Lukhnos: Gramambular // Modernization.
Co-Authored-By: Lukhnos Liu <lukhnos@lukhnos.org>
This commit is contained in:
parent
256a20d93f
commit
8bb1ad0be3
|
@ -17,8 +17,8 @@ THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABI
|
||||||
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
||||||
*/
|
*/
|
||||||
|
|
||||||
#ifndef Bigram_h
|
#ifndef BIGRAM_H_
|
||||||
#define Bigram_h
|
#define BIGRAM_H_
|
||||||
|
|
||||||
#include <vector>
|
#include <vector>
|
||||||
|
|
||||||
|
@ -34,65 +34,60 @@ namespace Taiyan {
|
||||||
KeyValuePair keyValue;
|
KeyValuePair keyValue;
|
||||||
double score;
|
double score;
|
||||||
|
|
||||||
bool operator==(const Bigram& inAnother) const;
|
bool operator==(const Bigram& another) const;
|
||||||
bool operator<(const Bigram& inAnother) const;
|
bool operator<(const Bigram& another) const;
|
||||||
};
|
};
|
||||||
|
|
||||||
inline ostream& operator<<(ostream& inStream, const Bigram& inGram)
|
inline std::ostream& operator<<(std::ostream& stream, const Bigram& gram) {
|
||||||
{
|
std::streamsize p = stream.precision();
|
||||||
streamsize p = inStream.precision();
|
stream.precision(6);
|
||||||
inStream.precision(6);
|
stream << "(" << gram.keyValue << "|" << gram.preceedingKeyValue << ","
|
||||||
inStream << "(" << inGram.keyValue << "|" <<inGram.preceedingKeyValue << "," << inGram.score << ")";
|
<< gram.score << ")";
|
||||||
inStream.precision(p);
|
stream.precision(p);
|
||||||
return inStream;
|
return stream;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline ostream& operator<<(ostream& inStream, const vector<Bigram>& inGrams)
|
inline std::ostream& operator<<(std::ostream& stream,
|
||||||
{
|
const std::vector<Bigram>& grams) {
|
||||||
inStream << "[" << inGrams.size() << "]=>{";
|
stream << "[" << grams.size() << "]=>{";
|
||||||
|
|
||||||
size_t index = 0;
|
size_t index = 0;
|
||||||
|
|
||||||
for (vector<Bigram>::const_iterator gi = inGrams.begin() ; gi != inGrams.end() ; ++gi, ++index) {
|
for (std::vector<Bigram>::const_iterator gi = grams.begin();
|
||||||
inStream << index << "=>";
|
gi != grams.end(); ++gi, ++index) {
|
||||||
inStream << *gi;
|
stream << index << "=>";
|
||||||
if (gi + 1 != inGrams.end()) {
|
stream << *gi;
|
||||||
inStream << ",";
|
if (gi + 1 != grams.end()) {
|
||||||
|
stream << ",";
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
inStream << "}";
|
stream << "}";
|
||||||
return inStream;
|
return stream;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline Bigram::Bigram()
|
inline Bigram::Bigram() : score(0.0) {}
|
||||||
: score(0.0)
|
|
||||||
{
|
inline bool Bigram::operator==(const Bigram& another) const {
|
||||||
|
return preceedingKeyValue == another.preceedingKeyValue &&
|
||||||
|
keyValue == another.keyValue && score == another.score;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline bool Bigram::operator==(const Bigram& inAnother) const
|
inline bool Bigram::operator<(const Bigram& another) const {
|
||||||
{
|
if (preceedingKeyValue < another.preceedingKeyValue) {
|
||||||
return preceedingKeyValue == inAnother.preceedingKeyValue && keyValue == inAnother.keyValue && score == inAnother.score;
|
|
||||||
}
|
|
||||||
|
|
||||||
inline bool Bigram::operator<(const Bigram& inAnother) const
|
|
||||||
{
|
|
||||||
if (preceedingKeyValue < inAnother.preceedingKeyValue) {
|
|
||||||
return true;
|
return true;
|
||||||
}
|
} else if (preceedingKeyValue == another.preceedingKeyValue) {
|
||||||
else if (preceedingKeyValue == inAnother.preceedingKeyValue) {
|
if (keyValue < another.keyValue) {
|
||||||
if (keyValue < inAnother.keyValue) {
