merge in jb-mr1-release history after reset to jb-mr1-dev
diff --git a/java/src/com/android/inputmethod/latin/BinaryDictionary.java b/java/src/com/android/inputmethod/latin/BinaryDictionary.java
index 8909526..c3ae81f 100644
--- a/java/src/com/android/inputmethod/latin/BinaryDictionary.java
+++ b/java/src/com/android/inputmethod/latin/BinaryDictionary.java
@@ -41,7 +41,7 @@
* It is necessary to keep it at this value because some languages e.g. German have
* really long words.
*/
- public static final int MAX_WORD_LENGTH = 48;
+ public static final int MAX_WORD_LENGTH = Constants.Dictionary.MAX_WORD_LENGTH;
public static final int MAX_WORDS = 18;
public static final int MAX_SPACES = 16;
diff --git a/java/src/com/android/inputmethod/latin/Constants.java b/java/src/com/android/inputmethod/latin/Constants.java
index d71c0f9..57e12a6 100644
--- a/java/src/com/android/inputmethod/latin/Constants.java
+++ b/java/src/com/android/inputmethod/latin/Constants.java
@@ -16,8 +16,6 @@
package com.android.inputmethod.latin;
-import android.view.inputmethod.EditorInfo;
-
public final class Constants {
public static final class Color {
/**
@@ -54,7 +52,7 @@
* The private IME option used to indicate that the given text field needs ASCII code points
* input.
*
- * @deprecated Use {@link EditorInfo#IME_FLAG_FORCE_ASCII}.
+ * @deprecated Use EditorInfo#IME_FLAG_FORCE_ASCII.
*/
@SuppressWarnings("dep-ann")
public static final String FORCE_ASCII = "forceAscii";
@@ -128,6 +126,14 @@
}
}
+ public static class Dictionary {
+ public static final int MAX_WORD_LENGTH = 48;
+
+ private Dictionary() {
+ // This utility class is no publicly instantiable.
+ }
+ }
+
public static final int NOT_A_CODE = -1;
// See {@link KeyboardActionListener.Adapter#isInvalidCoordinate(int)}.
diff --git a/java/src/com/android/inputmethod/latin/makedict/BinaryDictInputOutput.java b/java/src/com/android/inputmethod/latin/makedict/BinaryDictInputOutput.java
index 7de5cf3..6775144 100644
--- a/java/src/com/android/inputmethod/latin/makedict/BinaryDictInputOutput.java
+++ b/java/src/com/android/inputmethod/latin/makedict/BinaryDictInputOutput.java
@@ -16,6 +16,7 @@
package com.android.inputmethod.latin.makedict;
+import com.android.inputmethod.latin.Constants;
import com.android.inputmethod.latin.makedict.FusionDictionary.CharGroup;
import com.android.inputmethod.latin.makedict.FusionDictionary.DictionaryOptions;
import com.android.inputmethod.latin.makedict.FusionDictionary.Node;
@@ -44,9 +45,28 @@
*/
public class BinaryDictInputOutput {
- final static boolean DBG = MakedictLog.DBG;
+ private static final boolean DBG = MakedictLog.DBG;
- /* Node layout is as follows:
+ /*
+ * Array of Node(FusionDictionary.Node) layout is as follows:
+ *
+ * g |
+ * r | the number of groups, 1 or 2 bytes.
+ * o | 1 byte = bbbbbbbb match
+ * u | case 1xxxxxxx => xxxxxxx << 8 + next byte
+ * p | otherwise => bbbbbbbb
+ * c |
+ * ount
+ *
+ * g |
+ * r | sequence of groups,
+ * o | the layout of each group is described below.
+ * u |
+ * ps
+ *
+ */
+
+ /* Node(CharGroup) layout is as follows:
* | addressType xx : mask with MASK_GROUP_ADDRESS_TYPE
* 2 bits, 00 = no children : FLAG_GROUP_ADDRESS_TYPE_NOADDRESS
* f | 01 = 1 byte : FLAG_GROUP_ADDRESS_TYPE_ONEBYTE
@@ -59,6 +79,13 @@
* | is not a word ? 1 bit, 1 = yes, 0 = no : FLAG_IS_NOT_A_WORD
* | is blacklisted ? 1 bit, 1 = yes, 0 = no : FLAG_IS_BLACKLISTED
*
+ * p |
+ * a | IF HAS_PARENT_ADDRESS (defined in the file header)
+ * r | parent address, 3byte
+ * e | the address must be negative, so the absolute value of the address is stored.
+ * n |
+ * taddress
+ *
* c | IF FLAG_HAS_MULTIPLE_CHARS
* h | char, char, char, char n * (1 or 3 bytes) : use CharGroupInfo for i/o helpers
* a | end 1 byte, = 0
@@ -132,18 +159,22 @@
private static final int VERSION_1_MAGIC_NUMBER = 0x78B1;
public static final int VERSION_2_MAGIC_NUMBER = 0x9BC13AFE;
private static final int MINIMUM_SUPPORTED_VERSION = 1;
- private static final int MAXIMUM_SUPPORTED_VERSION = 2;
+ private static final int MAXIMUM_SUPPORTED_VERSION = 3;
private static final int NOT_A_VERSION_NUMBER = -1;
private static final int FIRST_VERSION_WITH_HEADER_SIZE = 2;
+ private static final int FIRST_VERSION_WITH_PARENT_ADDRESS = 3;
// These options need to be the same numeric values as the one in the native reading code.
private static final int GERMAN_UMLAUT_PROCESSING_FLAG = 0x1;
+ private static final int HAS_PARENT_ADDRESS = 0x2;
private static final int FRENCH_LIGATURE_PROCESSING_FLAG = 0x4;
private static final int CONTAINS_BIGRAMS_FLAG = 0x8;
// TODO: Make this value adaptative to content data, store it in the header, and
// use it in the reading code.
- private static final int MAX_WORD_LENGTH = 48;
+ private static final int MAX_WORD_LENGTH = Constants.Dictionary.MAX_WORD_LENGTH;
+
+ private static final int PARENT_ADDRESS_SIZE = 3;
private static final int MASK_GROUP_ADDRESS_TYPE = 0xC0;
private static final int FLAG_GROUP_ADDRESS_TYPE_NOADDRESS = 0x00;
@@ -178,6 +209,7 @@
private static final int GROUP_SHORTCUT_LIST_SIZE_SIZE = 2;
private static final int NO_CHILDREN_ADDRESS = Integer.MIN_VALUE;
+ private static final int NO_PARENT_ADDRESS = 0;
private static final int INVALID_CHARACTER = -1;
private static final int MAX_CHARGROUPS_FOR_ONE_BYTE_CHARGROUP_COUNT = 0x7F; // 127
@@ -246,8 +278,17 @@
*/
public static class FormatOptions {
public final int mVersion;
+ public final boolean mHasParentAddress;
public FormatOptions(final int version) {
+ this(version, false);
+ }
+ public FormatOptions(final int version, final boolean hasParentAddress) {
mVersion = version;
+ if (version < FIRST_VERSION_WITH_PARENT_ADDRESS && hasParentAddress) {
+ throw new RuntimeException("Parent addresses are only supported with versions "
+ + FIRST_VERSION_WITH_PARENT_ADDRESS + " and ulterior.");
+ }
+ mHasParentAddress = hasParentAddress;
}
}
@@ -277,7 +318,7 @@
/**
* Helper method to find out whether this code fits on one byte
*/
- private static boolean fitsOnOneByte(int character) {
+ private static boolean fitsOnOneByte(final int character) {
return character >= MINIMAL_ONE_BYTE_CHARACTER_VALUE
&& character <= MAXIMAL_ONE_BYTE_CHARACTER_VALUE;
}
@@ -299,7 +340,7 @@
* @param character the character code.
