misc optimizations, first "working" version of book serialization/deserialization (missing things are marked with TODO comments)

jni/NativeFormats/JavaNativeFormatPlugin.cpp

@@ -119,7 +119,7 @@ JNIEXPORT jboolean JNICALL Java_org_geometerplus_fbreader_formats_NativeFormatPl
 }
 
 extern "C"
-JNIEXPORT void JNICALL Java_org_geometerplus_fbreader_formats_NativeFormatPlugin_detectLanguageAndEncoding(JNIEnv* env, jobject thiz, jobject javaBook) {
+JNIEXPORT void JNICALL Java_org_geometerplus_fbreader_formats_NativeFormatPlugin_detectLanguageAndEncodingNative(JNIEnv* env, jobject thiz, jobject javaBook) {
 	shared_ptr<FormatPlugin> plugin = findCppPlugin(thiz);
 	if (plugin.isNull()) {
 		return;

src/org/geometerplus/fbreader/formats/NativeFormatPlugin.java

@@ -56,7 +56,11 @@ public class NativeFormatPlugin extends FormatPlugin {
 	private native boolean readMetaInfoNative(Book book);
 
 	@Override
-	public native void detectLanguageAndEncoding(Book book);
+	public void detectLanguageAndEncoding(Book book) {
+		detectLanguageAndEncodingNative(book);
+	}
+
+	public native void detectLanguageAndEncodingNative(Book book);
 
 	@Override
 	synchronized public void readModel(BookModel model) throws BookReadingException {

src/org/geometerplus/fbreader/formats/fb2/FB2NativePlugin.java

@@ -26,6 +26,7 @@ import org.geometerplus.zlibrary.core.filesystem.ZLFile;
 
 import org.geometerplus.fbreader.bookmodel.BookReadingException;
 import org.geometerplus.fbreader.formats.NativeFormatPlugin;
+import org.geometerplus.fbreader.library.Book;
 
 public class FB2NativePlugin extends NativeFormatPlugin {
 	public FB2NativePlugin() {
@@ -45,4 +46,9 @@ public class FB2NativePlugin extends NativeFormatPlugin {
 	public EncodingCollection supportedEncodings() {
 		return new AutoEncodingCollection();
 	}
+
+	@Override
+	public void detectLanguageAndEncoding(Book book) {
+		book.setEncoding("auto");
+	}
 }

src/org/geometerplus/fbreader/formats/oeb/OEBNativePlugin.java

@@ -29,19 +29,14 @@ import org.geometerplus.zlibrary.core.filesystem.ZLFile;
 
 import org.geometerplus.fbreader.bookmodel.BookModel;
 import org.geometerplus.fbreader.bookmodel.BookReadingException;
-
 import org.geometerplus.fbreader.formats.NativeFormatPlugin;
+import org.geometerplus.fbreader.library.Book;
 
 public class OEBNativePlugin extends NativeFormatPlugin {
 	public OEBNativePlugin() {
 		super("ePub");
 	}
 
-	@Override
-	public EncodingCollection supportedEncodings() {
-		return new AutoEncodingCollection();
-	}
-
 	@Override
 	public void readModel(BookModel model) throws BookReadingException {
 		super.readModel(model);
@@ -54,4 +49,14 @@ public class OEBNativePlugin extends NativeFormatPlugin {
 			}
 		});
 	}
+
+	@Override
+	public EncodingCollection supportedEncodings() {
+		return new AutoEncodingCollection();
+	}
+
+	@Override
+	public void detectLanguageAndEncoding(Book book) {
+		book.setEncoding("auto");
+	}
 }

src/org/geometerplus/fbreader/library/Book.java

@@ -321,7 +321,9 @@ public class Book {
 	public String getEncoding() {
 		if (myEncoding == null) {
 			try {
+				System.err.println("detect encoding for " + getId());
 				getPlugin().detectLanguageAndEncoding(this);
+				System.err.println("detected encoding for " + getId() + " = " + myEncoding);
 			} catch (BookReadingException e) {
 			}
 			if (myEncoding == null) {

src/org/geometerplus/fbreader/library/BookSerializerUtil.java

@@ -19,17 +19,28 @@
 
 package org.geometerplus.fbreader.library;
 