|
|
||||||
return true;
|
return true;
|
||||||
}
|
} else if (keyValue == another.keyValue) {
|
||||||
else if (keyValue == inAnother.keyValue) {
|
return score < another.score;
|
||||||
return score < inAnother.score;
|
|
||||||
}
|
}
|
||||||
return false;
|
return false;
|
||||||
}
|
}
|
||||||
|
|
||||||
return false;
|
return false;
|
||||||
}
|
}
|
||||||
}
|
} // namespace Gramambular
|
||||||
}
|
} // namespace Taiyan
|
||||||
|
|
||||||
#endif
|
#endif
|
||||||
|
|
|
@ -17,123 +17,113 @@ THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABI
|
||||||
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
||||||
*/
|
*/
|
||||||
|
|
||||||
#ifndef BlockReadingBuilder_h
|
#ifndef BLOCKREADINGBUILDER_H_
|
||||||
#define BlockReadingBuilder_h
|
#define BLOCKREADINGBUILDER_H_
|
||||||
|
|
||||||
|
#include <string>
|
||||||
#include <vector>
|
#include <vector>
|
||||||
|
|
||||||
#include "Grid.h"
|
#include "Grid.h"
|
||||||
#include "LanguageModel.h"
|
#include "LanguageModel.h"
|
||||||
|
|
||||||
namespace Taiyan {
|
namespace Taiyan {
|
||||||
namespace Gramambular {
|
namespace Gramambular {
|
||||||
using namespace std;
|
|
||||||
|
|
||||||
class BlockReadingBuilder {
|
class BlockReadingBuilder {
|
||||||
public:
|
public:
|
||||||
BlockReadingBuilder(LanguageModel *inLM);
|
explicit BlockReadingBuilder(LanguageModel* lm);
|
||||||
void clear();
|
void clear();
|
||||||
|
|
||||||
size_t length() const;
|
size_t length() const;
|
||||||
size_t cursorIndex() const;
|
size_t cursorIndex() const;
|
||||||
void setCursorIndex(size_t inNewIndex);
|
void setCursorIndex(size_t newIndex);
|
||||||
void insertReadingAtCursor(const string& inReading);
|
void insertReadingAtCursor(const std::string& reading);
|
||||||
bool deleteReadingBeforeCursor(); // backspace
|
bool deleteReadingBeforeCursor(); // backspace
|
||||||
bool deleteReadingAfterCursor(); // delete
|
bool deleteReadingAfterCursor(); // delete
|
||||||
|
|
||||||
bool removeHeadReadings(size_t count);
|
bool removeHeadReadings(size_t count);
|
||||||
|
|
||||||
void setJoinSeparator(const string& separator);
|
void setJoinSeparator(const std::string& separator);
|
||||||
const string joinSeparator() const;
|
const std::string joinSeparator() const;
|
||||||
|
|
||||||
vector<string> readings() const;
|
std::vector<std::string> readings() const;
|
||||||
|
|
||||||
Grid& grid();
|
Grid& grid();
|
||||||
|
|
||||||
protected:
|
protected:
|
||||||
void build();
|
void build();
|
||||||
|
|
||||||
static const string Join(vector<string>::const_iterator begin, vector<string>::const_iterator end, const string& separator);
|
static const std::string Join(std::vector<std::string>::const_iterator begin,
|
||||||
|
std::vector<std::string>::const_iterator end,
|
||||||
|
const std::string& separator);
|
||||||
|
|
||||||
// 最多使用六個字組成一個詞
|
// 最多使用六個字組成一個詞
|
||||||
static const size_t MaximumBuildSpanLength = 6;
|
static const size_t MaximumBuildSpanLength = 6;
|
||||||
|
|
||||||
size_t m_cursorIndex;
|
size_t m_cursorIndex;
|
||||||
vector<string> m_readings;
|
std::vector<std::string> m_readings;
|
||||||
|
|
||||||
Grid m_grid;
|
Grid m_grid;
|
||||||
LanguageModel* m_LM;
|
LanguageModel* m_LM;
|
||||||
string m_joinSeparator;
|
std::string m_joinSeparator;
|
||||||
};
|
};
|
||||||
|
|
||||||
inline BlockReadingBuilder::BlockReadingBuilder(LanguageModel *inLM)
|
inline BlockReadingBuilder::BlockReadingBuilder(LanguageModel* lm)
|
||||||
: m_LM(inLM)
|
: m_LM(lm), m_cursorIndex(0) {}
|
||||||
, m_cursorIndex(0)
|
|
||||||
{
|
|
||||||
}
|
|
||||||
|
|
||||||
inline void BlockReadingBuilder::clear()