* @return the size in binary encoded-form, either 1 or 3 bytes.
*/
- private static int getCharSize(int character) {
+ private static int getCharSize(final int character) {
// See char encoding in FusionDictionary.java
if (fitsOnOneByte(character)) return 1;
if (INVALID_CHARACTER == character) return 1;
@@ -372,7 +413,7 @@
* @param buffer the ByteArrayOutputStream to write to.
* @param word the string to write.
*/
- private static void writeString(ByteArrayOutputStream buffer, final String word) {
+ private static void writeString(final ByteArrayOutputStream buffer, final String word) {
final int length = word.length();
for (int i = 0; i < length; i = word.offsetByCodePoints(i, 1)) {
final int codePoint = word.codePointAt(i);
@@ -428,7 +469,7 @@
* @param group the group
* @return the size of the char array, including the terminator if any
*/
- private static int getGroupCharactersSize(CharGroup group) {
+ private static int getGroupCharactersSize(final CharGroup group) {
int size = CharEncoding.getCharArraySize(group.mChars);
if (group.hasSeveralChars()) size += GROUP_TERMINATOR_SIZE;
return size;
@@ -446,7 +487,7 @@
return 2;
} else {
throw new RuntimeException("Can't have more than " + MAX_CHARGROUPS_IN_A_NODE
- + " groups in a node (found " + count +")");
+ + " groups in a node (found " + count + ")");
}
}
@@ -493,10 +534,11 @@
* Compute the maximum size of a CharGroup, assuming 3-byte addresses for everything.
*
* @param group the CharGroup to compute the size of.
+ * @param options file format options.
* @return the maximum size of the group.
*/
- private static int getCharGroupMaximumSize(CharGroup group) {
- int size = getGroupCharactersSize(group) + GROUP_FLAGS_SIZE;
+ private static int getCharGroupMaximumSize(final CharGroup group, final FormatOptions options) {
+ int size = getGroupHeaderSize(group, options);
// If terminal, one byte for the frequency
if (group.isTerminal()) size += GROUP_FREQUENCY_SIZE;
size += GROUP_MAX_ADDRESS_SIZE; // For children address
@@ -513,11 +555,12 @@
* it in the 'actualSize' member of the node.
*
* @param node the node to compute the maximum size of.
+ * @param options file format options.
*/
- private static void setNodeMaximumSize(Node node) {
+ private static void setNodeMaximumSize(final Node node, final FormatOptions options) {
int size = getGroupCountSize(node);
for (CharGroup g : node.mData) {
- final int groupSize = getCharGroupMaximumSize(g);
+ final int groupSize = getCharGroupMaximumSize(g, options);
g.mCachedSize = groupSize;
size += groupSize;
}
@@ -527,11 +570,33 @@
/**
* Helper method to hide the actual value of the no children address.
*/
- private static boolean hasChildrenAddress(int address) {
+ private static boolean hasChildrenAddress(final int address) {
return NO_CHILDREN_ADDRESS != address;
}
/**
+ * Helper method to check whether the CharGroup has a parent address.
+ */
+ private static boolean hasParentAddress(final FormatOptions options) {
+ return options.mVersion >= FIRST_VERSION_WITH_PARENT_ADDRESS
+ && options.mHasParentAddress;
+ }
+
+ /**
+ * Compute the size of the header (flag + [parent address] + characters size) of a CharGroup.
+ *
+ * @param group the group of which to compute the size of the header
+ * @param options file format options.
+ */
+ private static int getGroupHeaderSize(final CharGroup group, final FormatOptions options) {
+ if (hasParentAddress(options)) {
+ return GROUP_FLAGS_SIZE + PARENT_ADDRESS_SIZE + getGroupCharactersSize(group);
+ } else {
+ return GROUP_FLAGS_SIZE + getGroupCharactersSize(group);
+ }
+ }
+
+ /**
* Compute the size, in bytes, that an address will occupy.
*
* This can be used either for children addresses (which are always positive) or for
@@ -541,7 +606,7 @@
* @param address the address
* @return the byte size.
*/
- private static int getByteSize(int address) {
+ private static int getByteSize(final int address) {
assert(address < 0x1000000);
if (!hasChildrenAddress(address)) {
return 0;
@@ -557,14 +622,14 @@
// This method is responsible for finding a nice ordering of the nodes that favors run-time
// cache performance and dictionary size.
- /* package for tests */ static ArrayList<Node> flattenTree(Node root) {
+ /* package for tests */ static ArrayList<Node> flattenTree(final Node root) {
final int treeSize = FusionDictionary.countCharGroups(root);
MakedictLog.i("Counted nodes : " + treeSize);
final ArrayList<Node> flatTree = new ArrayList<Node>(treeSize);
return flattenTreeInner(flatTree, root);
}
- private static ArrayList<Node> flattenTreeInner(ArrayList<Node> list, Node node) {
+ private static ArrayList<Node> flattenTreeInner(final ArrayList<Node> list, final Node node) {
// Removing the node is necessary if the tails are merged, because we would then
// add the same node several times when we only want it once. A number of places in
// the code also depends on any node being only once in the list.
@@ -614,9 +679,11 @@
*
* @param node the node to compute the size of.
* @param dict the dictionary in which the word/attributes are to be found.
+ * @param formatOptions file format options.
* @return false if none of the cached addresses inside the node changed, true otherwise.
*/
- private static boolean computeActualNodeSize(Node node, FusionDictionary dict) {
+ private static boolean computeActualNodeSize(final Node node, final FusionDictionary dict,
+ final FormatOptions formatOptions) {
boolean changed = false;
int size = getGroupCountSize(node);
for (CharGroup group : node.mData) {
@@ -624,11 +691,14 @@
changed = true;
group.mCachedAddress = node.mCachedAddress + size;
}
- int groupSize = GROUP_FLAGS_SIZE + getGroupCharactersSize(group);
+ int groupSize = getGroupHeaderSize(group, formatOptions);
if (group.isTerminal()) groupSize += GROUP_FREQUENCY_SIZE;
if (null != group.mChildren) {
- final int offsetBasePoint= groupSize + node.mCachedAddress + size;
+ final int offsetBasePoint = groupSize + node.mCachedAddress + size;
final int offset = group.mChildren.mCachedAddress - offsetBasePoint;
+ // assign my address to children's parent address
+ group.mChildren.mCachedParentAddress = group.mCachedAddress
+ - group.mChildren.mCachedAddress;
groupSize += getByteSize(offset);
}
groupSize += getShortcutListSize(group.mShortcutTargets);
@@ -657,7 +727,7 @@
* @param flatNodes the array of nodes.
* @return the byte size of the entire stack.