+import java.util.ArrayList;
+
+import org.geometerplus.zlibrary.core.constants.XMLNamespaces;
+import org.geometerplus.zlibrary.core.filesystem.ZLFile;
+import org.geometerplus.zlibrary.core.xml.ZLXMLReaderAdapter;
+import org.geometerplus.zlibrary.core.xml.ZLStringMap;
+
 public abstract class BookSerializerUtil {
 	private BookSerializerUtil() {
 	}
 
 	public static String serialize(Book book) {
 		final StringBuilder buffer = new StringBuilder();
-		buffer.append("<entry>\n");
+		appendTagWithAttributes(
+			buffer, "entry", false,
+			"xmlns:dc", XMLNamespaces.DublinCore
+		);
 
 		appendTagWithContent(buffer, "id", String.valueOf(book.getId()));
 		appendTagWithContent(buffer, "title", book.getTitle());
 		appendTagWithContent(buffer, "dc:language", book.getLanguage());
+		appendTagWithContent(buffer, "dc:encoding", book.getEncodingNoDetection());
 
 		for (Author author : book.authors()) {
 			buffer.append("<author>\n");
@@ -40,14 +51,18 @@ public abstract class BookSerializerUtil {
 
 		for (Tag tag : book.tags()) {
 			appendTagWithAttributes(
-				buffer, "category",
+				buffer, "category", true,
 				"term", tag.toString("/"),
 				"label", tag.Name
 			);
 		}
 
+		// TODO: serialize series info
+		// TODO: serialize description (?)
+		// TODO: serialize cover (?)
+
 		appendTagWithAttributes(
-			buffer, "link",
+			buffer, "link", true,
 			"href", book.File.getUrl(),
 			// TODO: real book mimetype
 			"type", "application/epub+zip",
@@ -59,22 +74,221 @@ public abstract class BookSerializerUtil {
 	}
 
 	public static Book deserialize(String xml) {
-		// TODO: implement
+		final Deserializer deserializer = new Deserializer();
-		return null;
+		deserializer.readQuietly(xml);
+		return deserializer.getBook();
 	}
 
 	private static void appendTagWithContent(StringBuilder buffer, String tag, String content) {
-		buffer
+		if (content != null) {
-			.append('<').append(tag).append('>')
+			buffer
-			.append(content)
+				.append('<').append(tag).append('>')
-			.append("</").append(tag).append(">\n");
+				.append(escapeForXml(content))
+				.append("</").append(tag).append(">\n");
+		}
 	}
 