|
inline void BlockReadingBuilder::clear() {
|
||||||
{
|
|
||||||
m_cursorIndex = 0;
|
m_cursorIndex = 0;
|
||||||
m_readings.clear();
|
m_readings.clear();
|
||||||
m_grid.clear();
|
m_grid.clear();
|
||||||
}
|
}
|
||||||
|
|
||||||
inline size_t BlockReadingBuilder::length() const
|
inline size_t BlockReadingBuilder::length() const { return m_readings.size(); }
|
||||||
{
|
|
||||||
return m_readings.size();
|
inline size_t BlockReadingBuilder::cursorIndex() const { return m_cursorIndex; }
|
||||||
|
|
||||||
|
inline void BlockReadingBuilder::setCursorIndex(size_t newIndex) {
|
||||||
|
m_cursorIndex = newIndex > m_readings.size() ? m_readings.size() : newIndex;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline size_t BlockReadingBuilder::cursorIndex() const
|
inline void BlockReadingBuilder::insertReadingAtCursor(
|
||||||
{
|
const std::string& reading) {
|
||||||
return m_cursorIndex;
|
m_readings.insert(m_readings.begin() + m_cursorIndex, reading);
|
||||||
}
|
|
||||||
|
|
||||||
inline void BlockReadingBuilder::setCursorIndex(size_t inNewIndex)
|
|
||||||
{
|
|
||||||
m_cursorIndex = inNewIndex > m_readings.size() ? m_readings.size() : inNewIndex;
|
|
||||||
}
|
|
||||||
|
|
||||||
inline void BlockReadingBuilder::insertReadingAtCursor(const string& inReading)
|
|
||||||
{
|
|
||||||
m_readings.insert(m_readings.begin() + m_cursorIndex, inReading);
|
|
||||||
|
|
||||||
m_grid.expandGridByOneAtLocation(m_cursorIndex);
|
m_grid.expandGridByOneAtLocation(m_cursorIndex);
|
||||||
build();
|
build();
|
||||||
m_cursorIndex++;
|
m_cursorIndex++;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline vector<string> BlockReadingBuilder::readings() const
|
inline std::vector<std::string> BlockReadingBuilder::readings() const {
|
||||||
{
|
|
||||||
return m_readings;
|
return m_readings;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline bool BlockReadingBuilder::deleteReadingBeforeCursor()
|
inline bool BlockReadingBuilder::deleteReadingBeforeCursor() {
|
||||||
{
|
|
||||||
if (!m_cursorIndex) {
|
if (!m_cursorIndex) {
|
||||||
return false;
|
return false;
|
||||||
}
|
}
|
||||||
|
|
||||||
m_readings.erase(m_readings.begin() + m_cursorIndex - 1, m_readings.begin() + m_cursorIndex);
|
m_readings.erase(m_readings.begin() + m_cursorIndex - 1,
|
||||||
|
m_readings.begin() + m_cursorIndex);
|
||||||
m_cursorIndex--;
|
m_cursorIndex--;
|
||||||
m_grid.shrinkGridByOneAtLocation(m_cursorIndex);
|
m_grid.shrinkGridByOneAtLocation(m_cursorIndex);
|
||||||
build();
|
build();
|
||||||
return true;
|
return true;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline bool BlockReadingBuilder::deleteReadingAfterCursor()
|
inline bool BlockReadingBuilder::deleteReadingAfterCursor() {
|
||||||
{
|
|
||||||
if (m_cursorIndex == m_readings.size()) {
|
if (m_cursorIndex == m_readings.size()) {
|
||||||
return false;
|
return false;
|
||||||
}
|
}
|
||||||
|
|
||||||
m_readings.erase(m_readings.begin() + m_cursorIndex, m_readings.begin() + m_cursorIndex + 1);
|
m_readings.erase(m_readings.begin() + m_cursorIndex,
|
||||||
|
m_readings.begin() + m_cursorIndex + 1);
|
||||||
m_grid.shrinkGridByOneAtLocation(m_cursorIndex);
|
m_grid.shrinkGridByOneAtLocation(m_cursorIndex);
|
||||||
build();
|
build();
|
||||||
return true;
|
return true;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline bool BlockReadingBuilder::removeHeadReadings(size_t count)
|
inline bool BlockReadingBuilder::removeHeadReadings(size_t count) {
|
||||||
{
|
|
||||||
if (count > length()) {
|
if (count > length()) {
|
||||||
return false;
|
return false;
|
||||||
}
