }
    * @param id name of the file to read
    *           this can be any name supported by Location,
    *           not necessarily a file on disk
+   * @param encoding the file encoding, e.g. "UTF-8"
    * @return the complete contents of the specified file
    * @throws IOException if the file cannot be read or is larger than 2GB
    * @see Location#getMappedId(String)
    */
-  public static String readFile(String id) throws IOException {
+  public static String readFile(String id, String encoding) throws IOException {
     RandomAccessInputStream in = new RandomAccessInputStream(id);
     long idLen = in.length();
     if (idLen > Integer.MAX_VALUE) {
@@ -88,11 +89,25 @@ public static String readFile(String id) throws IOException {
     byte[] b = new byte[len];
     in.readFully(b);
     in.close();
-    String data = new String(b, Constants.ENCODING);
+    String data = new String(b, encoding);
     b = null;
     return data;
   }
 
+  /**
+   * Reads the contents of the given file into a string.
+   *
+   * @param id name of the file to read
+   *           this can be any name supported by Location,
+   *           not necessarily a file on disk
+   * @return the complete contents of the specified file
+   * @throws IOException if the file cannot be read or is larger than 2GB
+   * @see Location#getMappedId(String)
+   */
+  public static String readFile(String id) throws IOException {
+    return readFile(id, Constants.ENCODING);
+  }
+
   // -- Word decoding - bytes to primitive types --
 
   /**

src/test/java/loci/common/utests/DataToolsTest.java

@@ -32,10 +32,13 @@
 
 package loci.common.utests;
 
+import loci.common.ByteArrayHandle;
 import loci.common.DataTools;
+import loci.common.Location;
 import java.util.Locale;
 
 import static org.testng.AssertJUnit.assertEquals;
+import static org.testng.AssertJUnit.assertFalse;
 import static org.testng.AssertJUnit.fail;
 import org.testng.annotations.DataProvider;
 import org.testng.annotations.Test;
@@ -236,4 +239,59 @@ public void testParseDouble(String locale) {
     public void testThreadSafety() {
       assertEquals(DataTools.parseDouble("1.0"), 1.0d);
     }
+
+
+  @Test
+  public void testReadFileDefaultEncoding() {
+    String src = "µ€œẞ";
+    String tmpFile = "utf-test.txt";
+    try {
+      byte[] bytes = src.getBytes("UTF-8");
+      Location.mapFile(tmpFile, new ByteArrayHandle(bytes));
+      String result = DataTools.readFile(tmpFile);
+      assertEquals(src, result);
+    }
+    catch (Exception e) {
+      fail(e.getMessage());
+    }
+    finally {
+      Location.mapFile(tmpFile, null);
+    }
+  }
+
+  @Test
+  public void testReadFileDifferentEncoding() {
+    String src = "µ€œ";
+    String tmpFile = "windows-1252-test.txt";
+    try {
+      byte[] bytes = src.getBytes("windows-1252");
+      Location.mapFile(tmpFile, new ByteArrayHandle(bytes));
+      String result = DataTools.readFile(tmpFile, "windows-1252");
+      assertEquals(src, result);
+    }
+    catch (Exception e) {
+      fail(e.getMessage());
+    }
+    finally {
+      Location.mapFile(tmpFile, null);
+    }
+  }
+
+  @Test
+  public void testReadFileWrongEncoding() {
+    String src = "µ€œ";
+    String tmpFile = "windows-1252-vs-utf8-test.txt";
+    try {
+      byte[] bytes = src.getBytes("UTF-8");
+      Location.mapFile(tmpFile, new ByteArrayHandle(bytes));
+      String result = DataTools.readFile(tmpFile, "windows-1252");
+      assertFalse(src.equals(result));
+    }
+    catch (Exception e) {
+      fail(e.getMessage());
+    }
+    finally {
+      Location.mapFile(tmpFile, null);
+    }
+  }
 }