am 109ba3ac
: Merge "Regenerate old version dictionaries using new format."
* commit '109ba3ace3815a7650a9fb0485b187968ddf8c29': Regenerate old version dictionaries using new format.
This commit is contained in:
commit
d749af1664
6 changed files with 13 additions and 30 deletions
|
@ -224,14 +224,10 @@ final public class BinaryDictionaryGetter {
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
// ## HACK ## we prevent usage of a dictionary before version 18 for English only. The reason
|
// ## HACK ## we prevent usage of a dictionary before version 18. The reason for this is, since
|
||||||
// for this is, since those do not include whitelist entries, the new code with an old version
|
// those do not include whitelist entries, the new code with an old version of the dictionary
|
||||||
// of the dictionary would lose whitelist functionality.
|
// would lose whitelist functionality.
|
||||||
private static boolean hackCanUseDictionaryFile(final Locale locale, final File f) {
|
private static boolean hackCanUseDictionaryFile(final Locale locale, final File f) {
|
||||||
// Only for English - other languages didn't have a whitelist, hence this
|
|
||||||
// ad-hoc ## HACK ##
|
|
||||||
if (!Locale.ENGLISH.getLanguage().equals(locale.getLanguage())) return true;
|
|
||||||
|
|
||||||
FileInputStream inStream = null;
|
FileInputStream inStream = null;
|
||||||
try {
|
try {
|
||||||
// Read the version of the file
|
// Read the version of the file
|
||||||
|
|
|
@ -92,7 +92,8 @@ abstract public class ExpandableBinaryDictionary extends Dictionary {
|
||||||
/** Controls access to the local binary dictionary for this instance. */
|
/** Controls access to the local binary dictionary for this instance. */
|
||||||
private final DictionaryController mLocalDictionaryController = new DictionaryController();
|
private final DictionaryController mLocalDictionaryController = new DictionaryController();
|
||||||
|
|
||||||
private static final int BINARY_DICT_VERSION = 1;
|
// TODO: Regenerate version 3 binary dictionary.
|
||||||
|
private static final int BINARY_DICT_VERSION = 2;
|
||||||
private static final FormatSpec.FormatOptions FORMAT_OPTIONS =
|
private static final FormatSpec.FormatOptions FORMAT_OPTIONS =
|
||||||
new FormatSpec.FormatOptions(BINARY_DICT_VERSION);
|
new FormatSpec.FormatOptions(BINARY_DICT_VERSION);
|
||||||
|
|
||||||
|
@ -415,6 +416,12 @@ abstract public class ExpandableBinaryDictionary extends Dictionary {
|
||||||
// shared dictionary.
|
// shared dictionary.
|
||||||
loadBinaryDictionary();
|
loadBinaryDictionary();
|
||||||
}
|
}
|
||||||
|
if (mBinaryDictionary != null && !mBinaryDictionary.isValidDictionary()) {
|
||||||
|
// Binary dictionary is not valid. Regenerate the dictionary file.
|
||||||
|
mSharedDictionaryController.mLastUpdateTime = time;
|
||||||
|
generateBinaryDictionary();
|
||||||
|
loadBinaryDictionary();
|
||||||
|
}
|
||||||
mLocalDictionaryController.mLastUpdateTime = time;
|
mLocalDictionaryController.mLastUpdateTime = time;
|
||||||
} finally {
|
} finally {
|
||||||
mSharedDictionaryController.unlock();
|
mSharedDictionaryController.unlock();
|
||||||
|
|
|
@ -27,10 +27,6 @@ const int BinaryDictionaryFormatUtils::DICTIONARY_MINIMUM_SIZE = 4;
|
||||||
/**
|
/**
|
||||||
* Format versions
|
* Format versions
|
||||||
*/
|
*/
|
||||||
// Originally, format version 1 had a 16-bit magic number, then the version number `01'
|
|
||||||
// then options that must be 0. Hence the first 32-bits of the format are always as follow
|
|
||||||
// and it's okay to consider them a magic number as a whole.
