Mandarin // Indentation Changes to Xcode Default.
This commit is contained in:
ShikiSuen
parent
a6fbeff683
commit
cbb565e9b6
File diff suppressed because it is too large
Load Diff
|
|
@ -29,176 +29,176 @@ namespace Taiyan {
|
||||||
namespace Mandarin {
|
namespace Mandarin {
|
||||||
|
|
||||||
class BopomofoSyllable {
|
class BopomofoSyllable {
|
||||||
public:
|
public:
|
||||||
typedef uint16_t Component;
|
typedef uint16_t Component;
|
||||||
|
|
||||||
explicit BopomofoSyllable(Component syllable = 0) : syllable_(syllable) {}
|
explicit BopomofoSyllable(Component syllable = 0) : syllable_(syllable) {}
|
||||||
|
|
||||||
BopomofoSyllable(const BopomofoSyllable&) = default;
|
BopomofoSyllable(const BopomofoSyllable&) = default;
|
||||||
BopomofoSyllable(BopomofoSyllable&& another) = default;
|
BopomofoSyllable(BopomofoSyllable&& another) = default;
|
||||||
BopomofoSyllable& operator=(const BopomofoSyllable&) = default;
|
BopomofoSyllable& operator=(const BopomofoSyllable&) = default;
|
||||||
BopomofoSyllable& operator=(BopomofoSyllable&&) = default;
|
BopomofoSyllable& operator=(BopomofoSyllable&&) = default;
|
||||||
|
|
||||||
// takes the ASCII-form, "v"-tolerant, TW-style Hanyu Pinyin (fong, pong, bong
|
// takes the ASCII-form, "v"-tolerant, TW-style Hanyu Pinyin (fong, pong, bong
|
||||||
// acceptable)
|
// acceptable)
|
||||||
static const BopomofoSyllable FromHanyuPinyin(const std::string& str);
|
static const BopomofoSyllable FromHanyuPinyin(const std::string& str);
|
||||||
|
|
||||||
// TO DO: Support accented vowels
|
// TO DO: Support accented vowels
|
||||||
const std::string HanyuPinyinString(bool includesTone,
|
const std::string HanyuPinyinString(bool includesTone,
|
||||||
bool useVForUUmlaut) const;
|
bool useVForUUmlaut) const;
|
||||||
// const std::string HanyuPinyinString(bool includesTone, bool useVForUUmlaut,
|
// const std::string HanyuPinyinString(bool includesTone, bool useVForUUmlaut,
|
||||||
// bool composeAccentedVowel) const;
|
// bool composeAccentedVowel) const;
|
||||||
|
|
||||||
// PHT = Pai-hua-tsi
|
// PHT = Pai-hua-tsi
|
||||||
static const BopomofoSyllable FromPHT(const std::string& str);
|
static const BopomofoSyllable FromPHT(const std::string& str);
|
||||||
const std::string PHTString(bool includesTone) const;
|
const std::string PHTString(bool includesTone) const;
|
||||||
|
|
||||||
static const BopomofoSyllable FromComposedString(const std::string& str);
|
static const BopomofoSyllable FromComposedString(const std::string& str);
|
||||||
const std::string composedString() const;
|
const std::string composedString() const;
|
||||||
|
|
||||||
void clear() { syllable_ = 0; }
|
void clear() { syllable_ = 0; }
|
||||||
|
|
||||||
bool isEmpty() const { return !syllable_; }
|
bool isEmpty() const { return !syllable_; }
|
||||||
|
|
||||||
bool hasConsonant() const { return !!(syllable_ & ConsonantMask); }
|
bool hasConsonant() const { return !!(syllable_ & ConsonantMask); }
|
||||||
|
|
||||||
bool hasMiddleVowel() const { return !!(syllable_ & MiddleVowelMask); }
|
bool hasMiddleVowel() const { return !!(syllable_ & MiddleVowelMask); }
|
||||||
bool hasVowel() const { return !!(syllable_ & VowelMask); }
|
bool hasVowel() const { return !!(syllable_ & VowelMask); }
|
||||||
|
|
||||||
bool hasToneMarker() const { return !!(syllable_ & ToneMarkerMask); }
|
bool hasToneMarker() const { return !!(syllable_ & ToneMarkerMask); }
|
||||||
|
|
||||||
Component consonantComponent() const { return syllable_ & ConsonantMask; }
|
Component consonantComponent() const { return syllable_ & ConsonantMask; }
|
||||||
|
|
||||||
Component middleVowelComponent() const {
|
Component middleVowelComponent() const {
|
||||||
return syllable_ & MiddleVowelMask;
|
return syllable_ & MiddleVowelMask;
|
||||||
}
|
}
|
||||||
|
|
||||||
Component vowelComponent() const { return syllable_ & VowelMask; }
|
Component vowelComponent() const { return syllable_ & VowelMask; }
|
||||||
|
|
||||||
Component toneMarkerComponent() const { return syllable_ & ToneMarkerMask; }
|
Component toneMarkerComponent() const { return syllable_ & ToneMarkerMask; }
|
||||||
|
|
||||||
bool operator==(const BopomofoSyllable& another) const {
|
bool operator==(const BopomofoSyllable& another) const {
|
||||||
return syllable_ == another.syllable_;