|
}
|
||||||
|
@ -150,23 +140,18 @@ namespace Taiyan {
|
||||||
return true;
|
return true;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline void BlockReadingBuilder::setJoinSeparator(const string& separator)
|
inline void BlockReadingBuilder::setJoinSeparator(
|
||||||
{
|
const std::string& separator) {
|
||||||
m_joinSeparator = separator;
|
m_joinSeparator = separator;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline const string BlockReadingBuilder::joinSeparator() const
|
inline const std::string BlockReadingBuilder::joinSeparator() const {
|
||||||
{
|
|
||||||
return m_joinSeparator;
|
return m_joinSeparator;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline Grid& BlockReadingBuilder::grid()
|
inline Grid& BlockReadingBuilder::grid() { return m_grid; }
|
||||||
{
|
|
||||||
return m_grid;
|
|
||||||
}
|
|
||||||
|
|
||||||
inline void BlockReadingBuilder::build()
|
inline void BlockReadingBuilder::build() {
|
||||||
{
|
|
||||||
if (!m_LM) {
|
if (!m_LM) {
|
||||||
return;
|
return;
|
||||||
}
|
}
|
||||||
|
@ -176,8 +161,7 @@ namespace Taiyan {
|
||||||
|
|
||||||
if (m_cursorIndex < MaximumBuildSpanLength) {
|
if (m_cursorIndex < MaximumBuildSpanLength) {
|
||||||
begin = 0;
|
begin = 0;
|
||||||
}
|
} else {
|
||||||
else {
|
|
||||||
begin = m_cursorIndex - MaximumBuildSpanLength;
|
begin = m_cursorIndex - MaximumBuildSpanLength;
|
||||||
}
|
}
|
||||||
|
|
||||||
|
@ -187,12 +171,14 @@ namespace Taiyan {
|
||||||
|
|
||||||
for (size_t p = begin; p < end; p++) {
|
for (size_t p = begin; p < end; p++) {
|
||||||
for (size_t q = 1; q <= MaximumBuildSpanLength && p + q <= end; q++) {
|
for (size_t q = 1; q <= MaximumBuildSpanLength && p + q <= end; q++) {
|
||||||
string combinedReading = Join(m_readings.begin() + p, m_readings.begin() + p + q, m_joinSeparator);
|
std::string combinedReading = Join(
|
||||||
if (!m_grid.hasNodeAtLocationSpanningLengthMatchingKey(p, q, combinedReading)) {
|
m_readings.begin() + p, m_readings.begin() + p + q, m_joinSeparator);
|
||||||
vector<Unigram> unigrams = m_LM->unigramsForKey(combinedReading);
|
if (!m_grid.hasNodeAtLocationSpanningLengthMatchingKey(p, q,
|
||||||
|
combinedReading)) {
|
||||||
|
std::vector<Unigram> unigrams = m_LM->unigramsForKey(combinedReading);
|
||||||
|
|
||||||
if (unigrams.size() > 0) {
|
if (unigrams.size() > 0) {
|
||||||
Node n(combinedReading, unigrams, vector<Bigram>());
|
Node n(combinedReading, unigrams, std::vector<Bigram>());
|
||||||
m_grid.insertNode(n, p, q);
|
m_grid.insertNode(n, p, q);
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
@ -200,10 +186,12 @@ namespace Taiyan {
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
inline const string BlockReadingBuilder::Join(vector<string>::const_iterator begin, vector<string>::const_iterator end, const string& separator)
|
inline const std::string BlockReadingBuilder::Join(
|
||||||
{
|
std::vector<std::string>::const_iterator begin,
|
||||||
string result;
|
std::vector<std::string>::const_iterator end,
|
||||||
for (vector<string>::const_iterator iter = begin ; iter != end ; ) {
|
const std::string& separator) {
|
||||||
|
std::string result;
|
||||||
|
for (std::vector<std::string>::const_iterator iter = begin; iter != end;) {
|
||||||
result += *iter;
|
result += *iter;
|
||||||
++iter;
|
++iter;
|
||||||
if (iter != end) {
|
if (iter != end) {
|
||||||
|
@ -212,7 +200,7 @@ namespace Taiyan {
|
||||||
}
|
}
|
||||||
return result;
|
return result;
|
||||||
}
|
}
|
||||||
}
|
} // namespace Gramambular
|
||||||
}
|
} // namespace Taiyan
|
||||||
|
|
||||||
#endif
|
#endif
|
||||||
|
|
|
