From f7425bb15be6514bb2daedbc62760ca5d52c08cf Mon Sep 17 00:00:00 2001
From: satok <satok@google.com>
Date: Wed, 5 Jan 2011 16:37:53 +0900
Subject: [PATCH] Supress overflow at mulitplying demotion rate
Change-Id: I2003c5f88a5062b11e2f21522095bb94b1eb4efd
---
native/src/unigram_dictionary.cpp | 21 ++++++++++++++-------
1 file changed, 14 insertions(+), 7 deletions(-)
diff --git a/native/src/unigram_dictionary.cpp b/native/src/unigram_dictionary.cpp
index 5e1212a5d7..af2cc97fce 100644
--- a/native/src/unigram_dictionary.cpp
+++ b/native/src/unigram_dictionary.cpp
@@ -269,6 +269,14 @@ void UnigramDictionary::getSuggestionCandidates(const int skipPos,
}
}
+inline static void multiplyRate(const int rate, int *freq) {
+ if (rate > 1000000) {
+ *freq = (*freq / 100) * rate;
+ } else {
+ *freq = *freq * rate / 100;
+ }
+}
+
bool UnigramDictionary::getMissingSpaceWords(const int inputLength, const int missingSpacePos) {
if (missingSpacePos <= 0 || missingSpacePos >= inputLength
|| inputLength >= MAX_WORD_LENGTH) return false;
@@ -294,7 +302,7 @@ bool UnigramDictionary::getMissingSpaceWords(const int inputLength, const int mi
int pairFreq = ((firstFreq + secondFreq) / 2);
for (int i = 0; i < inputLength; ++i) pairFreq *= TYPED_LETTER_MULTIPLIER;
- pairFreq = pairFreq * WORDS_WITH_MISSING_SPACE_CHARACTER_DEMOTION_RATE / 100;
+ multiplyRate(WORDS_WITH_MISSING_SPACE_CHARACTER_DEMOTION_RATE, &pairFreq);
addWord(word, newWordLength, pairFreq);
return true;
}
@@ -345,14 +353,13 @@ inline int UnigramDictionary::calculateFinalFreq(const int inputIndex, const int
const bool sameLength) {
// TODO: Demote by edit distance
int finalFreq = freq * snr;
- if (skipPos >= 0) finalFreq = finalFreq * WORDS_WITH_MISSING_CHARACTER_DEMOTION_RATE / 100;
- if (transposedPos >= 0) finalFreq = finalFreq
- * WORDS_WITH_TRANSPOSED_CHARACTERS_DEMOTION_RATE / 100;
+ if (skipPos >= 0) multiplyRate(WORDS_WITH_MISSING_CHARACTER_DEMOTION_RATE, &finalFreq);
+ if (transposedPos >= 0) multiplyRate(
+ WORDS_WITH_TRANSPOSED_CHARACTERS_DEMOTION_RATE, &finalFreq);
if (excessivePos >= 0) {
- finalFreq = finalFreq * WORDS_WITH_EXCESSIVE_CHARACTER_DEMOTION_RATE / 100;
+ multiplyRate(WORDS_WITH_EXCESSIVE_CHARACTER_DEMOTION_RATE, &finalFreq);
if (!existsAdjacentProximityChars(inputIndex, mInputLength)) {
- finalFreq = finalFreq
- * WORDS_WITH_EXCESSIVE_CHARACTER_OUT_OF_PROXIMITY_DEMOTION_RATE / 100;
+ multiplyRate(WORDS_WITH_EXCESSIVE_CHARACTER_OUT_OF_PROXIMITY_DEMOTION_RATE, &finalFreq);
}
}
if (sameLength && skipPos < 0) finalFreq *= FULL_WORD_MULTIPLIER;
--
GitLab