|
return syllable_ == another.syllable_;
|
||||||
}
|
}
|
||||||
|
|
||||||
bool operator!=(const BopomofoSyllable& another) const {
|
bool operator!=(const BopomofoSyllable& another) const {
|
||||||
return syllable_ != another.syllable_;
|
return syllable_ != another.syllable_;
|
||||||
}
|
}
|
||||||
|
|
||||||
bool isOverlappingWith(const BopomofoSyllable& another) const {
|
bool isOverlappingWith(const BopomofoSyllable& another) const {
|
||||||
#define IOW_SAND(mask) ((syllable_ & mask) && (another.syllable_ & mask))
|
#define IOW_SAND(mask) ((syllable_ & mask) && (another.syllable_ & mask))
|
||||||
return IOW_SAND(ConsonantMask) || IOW_SAND(MiddleVowelMask) ||
|
return IOW_SAND(ConsonantMask) || IOW_SAND(MiddleVowelMask) ||
|
||||||
IOW_SAND(VowelMask) || IOW_SAND(ToneMarkerMask);
|
IOW_SAND(VowelMask) || IOW_SAND(ToneMarkerMask);
|
||||||
#undef IOW_SAND
|
#undef IOW_SAND
|
||||||
}
|
}
|
||||||
|
|
||||||
// consonants J, Q, X all require the existence of vowel I or UE
|
// consonants J, Q, X all require the existence of vowel I or UE
|
||||||
bool belongsToJQXClass() const {
|
bool belongsToJQXClass() const {
|
||||||
Component consonant = syllable_ & ConsonantMask;
|
Component consonant = syllable_ & ConsonantMask;
|
||||||
return (consonant == J || consonant == Q || consonant == X);
|
return (consonant == J || consonant == Q || consonant == X);
|
||||||
}
|
}
|
||||||
|
|
||||||
// zi, ci, si, chi, chi, shi, ri
|
// zi, ci, si, chi, chi, shi, ri
|
||||||
bool belongsToZCSRClass() const {
|
bool belongsToZCSRClass() const {
|
||||||
Component consonant = syllable_ & ConsonantMask;
|
Component consonant = syllable_ & ConsonantMask;
|
||||||
return (consonant >= ZH && consonant <= S);
|
return (consonant >= ZH && consonant <= S);
|
||||||
}
|
}
|
||||||
|
|
||||||
Component maskType() const {
|
Component maskType() const {
|
||||||
Component mask = 0;
|
Component mask = 0;
|
||||||
mask |= (syllable_ & ConsonantMask) ? ConsonantMask : 0;
|
mask |= (syllable_ & ConsonantMask) ? ConsonantMask : 0;
|
||||||
mask |= (syllable_ & MiddleVowelMask) ? MiddleVowelMask : 0;
|
mask |= (syllable_ & MiddleVowelMask) ? MiddleVowelMask : 0;
|
||||||
mask |= (syllable_ & VowelMask) ? VowelMask : 0;
|
mask |= (syllable_ & VowelMask) ? VowelMask : 0;
|
||||||
mask |= (syllable_ & ToneMarkerMask) ? ToneMarkerMask : 0;
|
mask |= (syllable_ & ToneMarkerMask) ? ToneMarkerMask : 0;
|
||||||
return mask;
|
return mask;
|
||||||
}
|
}
|
||||||
|
|
||||||
const BopomofoSyllable operator+(const BopomofoSyllable& another) const {
|
const BopomofoSyllable operator+(const BopomofoSyllable& another) const {
|
||||||
Component newSyllable = syllable_;
|
Component newSyllable = syllable_;
|
||||||
#define OP_SOVER(mask) \
|
#define OP_SOVER(mask) \
|
||||||
if (another.syllable_ & mask) { \
|
if (another.syllable_ & mask) { \
|
||||||
newSyllable = (newSyllable & ~mask) | (another.syllable_ & mask); \
|
newSyllable = (newSyllable & ~mask) | (another.syllable_ & mask); \
|
||||||
}
|
}
|
||||||
OP_SOVER(ConsonantMask);
|
OP_SOVER(ConsonantMask);
|
||||||
OP_SOVER(MiddleVowelMask);
|
OP_SOVER(MiddleVowelMask);
|
||||||
OP_SOVER(VowelMask);
|
OP_SOVER(VowelMask);
|
||||||
OP_SOVER(ToneMarkerMask);
|
OP_SOVER(ToneMarkerMask);
|
||||||
#undef OP_SOVER
|
#undef OP_SOVER
|
||||||
return BopomofoSyllable(newSyllable);
|
return BopomofoSyllable(newSyllable);
|
||||||
}
|
}
|
||||||
|
|
||||||
BopomofoSyllable& operator+=(const BopomofoSyllable& another) {
|
BopomofoSyllable& operator+=(const BopomofoSyllable& another) {
|
||||||
#define OPE_SOVER(mask) \
|
#define OPE_SOVER(mask) \
|
||||||
if (another.syllable_ & mask) { \
|
if (another.syllable_ & mask) { \
|
||||||
syllable_ = (syllable_ & ~mask) | (another.syllable_ & mask); \
|
syllable_ = (syllable_ & ~mask) | (another.syllable_ & mask); \
|
||||||
}
|
}
|
||||||
OPE_SOVER(ConsonantMask);
|
OPE_SOVER(ConsonantMask);
|
||||||
OPE_SOVER(MiddleVowelMask);
|
OPE_SOVER(MiddleVowelMask);
|
||||||
OPE_SOVER(VowelMask);
|
OPE_SOVER(VowelMask);
|
||||||
OPE_SOVER(ToneMarkerMask);
|
OPE_SOVER(ToneMarkerMask);
|
||||||
#undef OPE_SOVER
|
#undef OPE_SOVER
|
||||||
return *this;
|
return *this;
|
||||||
}
|
}
|
||||||
|
|
||||||
uint16_t absoluteOrder() const {
|
uint16_t absoluteOrder() const {
|
||||||