*/
- private static int stackNodes(ArrayList<Node> flatNodes) {
+ private static int stackNodes(final ArrayList<Node> flatNodes) {
int nodeOffset = 0;
for (Node n : flatNodes) {
n.mCachedAddress = nodeOffset;
@@ -687,12 +757,13 @@
*
* @param dict the dictionary
* @param flatNodes the ordered array of nodes
+ * @param formatOptions file format options.
* @return the same array it was passed. The nodes have been updated for address and size.
*/
- private static ArrayList<Node> computeAddresses(FusionDictionary dict,
- ArrayList<Node> flatNodes) {
+ private static ArrayList<Node> computeAddresses(final FusionDictionary dict,
+ final ArrayList<Node> flatNodes, final FormatOptions formatOptions) {
// First get the worst sizes and offsets
- for (Node n : flatNodes) setNodeMaximumSize(n);
+ for (Node n : flatNodes) setNodeMaximumSize(n, formatOptions);
final int offset = stackNodes(flatNodes);
MakedictLog.i("Compressing the array addresses. Original size : " + offset);
@@ -704,7 +775,7 @@
changesDone = false;
for (Node n : flatNodes) {
final int oldNodeSize = n.mCachedSize;
- final boolean changed = computeActualNodeSize(n, dict);
+ final boolean changed = computeActualNodeSize(n, dict, formatOptions);
final int newNodeSize = n.mCachedSize;
if (oldNodeSize < newNodeSize) throw new RuntimeException("Increased size ?!");
changesDone |= changed;
@@ -732,7 +803,7 @@
*
* @param array the array node to check
*/
- private static void checkFlatNodeArray(ArrayList<Node> array) {
+ private static void checkFlatNodeArray(final ArrayList<Node> array) {
int offset = 0;
int index = 0;
for (Node n : array) {
@@ -890,12 +961,14 @@
/**
* Makes the 2-byte value for options flags.
*/
- private static final int makeOptionsValue(final FusionDictionary dictionary) {
+ private static final int makeOptionsValue(final FusionDictionary dictionary,
+ final FormatOptions formatOptions) {
final DictionaryOptions options = dictionary.mOptions;
final boolean hasBigrams = dictionary.hasBigrams();
return (options.mFrenchLigatureProcessing ? FRENCH_LIGATURE_PROCESSING_FLAG : 0)
+ (options.mGermanUmlautProcessing ? GERMAN_UMLAUT_PROCESSING_FLAG : 0)
- + (hasBigrams ? CONTAINS_BIGRAMS_FLAG : 0);
+ + (hasBigrams ? CONTAINS_BIGRAMS_FLAG : 0)
+ + (formatOptions.mHasParentAddress ? HAS_PARENT_ADDRESS : 0);
}
/**
@@ -918,13 +991,16 @@
* @param dict the dictionary the node is a part of (for relative offsets).
* @param buffer the memory buffer to write to.
* @param node the node to write.
+ * @param formatOptions file format options.
* @return the address of the END of the node.
*/
- private static int writePlacedNode(FusionDictionary dict, byte[] buffer, Node node) {
+ private static int writePlacedNode(final FusionDictionary dict, byte[] buffer,
+ final Node node, final FormatOptions formatOptions) {
int index = node.mCachedAddress;
final int groupCount = node.mData.size();
final int countSize = getGroupCountSize(node);
+ final int parentAddress = node.mCachedParentAddress;
if (1 == countSize) {
buffer[index++] = (byte)groupCount;
} else if (2 == countSize) {
@@ -941,7 +1017,7 @@
if (index != group.mCachedAddress) throw new RuntimeException("Bug: write index is not "
+ "the same as the cached address of the group : "
+ index + " <> " + group.mCachedAddress);
- groupAddress += GROUP_FLAGS_SIZE + getGroupCharactersSize(group);
+ groupAddress += getGroupHeaderSize(group, formatOptions);
// Sanity checks.
if (DBG && group.mFrequency > MAX_TERMINAL_FREQUENCY) {
throw new RuntimeException("A node has a frequency > " + MAX_TERMINAL_FREQUENCY
@@ -952,6 +1028,22 @@
? NO_CHILDREN_ADDRESS : group.mChildren.mCachedAddress - groupAddress;
byte flags = makeCharGroupFlags(group, groupAddress, childrenOffset);
buffer[index++] = flags;
+
+ if (hasParentAddress(formatOptions)) {
+ if (parentAddress == NO_PARENT_ADDRESS) {
+ // this node is the root node.
+ buffer[index] = buffer[index + 1] = buffer[index + 2] = 0;
+ } else {
+ // write parent address. (version 3)
+ final int actualParentAddress = Math.abs(parentAddress
+ + (node.mCachedAddress - group.mCachedAddress));
+ buffer[index] = (byte)((actualParentAddress >> 16) & 0xFF);
+ buffer[index + 1] = (byte)((actualParentAddress >> 8) & 0xFF);
+ buffer[index + 2] = (byte)(actualParentAddress & 0xFF);
+ }
+ index += 3;
+ }
+
index = CharEncoding.writeCharArray(group.mChars, buffer, index);
if (group.hasSeveralChars()) {
buffer[index++] = GROUP_CHARACTERS_TERMINATOR;
@@ -1076,7 +1168,7 @@
*
* @param destination the stream to write the binary data to.
* @param dict the dictionary to write.
- * @param formatOptions the options of file format.
+ * @param formatOptions file format options.
*/
public static void writeDictionaryBinary(final OutputStream destination,
final FusionDictionary dict, final FormatOptions formatOptions)
@@ -1115,7 +1207,7 @@
headerBuffer.write((byte) (0xFF & version));
}
// Options flags
- final int options = makeOptionsValue(dict);
+ final int options = makeOptionsValue(dict, formatOptions);
headerBuffer.write((byte) (0xFF & (options >> 8)));
headerBuffer.write((byte) (0xFF & options));
if (version >= FIRST_VERSION_WITH_HEADER_SIZE) {
@@ -1149,20 +1241,20 @@
ArrayList<Node> flatNodes = flattenTree(dict.mRoot);
MakedictLog.i("Computing addresses...");
- computeAddresses(dict, flatNodes);
+ computeAddresses(dict, flatNodes, formatOptions);
MakedictLog.i("Checking array...");
if (DBG) checkFlatNodeArray(flatNodes);
// Create a buffer that matches the final dictionary size.
final Node lastNode = flatNodes.get(flatNodes.size() - 1);
- final int bufferSize =(lastNode.mCachedAddress + lastNode.mCachedSize);
+ final int bufferSize = lastNode.mCachedAddress + lastNode.mCachedSize;
final byte[] buffer = new byte[bufferSize];
int index = 0;
MakedictLog.i("Writing file...");
int dataEndOffset = 0;
for (Node n : flatNodes) {
- dataEndOffset = writePlacedNode(dict, buffer, n);
+ dataEndOffset = writePlacedNode(dict, buffer, n, formatOptions);
}
if (DBG) showStatistics(flatNodes);
@@ -1177,23 +1269,36 @@
// Input methods: Read a binary dictionary to memory.
// readDictionaryBinary is the public entry point for them.
