Include the original file instead of a patch if the patch is larger than the original file.

This saves processing and disk space, as there is no point in applying a patch when the original file is smaller and can be included in its entirety.

src/main/java/org/qortal/arbitrary/patch/UnifiedDiffPatch.java

@@ -91,7 +91,8 @@ public class UnifiedDiffPatch {
     }
 
     /**
-     * Validate the patch to ensure it works correctly
+     * Validate the patch at the "destination" path to ensure
+     * it works correctly and is smaller than the original file
      *
      * @return true if valid, false if invalid
      * @throws IOException
@@ -110,9 +111,14 @@ public class UnifiedDiffPatch {
             byte[] inputDigest = Crypto.digest(after.toFile());
             byte[] outputDigest = Crypto.digest(tempPath.toFile());
             if (Arrays.equals(inputDigest, outputDigest)) {
-                // Patch is valid
+                // Patch is valid, but we might want to reject if it's larger than the original file
+                long originalSize = Files.size(after);
+                long patchSize = Files.size(destination);
+                if (patchSize < originalSize) {
+                    // Patch file is smaller than the original file size, so treat it as valid
                     return true;
                 }
+            }
             else {
                 LOGGER.info("Checksum mismatch when verifying patch for file {}", destination.toString());
                 return false;

src/test/java/org/qortal/test/arbitrary/ArbitraryDataMergeTests.java

@@ -68,23 +68,26 @@ public class ArbitraryDataMergeTests extends Common {
                 Crypto.digest(Paths.get(patchPath.toString(), "dir1", "dir2", "lorem5.txt").toFile())
         ));
 
-        // Ensure that the patch files differ from the second path (except for lorem3, which is missing)
+        // Ensure that the patch files 1, 4, and 5 differ from the second path
         assertFalse(Arrays.equals(
                 Crypto.digest(Paths.get(path2.toString(), "lorem1.txt").toFile()),
                 Crypto.digest(Paths.get(patchPath.toString(), "lorem1.txt").toFile())
         ));
-        assertFalse(Arrays.equals(
-                Crypto.digest(Paths.get(path2.toString(), "lorem2.txt").toFile()),
-                Crypto.digest(Paths.get(patchPath.toString(), "lorem2.txt").toFile())
-        ));
         assertFalse(Arrays.equals(
                 Crypto.digest(Paths.get(path2.toString(), "dir1", "lorem4.txt").toFile()),
                 Crypto.digest(Paths.get(patchPath.toString(), "dir1", "lorem4.txt").toFile())
         ));
-        assertFalse(Arrays.equals(
+
+        // Files 2 and 5 should match the original files, because their contents were
+        // too small to create a patch file smaller than to original file
+        assertArrayEquals(
+                Crypto.digest(Paths.get(path2.toString(), "lorem2.txt").toFile()),
+                Crypto.digest(Paths.get(patchPath.toString(), "lorem2.txt").toFile())
+        );
+        assertArrayEquals(
                 Crypto.digest(Paths.get(path2.toString(), "dir1", "dir2", "lorem5.txt").toFile()),
                 Crypto.digest(Paths.get(patchPath.toString(), "dir1", "dir2", "lorem5.txt").toFile())
-        ));
+        );
 
         // Now merge the patch with the original path
         ArbitraryDataMerge merge = new ArbitraryDataMerge(path1, patchPath);
@@ -227,6 +230,8 @@ public class ArbitraryDataMergeTests extends Common {
         // Write a random string to the first file
         BufferedWriter file1Writer = new BufferedWriter(new FileWriter(file1));
         String initialString = this.generateRandomString(1024);
+        // Add a newline every 50 chars
+        initialString = initialString.replaceAll("(.{50})", "$1\n");
         file1Writer.write(initialString);
         file1Writer.newLine();
         file1Writer.close();
@@ -289,6 +294,10 @@ public class ArbitraryDataMergeTests extends Common {
         // Write a random string to the first file
         BufferedWriter file1Writer = new BufferedWriter(new FileWriter(file1));
         String initialString = this.generateRandomString(1024);
+        // Add a newline every 50 chars
+        initialString = initialString.replaceAll("(.{50})", "$1\n");
+        // Remove newline at end of string
+        initialString = initialString.stripTrailing();
         file1Writer.write(initialString);
         // No newline
         file1Writer.close();
@@ -354,6 +363,8 @@ public class ArbitraryDataMergeTests extends Common {
         // Write a random string to the first file
         BufferedWriter file1Writer = new BufferedWriter(new FileWriter(file1));
         String initialString = this.generateRandomString(110 * 1024);
+        // Add a newline every 50 chars
+        initialString = initialString.replaceAll("(.{50})", "$1\n");
         file1Writer.write(initialString);
         file1Writer.newLine();
         file1Writer.close();

src/test/resources/arbitrary/demo1/lorem1.txt

@@ -1 +1,10 @@
 Lorem ipsum dolor sit amet, consectetur adipiscing elit.
+Ut ligula felis, imperdiet nec placerat at, placerat
+quis diam. Praesent a ultricies lacus.
+Aenean luctus blandit dui. Quisque vel augue
+diam. Nulla libero libero, condimentum sed
+accumsan eu, elementum sit amet turpis.
+In semper risus ac libero lobortis,
+ut consectetur urna euismod.
+Donec ut erat quis mi eleifend tincidunt
+aliquet vitae lacus.

src/test/resources/arbitrary/demo2/lorem1.txt

@@ -1 +1,10 @@
-Lorem ipsum dolor sit amet, adipiscing elit.
+Lorem ipsum dolor sit amet, consectetur adipiscing elit.
+Ut ligula felis, imperdiet nec placerat at, placerat
+quis diam. Praesent a ultricies lacus.
+Aenean luctus blandit dui. Quisque vel augue
+diam. Nulla libero libero; condimentum sed
+accumsan eu, elementum sit amet turpis.
+In semper risus ac libero lobortis,
+ut consectetur urna euismod.
+Donec ut erat quis mi eleifend tincidunt
+aliquet vitae lacus.