am db6984c7: Merge "Have random words stick to a restricted (random) charset"

* commit 'db6984c7bd0f3a933581cd0af201cf5e70643e45':
  Have random words stick to a restricted (random) charset
main
Jean Chalard 2013-07-04 19:49:47 -07:00 committed by Android Git Automerger
commit cbc1fd044b
1 changed files with 26 additions and 14 deletions

View File

@ -52,6 +52,7 @@ import java.util.Set;
public class BinaryDictIOTests extends AndroidTestCase { public class BinaryDictIOTests extends AndroidTestCase {
private static final String TAG = BinaryDictIOTests.class.getSimpleName(); private static final String TAG = BinaryDictIOTests.class.getSimpleName();
private static final int DEFAULT_MAX_UNIGRAMS = 100; private static final int DEFAULT_MAX_UNIGRAMS = 100;
private static final int DEFAULT_CODE_POINT_SET_SIZE = 50;
private static final int UNIGRAM_FREQ = 10; private static final int UNIGRAM_FREQ = 10;
private static final int BIGRAM_FREQ = 50; private static final int BIGRAM_FREQ = 50;
private static final int TOLERANCE_OF_BIGRAM_FREQ = 5; private static final int TOLERANCE_OF_BIGRAM_FREQ = 5;
@ -81,7 +82,8 @@ public class BinaryDictIOTests extends AndroidTestCase {
Log.e(TAG, "Testing dictionary: seed is " + seed); Log.e(TAG, "Testing dictionary: seed is " + seed);
final Random random = new Random(seed); final Random random = new Random(seed);
sWords.clear(); sWords.clear();
generateWords(maxUnigrams, random); final int[] codePointSet = generateCodePointSet(DEFAULT_CODE_POINT_SET_SIZE, random);
generateWords(maxUnigrams, random, codePointSet);
for (int i = 0; i < sWords.size(); ++i) { for (int i = 0; i < sWords.size(); ++i) {
sChainBigrams.put(i, new ArrayList<Integer>()); sChainBigrams.put(i, new ArrayList<Integer>());
@ -96,6 +98,23 @@ public class BinaryDictIOTests extends AndroidTestCase {
} }
} }
private int[] generateCodePointSet(final int codePointSetSize, final Random random) {
final int[] codePointSet = new int[codePointSetSize];
for (int i = codePointSet.length - 1; i >= 0; ) {
final int r = Math.abs(random.nextInt());
if (r < 0) continue;
// Don't insert 0~0x20, but insert any other code point.
// Code points are in the range 0~0x10FFFF.
final int candidateCodePoint = (int)(0x20 + r % (Character.MAX_CODE_POINT - 0x20));
// Code points between MIN_ and MAX_SURROGATE are not valid on their own.
if (candidateCodePoint >= Character.MIN_SURROGATE
&& candidateCodePoint <= Character.MAX_SURROGATE) continue;
codePointSet[i] = candidateCodePoint;
--i;
}
return codePointSet;
}
// Utilities for test // Utilities for test
/** /**
@ -131,28 +150,20 @@ public class BinaryDictIOTests extends AndroidTestCase {
/** /**
* Generates a random word. * Generates a random word.
*/ */
private String generateWord(final Random random) { private String generateWord(final Random random, final int[] codePointSet) {
StringBuilder builder = new StringBuilder("a"); StringBuilder builder = new StringBuilder("a");
int count = random.nextInt() % 30; // Arbitrarily 30 chars max int count = random.nextInt() % 30; // Arbitrarily 30 chars max
while (count > 0) { while (count > 0) {
final long r = Math.abs(random.nextInt()); builder.appendCodePoint(codePointSet[Math.abs(random.nextInt()) % codePointSet.length]);
if (r < 0) continue;
// Don't insert 0~0x20, but insert any other code point.
// Code points are in the range 0~0x10FFFF.
final int candidateCodePoint = (int)(0x20 + r % (Character.MAX_CODE_POINT - 0x20));
// Code points between MIN_ and MAX_SURROGATE are not valid on their own.
if (candidateCodePoint >= Character.MIN_SURROGATE
&& candidateCodePoint <= Character.MAX_SURROGATE) continue;
builder.appendCodePoint(candidateCodePoint);
--count; --count;
} }
return builder.toString(); return builder.toString();
} }
private void generateWords(final int number, final Random random) { private void generateWords(final int number, final Random random, final int[] codePointSet) {
final Set<String> wordSet = CollectionUtils.newHashSet(); final Set<String> wordSet = CollectionUtils.newHashSet();
while (wordSet.size() < number) { while (wordSet.size() < number) {
wordSet.add(generateWord(random)); wordSet.add(generateWord(random, codePointSet));
} }
sWords.addAll(wordSet); sWords.addAll(wordSet);
} }
@ -560,8 +571,9 @@ public class BinaryDictIOTests extends AndroidTestCase {
// Test a word that isn't contained within the dictionary. // Test a word that isn't contained within the dictionary.
final Random random = new Random((int)System.currentTimeMillis()); final Random random = new Random((int)System.currentTimeMillis());
final int[] codePointSet = generateCodePointSet(DEFAULT_CODE_POINT_SET_SIZE, random);
for (int i = 0; i < 1000; ++i) { for (int i = 0; i < 1000; ++i) {
final String word = generateWord(random); final String word = generateWord(random, codePointSet);
if (sWords.indexOf(word) != -1) continue; if (sWords.indexOf(word) != -1) continue;
runGetTerminalPosition(buffer, word, i, false); runGetTerminalPosition(buffer, word, i, false);
} }