diff --git a/native/jni/src/suggest/policyimpl/dictionary/header/header_policy.cpp b/native/jni/src/suggest/policyimpl/dictionary/header/header_policy.cpp index eb828b58c..439c3de1d 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/header/header_policy.cpp +++ b/native/jni/src/suggest/policyimpl/dictionary/header/header_policy.cpp @@ -16,24 +16,90 @@ #include "suggest/policyimpl/dictionary/header/header_policy.h" +#include + namespace latinime { -const char *const HeaderPolicy::MULTIPLE_WORDS_DEMOTION_RATE_KEY = - "MULTIPLE_WORDS_DEMOTION_RATE"; -const float HeaderPolicy::DEFAULT_MULTI_WORD_COST_MULTIPLIER = 1.0f; -const float HeaderPolicy::MULTI_WORD_COST_MULTIPLIER_SCALE = 100.0f; +const char *const HeaderPolicy::MULTIPLE_WORDS_DEMOTION_RATE_KEY = "MULTIPLE_WORDS_DEMOTION_RATE"; +const float HeaderPolicy::DEFAULT_MULTIPLE_WORD_COST_MULTIPLIER = 1.0f; +const float HeaderPolicy::MULTIPLE_WORD_COST_MULTIPLIER_SCALE = 100.0f; -float HeaderPolicy::readMultiWordCostMultiplier() const { - const int headerValue = HeaderReadingUtils::readHeaderValueInt( - mDictBuf, MULTIPLE_WORDS_DEMOTION_RATE_KEY); +// Used for logging. Question mark is used to indicate that the key is not found. +void HeaderPolicy::readHeaderValueOrQuestionMark(const char *const key, int *outValue, + int outValueSize) const { + if (outValueSize <= 0) return; + if (outValueSize == 1) { + outValue[0] = '\0'; + return; + } + std::vector keyCodePointVector; + insertCharactersIntoVector(key, &keyCodePointVector); + HeaderReadingUtils::AttributeMap::const_iterator it = mAttributeMap.find(keyCodePointVector); + if (it == mAttributeMap.end()) { + // The key was not found. + outValue[0] = '?'; + outValue[1] = '\0'; + return; + } + const int terminalIndex = min(static_cast(it->second.size()), outValueSize - 1); + for (int i = 0; i < terminalIndex; ++i) { + outValue[i] = it->second[i]; + } + outValue[terminalIndex] = '\0'; +} + +float HeaderPolicy::readMultipleWordCostMultiplier() const { + std::vector multipleWordsDemotionRateKeyVector; + insertCharactersIntoVector(MULTIPLE_WORDS_DEMOTION_RATE_KEY, + &multipleWordsDemotionRateKeyVector); + HeaderReadingUtils::AttributeMap::const_iterator it = + mAttributeMap.find(multipleWordsDemotionRateKeyVector); + if (it == mAttributeMap.end()) { + // The key was not found. + return DEFAULT_MULTIPLE_WORD_COST_MULTIPLIER; + } + const int headerValue = parseIntAttributeValue(&(it->second)); if (headerValue == S_INT_MIN) { - // not found - return DEFAULT_MULTI_WORD_COST_MULTIPLIER; + // Invalid value + return DEFAULT_MULTIPLE_WORD_COST_MULTIPLIER; } if (headerValue <= 0) { return static_cast(MAX_VALUE_FOR_WEIGHTING); } - return MULTI_WORD_COST_MULTIPLIER_SCALE / static_cast(headerValue); + return MULTIPLE_WORD_COST_MULTIPLIER_SCALE / static_cast(headerValue); +} + +/* static */ HeaderReadingUtils::AttributeMap HeaderPolicy::createAttributeMapAndReadAllAttributes( + const uint8_t *const dictBuf) { + HeaderReadingUtils::AttributeMap attributeMap; + HeaderReadingUtils::fetchAllHeaderAttributes(dictBuf, &attributeMap); + return attributeMap; +} + +/* static */ int HeaderPolicy::parseIntAttributeValue( + const std::vector *const attributeValue) { + int value = 0; + bool isNegative = false; + for (size_t i = 0; i < attributeValue->size(); ++i) { + if (i == 0 && attributeValue->at(i) == '-') { + isNegative = true; + } else { + if (!isdigit(attributeValue->at(i))) { + // If not a number, return S_INT_MIN + return S_INT_MIN; + } + value *= 10; + value += attributeValue->at(i) - '0'; + } + } + return isNegative ? -value : value; +} + +/* static */ void HeaderPolicy::insertCharactersIntoVector(const char *const characters, + std::vector *const vector) { + for (int i = 0; characters[i]; ++i) { + vector->push_back(characters[i]); + } } } // namespace latinime diff --git a/native/jni/src/suggest/policyimpl/dictionary/header/header_policy.h b/native/jni/src/suggest/policyimpl/dictionary/header/header_policy.h index e3e6fc077..571ff9e69 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/header/header_policy.h +++ b/native/jni/src/suggest/policyimpl/dictionary/header/header_policy.h @@ -17,6 +17,7 @@ #ifndef LATINIME_HEADER_POLICY_H #define LATINIME_HEADER_POLICY_H +#include #include #include "defines.h" @@ -30,7 +31,8 @@ class HeaderPolicy : public DictionaryHeaderStructurePolicy { explicit HeaderPolicy(const uint8_t *const dictBuf) : mDictBuf(dictBuf), mDictionaryFlags(HeaderReadingUtils::getFlags(dictBuf)), mSize(HeaderReadingUtils::getHeaderSize(dictBuf)), - mMultiWordCostMultiplier(readMultiWordCostMultiplier()) {} + mAttributeMap(createAttributeMapAndReadAllAttributes(mDictBuf)), + mMultiWordCostMultiplier(readMultipleWordCostMultiplier()) {} ~HeaderPolicy() {} @@ -55,34 +57,31 @@ class HeaderPolicy : public DictionaryHeaderStructurePolicy { return mMultiWordCostMultiplier; } - AK_FORCE_INLINE void readHeaderValueOrQuestionMark(const char *const key, - int *outValue, int outValueSize) const { - if (outValueSize <= 0) return; - if (outValueSize == 1) { - outValue[0] = '\0'; - return; - } - if (!HeaderReadingUtils::readHeaderValue(mDictBuf, - key, outValue, outValueSize)) { - outValue[0] = '?'; - outValue[1] = '\0'; - } - } + void readHeaderValueOrQuestionMark(const char *const key, + int *outValue, int outValueSize) const; private: DISALLOW_IMPLICIT_CONSTRUCTORS(HeaderPolicy); static const char *const MULTIPLE_WORDS_DEMOTION_RATE_KEY; - static const float DEFAULT_MULTI_WORD_COST_MULTIPLIER; - static const float MULTI_WORD_COST_MULTIPLIER_SCALE; + static const float DEFAULT_MULTIPLE_WORD_COST_MULTIPLIER; + static const float MULTIPLE_WORD_COST_MULTIPLIER_SCALE; const uint8_t *const mDictBuf; const HeaderReadingUtils::DictionaryFlags mDictionaryFlags; const int mSize; + HeaderReadingUtils::AttributeMap mAttributeMap; const float mMultiWordCostMultiplier; - float readMultiWordCostMultiplier() const; -}; + float readMultipleWordCostMultiplier() const; + static HeaderReadingUtils::AttributeMap createAttributeMapAndReadAllAttributes( + const uint8_t *const dictBuf); + + static int parseIntAttributeValue(const std::vector *const attributeValue); + + static void insertCharactersIntoVector( + const char *const characters, std::vector *const vector); +}; } // namespace latinime #endif /* LATINIME_HEADER_POLICY_H */ diff --git a/native/jni/src/suggest/policyimpl/dictionary/header/header_reading_utils.cpp b/native/jni/src/suggest/policyimpl/dictionary/header/header_reading_utils.cpp index f323876c4..186c043c1 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/header/header_reading_utils.cpp +++ b/native/jni/src/suggest/policyimpl/dictionary/header/header_reading_utils.cpp @@ -16,23 +16,22 @@ #include "suggest/policyimpl/dictionary/header/header_reading_utils.h" -#include -#include +#include #include "defines.h" #include "suggest/policyimpl/dictionary/utils/byte_array_utils.h" namespace latinime { -const int HeaderReadingUtils::MAX_OPTION_KEY_LENGTH = 256; +const int HeaderReadingUtils::MAX_ATTRIBUTE_KEY_LENGTH = 256; +const int HeaderReadingUtils::MAX_ATTRIBUTE_VALUE_LENGTH = 256; const int HeaderReadingUtils::HEADER_MAGIC_NUMBER_SIZE = 4; const int HeaderReadingUtils::HEADER_DICTIONARY_VERSION_SIZE = 2; const int HeaderReadingUtils::HEADER_FLAG_SIZE = 2; const int HeaderReadingUtils::HEADER_SIZE_FIELD_SIZE = 4; -const HeaderReadingUtils::DictionaryFlags - HeaderReadingUtils::NO_FLAGS = 0; +const HeaderReadingUtils::DictionaryFlags HeaderReadingUtils::NO_FLAGS = 0; // Flags for special processing // Those *must* match the flags in makedict (FormatSpec#*_PROCESSING_FLAG) or // something very bad (like, the apocalypse) will happen. Please update both at the same time. @@ -56,53 +55,27 @@ const HeaderReadingUtils::DictionaryFlags HEADER_MAGIC_NUMBER_SIZE + HEADER_DICTIONARY_VERSION_SIZE); } -// Returns if the key is found or not and reads the found value into outValue. -/* static */ bool HeaderReadingUtils::readHeaderValue(const uint8_t *const dictBuf, - const char *const key, int *outValue, const int outValueSize) { - if (outValueSize <= 0) { - return false; - } +/* static */ void HeaderReadingUtils::fetchAllHeaderAttributes(const uint8_t *const dictBuf, + AttributeMap *const headerAttributes) { const int headerSize = getHeaderSize(dictBuf); int pos = getHeaderOptionsPosition(); if (pos == NOT_A_DICT_POS) { // The header doesn't have header options. - return false; + return; } + int keyBuffer[MAX_ATTRIBUTE_KEY_LENGTH]; + int valueBuffer[MAX_ATTRIBUTE_VALUE_LENGTH]; while (pos < headerSize) { - if(ByteArrayUtils::compareStringInBufferWithCharArray( - dictBuf, key, headerSize - pos, &pos) == 0) { - // The key was found. - const int length = ByteArrayUtils::readStringAndAdvancePosition(dictBuf, outValueSize, - outValue, &pos); - // Add a 0 terminator to the string. - outValue[length < outValueSize ? length : outValueSize - 1] = '\0'; - return true; - } - ByteArrayUtils::advancePositionToBehindString(dictBuf, headerSize - pos, &pos); + const int keyLength = ByteArrayUtils::readStringAndAdvancePosition(dictBuf, + MAX_ATTRIBUTE_KEY_LENGTH, keyBuffer, &pos); + std::vector key; + key.insert(key.end(), keyBuffer, keyBuffer + keyLength); + const int valueLength = ByteArrayUtils::readStringAndAdvancePosition(dictBuf, + MAX_ATTRIBUTE_VALUE_LENGTH, valueBuffer, &pos); + std::vector value; + value.insert(value.end(), valueBuffer, valueBuffer + valueLength); + headerAttributes->insert(AttributeMap::value_type(key, value)); } - // The key was not found. - return false; -} - -/* static */ int HeaderReadingUtils::readHeaderValueInt( - const uint8_t *const dictBuf, const char *const key) { - const int bufferSize = LARGEST_INT_DIGIT_COUNT; - int intBuffer[bufferSize]; - char charBuffer[bufferSize]; - if (!readHeaderValue(dictBuf, key, intBuffer, bufferSize)) { - return S_INT_MIN; - } - for (int i = 0; i < bufferSize; ++i) { - charBuffer[i] = intBuffer[i]; - if (charBuffer[i] == '0') { - break; - } - if (!isdigit(charBuffer[i])) { - // If not a number, return S_INT_MIN - return S_INT_MIN; - } - } - return