@ -17,8 +17,8 @@ THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABI
|
||||||
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
||||||
*/
|
*/
|
||||||
|
|
||||||
#ifndef Gramambular_h
|
#ifndef GRAMAMBULAR_H_
|
||||||
#define Gramambular_h
|
#define GRAMAMBULAR_H_
|
||||||
|
|
||||||
#include "Bigram.h"
|
#include "Bigram.h"
|
||||||
#include "BlockReadingBuilder.h"
|
#include "BlockReadingBuilder.h"
|
||||||
|
|
|
@ -17,10 +17,13 @@ THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABI
|
||||||
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
||||||
*/
|
*/
|
||||||
|
|
||||||
#ifndef Grid_h
|
#ifndef GRID_H_
|
||||||
#define Grid_h
|
#define GRID_H_
|
||||||
|
|
||||||
#include <map>
|
#include <map>
|
||||||
|
#include <string>
|
||||||
|
#include <vector>
|
||||||
|
|
||||||
#include "NodeAnchor.h"
|
#include "NodeAnchor.h"
|
||||||
#include "Span.h"
|
#include "Span.h"
|
||||||
|
|
||||||
|
@ -30,112 +33,108 @@ namespace Taiyan {
|
||||||
class Grid {
|
class Grid {
|
||||||
public:
|
public:
|
||||||
void clear();
|
void clear();
|
||||||
void insertNode(const Node& inNode, size_t inLocation, size_t inSpanningLength);
|
void insertNode(const Node& node, size_t location, size_t spanningLength);
|
||||||
bool hasNodeAtLocationSpanningLengthMatchingKey(size_t inLocation, size_t inSpanningLength, const string& inKey);
|
bool hasNodeAtLocationSpanningLengthMatchingKey(size_t location,
|
||||||
|
size_t spanningLength,
|
||||||
|
const std::string& key);
|
||||||
|
|
||||||
void expandGridByOneAtLocation(size_t inLocation);
|
void expandGridByOneAtLocation(size_t location);
|
||||||
void shrinkGridByOneAtLocation(size_t inLocation);
|
void shrinkGridByOneAtLocation(size_t location);
|
||||||
|
|
||||||
size_t width() const;
|
size_t width() const;
|
||||||
vector<NodeAnchor> nodesEndingAt(size_t inLocation);
|
std::vector<NodeAnchor> nodesEndingAt(size_t location);
|
||||||
vector<NodeAnchor> nodesCrossingOrEndingAt(size_t inLocation);
|
std::vector<NodeAnchor> nodesCrossingOrEndingAt(size_t location);
|
||||||
|
|
||||||
// "Freeze" the node with the unigram that represents the selected candidate value.
|
// "Freeze" the node with the unigram that represents the selected candidate
|
||||||
// After this, the node that contains the unigram will always be evaluated to that
|
// value. After this, the node that contains the unigram will always be
|
||||||
// unigram, while all other overlapping nodes will be reset to their initial state
|
// evaluated to that unigram, while all other overlapping nodes will be reset
|
||||||
// (that is, if any of those nodes were "frozen" or fixed, they will be unfrozen.)
|
// to their initial state (that is, if any of those nodes were "frozen" or
|
||||||
NodeAnchor fixNodeSelectedCandidate(size_t location, const string& value);
|
// fixed, they will be unfrozen.)
|
||||||
|
NodeAnchor fixNodeSelectedCandidate(size_t location,
|
||||||
|
const std::string& value);
|
||||||
|
|
||||||
// Similar to fixNodeSelectedCandidate, but instead of "freezing" the node, only
|
// Similar to fixNodeSelectedCandidate, but instead of "freezing" the node,
|
||||||
// boost the unigram that represents the value with an overriding score. This
|
// only boost the unigram that represents the value with an overriding score.
|
||||||
// has the same side effect as fixNodeSelectedCandidate, which is that all other
|
// This has the same side effect as fixNodeSelectedCandidate, which is that
|
||||||
// overlapping nodes will be reset to their initial state.
|
// all other overlapping nodes will be reset to their initial state.
