Issue #106 - First chop at AsciiReader

src/main/java/org/gedcom4j/io/reader/AsciiReader.java

@@ -23,21 +23,37 @@
 
 import java.io.IOException;
 import java.io.InputStream;
-import java.util.ArrayList;
-import java.util.List;
 
-import org.gedcom4j.exception.ParserCancelledException;
-import org.gedcom4j.io.event.FileProgressEvent;
+import org.gedcom4j.exception.GedcomParserException;
 import org.gedcom4j.parser.GedcomParser;
 
 /**
- * A reader that loads from an input stream and gives back a collection of strings representing the data therein. This
- * implementation handles ASCII encoding (1 byte per character, no extended character support).
+ * A reader that reads a single line from an Ascii-encoded file.
  * 
  * @author frizbog
  */
 class AsciiReader extends AbstractEncodingSpecificReader {
 
+    /**
+     * Are we at the end of file yet?
+     */
+    private boolean eof = false;
+
+    /**
+     * The prior character we read
+     */
+    private int lastChar = -1;
+
+    /**
+     * The current character we've just read
+     */
+    private int currChar = -1;
+
+    /**
+     * The line buffer for the current line
+     */
+    private final StringBuilder lineBuffer = new StringBuilder();
+
     /**
      * Constructor
      * 
@@ -51,87 +67,50 @@ protected AsciiReader(GedcomParser parser, InputStream byteStream) {
         super(parser, byteStream);
     }
 
-    /**
-     * {@inheritDoc}
-     */
     @Override
-    protected List<String> load() throws IOException, ParserCancelledException {
-        List<String> result = new ArrayList<String>();
-        StringBuilder lineBuffer = new StringBuilder();
-
-        int lastChar;
-        int b = -1;
-        boolean eof = false;
-
+    public String nextLine() throws IOException, GedcomParserException {
+        String result = null;
         while (!eof) {
-            lastChar = b;
-            b = byteStream.read();
+            lastChar = currChar;
+            currChar = byteStream.read();
 
             // Check for EOF
-            if (b < 0) {
+            if (currChar < 0) {
                 // hit EOF - add final line buffer (last line) and get out
-                addNonBlankLine(result, lineBuffer);
+                eof = true;
+                result = lineBuffer.toString();
                 break;
             }
 
             // Check for carriage returns - signify EOL
-            if (b == 0x0D) {
-                addNonBlankLine(result, lineBuffer);
+            if (currChar == 0x0D) {
+                result = lineBuffer.toString();
                 lineBuffer.setLength(0);
-                continue;
+                break;
             }
 
             // Check for line feeds - signify EOL (unless prev char was a
             // CR)
-            if (b == 0x0A) {
+            if (currChar == 0x0A) {
                 if (lastChar != 0x0D) {
-                    addNonBlankLine(result, lineBuffer);
+                    result = lineBuffer.toString();
                     lineBuffer.setLength(0);
                 }
-                continue;
+                break;
             }
 
             // All other characters in 0x00 to 0x7F range are treated the
             // same,
             // regardless of encoding, and added as is
-            if (b < 0x80) {
-                lineBuffer.append(Character.valueOf((char) b));
+            if (currChar < 0x80) {
+                lineBuffer.append(Character.valueOf((char) currChar));
                 continue;
             }
 
             // If we fell through to here, we have an extended character
-            throw new IOException("Extended characters not supported in ASCII: 0x" + Integer.toHexString(b));
+            throw new IOException("Extended characters not supported in ASCII: 0x" + Integer.toHexString(currChar));
         }
-        parser.notifyFileObservers(new FileProgressEvent(this, linesRead, true));
         return result;
     }
 
-    /**
-     * Add line to result if it is not blank
-     * 
-     * @param result
-     *            the resulting list of lines
-     * @param lineBuffer
-     *            the line buffer
-     * @throws ParserCancelledException
-     *             if the file load is cancelled
-     */
-    private void addNonBlankLine(List<String> result, StringBuilder lineBuffer) throws ParserCancelledException {
-        if (parser.isCancelled()) {
-            throw new ParserCancelledException("File load is cancelled");
-        }
-        if (lineBuffer.length() > 0) {
-            String s = lineBuffer.toString();
-            if (STRINGS_TO_INTERN.contains(s)) {
-                result.add(s.intern());
-            } else {
-                result.add(s);
-            }
-        }
-        linesRead++;
-        if (linesRead % parser.getReadNotificationRate() == 0) {
-            parser.notifyFileObservers(new FileProgressEvent(this, linesRead, false));
-        }
-    }
-
 }