- static final int[] characterBuffer = new int[MAX_WORD_LENGTH];
+ private static final int[] CHARACTER_BUFFER = new int[MAX_WORD_LENGTH];
private static CharGroupInfo readCharGroup(final FusionDictionaryBufferInterface buffer,
- final int originalGroupAddress) {
+ final int originalGroupAddress, final FormatOptions options) {
int addressPointer = originalGroupAddress;
final int flags = buffer.readUnsignedByte();
++addressPointer;
+
+ final int parentAddress;
+ if (hasParentAddress(options)) {
+ // read the parent address. (version 3)
+ parentAddress = -buffer.readUnsignedInt24();
+ addressPointer += 3;
+ } else {
+ parentAddress = NO_PARENT_ADDRESS;
+ }
+
final int characters[];
if (0 != (flags & FLAG_HAS_MULTIPLE_CHARS)) {
int index = 0;
int character = CharEncoding.readChar(buffer);
addressPointer += CharEncoding.getCharSize(character);
while (-1 != character) {
- characterBuffer[index++] = character;
+ // FusionDictionary is making sure that the length of the word is smaller than
+ // MAX_WORD_LENGTH.
+ // So we'll never write past the end of CHARACTER_BUFFER.
+ CHARACTER_BUFFER[index++] = character;
character = CharEncoding.readChar(buffer);
addressPointer += CharEncoding.getCharSize(character);
}
- characters = Arrays.copyOfRange(characterBuffer, 0, index);
+ characters = Arrays.copyOfRange(CHARACTER_BUFFER, 0, index);
} else {
final int character = CharEncoding.readChar(buffer);
addressPointer += CharEncoding.getCharSize(character);
@@ -1271,7 +1376,7 @@
}
}
return new CharGroupInfo(originalGroupAddress, addressPointer, flags, characters, frequency,
- childrenAddress, shortcutTargets, bigrams);
+ parentAddress, childrenAddress, shortcutTargets, bigrams);
}
/**
@@ -1298,13 +1403,56 @@
* @param buffer the buffer to read from.
* @param headerSize the size of the header.
* @param address the address to seek.
+ * @param formatOptions file format options.
* @return the word, as a string.
*/
private static String getWordAtAddress(final FusionDictionaryBufferInterface buffer,
- final int headerSize, final int address) {
+ final int headerSize, final int address, final FormatOptions formatOptions) {
final String cachedString = wordCache.get(address);
if (null != cachedString) return cachedString;
+
+ final String result;
final int originalPointer = buffer.position();
+
+ if (hasParentAddress(formatOptions)) {
+ result = getWordAtAddressWithParentAddress(buffer, headerSize, address, formatOptions);
+ } else {
+ result = getWordAtAddressWithoutParentAddress(buffer, headerSize, address,
+ formatOptions);
+ }
+
+ wordCache.put(address, result);
+ buffer.position(originalPointer);
+ return result;
+ }
+
+ private static int[] sGetWordBuffer = new int[MAX_WORD_LENGTH];
+ private static String getWordAtAddressWithParentAddress(
+ final FusionDictionaryBufferInterface buffer, final int headerSize, final int address,
+ final FormatOptions options) {
+ final StringBuilder builder = new StringBuilder();
+
+ int currentAddress = address;
+ int index = MAX_WORD_LENGTH - 1;
+ // the length of the path from the root to the leaf is limited by MAX_WORD_LENGTH
+ for (int count = 0; count < MAX_WORD_LENGTH; ++count) {
+ buffer.position(currentAddress + headerSize);
+ final CharGroupInfo currentInfo = readCharGroup(buffer, currentAddress, options);
+ for (int i = 0; i < currentInfo.mCharacters.length; ++i) {
+ sGetWordBuffer[index--] =
+ currentInfo.mCharacters[currentInfo.mCharacters.length - i - 1];
+ }
+
+ if (currentInfo.mParentAddress == NO_PARENT_ADDRESS) break;
+ currentAddress = currentInfo.mParentAddress + currentInfo.mOriginalAddress;
+ }
+
+ return new String(sGetWordBuffer, index + 1, MAX_WORD_LENGTH - index - 1);
+ }
+
+ private static String getWordAtAddressWithoutParentAddress(
+ final FusionDictionaryBufferInterface buffer, final int headerSize, final int address,
+ final FormatOptions options) {
buffer.position(headerSize);
final int count = readCharGroupCount(buffer);
int groupOffset = getGroupCountSize(count);
@@ -1313,7 +1461,7 @@
CharGroupInfo last = null;
for (int i = count - 1; i >= 0; --i) {
- CharGroupInfo info = readCharGroup(buffer, groupOffset);
+ CharGroupInfo info = readCharGroup(buffer, groupOffset, options);
groupOffset = info.mEndAddress;
if (info.mOriginalAddress == address) {
builder.append(new String(info.mCharacters, 0, info.mCharacters.length));
@@ -1341,8 +1489,6 @@
continue;
}
}
- buffer.position(originalPointer);
- wordCache.put(address, result);
return result;
}
@@ -1358,24 +1504,26 @@
* @param headerSize the size, in bytes, of the file header.
* @param reverseNodeMap a mapping from addresses to already read nodes.
* @param reverseGroupMap a mapping from addresses to already read character groups.
+ * @param options file format options.
* @return the read node with all his children already read.