// turn BPMF syllable into a 4*14*4*22 number
|
// turn BPMF syllable into a 4*14*4*22 number
|
||||||
return (uint16_t)(syllable_ & ConsonantMask) +
|
return (uint16_t)(syllable_ & ConsonantMask) +
|
||||||
(uint16_t)((syllable_ & MiddleVowelMask) >> 5) * 22 +
|
(uint16_t)((syllable_ & MiddleVowelMask) >> 5) * 22 +
|
||||||
(uint16_t)((syllable_ & VowelMask) >> 7) * 22 * 4 +
|
(uint16_t)((syllable_ & VowelMask) >> 7) * 22 * 4 +
|
||||||
(uint16_t)((syllable_ & ToneMarkerMask) >> 11) * 22 * 4 * 14;
|
(uint16_t)((syllable_ & ToneMarkerMask) >> 11) * 22 * 4 * 14;
|
||||||
}
|
}
|
||||||
|
|
||||||
const std::string absoluteOrderString() const {
|
const std::string absoluteOrderString() const {
|
||||||
// 5*14*4*22 = 6160, we use a 79*79 encoding to represent that
|
// 5*14*4*22 = 6160, we use a 79*79 encoding to represent that
|
||||||
uint16_t order = absoluteOrder();
|
uint16_t order = absoluteOrder();
|
||||||
char low = 48 + static_cast<char>(order % 79);
|
char low = 48 + static_cast<char>(order % 79);
|
||||||
char high = 48 + static_cast<char>(order / 79);
|
char high = 48 + static_cast<char>(order / 79);
|
||||||
std::string result(2, ' ');
|
std::string result(2, ' ');
|
||||||
result[0] = low;
|
result[0] = low;
|
||||||
result[1] = high;
|
result[1] = high;
|
||||||
return result;
|
return result;
|
||||||
}
|
}
|
||||||
|
|
||||||
static BopomofoSyllable FromAbsoluteOrder(uint16_t order) {
|
static BopomofoSyllable FromAbsoluteOrder(uint16_t order) {
|
||||||
return BopomofoSyllable((order % 22) | ((order / 22) % 4) << 5 |
|
return BopomofoSyllable((order % 22) | ((order / 22) % 4) << 5 |
|
||||||
((order / (22 * 4)) % 14) << 7 |
|
((order / (22 * 4)) % 14) << 7 |
|
||||||
((order / (22 * 4 * 14)) % 5) << 11);
|
((order / (22 * 4 * 14)) % 5) << 11);
|
||||||
}
|
}
|
||||||
|
|
||||||
static BopomofoSyllable FromAbsoluteOrderString(const std::string& str) {
|
static BopomofoSyllable FromAbsoluteOrderString(const std::string& str) {
|
||||||
if (str.length() != 2) return BopomofoSyllable();
|
if (str.length() != 2) return BopomofoSyllable();
|
||||||
|
|
||||||
return FromAbsoluteOrder((uint16_t)(str[1] - 48) * 79 +
|
return FromAbsoluteOrder((uint16_t)(str[1] - 48) * 79 +
|
||||||
(uint16_t)(str[0] - 48));
|
(uint16_t)(str[0] - 48));
|
||||||
}
|
}
|
||||||
|
|
||||||
friend std::ostream& operator<<(std::ostream& stream,
|
friend std::ostream& operator<<(std::ostream& stream,
|
||||||
const BopomofoSyllable& syllable);
|
const BopomofoSyllable& syllable);
|
||||||
|
|
||||||
static constexpr Component
|
static constexpr Component
|
||||||
ConsonantMask = 0x001f, // 0000 0000 0001 1111, 21 consonants
|
ConsonantMask = 0x001f, // 0000 0000 0001 1111, 21 consonants
|
||||||
MiddleVowelMask = 0x0060, // 0000 0000 0110 0000, 3 middle vowels
|
MiddleVowelMask = 0x0060, // 0000 0000 0110 0000, 3 middle vowels
|
||||||
VowelMask = 0x0780, // 0000 0111 1000 0000, 13 vowels
|
VowelMask = 0x0780, // 0000 0111 1000 0000, 13 vowels
|
||||||
ToneMarkerMask = 0x3800, // 0011 1000 0000 0000, 5 tones (tone1 = 0x00)
|
ToneMarkerMask = 0x3800, // 0011 1000 0000 0000, 5 tones (tone1 = 0x00)
|
||||||
B = 0x0001, P = 0x0002, M = 0x0003, F = 0x0004, D = 0x0005, T = 0x0006,
|
B = 0x0001, P = 0x0002, M = 0x0003, F = 0x0004, D = 0x0005, T = 0x0006,
|
||||||
N = 0x0007, L = 0x0008, G = 0x0009, K = 0x000a, H = 0x000b, J = 0x000c,
|
N = 0x0007, L = 0x0008, G = 0x0009, K = 0x000a, H = 0x000b, J = 0x000c,
|
||||||
Q = 0x000d, X = 0x000e, ZH = 0x000f, CH = 0x0010, SH = 0x0011, R = 0x0012,
|
Q = 0x000d, X = 0x000e, ZH = 0x000f, CH = 0x0010, SH = 0x0011, R = 0x0012,
|
||||||
Z = 0x0013, C = 0x0014, S = 0x0015, I = 0x0020, U = 0x0040,
|
Z = 0x0013, C = 0x0014, S = 0x0015, I = 0x0020, U = 0x0040,
|
||||||
UE = 0x0060, // ue = u umlaut (we use the German convention here as an
|
UE = 0x0060, // ue = u umlaut (we use the German convention here as an
|
||||||
// ersatz to the /ju:/ sound)
|
// ersatz to the /ju:/ sound)
|
||||||
A = 0x0080, O = 0x0100, ER = 0x0180, E = 0x0200, AI = 0x0280, EI = 0x0300,
|
A = 0x0080, O = 0x0100, ER = 0x0180, E = 0x0200, AI = 0x0280, EI = 0x0300,
|
||||||
AO = 0x0380, OU = 0x0400, AN = 0x0480, EN = 0x0500, ANG = 0x0580,
|
AO = 0x0380, OU = 0x0400, AN = 0x0480, EN = 0x0500, ANG = 0x0580,
|
||||||
ENG = 0x0600, ERR = 0x0680, Tone1 = 0x0000, Tone2 = 0x0800,