atoi(charBuffer); } } // namespace latinime diff --git a/native/jni/src/suggest/policyimpl/dictionary/header/header_reading_utils.h b/native/jni/src/suggest/policyimpl/dictionary/header/header_reading_utils.h index c94919640..5716198fb 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/header/header_reading_utils.h +++ b/native/jni/src/suggest/policyimpl/dictionary/header/header_reading_utils.h @@ -17,7 +17,9 @@ #ifndef LATINIME_HEADER_READING_UTILS_H #define LATINIME_HEADER_READING_UTILS_H +#include #include +#include #include "defines.h" @@ -26,8 +28,7 @@ namespace latinime { class HeaderReadingUtils { public: typedef uint16_t DictionaryFlags; - - static const int MAX_OPTION_KEY_LENGTH; + typedef std::map, std::vector > AttributeMap; static int getHeaderSize(const uint8_t *const dictBuf); @@ -50,14 +51,15 @@ class HeaderReadingUtils { + HEADER_SIZE_FIELD_SIZE; } - static bool readHeaderValue(const uint8_t *const dictBuf, - const char *const key, int *outValue, const int outValueSize); - - static int readHeaderValueInt(const uint8_t *const dictBuf, const char *const key); + static void fetchAllHeaderAttributes(const uint8_t *const dictBuf, + AttributeMap *const headerAttributes); private: DISALLOW_IMPLICIT_CONSTRUCTORS(HeaderReadingUtils); + static const int MAX_ATTRIBUTE_KEY_LENGTH; + static const int MAX_ATTRIBUTE_VALUE_LENGTH; + static const int HEADER_MAGIC_NUMBER_SIZE; static const int HEADER_DICTIONARY_VERSION_SIZE; static const int HEADER_FLAG_SIZE; diff --git a/native/jni/src/suggest/policyimpl/dictionary/utils/byte_array_utils.h b/native/jni/src/suggest/policyimpl/dictionary/utils/byte_array_utils.h index 1d14929c7..e2cb9a065 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/utils/byte_array_utils.h +++ b/native/jni/src/suggest/policyimpl/dictionary/utils/byte_array_utils.h @@ -176,39 +176,6 @@ class ByteArrayUtils { return length; } - // Returns an integer less than, equal to, or greater than zero when string starting from pos - // in buffer is less than, match, or is greater than charArray. - static AK_FORCE_INLINE int compareStringInBufferWithCharArray(const uint8_t *const buffer, - const char *const charArray, const int maxLength, int *const pos) { - int index = 0; - int codePoint = readCodePointAndAdvancePosition(buffer, pos); - const uint8_t *const uint8CharArrayForComparison = - reinterpret_cast(charArray); - while (NOT_A_CODE_POINT != codePoint - && '\0' != uint8CharArrayForComparison[index] && index < maxLength) { - if (codePoint != uint8CharArrayForComparison[index]) { - // Different character is found. - // Skip the rest of the string in the buffer. - advancePositionToBehindString(buffer, maxLength - index, pos); - return codePoint - uint8CharArrayForComparison[index]; - } - // Advance - codePoint = readCodePointAndAdvancePosition(buffer, pos); - ++index; - } - if (NOT_A_CODE_POINT != codePoint && index < maxLength) { - // Skip the rest of the string in the buffer. - advancePositionToBehindString(buffer, maxLength - index, pos); - } - if (NOT_A_CODE_POINT == codePoint && '\0' == uint8CharArrayForComparison[index]) { - // When both of the last characters are terminals, we consider the string in the buffer - // matches the given char array - return 0; - } else { - return codePoint - uint8CharArrayForComparison[index]; - } - } - private: DISALLOW_IMPLICIT_CONSTRUCTORS(ByteArrayUtils);