refs #92 - check for blank lines in fetch file and throw error if there are any

Author: johnscancella

src/main/java/gov/loc/repository/bagit/reader/FetchReader.java

@@ -23,6 +23,7 @@
 public final class FetchReader {
   private static final Logger logger = LoggerFactory.getLogger(FetchReader.class);
   private static final ResourceBundle messages = ResourceBundle.getBundle("MessageBundle");
+  private static final String FETCH_LINE_REGEX = ".*[ \t]*(\\d*|-)[ \t]*.*";
 
   private FetchReader(){
     //intentionally left empty
@@ -51,14 +52,19 @@ public static List<FetchItem> readFetch(final Path fetchFile, final Charset enco
       long length = 0;
       URL url = null;
       while(line != null){
-        parts = line.split("\\s+", 3);
-        final Path path = TagFileReader.createFileFromManifest(bagRootDir, parts[2]);
-        length = parts[1].equals("-") ? -1 : Long.decode(parts[1]);
-        url = new URL(parts[0]);
-        
-        logger.debug(messages.getString("read_fetch_file_line"), url, length, parts[2], fetchFile);
-        final FetchItem itemToFetch = new FetchItem(url, length, path);
-        itemsToFetch.add(itemToFetch);
+        if(line.matches(FETCH_LINE_REGEX) && !line.matches("\\s*")){
+          parts = line.split("\\s+", 3);
+          final Path path = TagFileReader.createFileFromManifest(bagRootDir, parts[2]);
+          length = parts[1].equals("-") ? -1 : Long.decode(parts[1]);
+          url = new URL(parts[0]);
+          
+          logger.debug(messages.getString("read_fetch_file_line"), url, length, parts[2], fetchFile);
+          final FetchItem itemToFetch = new FetchItem(url, length, path);
+          itemsToFetch.add(itemToFetch);
+        }
+        else{
+          throw new InvalidBagitFileFormatException(messages.getString("invalid_fetch_file_line_error").replace("{}", line));
+        }
 
         line = reader.readLine();
       }

src/main/resources/MessageBundle.properties

@@ -115,6 +115,7 @@ unparsable_version_error=Version must be in format MAJOR.MINOR but was [{}]!
 #for FetchReader.java
 reading_fetch_file=Attempting to read [{}].
 read_fetch_file_line=Read URL [{}] length [{}] path [{}] from fetch file [{}].
+invalid_fetch_file_line_error=The Line [{}] is invalid for fetch.txt. Each line must take the form of <URL> <LENGTH> <FILENAM>.
 
 #for KeyValueReader.java
 read_key_value_line=Found key [{}] value [{}] in file [{}] using split regex [{}].

src/test/java/gov/loc/repository/bagit/reader/FetchReaderTest.java

@@ -66,6 +66,12 @@ public void testReadFetchWithSizeSpecified() throws Exception{
     }
   }
 
+  @Test(expected=InvalidBagitFileFormatException.class)
+  public void testReadBlankLinesThrowsException() throws Exception{
+    Path fetchFile = Paths.get(getClass().getClassLoader().getResource("fetchFiles/fetchWithBlankLines.txt").toURI());
+    FetchReader.readFetch(fetchFile, StandardCharsets.UTF_8, Paths.get("/foo"));
+  }
+  
   @Test(expected=InvalidBagitFileFormatException.class)
   public void testReadWindowsSpecialDirMaliciousFetchThrowsException() throws Exception{
     Path fetchFile = Paths.get(getClass().getClassLoader().getResource("maliciousFetchFile/windowsSpecialDirectoryName.txt").toURI());
@@ -92,4 +98,16 @@ public void testReadFileUrlMaliciousFetchThrowsException() throws Exception{
     }
     throw new MaliciousPathException("Skipping for windows cause it isn't valid");
   }
+  
+  @Test
+  public void foo(){
+    String regex = ".*[ \t]*(\\d*|-)[ \t]*.*";
+    String test1 = "http://localhost/foo/data/test2.txt - ~/foo/bar/ham.txt";
+    String test2 = "http://localhost/foo/data/dir1/test3.txt 100057 data/dir1/test3.txt";
+    String test3 = "http://localhost/foo/data/dir1/test3.txt \t 100057 \t data/dir1/test3.txt";
+    
+    System.err.println(test1.matches(regex));
+    System.err.println(test2.matches(regex));
+    System.err.println(test3.matches(regex));
+  }
 }

src/test/resources/fetchFiles/fetchWithBlankLines.txt

@@ -0,0 +1,7 @@
+http://localhost/foo/data/dir1/test3.txt 1 data/dir1/test3.txt
+http://localhost/foo/data/dir2/dir3/test5.txt 2 data/dir2/dir3/test5.txt
+http://localhost/foo/data/dir2/test4.txt 3 data/dir2/test4.txt
+http://localhost/foo/data/test%201.txt 4 data/test 1.txt
+http://localhost/foo/data/test2.txt 5 data/test2.txt
+ 
+