|
ENG = 0x0600, ERR = 0x0680, Tone1 = 0x0000, Tone2 = 0x0800,
|
||||||
Tone3 = 0x1000, Tone4 = 0x1800, Tone5 = 0x2000;
|
Tone3 = 0x1000, Tone4 = 0x1800, Tone5 = 0x2000;
|
||||||
|
|
||||||
protected:
|
protected:
|
||||||
Component syllable_;
|
Component syllable_;
|
||||||
};
|
};
|
||||||
|
|
||||||
inline std::ostream& operator<<(std::ostream& stream,
|
inline std::ostream& operator<<(std::ostream& stream,
|
||||||
const BopomofoSyllable& syllable) {
|
const BopomofoSyllable& syllable) {
|
||||||
stream << syllable.composedString();
|
stream << syllable.composedString();
|
||||||
return stream;
|
return stream;
|
||||||
}
|
}
|
||||||
|
|
||||||
typedef BopomofoSyllable BPMF;
|
typedef BopomofoSyllable BPMF;
|
||||||
|
|
@ -207,309 +207,308 @@ typedef std::map<char, std::vector<BPMF::Component> > BopomofoKeyToComponentMap;
|
||||||
typedef std::map<BPMF::Component, char> BopomofoComponentToKeyMap;
|
typedef std::map<BPMF::Component, char> BopomofoComponentToKeyMap;
|
||||||
|
|
||||||
class BopomofoKeyboardLayout {
|
class BopomofoKeyboardLayout {
|
||||||
public:
|
public:
|
||||||
static const BopomofoKeyboardLayout* StandardLayout();
|
static const BopomofoKeyboardLayout* StandardLayout();
|
||||||
static const BopomofoKeyboardLayout* ETenLayout();
|
static const BopomofoKeyboardLayout* ETenLayout();
|
||||||
static const BopomofoKeyboardLayout* HsuLayout();
|
static const BopomofoKeyboardLayout* HsuLayout();
|
||||||
static const BopomofoKeyboardLayout* ETen26Layout();
|
static const BopomofoKeyboardLayout* ETen26Layout();
|
||||||
static const BopomofoKeyboardLayout* IBMLayout();
|
static const BopomofoKeyboardLayout* IBMLayout();
|
||||||
static const BopomofoKeyboardLayout* HanyuPinyinLayout();
|
static const BopomofoKeyboardLayout* HanyuPinyinLayout();
|
||||||
|
|
||||||
BopomofoKeyboardLayout(const BopomofoKeyToComponentMap& ktcm,
|
BopomofoKeyboardLayout(const BopomofoKeyToComponentMap& ktcm,
|
||||||
const std::string& name)
|
const std::string& name)
|
||||||
: m_keyToComponent(ktcm), m_name(name) {
|
: m_keyToComponent(ktcm), m_name(name) {
|
||||||
for (BopomofoKeyToComponentMap::const_iterator miter =
|
for (BopomofoKeyToComponentMap::const_iterator miter =
|
||||||
m_keyToComponent.begin();
|
m_keyToComponent.begin();
|
||||||
miter != m_keyToComponent.end(); ++miter)
|
miter != m_keyToComponent.end(); ++miter)
|
||||||
for (std::vector<BPMF::Component>::const_iterator viter =
|
for (std::vector<BPMF::Component>::const_iterator viter =
|
||||||
(*miter).second.begin();
|
(*miter).second.begin();
|
||||||
viter != (*miter).second.end(); ++viter)
|
viter != (*miter).second.end(); ++viter)
|
||||||
m_componentToKey[*viter] = (*miter).first;
|
m_componentToKey[*viter] = (*miter).first;
|
||||||
}
|
}
|
||||||
|
|
||||||
const std::string name() const { return m_name; }
|
const std::string name() const { return m_name; }
|
||||||
|
|
||||||
char componentToKey(BPMF::Component component) const {
|
char componentToKey(BPMF::Component component) const {
|
||||||
BopomofoComponentToKeyMap::const_iterator iter =
|
BopomofoComponentToKeyMap::const_iterator iter =
|
||||||
m_componentToKey.find(component);
|
m_componentToKey.find(component);
|
||||||
return (iter == m_componentToKey.end()) ? 0 : (*iter).second;
|
return (iter == m_componentToKey.end()) ? 0 : (*iter).second;
|
||||||
}
|
}
|
||||||
|
|
||||||
const std::vector<BPMF::Component> keyToComponents(char key) const {
|
const std::vector<BPMF::Component> keyToComponents(char key) const {
|
||||||
BopomofoKeyToComponentMap::const_iterator iter = m_keyToComponent.find(key);
|
BopomofoKeyToComponentMap::const_iterator iter = m_keyToComponent.find(key);
|
||||||
return (iter == m_keyToComponent.end()) ? std::vector<BPMF::Component>()
|
return (iter == m_keyToComponent.end()) ? std::vector<BPMF::Component>()
|
||||||
: (*iter).second;
|
: (*iter).second;
|
||||||
}
|
}
|
||||||
|
|
||||||
const std::string keySequenceFromSyllable(BPMF syllable) const {
|
const std::string keySequenceFromSyllable(BPMF syllable) const {
|
||||||
std::string sequence;
|
std::string sequence;
|
||||||
|
|
||||||
BPMF::Component c;
|
BPMF::Component c;
|
||||||
char k;
|
char k;
|
||||||
#define STKS_COMBINE(component) \
|
#define STKS_COMBINE(component) \
|
||||||
if ((c = component)) { \