|
|
||||||
const uint32_t BinaryDictionaryFormatUtils::FORMAT_VERSION_1_MAGIC_NUMBER = 0x78B10100;
|
|
||||||
|
|
||||||
// The versions of Latin IME that only handle format version 1 only test for the magic
|
// The versions of Latin IME that only handle format version 1 only test for the magic
|
||||||
// number, so we had to change it so that version 2 files would be rejected by older
|
// number, so we had to change it so that version 2 files would be rejected by older
|
||||||
|
@ -50,12 +46,6 @@ const int BinaryDictionaryFormatUtils::FORMAT_VERSION_2_MINIMUM_SIZE = 12;
|
||||||
}
|
}
|
||||||
const uint32_t magicNumber = ByteArrayUtils::readUint32(dict, 0);
|
const uint32_t magicNumber = ByteArrayUtils::readUint32(dict, 0);
|
||||||
switch (magicNumber) {
|
switch (magicNumber) {
|
||||||
case FORMAT_VERSION_1_MAGIC_NUMBER:
|
|
||||||
// Format 1 header is exactly 5 bytes long and looks like:
|
|
||||||
// Magic number (2 bytes) 0x78 0xB1
|
|
||||||
// Version number (1 byte) 0x01
|
|
||||||
// Options (2 bytes) must be 0x00 0x00
|
|
||||||
return VERSION_1;
|
|
||||||
case FORMAT_VERSION_2_MAGIC_NUMBER:
|
case FORMAT_VERSION_2_MAGIC_NUMBER:
|
||||||
// Version 2 dictionaries are at least 12 bytes long.
|
// Version 2 dictionaries are at least 12 bytes long.
|
||||||
// If this dictionary has the version 2 magic number but is less than 12 bytes long,
|
// If this dictionary has the version 2 magic number but is less than 12 bytes long,
|
||||||
|
|
|
@ -33,10 +33,9 @@ namespace latinime {
|
||||||
*/
|
*/
|
||||||
class BinaryDictionaryFormatUtils {
|
class BinaryDictionaryFormatUtils {
|
||||||
public:
|
public:
|
||||||
// TODO: Remove obsolete version logic
|
// TODO: Support version 3 format.
|
||||||
enum FORMAT_VERSION {
|
enum FORMAT_VERSION {
|
||||||
VERSION_1,
|
VERSION_2 = 1,
|
||||||
VERSION_2,
|
|
||||||
UNKNOWN_VERSION
|
UNKNOWN_VERSION
|
||||||
};
|
};
|
||||||
|
|
||||||
|
@ -46,7 +45,6 @@ class BinaryDictionaryFormatUtils {
|
||||||
DISALLOW_IMPLICIT_CONSTRUCTORS(BinaryDictionaryFormatUtils);
|
DISALLOW_IMPLICIT_CONSTRUCTORS(BinaryDictionaryFormatUtils);
|
||||||
|
|
||||||
static const int DICTIONARY_MINIMUM_SIZE;
|
static const int DICTIONARY_MINIMUM_SIZE;
|
||||||
static const uint32_t FORMAT_VERSION_1_MAGIC_NUMBER;
|
|
||||||
static const uint32_t FORMAT_VERSION_2_MAGIC_NUMBER;
|
static const uint32_t FORMAT_VERSION_2_MAGIC_NUMBER;
|
||||||
static const int FORMAT_VERSION_2_MINIMUM_SIZE;
|
static const int FORMAT_VERSION_2_MINIMUM_SIZE;
|
||||||
};
|
};
|
||||||
|
|
|
@ -26,8 +26,6 @@ namespace latinime {
|
||||||
|
|
||||||
const int BinaryDictionaryHeaderReadingUtils::MAX_OPTION_KEY_LENGTH = 256;
|
const int BinaryDictionaryHeaderReadingUtils::MAX_OPTION_KEY_LENGTH = 256;
|
||||||
|
|
||||||
const int BinaryDictionaryHeaderReadingUtils::FORMAT_VERSION_1_HEADER_SIZE = 5;
|
|
||||||
|
|
||||||
const int BinaryDictionaryHeaderReadingUtils::VERSION_2_MAGIC_NUMBER_SIZE = 4;
