Merge "Some obvious optimizations to makedict" into jb-dev
diff --git a/java/src/com/android/inputmethod/latin/makedict/FusionDictionary.java b/java/src/com/android/inputmethod/latin/makedict/FusionDictionary.java
index b08702e..55ceb1c 100644
--- a/java/src/com/android/inputmethod/latin/makedict/FusionDictionary.java
+++ b/java/src/com/android/inputmethod/latin/makedict/FusionDictionary.java
@@ -28,6 +28,8 @@
*/
public class FusionDictionary implements Iterable<Word> {
+ private static final boolean DBG = MakedictLog.DBG;
+
/**
* A node of the dictionary, containing several CharGroups.
*
@@ -159,6 +161,7 @@
* shortcut list.
*/
public WeightedString getShortcut(final String word) {
+ // TODO: Don't do a linear search
if (mShortcutTargets != null) {
final int size = mShortcutTargets.size();
for (int i = 0; i < size; ++i) {
@@ -176,6 +179,7 @@
* Returns null if the word is not in the bigrams list.
*/
public WeightedString getBigram(final String word) {
+ // TODO: Don't do a linear search
if (mBigrams != null) {
final int size = mBigrams.size();
for (int i = 0; i < size; ++i) {
@@ -265,31 +269,21 @@
/**
* Helper method to convert a String to an int array.
*/
- static private int[] getCodePoints(String word) {
- final int wordLength = word.length();
- int[] array = new int[word.codePointCount(0, wordLength)];
- for (int i = 0; i < wordLength; i = word.offsetByCodePoints(i, 1)) {
- array[i] = word.codePointAt(i);
+ static private int[] getCodePoints(final String word) {
+ // TODO: this is a copy-paste of the contents of StringUtils.toCodePointArray,
+ // which is not visible from the makedict package. Factor this code.
+ final char[] characters = word.toCharArray();
+ final int length = characters.length;
+ final int[] codePoints = new int[Character.codePointCount(characters, 0, length)];
+ int codePoint = Character.codePointAt(characters, 0);
+ int dsti = 0;
+ for (int srci = Character.charCount(codePoint);
+ srci < length; srci += Character.charCount(codePoint), ++dsti) {
+ codePoints[dsti] = codePoint;
+ codePoint = Character.codePointAt(characters, srci);
}
- return array;
- }
-
- /**
- * Helper method to add all words in a list as 0-frequency entries
- *
- * These words are added when shortcuts targets or bigrams are not found in the dictionary
- * yet. The same words may be added later with an actual frequency - this is handled by
- * the private version of add().
- */
- private void addNeutralWords(final ArrayList<WeightedString> words) {
- if (null != words) {
- for (WeightedString word : words) {
- final CharGroup t = findWordInTree(mRoot, word.mWord);
- if (null == t) {
- add(getCodePoints(word.mWord), 0, null);
- }
- }
- }
+ codePoints[dsti] = codePoint;
+ return codePoints;
}
/**
@@ -339,7 +333,6 @@
if (charGroup != null) {
final CharGroup charGroup2 = findWordInTree(mRoot, word2);
if (charGroup2 == null) {
- // TODO: refactor with the identical code in addNeutralWords
add(getCodePoints(word2), 0, null);
}
charGroup.addBigram(word2, frequency);
@@ -386,7 +379,7 @@
Arrays.copyOfRange(word, charIndex, word.length),
shortcutTargets, null /* bigrams */, frequency);
currentNode.mData.add(insertionIndex, newGroup);
- checkStack(currentNode);
+ if (DBG) checkStack(currentNode);
} else {
// There is a word with a common prefix.
if (differentCharIndex == currentGroup.mChars.length) {
@@ -437,7 +430,7 @@
}
currentNode.mData.set(nodeIndex, newParent);
}
- checkStack(currentNode);
+ if (DBG) checkStack(currentNode);
}
}
}
diff --git a/java/src/com/android/inputmethod/latin/makedict/MakedictLog.java b/java/src/com/android/inputmethod/latin/makedict/MakedictLog.java
index 1281c7e..3f0cd07 100644
--- a/java/src/com/android/inputmethod/latin/makedict/MakedictLog.java
+++ b/java/src/com/android/inputmethod/latin/makedict/MakedictLog.java
@@ -22,7 +22,7 @@
* Wrapper to redirect log events to the right output medium.
*/
public class MakedictLog {
- private static final boolean DBG = false;
+ public static final boolean DBG = false;
private static final String TAG = MakedictLog.class.getSimpleName();
public static void d(String message) {
diff --git a/tools/makedict/src/com/android/inputmethod/latin/makedict/MakedictLog.java b/tools/makedict/src/com/android/inputmethod/latin/makedict/MakedictLog.java
index 1ab9f42..7eccff2 100644
--- a/tools/makedict/src/com/android/inputmethod/latin/makedict/MakedictLog.java
+++ b/tools/makedict/src/com/android/inputmethod/latin/makedict/MakedictLog.java
@@ -20,6 +20,8 @@
* Wrapper to redirect log events to the right output medium.
*/
public class MakedictLog {
+ public static final boolean DBG = true;
+
private static void print(String message) {
System.out.println(message);
}