|
if ((c = component)) { \
|
||||||
if ((k = componentToKey(c))) sequence += std::string(1, k); \
|
if ((k = componentToKey(c))) sequence += std::string(1, k); \
|
||||||
}
|
}
|
||||||
STKS_COMBINE(syllable.consonantComponent());
|
STKS_COMBINE(syllable.consonantComponent());
|
||||||
STKS_COMBINE(syllable.middleVowelComponent());
|
STKS_COMBINE(syllable.middleVowelComponent());
|
||||||
STKS_COMBINE(syllable.vowelComponent());
|
STKS_COMBINE(syllable.vowelComponent());
|
||||||
STKS_COMBINE(syllable.toneMarkerComponent());
|
STKS_COMBINE(syllable.toneMarkerComponent());
|
||||||
#undef STKS_COMBINE
|
#undef STKS_COMBINE
|
||||||
return sequence;
|
return sequence;
|
||||||
}
|
|
||||||
|
|
||||||
const BPMF syllableFromKeySequence(const std::string& sequence) const {
|
|
||||||
BPMF syllable;
|
|
||||||
|
|
||||||
for (std::string::const_iterator iter = sequence.begin();
|
|
||||||
iter != sequence.end(); ++iter) {
|
|
||||||
bool beforeSeqHasIorUE = sequenceContainsIorUE(sequence.begin(), iter);
|
|
||||||
bool aheadSeqHasIorUE = sequenceContainsIorUE(iter + 1, sequence.end());
|
|
||||||
|
|
||||||
std::vector<BPMF::Component> components = keyToComponents(*iter);
|
|
||||||
|
|
||||||
if (!components.size()) continue;
|
|
||||||
|
|
||||||
if (components.size() == 1) {
|
|
||||||
syllable += BPMF(components[0]);
|
|
||||||
continue;
|
|
||||||
}
|
|
||||||
|
|
||||||
BPMF head = BPMF(components[0]);
|
|
||||||
BPMF follow = BPMF(components[1]);
|
|
||||||
BPMF ending = components.size() > 2 ? BPMF(components[2]) : follow;
|
|
||||||
|
|
||||||
// apply the I/UE + E rule
|
|
||||||
if (head.vowelComponent() == BPMF::E &&
|
|
||||||
follow.vowelComponent() != BPMF::E) {
|
|
||||||
syllable += beforeSeqHasIorUE ? head : follow;
|
|
||||||
continue;
|
|
||||||
}
|
|
||||||
|
|
||||||
if (head.vowelComponent() != BPMF::E &&
|
|
||||||
follow.vowelComponent() == BPMF::E) {
|
|
||||||
syllable += beforeSeqHasIorUE ? follow : head;
|
|
||||||
continue;
|
|
||||||
}
|
|
||||||
|
|
||||||
// apply the J/Q/X + I/UE rule, only two components are allowed in the
|
|
||||||
// components vector here
|
|
||||||
if (head.belongsToJQXClass() && !follow.belongsToJQXClass()) {
|
|
||||||
if (!syllable.isEmpty()) {
|
|
||||||
if (ending != follow) syllable += ending;
|
|
||||||
} else {
|
|
||||||
syllable += aheadSeqHasIorUE ? head : follow;
|
|
||||||
}
|
|
||||||
|
|
||||||
continue;
|
|
||||||
}
|
|
||||||
|
|
||||||
if (!head.belongsToJQXClass() && follow.belongsToJQXClass()) {
|
|
||||||
if (!syllable.isEmpty()) {
|
|
||||||
if (ending != follow) syllable += ending;
|
|
||||||
} else {
|
|
||||||
syllable += aheadSeqHasIorUE ? follow : head;
|
|
||||||
}
|
|
||||||
|
|
||||||
continue;
|
|
||||||
}
|
|
||||||
|
|
||||||
// the nasty issue of only one char in the buffer
|
|
||||||
if (iter == sequence.begin() && iter + 1 == sequence.end()) {
|
|
||||||
if (head.hasVowel() || follow.hasToneMarker() ||
|
|
||||||
head.belongsToZCSRClass()) {
|
|
||||||
syllable += head;
|
|
||||||
} else {
|
|
||||||
if (follow.hasVowel() || ending.hasToneMarker()) {
|
|
||||||
syllable += follow;
|
|
||||||
} else {
|
|
||||||
syllable += ending;
|
|
||||||
}
|
|
||||||
}
|
|
||||||
|
|
||||||
continue;
|
|
||||||
}
|
|
||||||
|
|
||||||
if (!(syllable.maskType() & head.maskType()) &&
|
|
||||||
!endAheadOrAheadHasToneMarkKey(iter + 1, sequence.end())) {
|
|
||||||
syllable += head;
|
|
||||||
} else {
|
|
||||||
if (endAheadOrAheadHasToneMarkKey(iter + 1, sequence.end()) &&
|
|
||||||
head.belongsToZCSRClass() && syllable.isEmpty()) {
|
|
||||||
syllable += head;
|
|
||||||
} else if (syllable.maskType() < follow.maskType()) {
|
|
||||||
syllable += follow;
|
|
||||||
} else {
|
|
||||||
syllable += ending;
|
|
||||||
}
|
|
||||||
}
|
|
||||||
}
|
}
|
||||||
|
|
||||||
// heuristics for Hsu keyboard layout
|
const BPMF syllableFromKeySequence(const std::string& sequence) const {
|
||||||
if (this == HsuLayout()) {
|
BPMF syllable;
|
||||||
// fix the left out L to ERR when it has sound, and GI, GUE -> JI, JUE
|
|
||||||
if (syllable.vowelComponent() == BPMF::ENG && !syllable.hasConsonant() &&
|
for (std::string::const_iterator iter = sequence.begin();
|
||||||
!syllable.hasMiddleVowel()) {
|
iter != sequence.end(); ++iter) {
|
||||||
syllable += BPMF(BPMF::ERR);