*/
private static Node readNode(final FusionDictionaryBufferInterface buffer, final int headerSize,
- final Map<Integer, Node> reverseNodeMap, final Map<Integer, CharGroup> reverseGroupMap)
+ final Map<Integer, Node> reverseNodeMap, final Map<Integer, CharGroup> reverseGroupMap,
+ final FormatOptions options)
throws IOException {
final int nodeOrigin = buffer.position() - headerSize;
final int count = readCharGroupCount(buffer);
final ArrayList<CharGroup> nodeContents = new ArrayList<CharGroup>();
int groupOffset = nodeOrigin + getGroupCountSize(count);
for (int i = count; i > 0; --i) {
- CharGroupInfo info = readCharGroup(buffer, groupOffset);
+ CharGroupInfo info = readCharGroup(buffer, groupOffset, options);
ArrayList<WeightedString> shortcutTargets = info.mShortcutTargets;
ArrayList<WeightedString> bigrams = null;
if (null != info.mBigrams) {
bigrams = new ArrayList<WeightedString>();
for (PendingAttribute bigram : info.mBigrams) {
final String word = getWordAtAddress(
- buffer, headerSize, bigram.mAddress);
+ buffer, headerSize, bigram.mAddress, options);
bigrams.add(new WeightedString(word, bigram.mFrequency));
}
}
@@ -1385,7 +1533,7 @@
final int currentPosition = buffer.position();
buffer.position(info.mChildrenAddress + headerSize);
children = readNode(
- buffer, headerSize, reverseNodeMap, reverseGroupMap);
+ buffer, headerSize, reverseNodeMap, reverseGroupMap, options);
buffer.position(currentPosition);
}
nodeContents.add(
@@ -1429,7 +1577,8 @@
private static void readUnigramsAndBigramsBinaryInner(
final FusionDictionaryBufferInterface buffer, final int headerSize,
final Map<Integer, String> words, final Map<Integer, Integer> frequencies,
- final Map<Integer, ArrayList<PendingAttribute>> bigrams) {
+ final Map<Integer, ArrayList<PendingAttribute>> bigrams,
+ final FormatOptions formatOptions) {
int[] pushedChars = new int[MAX_WORD_LENGTH + 1];
Stack<Position> stack = new Stack<Position>();
@@ -1455,7 +1604,7 @@
p.mPosition = 0;
}
- CharGroupInfo info = readCharGroup(buffer, p.mAddress - headerSize);
+ CharGroupInfo info = readCharGroup(buffer, p.mAddress - headerSize, formatOptions);
for (int i = 0; i < info.mCharacters.length; ++i) {
pushedChars[index++] = info.mCharacters[i];
}
@@ -1497,11 +1646,9 @@
final Map<Integer, ArrayList<PendingAttribute>> bigrams) throws IOException,
UnsupportedFormatException {
// Read header
- FormatOptions formatOptions = null;
- DictionaryOptions dictionaryOptions = null;
final FileHeader header = readHeader(buffer);
-
- readUnigramsAndBigramsBinaryInner(buffer, header.mHeaderSize, words, frequencies, bigrams);
+ readUnigramsAndBigramsBinaryInner(buffer, header.mHeaderSize, words, frequencies, bigrams,
+ header.mFormatOptions);
}
/**
@@ -1562,7 +1709,8 @@
new FusionDictionary.DictionaryOptions(attributes,
0 != (optionsFlags & GERMAN_UMLAUT_PROCESSING_FLAG),
0 != (optionsFlags & FRENCH_LIGATURE_PROCESSING_FLAG)),
- new FormatOptions(version));
+ new FormatOptions(version,
+ 0 != (optionsFlags & HAS_PARENT_ADDRESS)));
return header;
}
@@ -1609,7 +1757,7 @@
Map<Integer, Node> reverseNodeMapping = new TreeMap<Integer, Node>();
Map<Integer, CharGroup> reverseGroupMapping = new TreeMap<Integer, CharGroup>();
final Node root = readNode(buffer, header.mHeaderSize, reverseNodeMapping,
- reverseGroupMapping);
+ reverseGroupMapping, header.mFormatOptions);
FusionDictionary newDict = new FusionDictionary(root, header.mDictionaryOptions);
if (null != dict) {
diff --git a/java/src/com/android/inputmethod/latin/makedict/CharGroupInfo.java b/java/src/com/android/inputmethod/latin/makedict/CharGroupInfo.java
index ef7dbb2..ed93884 100644
--- a/java/src/com/android/inputmethod/latin/makedict/CharGroupInfo.java
+++ b/java/src/com/android/inputmethod/latin/makedict/CharGroupInfo.java
@@ -31,18 +31,20 @@
public final int[] mCharacters;
public final int mFrequency;
public final int mChildrenAddress;
+ public final int mParentAddress;
public final ArrayList<WeightedString> mShortcutTargets;
public final ArrayList<PendingAttribute> mBigrams;
public CharGroupInfo(final int originalAddress, final int endAddress, final int flags,
- final int[] characters, final int frequency, final int childrenAddress,
- final ArrayList<WeightedString> shortcutTargets,
+ final int[] characters, final int frequency, final int parentAddress,
+ final int childrenAddress, final ArrayList<WeightedString> shortcutTargets,
final ArrayList<PendingAttribute> bigrams) {
mOriginalAddress = originalAddress;
mEndAddress = endAddress;
mFlags = flags;
mCharacters = characters;
mFrequency = frequency;
+ mParentAddress = parentAddress;
mChildrenAddress = childrenAddress;
mShortcutTargets = shortcutTargets;
mBigrams = bigrams;
diff --git a/java/src/com/android/inputmethod/latin/makedict/FusionDictionary.java b/java/src/com/android/inputmethod/latin/makedict/FusionDictionary.java
index 61f7371..6775de8 100644
--- a/java/src/com/android/inputmethod/latin/makedict/FusionDictionary.java
+++ b/java/src/com/android/inputmethod/latin/makedict/FusionDictionary.java
@@ -16,6 +16,8 @@
package com.android.inputmethod.latin.makedict;
+import com.android.inputmethod.latin.Constants;
+
import java.util.ArrayList;
import java.util.Arrays;
import java.util.Collections;
@@ -41,17 +43,15 @@
public static class Node {
ArrayList<CharGroup> mData;
// To help with binary generation
- int mCachedSize;
- int mCachedAddress;
+ int mCachedSize = Integer.MIN_VALUE;
+ int mCachedAddress = Integer.MIN_VALUE;
+ int mCachedParentAddress = 0;
+
public Node() {
mData = new ArrayList<CharGroup>();
- mCachedSize = Integer.MIN_VALUE;
- mCachedAddress = Integer.MIN_VALUE;
}
public Node(ArrayList<CharGroup> data) {
mData = data;
- mCachedSize = Integer.MIN_VALUE;
- mCachedAddress = Integer.MIN_VALUE;
}
}
@@ -382,6 +382,11 @@
final ArrayList<WeightedString> shortcutTargets,
final boolean isNotAWord, final boolean isBlacklistEntry) {
assert(frequency >= 0 && frequency <= 255);
+ if (word.length >= Constants.Dictionary.MAX_WORD_LENGTH) {
+ MakedictLog.w("Ignoring a word that is too long: word.length = " + word.length);
+ return;
+ }
+
Node currentNode = mRoot;
int charIndex = 0;
diff --git a/native/jni/src/defines.h b/native/jni/src/defines.h
index 0286365..95a9027 100644
--- a/native/jni/src/defines.h
+++ b/native/jni/src/defines.h
@@ -31,6 +31,9 @@
dumpResult(words, frequencies, maxWordCount, maxWordLength); } while (0)
#define DUMP_WORD(word, length) do { dumpWord(word, length); } while (0)
#define DUMP_WORD_INT(word, length) do { dumpWordInt(word, length); } while (0)