-	private static void appendTagWithAttributes(StringBuilder buffer, String tag, String ... attrs) {
+	private static void appendTagWithAttributes(StringBuilder buffer, String tag, boolean close, String ... attrs) {
 		buffer.append('<').append(tag);
 		for (int i = 0; i < attrs.length - 1; i += 2) {
-			buffer.append(' ').append(attrs[i]).append("=\"").append(attrs[i + 1]).append('"');
+			buffer.append(' ')
+				.append(escapeForXml(attrs[i])).append("=\"")
+				.append(escapeForXml(attrs[i + 1])).append('"');
+		}
+		if (close) {
+			buffer.append('/');
+		}
+		buffer.append(">\n");
+	}
+
+	private static String escapeForXml(String data) {
+		if (data.indexOf('&') != -1) {
+			data = data.replaceAll("&", "&amp;");
+		}
+		if (data.indexOf('<') != -1) {
+			data = data.replaceAll("&", "&lt;");
+		}
+		if (data.indexOf('>') != -1) {
+			data = data.replaceAll("&", "&gt;");
+		}
+		if (data.indexOf('\'') != -1) {
+			data = data.replaceAll("&", "&apos;");
+		}
+		if (data.indexOf('"') != -1) {
+			data = data.replaceAll("&", "&quot;");
+		}
+		return data;
+	}
+
+	private static final class Deserializer extends ZLXMLReaderAdapter {
+		private static enum State {
+			READ_NOTHING,
+			READ_ENTRY,
+			READ_ID,
+			READ_TITLE,
+			READ_LANGUAGE,
+			READ_ENCODING,
+			READ_AUTHOR,
+			READ_AUTHOR_URI,
+			READ_AUTHOR_NAME,
+		}
+
+		private State myState = State.READ_NOTHING;
+		
+		private long myId = -1;
+		private String myUrl;
+		private StringBuilder myTitle = new StringBuilder();
+		private StringBuilder myLanguage = new StringBuilder();
+		private StringBuilder myEncoding = new StringBuilder();
+		private ArrayList<Author> myAuthors = new ArrayList<Author>();
+		private StringBuilder myAuthorSortKey = new StringBuilder();
+		private StringBuilder myAuthorName = new StringBuilder();
+
+		private Book myBook;
+
+		public Book getBook() {
+			return myState == State.READ_NOTHING ? myBook : null;
+		}
+
+		private static void clear(StringBuilder buffer) {
+			buffer.delete(0, buffer.length());
+		}
+
+		private static String string(StringBuilder buffer) {
+			return buffer.length() != 0 ? buffer.toString() : null;
+		}
+
+		@Override
+		public void startDocumentHandler() {
+			myBook = null;
+
+			myId = -1;
+			myUrl = null;
+			clear(myTitle);
+			clear(myLanguage);
+			clear(myEncoding);
+			myAuthors.clear();
+
+			myState = State.READ_NOTHING;
+		}
+
+		@Override
+		public void endDocumentHandler() {
+			if (myId == -1) {
+				return;
+			}
+			myBook = new Book(
+				myId,
+				ZLFile.createFileByUrl(myUrl),
+				string(myTitle),
+				string(myEncoding),
+				string(myLanguage)
+			);
+			for (Author author : myAuthors) {
+				myBook.addAuthorWithNoCheck(author);
+			}
+			// TODO: add tags
+			// TODO: add series info
+		}
+
+		@Override
+		public boolean startElementHandler(String tag, ZLStringMap attributes) {
+			switch (myState) {
+				case READ_NOTHING:
+					if (!"entry".equals(tag)) {
+						return true;
+					}
+					myState = State.READ_ENTRY;
+					break;
+				case READ_ENTRY:
+					if ("id".equals(tag)) {
+						myState = State.READ_ID;
+					} else if ("title".equals(tag)) {
+						myState = State.READ_TITLE;
+					} else if ("dc:language".equals(tag)) {
+						myState = State.READ_LANGUAGE;
+					} else if ("dc:encoding".equals(tag)) {
+						myState = State.READ_ENCODING;
+					} else if ("author".equals(tag)) {
+						myState = State.READ_AUTHOR;
+						clear(myAuthorName);
+						clear(myAuthorSortKey);
+					} else if ("category".equals(tag)) {
+						// TODO: implement
+					} else if ("link".equals(tag)) {
+						// TODO: use "rel" attribute
+						myUrl = attributes.getValue("href");
+					} else {
+						return true;
+					}
+					break;
+				case READ_AUTHOR:
+					if ("uri".equals(tag)) {
+						myState = State.READ_AUTHOR_URI;
+					} else if ("name".equals(tag)) {
+						myState = State.READ_AUTHOR_NAME;
+					} else {
+						return true;
+					}
+					break;
+			}
+			return false;
+		}
+		
+		@Override
+		public boolean endElementHandler(String tag) {
+			switch (myState) {
+				case READ_NOTHING:
+					return true;
+				case READ_ENTRY:
+					if ("entry".equals(tag)) {
+						myState = State.READ_NOTHING;
+					}	
+					break;
+				case READ_AUTHOR_URI:
+				case READ_AUTHOR_NAME:
+					myState = State.READ_AUTHOR;
+					break;
+				case READ_AUTHOR:
+					if (myAuthorSortKey.length() > 0 && myAuthorName.length() > 0) {
+						myAuthors.add(
+							new Author(myAuthorName.toString(), myAuthorSortKey.toString())
+						);
+					}
+					myState = State.READ_ENTRY;
+					break;
+				default:
+					myState = State.READ_ENTRY;
+					break;
+			}
+			return false;
+		}
+		
+		@Override
+		public void characterDataHandler(char[] ch, int start, int length) {
+			switch (myState) {
+				case READ_ID:
+					try {
+						myId = Long.parseLong(new String(ch, start, length));
+					} catch (NumberFormatException e) {
+					}
+					break;
+				case READ_TITLE:
+					myTitle.append(ch, start, length);
+					break;
+				case READ_LANGUAGE:
+					myLanguage.append(ch, start, length);
+					break;
+				case READ_ENCODING:
+					myEncoding.append(ch, start, length);
+					break;
+				case READ_AUTHOR_URI:
+					myAuthorSortKey.append(ch, start, length);
+					break;
+				case READ_AUTHOR_NAME:
+					myAuthorName.append(ch, start, length);
+					break;
+			}
 		}
-		buffer.append("/>\n");
 	}
 }

src/org/geometerplus/fbreader/library/Library.java

@@ -237,6 +237,10 @@ public final class Library {
 		if (myBooks.containsKey(book.File)) {
 			return;
 		}
+
+		final String xml = BookSerializerUtil.serialize(book);
+		BookSerializerUtil.deserialize(xml);
+
 		myBooks.put(book.File, book);
 
 		List<Author> authors = book.authors();

src/org/geometerplus/fbreader/library/Tag.java

@@ -58,7 +58,13 @@ public final class Tag {
 	}
 
 	public String toString(String delimiter) {
-		return Parent == null ? Name : Parent.toString(delimiter) + delimiter + Name;
+		return toStringBuilder(delimiter).toString();
+	}
+
+	protected StringBuilder toStringBuilder(String delimiter) {
+		return Parent == null
+			? new StringBuilder(Name)
+			: Parent.toStringBuilder(delimiter).append(delimiter).append(Name);
 	}
 