|
bool beforeSeqHasIorUE = sequenceContainsIorUE(sequence.begin(), iter);
|
||||||
} else if (syllable.consonantComponent() == BPMF::G &&
|
bool aheadSeqHasIorUE = sequenceContainsIorUE(iter + 1, sequence.end());
|
||||||
(syllable.middleVowelComponent() == BPMF::I ||
|
|
||||||
syllable.middleVowelComponent() == BPMF::UE)) {
|
std::vector<BPMF::Component> components = keyToComponents(*iter);
|
||||||
syllable += BPMF(BPMF::J);
|
|
||||||
}
|
if (!components.size()) continue;
|
||||||
|
|
||||||
|
if (components.size() == 1) {
|
||||||
|
syllable += BPMF(components[0]);
|
||||||
|
continue;
|
||||||
|
}
|
||||||
|
|
||||||
|
BPMF head = BPMF(components[0]);
|
||||||
|
BPMF follow = BPMF(components[1]);
|
||||||
|
BPMF ending = components.size() > 2 ? BPMF(components[2]) : follow;
|
||||||
|
|
||||||
|
// apply the I/UE + E rule
|
||||||
|
if (head.vowelComponent() == BPMF::E &&
|
||||||
|
follow.vowelComponent() != BPMF::E) {
|
||||||
|
syllable += beforeSeqHasIorUE ? head : follow;
|
||||||
|
continue;
|
||||||
|
}
|
||||||
|
|
||||||
|
if (head.vowelComponent() != BPMF::E &&
|
||||||
|
follow.vowelComponent() == BPMF::E) {
|
||||||
|
syllable += beforeSeqHasIorUE ? follow : head;
|
||||||
|
continue;
|
||||||
|
}
|
||||||
|
|
||||||
|
// apply the J/Q/X + I/UE rule, only two components are allowed in the
|
||||||
|
// components vector here
|
||||||
|
if (head.belongsToJQXClass() && !follow.belongsToJQXClass()) {
|
||||||
|
if (!syllable.isEmpty()) {
|
||||||
|
if (ending != follow) syllable += ending;
|
||||||
|
} else {
|
||||||
|
syllable += aheadSeqHasIorUE ? head : follow;
|
||||||
|
}
|
||||||
|
|
||||||
|
continue;
|
||||||
|
}
|
||||||
|
|
||||||
|
if (!head.belongsToJQXClass() && follow.belongsToJQXClass()) {
|
||||||
|
if (!syllable.isEmpty()) {
|
||||||
|
if (ending != follow) syllable += ending;
|
||||||
|
} else {
|
||||||
|
syllable += aheadSeqHasIorUE ? follow : head;
|
||||||
|
}
|
||||||
|
|
||||||
|
continue;
|
||||||
|
}
|
||||||
|
|
||||||
|
// the nasty issue of only one char in the buffer
|
||||||
|
if (iter == sequence.begin() && iter + 1 == sequence.end()) {
|
||||||
|
if (head.hasVowel() || follow.hasToneMarker() ||
|
||||||
|
head.belongsToZCSRClass()) {
|
||||||
|
syllable += head;
|
||||||
|
} else {
|
||||||
|
if (follow.hasVowel() || ending.hasToneMarker()) {
|
||||||
|
syllable += follow;
|
||||||
|
} else {
|
||||||
|
syllable += ending;
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
continue;
|
||||||
|
}
|
||||||
|
|
||||||
|
if (!(syllable.maskType() & head.maskType()) &&
|
||||||
|
!endAheadOrAheadHasToneMarkKey(iter + 1, sequence.end())) {
|
||||||
|
syllable += head;
|
||||||
|
} else {
|
||||||
|
if (endAheadOrAheadHasToneMarkKey(iter + 1, sequence.end()) &&
|
||||||
|
head.belongsToZCSRClass() && syllable.isEmpty()) {
|
||||||
|
syllable += head;
|
||||||
|
} else if (syllable.maskType() < follow.maskType()) {
|
||||||
|
syllable += follow;
|
||||||
|
} else {
|
||||||
|
syllable += ending;
|
||||||
|
}
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
// heuristics for Hsu keyboard layout
|
||||||
|
if (this == HsuLayout()) {
|
||||||
|
// fix the left out L to ERR when it has sound, and GI, GUE -> JI, JUE
|
||||||
|
if (syllable.vowelComponent() == BPMF::ENG && !syllable.hasConsonant() &&
|
||||||
|
!syllable.hasMiddleVowel()) {
|
||||||
|
syllable += BPMF(BPMF::ERR);
|
||||||
|
} else if (syllable.consonantComponent() == BPMF::G &&
|
||||||
|
(syllable.middleVowelComponent() == BPMF::I ||
|
||||||
|
syllable.middleVowelComponent() == BPMF::UE)) {
|
||||||
|
syllable += BPMF(BPMF::J);
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
return syllable;
|
||||||
}
|
}
|
||||||
|
|
||||||
return syllable;
|
protected:
|
||||||
}
|
bool endAheadOrAheadHasToneMarkKey(std::string::const_iterator ahead,
|
||||||
|
std::string::const_iterator end) const {
|
||||||
|
if (ahead == end) return true;
|
||||||
|
|
||||||
protected:
|
char tone1 = componentToKey(BPMF::Tone1);
|
||||||
bool endAheadOrAheadHasToneMarkKey(std::string::const_iterator ahead,
|
char tone2 = componentToKey(BPMF::Tone2);
|
||||||
std::string::const_iterator end) const {
|
char tone3 = componentToKey(BPMF::Tone3);
|
||||||
if (ahead == end) return true;
|
char tone4 = componentToKey(BPMF::Tone4);
|
||||||
|