|
||||||
void overrideNodeScoreForSelectedCandidate(size_t location, const string& value, float overridingScore);
|
void overrideNodeScoreForSelectedCandidate(size_t location,
|
||||||
|
const std::string& value,
|
||||||
|
float overridingScore);
|
||||||
|
|
||||||
const string dumpDOT();
|
std::string dumpDOT();
|
||||||
|
|
||||||
protected:
|
protected:
|
||||||
vector<Span> m_spans;
|
std::vector<Span> m_spans;
|
||||||
};
|
};
|
||||||
|
|
||||||
inline void Grid::clear()
|
inline void Grid::clear() { m_spans.clear(); }
|
||||||
{
|
|
||||||
m_spans.clear();
|
|
||||||
}
|
|
||||||
|
|
||||||
inline void Grid::insertNode(const Node& inNode, size_t inLocation, size_t inSpanningLength)
|
inline void Grid::insertNode(const Node& node, size_t location,
|
||||||
{
|
size_t spanningLength) {
|
||||||
if (inLocation >= m_spans.size()) {
|
if (location >= m_spans.size()) {
|
||||||
size_t diff = inLocation - m_spans.size() + 1;
|
size_t diff = location - m_spans.size() + 1;
|
||||||
|
|
||||||
for (size_t i = 0; i < diff; i++) {
|
for (size_t i = 0; i < diff; i++) {
|
||||||
m_spans.push_back(Span());
|
m_spans.push_back(Span());
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
m_spans[inLocation].insertNodeOfLength(inNode, inSpanningLength);
|
m_spans[location].insertNodeOfLength(node, spanningLength);
|
||||||
}
|
}
|
||||||
|
|
||||||
inline bool Grid::hasNodeAtLocationSpanningLengthMatchingKey(size_t inLocation, size_t inSpanningLength, const string& inKey)
|
inline bool Grid::hasNodeAtLocationSpanningLengthMatchingKey(
|
||||||
{
|
size_t location, size_t spanningLength, const std::string& key) {
|
||||||
if (inLocation > m_spans.size()) {
|
if (location > m_spans.size()) {
|
||||||
return false;
|
return false;
|
||||||
}
|
}
|
||||||
|
|
||||||
const Node *n = m_spans[inLocation].nodeOfLength(inSpanningLength);
|
const Node* n = m_spans[location].nodeOfLength(spanningLength);
|
||||||
if (!n) {
|
if (!n) {
|
||||||
return false;
|
return false;
|
||||||
}
|
}
|
||||||
|
|
||||||
return inKey == n->key();
|
return key == n->key();
|
||||||
}
|
}
|
||||||
|
|
||||||
inline void Grid::expandGridByOneAtLocation(size_t inLocation)
|
inline void Grid::expandGridByOneAtLocation(size_t location) {
|
||||||
{
|
if (!location || location == m_spans.size()) {
|
||||||
if (!inLocation || inLocation == m_spans.size()) {
|
m_spans.insert(m_spans.begin() + location, Span());
|
||||||
m_spans.insert(m_spans.begin() + inLocation, Span());
|
} else {
|
||||||
}
|
m_spans.insert(m_spans.begin() + location, Span());
|
||||||
else {
|
for (size_t i = 0; i < location; i++) {
|
||||||
m_spans.insert(m_spans.begin() + inLocation, Span());
|
|
||||||
for (size_t i = 0 ; i < inLocation ; i++) {
|
|
||||||
// zaps overlapping spans
|
// zaps overlapping spans
|
||||||
m_spans[i].removeNodeOfLengthGreaterThan(inLocation - i);
|
m_spans[i].removeNodeOfLengthGreaterThan(location - i);
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
inline void Grid::shrinkGridByOneAtLocation(size_t inLocation)
|
inline void Grid::shrinkGridByOneAtLocation(size_t location) {
|
||||||
{
|
if (location >= m_spans.size()) {
|
||||||
if (inLocation >= m_spans.size()) {
|
|
||||||
return;
|
return;
|
||||||
}
|
}
|
||||||
|
|
||||||
m_spans.erase(m_spans.begin() + inLocation);
|
m_spans.erase(m_spans.begin() + location);
|
||||||