+// TODO: INTS_TO_CHARS
+#define SHORTS_TO_CHARS(input, length, output) do { \
+ shortArrayToCharArray(input, length, output); } while (0)
static inline void dumpWordInfo(const unsigned short *word, const int length,
const int rank, const int frequency) {
@@ -87,12 +90,29 @@
AKLOGI("i[ %s ]", charBuf);
}
-#ifndef __ANDROID__
-#define ASSERT(success) do { if (!success) { showStackTrace(); assert(success);};} while (0)
-#define SHOW_STACK_TRACE do { showStackTrace(); } while (0)
+// TODO: Change this to intArrayToCharArray
+static inline void shortArrayToCharArray(
+ const unsigned short *input, const int length, char *output) {
+ int i = 0;
+ for (;i < length; ++i) {
+ const unsigned short c = input[i];
+ if (c == 0) {
+ break;
+ }
+ // static_cast only for debugging
+ output[i] = static_cast<char>(c);
+ }
+ output[i] = 0;
+}
+#ifndef __ANDROID__
+#include <cassert>
#include <execinfo.h>
#include <stdlib.h>
+
+#define ASSERT(success) do { if (!(success)) { showStackTrace(); assert(success);} } while (0)
+#define SHOW_STACK_TRACE do { showStackTrace(); } while (0)
+
static inline void showStackTrace() {
void *callstack[128];
int i, frames = backtrace(callstack, 128);
@@ -107,7 +127,8 @@
free(strs);
}
#else
-#define ASSERT(success)
+#include <cassert>
+#define ASSERT(success) assert(success)
#define SHOW_STACK_TRACE
#endif
@@ -119,6 +140,8 @@
#define DUMP_WORD_INT(word, length)
#define ASSERT(success)
#define SHOW_STACK_TRACE
+// TODO: INTS_TO_CHARS
+#define SHORTS_TO_CHARS(input, length, output)
#endif
#ifdef FLAG_DO_PROFILE
@@ -197,6 +220,8 @@
#define DEBUG_CORRECTION_FREQ false
#define DEBUG_WORDS_PRIORITY_QUEUE false
+#define DEBUG_GEO_FULL false
+
#else // FLAG_DBG
#define DEBUG_DICT false
@@ -211,6 +236,8 @@
#define DEBUG_CORRECTION_FREQ false
#define DEBUG_WORDS_PRIORITY_QUEUE false
+#define DEBUG_GEO_FULL false
+
#endif // FLAG_DBG
#ifndef U_SHORT_MAX
diff --git a/native/jni/src/proximity_info.cpp b/native/jni/src/proximity_info.cpp
index a8c0430..9bb8b29 100644
--- a/native/jni/src/proximity_info.cpp
+++ b/native/jni/src/proximity_info.cpp
@@ -275,22 +275,22 @@
}
}
-float ProximityInfo::getKeyCenterXOfCodePointG(int charCode) const {
+int ProximityInfo::getKeyCenterXOfCodePointG(int charCode) const {
return getKeyCenterXOfKeyIdG(getKeyIndexOf(charCode));
}
-float ProximityInfo::getKeyCenterYOfCodePointG(int charCode) const {
+int ProximityInfo::getKeyCenterYOfCodePointG(int charCode) const {
return getKeyCenterYOfKeyIdG(getKeyIndexOf(charCode));
}
-float ProximityInfo::getKeyCenterXOfKeyIdG(int keyId) const {
+int ProximityInfo::getKeyCenterXOfKeyIdG(int keyId) const {
if (keyId >= 0) {
return mCenterXsG[keyId];
}
return 0;
}
-float ProximityInfo::getKeyCenterYOfKeyIdG(int keyId) const {
+int ProximityInfo::getKeyCenterYOfKeyIdG(int keyId) const {
if (keyId >= 0) {
return mCenterYsG[keyId];
}
diff --git a/native/jni/src/proximity_info.h b/native/jni/src/proximity_info.h
index 7c22e10..45df6ff 100644
--- a/native/jni/src/proximity_info.h
+++ b/native/jni/src/proximity_info.h
@@ -104,11 +104,10 @@
return KEYBOARD_HEIGHT;
}
- // TODO: These should return int.
- float getKeyCenterXOfCodePointG(int charCode) const;
- float getKeyCenterYOfCodePointG(int charCode) const;
- float getKeyCenterXOfKeyIdG(int keyId) const;
- float getKeyCenterYOfKeyIdG(int keyId) const;
+ int getKeyCenterXOfCodePointG(int charCode) const;
+ int getKeyCenterYOfCodePointG(int charCode) const;
+ int getKeyCenterXOfKeyIdG(int keyId) const;
+ int getKeyCenterYOfKeyIdG(int keyId) const;
int getKeyKeyDistanceG(int key0, int key1) const;
private:
diff --git a/native/jni/src/proximity_info_state.cpp b/native/jni/src/proximity_info_state.cpp
index 8c3174c..a4eb7e3 100644
--- a/native/jni/src/proximity_info_state.cpp
+++ b/native/jni/src/proximity_info_state.cpp
@@ -29,6 +29,14 @@
const ProximityInfo *proximityInfo, const int32_t *const inputCodes, const int inputSize,
const int *const xCoordinates, const int *const yCoordinates, const int *const times,
const int *const pointerIds, const bool isGeometric) {
+
+ if (isGeometric) {
+ mIsContinuationPossible = checkAndReturnIsContinuationPossible(
+ inputSize, xCoordinates, yCoordinates, times);
+ } else {
+ mIsContinuationPossible = false;
+ }
+
mProximityInfo = proximityInfo;
mHasTouchPositionCorrectionData = proximityInfo->hasTouchPositionCorrectionData();
mMostCommonKeyWidthSquare = proximityInfo->getMostCommonKeyWidthSquare();
@@ -70,19 +78,32 @@
///////////////////////
// Setup touch points
+ int pushTouchPointStartIndex = 0;
+ int lastSavedInputSize = 0;
mMaxPointToKeyLength = maxPointToKeyLength;
- mInputXs.clear();
- mInputYs.clear();
- mTimes.clear();
- mLengthCache.clear();
- mDistanceCache.clear();
- mNearKeysVector.clear();
+ if (mIsContinuationPossible && mInputIndice.size() > 1) {
+ // Just update difference.
+ // Two points prior is never skipped. Thus, we pop 2 input point data here.
+ pushTouchPointStartIndex = mInputIndice[mInputIndice.size() - 2];
+ popInputData();
+ popInputData();
+ lastSavedInputSize = mInputXs.size();
+ } else {
+ // Clear all data.
+ mInputXs.clear();
+ mInputYs.clear();
+ mTimes.clear();
+ mInputIndice.clear();
+ mLengthCache.clear();
+ mDistanceCache.clear();
+ mNearKeysVector.clear();
+ }
mInputSize = 0;
if (xCoordinates && yCoordinates) {
const bool proximityOnly = !isGeometric && (xCoordinates[0] < 0 || yCoordinates[0] < 0);
- int lastInputIndex = 0;
- for (int i = 0; i < inputSize; ++i) {
+ int lastInputIndex = pushTouchPointStartIndex;
+ for (int i = lastInputIndex; i < inputSize; ++i) {
const int pid = pointerIds ? pointerIds[i] : 0;
if (pointerId == pid) {
lastInputIndex = i;
@@ -95,7 +116,7 @@
NearKeysDistanceMap *prevNearKeysDistances = &nearKeysDistances[1];
NearKeysDistanceMap *prevPrevNearKeysDistances = &nearKeysDistances[2];
- for (int i = 0; i < inputSize; ++i) {
+ for (int i = pushTouchPointStartIndex; i <= lastInputIndex; ++i) {
// Assuming pointerId == 0 if pointerIds is null.
const int pid = pointerIds ? pointerIds[i] : 0;
if (pointerId == pid) {
@@ -103,7 +124,7 @@
const int x = proximityOnly ? NOT_A_COORDINATE : xCoordinates[i];
const int y = proximityOnly ? NOT_A_COORDINATE : yCoordinates[i];
const int time = times ? times[i] : -1;
- if (pushTouchPoint(c, x, y, time, isGeometric, i == lastInputIndex,
+ if (pushTouchPoint(i, c, x, y, time, isGeometric, i == lastInputIndex,
currentNearKeysDistances, prevNearKeysDistances,
prevPrevNearKeysDistances)) {
// Previous point information was popped.