 	@Override

src/org/geometerplus/zlibrary/core/filesystem/ZLFile.java

@@ -103,6 +103,14 @@ public abstract class ZLFile {
 		return file;
 	}
 
+	public static ZLFile createFileByUrl(String url) {
+		System.err.println("url = " + url);
+		if (url == null || !url.startsWith("file://")) {
+			return null;
+		}
+		return createFileByPath(url.substring("file://".length()));
+	}
+
 	public static ZLFile createFileByPath(String path) {
 		if (path == null) {
 			return null;

src/org/geometerplus/zlibrary/core/xml/ZLXMLParser.java

@@ -67,7 +67,7 @@ final class ZLXMLParser {
 		return s;
 	}
 
-	private final InputStreamReader myStreamReader;
+	private final Reader myStreamReader;
 	private final ZLXMLReader myXMLReader;
 	private final boolean myProcessNamespaces;
 
@@ -119,7 +119,15 @@ final class ZLXMLParser {
 		storeString(myEntityName);
 	}
 
-	public ZLXMLParser(ZLXMLReader xmlReader, InputStream stream, int bufferSize) throws IOException {
+	ZLXMLParser(ZLXMLReader xmlReader, Reader reader, int bufferSize) throws IOException {
+		myXMLReader = xmlReader;
+		myProcessNamespaces = xmlReader.processNamespaces();
+		myBuffer = getBuffer(bufferSize);
+		myBufferDescriptionLength = 0;
+		myStreamReader = reader;
+	}
+
+	ZLXMLParser(ZLXMLReader xmlReader, InputStream stream, int bufferSize) throws IOException {
 		myXMLReader = xmlReader;
 		myProcessNamespaces = xmlReader.processNamespaces();
 
@@ -204,7 +212,7 @@ final class ZLXMLParser {
 		final ZLXMLReader xmlReader = myXMLReader;
 		final HashMap<String,char[]> entityMap = getDTDMap(xmlReader.externalDTDs());
 		xmlReader.collectExternalEntities(entityMap);
-		final InputStreamReader streamReader = myStreamReader;
+		final Reader streamReader = myStreamReader;
 		final boolean processNamespaces = myProcessNamespaces;
 		HashMap<String,String> oldNamespaceMap = processNamespaces ? new HashMap<String,String>() : null;
 		HashMap<String,String> currentNamespaceMap = null;

src/org/geometerplus/zlibrary/core/xml/ZLXMLProcessor.java

@@ -33,13 +33,27 @@ public abstract class ZLXMLProcessor {
 		}
 	}
 
-	public static void read(ZLXMLReader reader, InputStream stream, int bufferSize) throws IOException {
+	public static void read(ZLXMLReader xmlReader, InputStream stream, int bufferSize) throws IOException {
 		ZLXMLParser parser = null;
 		try {
-			parser = new ZLXMLParser(reader, stream, bufferSize);
+			parser = new ZLXMLParser(xmlReader, stream, bufferSize);
-			reader.startDocumentHandler();
+			xmlReader.startDocumentHandler();
 			parser.doIt();
-			reader.endDocumentHandler();
+			xmlReader.endDocumentHandler();
+		} finally {
+			if (parser != null) {
+				parser.finish();
+			}
+		}
+	}
+
+	public static void read(ZLXMLReader xmlReader, Reader reader, int bufferSize) throws IOException {
+		ZLXMLParser parser = null;
+		try {
+			parser = new ZLXMLParser(xmlReader, reader, bufferSize);
+			xmlReader.startDocumentHandler();
+			parser.doIt();
+			xmlReader.endDocumentHandler();
 		} finally {
 			if (parser != null) {
 				parser.finish();

src/org/geometerplus/zlibrary/core/xml/ZLXMLReaderAdapter.java

@@ -19,9 +19,8 @@
 
 package org.geometerplus.zlibrary.core.xml;
 
+import java.io.*;
 import java.util.*;
-import java.io.InputStream;
-import java.io.IOException;
 
 import org.geometerplus.zlibrary.core.filesystem.ZLFile;
 
@@ -30,7 +29,16 @@ public abstract class ZLXMLReaderAdapter implements ZLXMLReader {
 
  	public boolean readQuietly(ZLFile file) {
 		try {
-			ZLXMLProcessor.read(this, file);
+			read(file);
+			return true;
+		} catch (IOException e) {
+			return false;
+		}
+	}
+	
+ 	public boolean readQuietly(String string) {
+		try {
+			read(string);
 			return true;
 		} catch (IOException e) {
 			return false;
@@ -44,7 +52,15 @@ public abstract class ZLXMLReaderAdapter implements ZLXMLReader {
 	public void read(InputStream stream) throws IOException {
 		ZLXMLProcessor.read(this, stream, 65536);
 	}
-	
+
+	public void read(String string) throws IOException {
+		ZLXMLProcessor.read(this, new StringReader(string), 65536);
+	}
+
+	public void read(Reader reader) throws IOException {
+		ZLXMLProcessor.read(this, reader, 65536);
+	}
+
 	public boolean dontCacheAttributeValues() {
 		return false;
 	}