char tone5 = componentToKey(BPMF::Tone5);
|
||||||
|
|
||||||
char tone1 = componentToKey(BPMF::Tone1);
|
if (tone1)
|
||||||
char tone2 = componentToKey(BPMF::Tone2);
|
if (*ahead == tone1) return true;
|
||||||
char tone3 = componentToKey(BPMF::Tone3);
|
|
||||||
char tone4 = componentToKey(BPMF::Tone4);
|
|
||||||
char tone5 = componentToKey(BPMF::Tone5);
|
|
||||||
|
|
||||||
if (tone1)
|
if (*ahead == tone2 || *ahead == tone3 || *ahead == tone4 ||
|
||||||
if (*ahead == tone1) return true;
|
*ahead == tone5)
|
||||||
|
return true;
|
||||||
|
|
||||||
if (*ahead == tone2 || *ahead == tone3 || *ahead == tone4 ||
|
return false;
|
||||||
*ahead == tone5)
|
}
|
||||||
return true;
|
|
||||||
|
|
||||||
return false;
|
bool sequenceContainsIorUE(std::string::const_iterator start,
|
||||||
}
|
std::string::const_iterator end) const {
|
||||||
|
char iChar = componentToKey(BPMF::I);
|
||||||
|
char ueChar = componentToKey(BPMF::UE);
|
||||||
|
|
||||||
bool sequenceContainsIorUE(std::string::const_iterator start,
|
for (; start != end; ++start)
|
||||||
std::string::const_iterator end) const {
|
if (*start == iChar || *start == ueChar) return true;
|
||||||
char iChar = componentToKey(BPMF::I);
|
return false;
|
||||||
char ueChar = componentToKey(BPMF::UE);
|
}
|
||||||
|
|
||||||
for (; start != end; ++start)
|
std::string m_name;
|
||||||
if (*start == iChar || *start == ueChar) return true;
|
BopomofoKeyToComponentMap m_keyToComponent;
|
||||||
return false;
|
BopomofoComponentToKeyMap m_componentToKey;
|
||||||
}
|
|
||||||
|
|
||||||
std::string m_name;
|
|
||||||
BopomofoKeyToComponentMap m_keyToComponent;
|
|
||||||
BopomofoComponentToKeyMap m_componentToKey;
|
|
||||||
};
|
};
|
||||||
|
|
||||||
class BopomofoReadingBuffer {
|
class BopomofoReadingBuffer {
|
||||||
public:
|
public:
|
||||||
explicit BopomofoReadingBuffer(const BopomofoKeyboardLayout* layout)
|
explicit BopomofoReadingBuffer(const BopomofoKeyboardLayout* layout)
|
||||||
: layout_(layout), pinyin_mode_(false) {
|
: layout_(layout), pinyin_mode_(false) {
|
||||||
if (layout == BopomofoKeyboardLayout::HanyuPinyinLayout()) {
|
if (layout == BopomofoKeyboardLayout::HanyuPinyinLayout()) {
|
||||||
pinyin_mode_ = true;
|
pinyin_mode_ = true;
|
||||||
pinyin_sequence_ = "";
|
pinyin_sequence_ = "";
|
||||||
}
|
|
||||||
}
|
|
||||||
|
|
||||||
void setKeyboardLayout(const BopomofoKeyboardLayout* layout) {
|
|
||||||
layout_ = layout;
|
|
||||||
|
|
||||||
if (layout == BopomofoKeyboardLayout::HanyuPinyinLayout()) {
|
|
||||||
pinyin_mode_ = true;
|
|
||||||
pinyin_sequence_ = "";
|
|
||||||
}
|
|
||||||
}
|
|
||||||
|
|
||||||
bool isValidKey(char k) const {
|
|
||||||
if (!pinyin_mode_) {
|
|
||||||
return layout_ ? (layout_->keyToComponents(k)).size() > 0 : false;
|
|
||||||
}
|
|
||||||
|
|
||||||
char lk = tolower(k);
|
|
||||||
if (lk >= 'a' && lk <= 'z') {
|
|
||||||
// if a tone marker is already in place
|
|
||||||
if (pinyin_sequence_.length()) {
|
|
||||||
char lastc = pinyin_sequence_[pinyin_sequence_.length() - 1];
|
|
||||||
if (lastc >= '2' && lastc <= '5') {
|
|
||||||
return false;
|
|
||||||
}
|
}
|
||||||
return true;
|
|
||||||
}
|
|
||||||
return true;
|
|
||||||
}
|
}
|
||||||
|
|
||||||
if (pinyin_sequence_.length() && (lk >= '2' && lk <= '5')) {
|
void setKeyboardLayout(const BopomofoKeyboardLayout* layout) {
|
||||||
return true;
|
layout_ = layout;
|
||||||
|
|
||||||
|
if (layout == BopomofoKeyboardLayout::HanyuPinyinLayout()) {
|
||||||
|
pinyin_mode_ = true;
|
||||||
|
pinyin_sequence_ = "";
|
||||||
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
return false;
|
bool isValidKey(char k) const {
|
||||||
}
|
if (!pinyin_mode_) {
|
||||||
|
return layout_ ? (layout_->keyToComponents(k)).size() > 0 : false;
|
||||||
|
}
|
||||||
|
|
||||||
bool combineKey(char k) {
|
char lk = tolower(k);
|
||||||
if (!isValidKey(k)) return false;
|
if (lk >= 'a' && lk <= 'z') {
|
||||||
|
// if a tone marker is already in place
|
||||||
|
if (pinyin_sequence_.length()) {
|
||||||
|
char lastc = pinyin_sequence_[pinyin_sequence_.length() - 1];
|
||||||
|
if (lastc >= '2' && lastc <= '5') {
|
||||||
|
return false;
|
||||||
|
}
|
||||||
|
return true;
|
||||||
|
}
|
||||||
|
return true;
|
||||||
|
}
|
||||||
|
|
||||||
if (pinyin_mode_) {
|