@@ -125,7 +146,7 @@
const int keyCount = mProximityInfo->getKeyCount();
mNearKeysVector.resize(mInputSize);
mDistanceCache.resize(mInputSize * keyCount);
- for (int i = 0; i < mInputSize; ++i) {
+ for (int i = lastSavedInputSize; i < mInputSize; ++i) {
mNearKeysVector[i].reset();
static const float NEAR_KEY_NORMALIZED_SQUARED_THRESHOLD = 4.0f;
for (int k = 0; k < keyCount; ++k) {
@@ -146,7 +167,7 @@
hypotf(mProximityInfo->getKeyboardWidth(), mProximityInfo->getKeyboardHeight())
* READ_FORWORD_LENGTH_SCALE);
for (int i = 0; i < mInputSize; ++i) {
- for (int j = i + 1; j < mInputSize; ++j) {
+ for (int j = max(i + 1, lastSavedInputSize); j < mInputSize; ++j) {
if (mLengthCache[j] - mLengthCache[i] >= readForwordLength) {
break;
}
@@ -199,6 +220,18 @@
}
}
+bool ProximityInfoState::checkAndReturnIsContinuationPossible(const int inputSize,
+ const int *const xCoordinates, const int *const yCoordinates, const int *const times) {
+ for (int i = 0; i < mInputSize; ++i) {
+ const int index = mInputIndice[i];
+ if (index > inputSize || xCoordinates[index] != mInputXs[i] ||
+ yCoordinates[index] != mInputYs[i] || times[index] != mTimes[i]) {
+ return false;
+ }
+ }
+ return true;
+}
+
// Calculating point to key distance for all near keys and returning the distance between
// the given point and the nearest key position.
float ProximityInfoState::updateNearKeysDistances(const int x, const int y,
@@ -305,8 +338,8 @@
// Sampling touch point and pushing information to vectors.
// Returning if previous point is popped or not.
-bool ProximityInfoState::pushTouchPoint(const int nodeChar, int x, int y, const int time,
- const bool sample, const bool isLastPoint,
+bool ProximityInfoState::pushTouchPoint(const int inputIndex, const int nodeChar, int x, int y,
+ const int time, const bool sample, const bool isLastPoint,
NearKeysDistanceMap *const currentNearKeysDistances,
const NearKeysDistanceMap *const prevNearKeysDistances,
const NearKeysDistanceMap *const prevPrevNearKeysDistances) {
@@ -320,10 +353,7 @@
currentNearKeysDistances, prevNearKeysDistances, prevPrevNearKeysDistances);
if (score < 0) {
// Pop previous point because it would be useless.
- mInputXs.pop_back();
- mInputYs.pop_back();
- mTimes.pop_back();
- mLengthCache.pop_back();
+ popInputData();
size = mInputXs.size();
popped = true;
} else {
@@ -371,6 +401,7 @@
mInputXs.push_back(x);
mInputYs.push_back(y);
mTimes.push_back(time);
+ mInputIndice.push_back(inputIndex);
return popped;
}
@@ -461,4 +492,12 @@
return static_cast<float>(mTimes[mInputSize - 1] - mTimes[0]) / static_cast<float>(mInputSize);
}
+void ProximityInfoState::popInputData() {
+ mInputXs.pop_back();
+ mInputYs.pop_back();
+ mTimes.pop_back();
+ mLengthCache.pop_back();
+ mInputIndice.pop_back();
+}
+
} // namespace latinime
diff --git a/native/jni/src/proximity_info_state.h b/native/jni/src/proximity_info_state.h
index 1d57773..fce4b5b 100644
--- a/native/jni/src/proximity_info_state.h
+++ b/native/jni/src/proximity_info_state.h
@@ -57,9 +57,9 @@
: mProximityInfo(0), mMaxPointToKeyLength(0),
mHasTouchPositionCorrectionData(false), mMostCommonKeyWidthSquare(0), mLocaleStr(),
mKeyCount(0), mCellHeight(0), mCellWidth(0), mGridHeight(0), mGridWidth(0),
- mInputXs(), mInputYs(), mTimes(), mDistanceCache(), mLengthCache(),
- mNearKeysVector(), mTouchPositionCorrectionEnabled(false),
- mInputSize(0) {
+ mIsContinuationPossible(false), mInputXs(), mInputYs(), mTimes(), mInputIndice(),
+ mDistanceCache(), mLengthCache(), mNearKeysVector(),
+ mTouchPositionCorrectionEnabled(false), mInputSize(0) {
memset(mInputCodes, 0, sizeof(mInputCodes));
memset(mNormalizedSquaredDistances, 0, sizeof(mNormalizedSquaredDistances));
memset(mPrimaryInputWord, 0, sizeof(mPrimaryInputWord));
@@ -212,6 +212,10 @@
return mLengthCache[index];
}
+ bool isContinuationPossible() const {
+ return mIsContinuationPossible;
+ }
+
float getPointToKeyLength(const int inputIndex, const int charCode, const float scale) const;
int getSpaceY() const;
@@ -231,7 +235,7 @@
float calculateSquaredDistanceFromSweetSpotCenter(
const int keyIndex, const int inputIndex) const;
- bool pushTouchPoint(const int nodeChar, int x, int y, const int time,
+ bool pushTouchPoint(const int inputIndex, const int nodeChar, int x, int y, const int time,
const bool sample, const bool isLastPoint,
NearKeysDistanceMap *const currentNearKeysDistances,
const NearKeysDistanceMap *const prevNearKeysDistances,
@@ -259,6 +263,9 @@
const NearKeysDistanceMap *const currentNearKeysDistances,
const NearKeysDistanceMap *const prevNearKeysDistances,
const NearKeysDistanceMap *const prevPrevNearKeysDistances) const;
+ bool checkAndReturnIsContinuationPossible(const int inputSize, const int *const xCoordinates,
+ const int *const yCoordinates, const int *const times);
+ void popInputData();
// const
const ProximityInfo *mProximityInfo;
@@ -271,10 +278,12 @@
int mCellWidth;
int mGridHeight;
int mGridWidth;
+ bool mIsContinuationPossible;
std::vector<int> mInputXs;
std::vector<int> mInputYs;
std::vector<int> mTimes;
+ std::vector<int> mInputIndice;
std::vector<float> mDistanceCache;
std::vector<int> mLengthCache;
std::vector<NearKeycodesSet> mNearKeysVector;
diff --git a/tools/dicttool/Android.mk b/tools/dicttool/Android.mk
index b0b47af..5bd836a 100644
--- a/tools/dicttool/Android.mk
+++ b/tools/dicttool/Android.mk
@@ -16,13 +16,16 @@
LOCAL_PATH := $(call my-dir)
include $(CLEAR_VARS)
-MAKEDICT_CORE_SOURCE_DIRECTORY := ../../java/src/com/android/inputmethod/latin/makedict
+LATINIME_CORE_SOURCE_DIRECTORY := ../../java/src/com/android/inputmethod/latin
+MAKEDICT_CORE_SOURCE_DIRECTORY := $(LATINIME_CORE_SOURCE_DIRECTORY)/makedict
LOCAL_MAIN_SRC_FILES := $(call all-java-files-under,$(MAKEDICT_CORE_SOURCE_DIRECTORY))
LOCAL_TOOL_SRC_FILES := $(call all-java-files-under,src)
LOCAL_SRC_FILES := $(LOCAL_TOOL_SRC_FILES) \