|
const int BinaryDictionaryHeaderReadingUtils::VERSION_2_MAGIC_NUMBER_SIZE = 4;
|
||||||
const int BinaryDictionaryHeaderReadingUtils::VERSION_2_DICTIONARY_VERSION_SIZE = 2;
|
const int BinaryDictionaryHeaderReadingUtils::VERSION_2_DICTIONARY_VERSION_SIZE = 2;
|
||||||
const int BinaryDictionaryHeaderReadingUtils::VERSION_2_DICTIONARY_FLAG_SIZE = 2;
|
const int BinaryDictionaryHeaderReadingUtils::VERSION_2_DICTIONARY_FLAG_SIZE = 2;
|
||||||
|
@ -48,8 +46,6 @@ const BinaryDictionaryHeaderReadingUtils::DictionaryFlags
|
||||||
/* static */ int BinaryDictionaryHeaderReadingUtils::getHeaderSize(
|
/* static */ int BinaryDictionaryHeaderReadingUtils::getHeaderSize(
|
||||||
const BinaryDictionaryInfo *const binaryDictionaryInfo) {
|
const BinaryDictionaryInfo *const binaryDictionaryInfo) {
|
||||||
switch (binaryDictionaryInfo->getFormat()) {
|
switch (binaryDictionaryInfo->getFormat()) {
|
||||||
case BinaryDictionaryFormatUtils::VERSION_1:
|
|
||||||
return FORMAT_VERSION_1_HEADER_SIZE;
|
|
||||||
case BinaryDictionaryFormatUtils::VERSION_2:
|
case BinaryDictionaryFormatUtils::VERSION_2:
|
||||||
// See the format of the header in the comment in
|
// See the format of the header in the comment in
|
||||||
// BinaryDictionaryFormatUtils::detectFormatVersion()
|
// BinaryDictionaryFormatUtils::detectFormatVersion()
|
||||||
|
@ -65,8 +61,6 @@ const BinaryDictionaryHeaderReadingUtils::DictionaryFlags
|
||||||
BinaryDictionaryHeaderReadingUtils::getFlags(
|
BinaryDictionaryHeaderReadingUtils::getFlags(
|
||||||
const BinaryDictionaryInfo *const binaryDictionaryInfo) {
|
const BinaryDictionaryInfo *const binaryDictionaryInfo) {
|
||||||
switch (binaryDictionaryInfo->getFormat()) {
|
switch (binaryDictionaryInfo->getFormat()) {
|
||||||
case BinaryDictionaryFormatUtils::VERSION_1:
|
|
||||||
return NO_FLAGS;
|
|
||||||
case BinaryDictionaryFormatUtils::VERSION_2:
|
case BinaryDictionaryFormatUtils::VERSION_2:
|
||||||
return ByteArrayUtils::readUint16(binaryDictionaryInfo->getDictBuf(),
|
return ByteArrayUtils::readUint16(binaryDictionaryInfo->getDictBuf(),
|
||||||
VERSION_2_MAGIC_NUMBER_SIZE + VERSION_2_DICTIONARY_VERSION_SIZE);
|
VERSION_2_MAGIC_NUMBER_SIZE + VERSION_2_DICTIONARY_VERSION_SIZE);
|
||||||
|
|
|
@ -82,8 +82,6 @@ class BinaryDictionaryHeaderReadingUtils {
|
||||||
private:
|
private:
|
||||||
DISALLOW_IMPLICIT_CONSTRUCTORS(BinaryDictionaryHeaderReadingUtils);
|
DISALLOW_IMPLICIT_CONSTRUCTORS(BinaryDictionaryHeaderReadingUtils);
|
||||||
|
|
||||||
static const int FORMAT_VERSION_1_HEADER_SIZE;
|
|
||||||
|
|
||||||
static const int VERSION_2_MAGIC_NUMBER_SIZE;
|
static const int VERSION_2_MAGIC_NUMBER_SIZE;
|
||||||
static const int VERSION_2_DICTIONARY_VERSION_SIZE;
|
static const int VERSION_2_DICTIONARY_VERSION_SIZE;
|
||||||
static const int VERSION_2_DICTIONARY_FLAG_SIZE;
|
static const int VERSION_2_DICTIONARY_FLAG_SIZE;
|
||||||
|
|
Loading…
Reference in a new issue