if (pinyin_sequence_.length() && (lk >= '2' && lk <= '5')) {
|
||||||
pinyin_sequence_ += std::string(1, tolower(k));
|
return true;
|
||||||
syllable_ = BPMF::FromHanyuPinyin(pinyin_sequence_);
|
}
|
||||||
return true;
|
|
||||||
|
return false;
|
||||||
}
|
}
|
||||||
|
|
||||||
std::string sequence =
|
bool combineKey(char k) {
|
||||||
|
if (!isValidKey(k)) return false;
|
||||||
|
|
||||||
|
if (pinyin_mode_) {
|
||||||
|
pinyin_sequence_ += std::string(1, tolower(k));
|
||||||
|
syllable_ = BPMF::FromHanyuPinyin(pinyin_sequence_);
|
||||||
|
return true;
|
||||||
|
}
|
||||||
|
|
||||||
|
std::string sequence =
|
||||||
layout_->keySequenceFromSyllable(syllable_) + std::string(1, k);
|
layout_->keySequenceFromSyllable(syllable_) + std::string(1, k);
|
||||||
syllable_ = layout_->syllableFromKeySequence(sequence);
|
syllable_ = layout_->syllableFromKeySequence(sequence);
|
||||||
return true;
|
return true;
|
||||||
}
|
|
||||||
|
|
||||||
void clear() {
|
|
||||||
pinyin_sequence_.clear();
|
|
||||||
syllable_.clear();
|
|
||||||
}
|
|
||||||
|
|
||||||
void backspace() {
|
|
||||||
if (!layout_) return;
|
|
||||||
|
|
||||||
if (pinyin_mode_) {
|
|
||||||
if (pinyin_sequence_.length()) {
|
|
||||||
pinyin_sequence_ =
|
|
||||||
pinyin_sequence_.substr(0, pinyin_sequence_.length() - 1);
|
|
||||||
}
|
|
||||||
|
|
||||||
syllable_ = BPMF::FromHanyuPinyin(pinyin_sequence_);
|
|
||||||
return;
|
|
||||||
}
|
}
|
||||||
|
|
||||||
std::string sequence = layout_->keySequenceFromSyllable(syllable_);
|
void clear() {
|
||||||
if (sequence.length()) {
|
pinyin_sequence_.clear();
|
||||||
sequence = sequence.substr(0, sequence.length() - 1);
|
syllable_.clear();
|
||||||
syllable_ = layout_->syllableFromKeySequence(sequence);
|
|
||||||
}
|
|
||||||
}
|
|
||||||
|
|
||||||
bool isEmpty() const { return syllable_.isEmpty(); }
|
|
||||||
|
|
||||||
const std::string composedString() const {
|
|
||||||
if (pinyin_mode_) {
|
|
||||||
return pinyin_sequence_;
|
|
||||||
}
|
}
|
||||||
|
|
||||||
return syllable_.composedString();
|
void backspace() {
|
||||||
}
|
if (!layout_) return;
|
||||||
|
|
||||||
const BPMF syllable() const { return syllable_; }
|
if (pinyin_mode_) {
|
||||||
|
if (pinyin_sequence_.length()) {
|
||||||
|
pinyin_sequence_ =
|
||||||
|
pinyin_sequence_.substr(0, pinyin_sequence_.length() - 1);
|
||||||
|
}
|
||||||
|
|
||||||
const std::string standardLayoutQueryString() const {
|
syllable_ = BPMF::FromHanyuPinyin(pinyin_sequence_);
|
||||||
return BopomofoKeyboardLayout::StandardLayout()->keySequenceFromSyllable(
|
return;
|
||||||
syllable_);
|
}
|
||||||
}
|
|
||||||
|
|
||||||
const std::string absoluteOrderQueryString() const {
|
std::string sequence = layout_->keySequenceFromSyllable(syllable_);
|
||||||
return syllable_.absoluteOrderString();
|
if (sequence.length()) {
|
||||||
}
|
sequence = sequence.substr(0, sequence.length() - 1);
|
||||||
|
syllable_ = layout_->syllableFromKeySequence(sequence);
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
bool hasToneMarker() const { return syllable_.hasToneMarker(); }
|
bool isEmpty() const { return syllable_.isEmpty(); }
|
||||||
|
|
||||||
protected:
|
const std::string composedString() const {
|
||||||
const BopomofoKeyboardLayout* layout_;
|
if (pinyin_mode_) {
|
||||||
BPMF syllable_;
|
return pinyin_sequence_;
|
||||||
|
}
|
||||||
|
|
||||||
bool pinyin_mode_;
|
return syllable_.composedString();
|
||||||
std::string pinyin_sequence_;
|
}
|
||||||
|
|
||||||
|
const BPMF syllable() const { return syllable_; }
|
||||||
|
|
||||||
|
const std::string standardLayoutQueryString() const {
|
||||||
|
return BopomofoKeyboardLayout::StandardLayout()->keySequenceFromSyllable(syllable_);
|
||||||
|
}
|
||||||
|
|
||||||
|
const std::string absoluteOrderQueryString() const {
|
||||||
|
return syllable_.absoluteOrderString();
|
||||||
|
}
|
||||||
|
|
||||||
|
bool hasToneMarker() const { return syllable_.hasToneMarker(); }
|
||||||
|
|
||||||
|
protected:
|
||||||
|
const BopomofoKeyboardLayout* layout_;
|
||||||
|
BPMF syllable_;
|
||||||
|
|
||||||
|
bool pinyin_mode_;
|
||||||
|
std::string pinyin_sequence_;
|
||||||
};
|
};
|
||||||
} // namespace Mandarin
|
} // namespace Mandarin
|
||||||
} // namespace Taiyan
|
} // namespace Taiyan
|
||||||
|
|
|
||||||
Loading…
Reference in New Issue