$(filter-out $(addprefix %/, $(notdir $(LOCAL_TOOL_SRC_FILES))), $(LOCAL_MAIN_SRC_FILES)) \
- $(call all-java-files-under,tests)
+ $(call all-java-files-under,tests) \
+ $(LATINIME_CORE_SOURCE_DIRECTORY)/Constants.java
+
LOCAL_JAR_MANIFEST := etc/manifest.txt
LOCAL_MODULE := dicttool_aosp
LOCAL_JAVA_LIBRARIES := junit
diff --git a/tools/dicttool/src/android/inputmethod/latin/dicttool/DictionaryMaker.java b/tools/dicttool/src/android/inputmethod/latin/dicttool/DictionaryMaker.java
index f723852..7cd4564 100644
--- a/tools/dicttool/src/android/inputmethod/latin/dicttool/DictionaryMaker.java
+++ b/tools/dicttool/src/android/inputmethod/latin/dicttool/DictionaryMaker.java
@@ -42,40 +42,35 @@
public class DictionaryMaker {
static class Arguments {
- private final static String OPTION_VERSION_2 = "-2";
- private final static String OPTION_INPUT_SOURCE = "-s";
- private final static String OPTION_INPUT_BIGRAM_XML = "-b";
- private final static String OPTION_INPUT_SHORTCUT_XML = "-c";
- private final static String OPTION_OUTPUT_BINARY = "-d";
- private final static String OPTION_OUTPUT_BINARY_FORMAT_VERSION_1 = "-d1";
- private final static String OPTION_OUTPUT_XML = "-x";
- private final static String OPTION_HELP = "-h";
+ private static final String OPTION_VERSION_1 = "-1";
+ private static final String OPTION_VERSION_2 = "-2";
+ private static final String OPTION_VERSION_3 = "-3";
+ private static final String OPTION_INPUT_SOURCE = "-s";
+ private static final String OPTION_INPUT_BIGRAM_XML = "-b";
+ private static final String OPTION_INPUT_SHORTCUT_XML = "-c";
+ private static final String OPTION_OUTPUT_BINARY = "-d";
+ private static final String OPTION_OUTPUT_XML = "-x";
+ private static final String OPTION_HELP = "-h";
public final String mInputBinary;
public final String mInputUnigramXml;
public final String mInputShortcutXml;
public final String mInputBigramXml;
public final String mOutputBinary;
- public final String mOutputBinaryFormat1;
public final String mOutputXml;
+ public final int mOutputBinaryFormatVersion;
private void checkIntegrity() throws IOException {
checkHasExactlyOneInput();
checkHasAtLeastOneOutput();
checkNotSameFile(mInputBinary, mOutputBinary);
- checkNotSameFile(mInputBinary, mOutputBinaryFormat1);
checkNotSameFile(mInputBinary, mOutputXml);
checkNotSameFile(mInputUnigramXml, mOutputBinary);
- checkNotSameFile(mInputUnigramXml, mOutputBinaryFormat1);
checkNotSameFile(mInputUnigramXml, mOutputXml);
checkNotSameFile(mInputShortcutXml, mOutputBinary);
- checkNotSameFile(mInputShortcutXml, mOutputBinaryFormat1);
checkNotSameFile(mInputShortcutXml, mOutputXml);
checkNotSameFile(mInputBigramXml, mOutputBinary);
- checkNotSameFile(mInputBigramXml, mOutputBinaryFormat1);
checkNotSameFile(mInputBigramXml, mOutputXml);
- checkNotSameFile(mOutputBinary, mOutputBinaryFormat1);
checkNotSameFile(mOutputBinary, mOutputXml);
- checkNotSameFile(mOutputBinaryFormat1, mOutputXml);
}
private void checkHasExactlyOneInput() {
@@ -90,7 +85,7 @@
}
private void checkHasAtLeastOneOutput() {
- if (null == mOutputBinary && null == mOutputBinaryFormat1 && null == mOutputXml) {
+ if (null == mOutputBinary && null == mOutputXml) {
throw new RuntimeException("No output specified");
}
}
@@ -114,13 +109,13 @@
public static String getHelp() {
return "Usage: makedict "
+ "[-s <unigrams.xml> [-b <bigrams.xml>] [-c <shortcuts_and_whitelist.xml>] "
- + "| -s <binary input>] [-d <binary output format version 2>] "
- + "[-d1 <binary output format version 1>] [-x <xml output>] [-2]\n"
+ + "| [-s <binary input>] [-d <binary output>] [-x <xml output>] "
+ + "[-1] [-2] [-3]\n"
+ "\n"
+ " Converts a source dictionary file to one or several outputs.\n"
+ " Source can be an XML file, with an optional XML bigrams file, or a\n"
+ " binary dictionary file.\n"
- + " Binary version 1 (Ice Cream Sandwich), 2 (Jelly Bean) and XML outputs\n"
+ + " Binary version 1 (Ice Cream Sandwich), 2 (Jelly Bean), 3 and XML outputs\n"
+ " are supported. All three can be output at the same time, but the same\n"
+ " output format cannot be specified several times. The behavior is\n"
+ " unspecified if the same file is specified for input and output, or for\n"
@@ -137,8 +132,8 @@
String inputShortcutXml = null;
String inputBigramXml = null;
String outputBinary = null;
- String outputBinaryFormat1 = null;
String outputXml = null;
+ int outputBinaryFormatVersion = 2; // the default version is 2.
while (!args.isEmpty()) {
final String arg = args.get(0);
@@ -146,6 +141,10 @@
if (arg.charAt(0) == '-') {
if (OPTION_VERSION_2.equals(arg)) {
// Do nothing, this is the default
+ } else if (OPTION_VERSION_3.equals(arg)) {
+ outputBinaryFormatVersion = 3;
+ } else if (OPTION_VERSION_1.equals(arg)) {
+ outputBinaryFormatVersion = 1;
} else if (OPTION_HELP.equals(arg)) {
displayHelp();
} else {
@@ -168,8 +167,6 @@
inputBigramXml = filename;
} else if (OPTION_OUTPUT_BINARY.equals(arg)) {
outputBinary = filename;
- } else if (OPTION_OUTPUT_BINARY_FORMAT_VERSION_1.equals(arg)) {
- outputBinaryFormat1 = filename;
} else if (OPTION_OUTPUT_XML.equals(arg)) {
outputXml = filename;
} else {
@@ -196,8 +193,8 @@
mInputShortcutXml = inputShortcutXml;
mInputBigramXml = inputBigramXml;
mOutputBinary = outputBinary;
- mOutputBinaryFormat1 = outputBinaryFormat1;
mOutputXml = outputXml;
+ mOutputBinaryFormatVersion = outputBinaryFormatVersion;
checkIntegrity();
}
}
@@ -294,10 +291,7 @@
throws FileNotFoundException, IOException, UnsupportedFormatException,
IllegalArgumentException {
if (null != args.mOutputBinary) {
- writeBinaryDictionary(args.mOutputBinary, dict, 2);
- }
- if (null != args.mOutputBinaryFormat1) {
- writeBinaryDictionary(args.mOutputBinaryFormat1, dict, 1);
+ writeBinaryDictionary(args.mOutputBinary, dict, args.mOutputBinaryFormatVersion);
}
if (null != args.mOutputXml) {
writeXmlDictionary(args.mOutputXml, dict);