for (size_t i = 0 ; i < inLocation ; i++) {
|
for (size_t i = 0; i < location; i++) {
|
||||||
// zaps overlapping spans
|
// zaps overlapping spans
|
||||||
m_spans[i].removeNodeOfLengthGreaterThan(inLocation - i);
|
m_spans[i].removeNodeOfLengthGreaterThan(location - i);
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
inline size_t Grid::width() const
|
inline size_t Grid::width() const { return m_spans.size(); }
|
||||||
{
|
|
||||||
return m_spans.size();
|
|
||||||
}
|
|
||||||
|
|
||||||
inline vector<NodeAnchor> Grid::nodesEndingAt(size_t inLocation)
|
inline std::vector<NodeAnchor> Grid::nodesEndingAt(size_t location) {
|
||||||
{
|
std::vector<NodeAnchor> result;
|
||||||
vector<NodeAnchor> result;
|
|
||||||
|
|
||||||
if (m_spans.size() && inLocation <= m_spans.size()) {
|
if (m_spans.size() && location <= m_spans.size()) {
|
||||||
for (size_t i = 0 ; i < inLocation ; i++) {
|
for (size_t i = 0; i < location; i++) {
|
||||||
Span& span = m_spans[i];
|
Span& span = m_spans[i];
|
||||||
if (i + span.maximumLength() >= inLocation) {
|
if (i + span.maximumLength() >= location) {
|
||||||
Node *np = span.nodeOfLength(inLocation - i);
|
Node* np = span.nodeOfLength(location - i);
|
||||||
if (np) {
|
if (np) {
|
||||||
NodeAnchor na;
|
NodeAnchor na;
|
||||||
na.node = np;
|
na.node = np;
|
||||||
na.location = i;
|
na.location = i;
|
||||||
na.spanningLength = inLocation - i;
|
na.spanningLength = location - i;
|
||||||
|
|
||||||
result.push_back(na);
|
result.push_back(na);
|
||||||
}
|
}
|
||||||
|
@ -146,19 +145,16 @@ namespace Taiyan {
|
||||||
return result;
|
return result;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline vector<NodeAnchor> Grid::nodesCrossingOrEndingAt(size_t inLocation)
|
inline std::vector<NodeAnchor> Grid::nodesCrossingOrEndingAt(size_t location) {
|
||||||
{
|
std::vector<NodeAnchor> result;
|
||||||
vector<NodeAnchor> result;
|
|
||||||
|
|
||||||
if (m_spans.size() && inLocation <= m_spans.size()) {
|
if (m_spans.size() && location <= m_spans.size()) {
|
||||||
for (size_t i = 0 ; i < inLocation ; i++) {
|
for (size_t i = 0; i < location; i++) {
|
||||||
Span& span = m_spans[i];
|
Span& span = m_spans[i];
|
||||||
|
|
||||||
if (i + span.maximumLength() >= inLocation) {
|
if (i + span.maximumLength() >= location) {
|
||||||
|
|
||||||
for (size_t j = 1, m = span.maximumLength(); j <= m; j++) {
|
for (size_t j = 1, m = span.maximumLength(); j <= m; j++) {
|
||||||
|
if (i + j < location) {
|
||||||
if (i + j < inLocation) {
|
|
||||||
continue;
|
continue;
|
||||||
}
|
}
|
||||||
|
|
||||||
|
@ -167,7 +163,7 @@ namespace Taiyan {
|
||||||
NodeAnchor na;
|
NodeAnchor na;
|
||||||
na.node = np;
|
na.node = np;
|
||||||
na.location = i;
|
na.location = i;
|
||||||
na.spanningLength = inLocation - i;
|
na.spanningLength = location - i;
|
||||||
|
|
||||||
result.push_back(na);
|
result.push_back(na);
|
||||||
}
|
}
|
||||||
|
@ -179,10 +175,11 @@ namespace Taiyan {
|
||||||
return result;
|
return result;
|
||||||
}
|
}
|
||||||
|
|
||||||
// For nodes found at the location, fix their currently-selected candidate using the supplied string value.
|
// For nodes found at the location, fix their currently-selected candidate using
|
||||||
inline NodeAnchor Grid::fixNodeSelectedCandidate(size_t location, const string& value)
|
// the supplied string value.