src/main/java/org/gedcom4j/io/reader/GedcomFileReader.java

@@ -26,6 +26,7 @@
 import java.util.List;
 
 import org.gedcom4j.exception.GedcomParserException;
+import org.gedcom4j.exception.ParserCancelledException;
 import org.gedcom4j.exception.UnsupportedGedcomCharsetException;
 import org.gedcom4j.io.event.FileProgressEvent;
 import org.gedcom4j.parser.GedcomParser;
@@ -110,7 +111,7 @@ public List<String> getLines() throws IOException, GedcomParserException {
         String s = null;
         do {
             s = nextLine();
-            if (s != null) {
+            if (s != null && s.length() > 0) {
                 result.add(s);
             }
         } while (s != null);
@@ -127,6 +128,9 @@ public List<String> getLines() throws IOException, GedcomParserException {
      *             if the file is malformed and cannot be processed as a result
      */
     public String nextLine() throws IOException, GedcomParserException {
+        if (parser.isCancelled()) {
+            throw new ParserCancelledException("File load is cancelled");
+        }
         String result = encodingSpecificReader.nextLine();
         linesProcessed++;
         if (linesProcessed % parser.getReadNotificationRate() == 0 || result == null) {

src/main/java/org/gedcom4j/parser/StringTreeBuilder.java

@@ -28,7 +28,7 @@ static String leftTrim(String line) {
         if (line == null) {
             return null;
         }
-        if (line == "") {
+        if (line.length() == 0) {
             return "";
         }
         if (!Character.isWhitespace(line.charAt(0))) {

src/test/java/org/gedcom4j/io/reader/AsciiReaderListenerTest.java

@@ -33,7 +33,7 @@ public class AsciiReaderListenerTest extends AbstractReaderListenerTest {
      * Constructor
      */
     public AsciiReaderListenerTest() {
-        super("sample/willis-ascii.ged", 20036, 41);
+        super("sample/willis-ascii.ged", 20037, 41);
     }
 
 }

src/test/java/org/gedcom4j/io/reader/GedcomFileReaderTest.java

@@ -29,6 +29,7 @@
 import java.util.List;
 
 import org.gedcom4j.exception.GedcomParserException;
+import org.gedcom4j.exception.UnsupportedGedcomCharsetException;
 import org.gedcom4j.parser.GedcomParser;
 import org.junit.Test;
 
@@ -177,9 +178,11 @@ public void testAnselLfOnly() throws IOException, GedcomParserException {
      * 
      * @throws IOException
      *             if the first few bytes of the "file" cannot be read
+     * @throws UnsupportedGedcomCharsetException
+     *             if the character set is not supported
      */
     @Test
-    public void testFirstNBytes() throws IOException {
+    public void testFirstNBytes() throws IOException, UnsupportedGedcomCharsetException {
         GedcomFileReader gfr = new GedcomFileReader(new GedcomParser(),
                 new BufferedInputStream(new ByteArrayInputStream(new byte[] { 0x12, 0x34, 0x56, 0x78 })));
         // Haven't save the first chunk yet
@@ -459,9 +462,11 @@ public void testUtf8LfNoBOM() throws IOException, GedcomParserException {
      */
     void testUtf8File(String fileName) throws IOException, FileNotFoundException, GedcomParserException {
         FileInputStream fileInputStream = null;
+        BufferedInputStream bufferedInputStream = null;
         try {
             fileInputStream = new FileInputStream(fileName);
-            GedcomFileReader gr = new GedcomFileReader(new GedcomParser(), new BufferedInputStream(fileInputStream));
+            bufferedInputStream = new BufferedInputStream(fileInputStream);
+            GedcomFileReader gr = new GedcomFileReader(new GedcomParser(), bufferedInputStream);
             List<String> lines = gr.getLines();
             assertNotNull(lines);
             assertEquals(77, lines.size());
@@ -481,6 +486,9 @@ void testUtf8File(String fileName) throws IOException, FileNotFoundException, Ge
             assertEquals("1 NAME Ellen /\u0141owenst\u0117in/", lines.get(49));
             assertEquals("1 NAME Fred /\u00DBlrich/", lines.get(53));
         } finally {
+            if (bufferedInputStream != null) {
+                bufferedInputStream.close();
+            }
             if (fileInputStream != null) {
                 fileInputStream.close();
             }