Percent encodes file pathnames in manifests. Closes #33

src/main/java/edu/mit/lib/bagit/Bag.java

@@ -49,7 +49,7 @@ public class Bag {
     // coding constants
     static final String DEFAULT_CS_ALGO = "SHA-512";
     static final String BAGIT_VSN = "1.0";
-    static final String LIB_VSN = "1.0";
+    static final String LIB_VSN = "1.1";
     static final String DFLT_FMT = "zip";
     static final String TGZIP_FMT = "tgz";
     static final String SPACER = " ";
@@ -526,7 +526,7 @@ public Map<String, String> manifest(String relPath) throws IOException {
             String line;
             while((line = reader.readLine()) != null) {
                 String[] parts = line.split("\\s+", 2);
-                mfMap.put(parts[1], parts[0]);
+                mfMap.put(decodePath(parts[1]), parts[0]);
             }
         }
         return mfMap;
@@ -652,6 +652,14 @@ static byte[] filterBytes(String data, Charset encoding, AtomicBoolean bomOut) {
         }
     }
 
+    static String encodePath(String path) {
+        return path.replaceAll("%", "%25").replaceAll("\n", "%0A").replaceAll("\r", "%0D");
+    }
+
+    static String decodePath(String path) {
+        return path.replaceAll("%25", "%").replaceAll("%0A", "\n").replaceAll("%0D", "\r");
+    }
+
     private static final char[] HEX_CHARS = "0123456789abcdef".toCharArray();
     static String toHex(byte[] data) {
         if ((data == null) || (data.length == 0)) {

src/main/java/edu/mit/lib/bagit/Filler.java

@@ -404,7 +404,7 @@ public Filler payloadRef(String relPath, InputStream in, URI uri) throws IOExcep
         var destFile = destDir.resolve("foo");
         long size = digestCopy(in, destFile, DATA_PATH + relPath, manWriters);
         var sizeStr = (size > 0L) ? Long.toString(size) : "-";
-        refWriter.writeLine(uri.toString() + " " + sizeStr + " " + DATA_PATH + relPath);
+        refWriter.writeLine(uri.toString() + " " + sizeStr + " " + DATA_PATH + encodePath(relPath));
         Files.delete(destFile);
         Files.delete(destDir);
         return this;
@@ -434,11 +434,11 @@ public Filler payloadRefUnsafe(String relPath, long size, URI uri, Map<String, S
             throw new IOException("checksums do not match bags");
         }
         for (String alg : manWriters.keySet()) {
-            manWriters.get(alg).writeLine(checksums.get(alg) + " " + relPath);
+            manWriters.get(alg).writeLine(checksums.get(alg) + " " + encodePath(relPath));
         }
         var sizeStr = (size > 0L) ? Long.toString(size) : "-";
         FlatWriter refWriter = getWriter(REF_FILE);
-        refWriter.writeLine(uri.toString() + " " + sizeStr + " " + DATA_PATH + relPath);
+        refWriter.writeLine(uri.toString() + " " + sizeStr + " " + DATA_PATH + encodePath(relPath));
         return this;
     }
 
@@ -619,10 +619,12 @@ public void writeProperty(String key, String value) throws IOException {
         }
 
         public void writeLine(String line) throws IOException {
+            var encLine = line;
             if (record) {
                 lines.add(line);
+                encLine = encodePath(line);
             }
-            write(filterBytes(line + lineSeparator, encoding, bomOut));
+            write(filterBytes(encLine + lineSeparator, encoding, bomOut));
         }
 
         public List<String> getLines() {

src/test/java/edu/mit/lib/bagit/BagTest.java

@@ -19,6 +19,7 @@
 import java.time.temporal.ChronoUnit;
 import java.util.Scanner;
 import java.util.HashSet;
+import java.util.List;
 import java.util.Map;
 import java.util.Set;
 
@@ -853,6 +854,45 @@ public void validAndInvalidBagUTF16() throws IOException {
         assertTrue(!bag.isValid());
     }
 
+    @Test
+    public void percentPathNamePayloadBag() throws IOException {
+        Path bagFile = tempFolder.newFolder("bag37").toPath();
+        new Filler(bagFile).payload("first%1.pdf", payload1).payload("second/second%2.pdf", payload2).toDirectory();
+        Path payloadDir = bagFile.resolve(DATA_DIR);
+        assertTrue(Files.isDirectory(payloadDir));
+        Path pload1 = payloadDir.resolve("first%1.pdf");
+        assertTrue(Files.exists(pload1));
+        Path pload2 = payloadDir.resolve("second/second%2.pdf");
+        assertTrue(Files.exists(pload2));
+        // look in manifest files for encoded path names
+        String manifLine = readTextLine(bagFile.resolve("manifest-sha512.txt"), 0);
+        assertTrue(manifLine.endsWith("first%251.pdf"));
+        String manifLine2 = readTextLine(bagFile.resolve("manifest-sha512.txt"), 1);
+        assertTrue(manifLine2.endsWith("second%252.pdf"));
+         // assure completeness
+        Bag bag = load(bagFile);
+        assertTrue(bag.isComplete());
+        assertTrue(bag.isValid());
+    }
+
+    @Test
+    public void percentPathNameFetchBag() throws IOException, URISyntaxException {
+        Path bagFile = tempFolder.newFolder("bag38").toPath();
+        URI location = new URI("http://www.example.com/foo");
+        URI location2 = new URI("http://www.example.com/foo");
+        new Filler(bagFile).payloadRef("first%1.pdf", payload1, location).payloadRef("second/second%2.pdf", payload2, location2).toDirectory();
+        // look in fetch file for encoded path names
+        String manifLine = readTextLine(bagFile.resolve("fetch.txt"), 0);
+        assertTrue(manifLine.endsWith("first%251.pdf"));
+        String manifLine2 = readTextLine(bagFile.resolve("fetch.txt"), 1);
+        assertTrue(manifLine2.endsWith("second%252.pdf"));
+    }
+
+    private String readTextLine(Path file, int lineNum) throws IOException {
+        final List<String> lines = Files.readAllLines(file);
+        return lines.get(lineNum);
+    }
+
     private String findSeparator(Path file) throws IOException {
         try (Scanner scanner = new Scanner(file)) {
             // it's one or the other