|
||||||
{
|
inline NodeAnchor Grid::fixNodeSelectedCandidate(size_t location,
|
||||||
vector<NodeAnchor> nodes = nodesCrossingOrEndingAt(location);
|
const std::string& value) {
|
||||||
|
std::vector<NodeAnchor> nodes = nodesCrossingOrEndingAt(location);
|
||||||
NodeAnchor node;
|
NodeAnchor node;
|
||||||
for (auto nodeAnchor : nodes) {
|
for (auto nodeAnchor : nodes) {
|
||||||
auto candidates = nodeAnchor.node->candidates();
|
auto candidates = nodeAnchor.node->candidates();
|
||||||
|
@ -194,16 +191,16 @@ namespace Taiyan {
|
||||||
if (candidates[i].value == value) {
|
if (candidates[i].value == value) {
|
||||||
const_cast<Node*>(nodeAnchor.node)->selectCandidateAtIndex(i);
|
const_cast<Node*>(nodeAnchor.node)->selectCandidateAtIndex(i);
|
||||||
node = nodeAnchor;
|
node = nodeAnchor;
|
||||||
break;;
|
break;
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
return node;
|
return node;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline void Grid::overrideNodeScoreForSelectedCandidate(size_t location, const string& value, float overridingScore)
|
inline void Grid::overrideNodeScoreForSelectedCandidate(
|
||||||
{
|
size_t location, const std::string& value, float overridingScore) {
|
||||||
vector<NodeAnchor> nodes = nodesCrossingOrEndingAt(location);
|
std::vector<NodeAnchor> nodes = nodesCrossingOrEndingAt(location);
|
||||||
for (auto nodeAnchor : nodes) {
|
for (auto nodeAnchor : nodes) {
|
||||||
auto candidates = nodeAnchor.node->candidates();
|
auto candidates = nodeAnchor.node->candidates();
|
||||||
|
|
||||||
|
@ -212,53 +209,15 @@ namespace Taiyan {
|
||||||
|
|
||||||
for (size_t i = 0, c = candidates.size(); i < c; ++i) {
|
for (size_t i = 0, c = candidates.size(); i < c; ++i) {
|
||||||
if (candidates[i].value == value) {
|
if (candidates[i].value == value) {
|
||||||
const_cast<Node*>(nodeAnchor.node)->selectFloatingCandidateAtIndex(i, overridingScore);
|
const_cast<Node*>(nodeAnchor.node)
|
||||||
|
->selectFloatingCandidateAtIndex(i, overridingScore);
|
||||||
break;
|
break;
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
inline const string Grid::dumpDOT()
|
} // namespace Gramambular
|
||||||
{
|
} // namespace Taiyan
|
||||||
stringstream sst;
|
|
||||||
sst << "digraph {" << endl;
|
|
||||||
sst << "graph [ rankdir=LR ];" << endl;
|
|
||||||
sst << "BOS;" << endl;
|
|
||||||
|
|
||||||
for (size_t p = 0 ; p < m_spans.size() ; p++) {
|
|
||||||
Span& span = m_spans[p];
|
|
||||||
for (size_t ni = 0 ; ni <= span.maximumLength() ; ni++) {
|
|
||||||
Node* np = span.nodeOfLength(ni);
|
|
||||||
if (np) {
|
|
||||||
if (!p) {
|
|
||||||
sst << "BOS -> " << np->currentKeyValue().value << ";" << endl;
|
|
||||||
}
|
|
||||||
|
|
||||||
sst << np->currentKeyValue().value << ";" << endl;
|
|
||||||
|
|
||||||
if (p + ni < m_spans.size()) {
|
|
||||||
Span& dstSpan = m_spans[p+ni];
|
|
||||||
for (size_t q = 0 ; q <= dstSpan.maximumLength() ; q++) {
|
|
||||||
Node *dn = dstSpan.nodeOfLength(q);
|
|
||||||
if (dn) {
|
|
||||||
sst << np->currentKeyValue().value << " -> " << dn->currentKeyValue().value << ";" << endl;
|
|
||||||
}
|
|
||||||
}
|
|
||||||
}
|
|
||||||
|
|
||||||
if (p + ni == m_spans.size()) {
|
|
||||||
sst << np->currentKeyValue().value << " -> " << "EOS;" << endl;
|
|
||||||
}
|
|
||||||
}
|
|
||||||
}
|
|
||||||
}
|
|
||||||
|
|
||||||
sst << "EOS;" << endl;
|
|
||||||
sst << "}";
|
|
||||||
return sst.str();
|
|
||||||
}
|
|
||||||
}
|
|
||||||
}
|
|
||||||
|
|
||||||
#endif
|
#endif
|
||||||
|
|
|
@ -0,0 +1,70 @@
|
||||||
|
// Copyright (c) 2011 and onwards The OpenVanilla Project (MIT License).
|
||||||
|
// All possible vChewing-specific modifications are (c) 2021 and onwards The vChewing Project (MIT-NTL License).
|
||||||
|
/*
|
||||||
|
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated
|
||||||
|
documentation files (the "Software"), to deal in the Software without restriction, including without limitation
|
||||||
|
the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and
|
||||||
|
to permit persons to whom the Software is furnished to do so, subject to the following conditions:
|
||||||
|
|
||||||
|
1. The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
|
||||||
|
|
||||||
|
2. No trademark license is granted to use the trade names, trademarks, service marks, or product names of Contributor,
|
||||||
|
except as required to fulfill notice requirements above.
|
||||||
|
|
||||||
|
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED
|
||||||
|
TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
|
||||||
|
THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
|
||||||
|
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
||||||
|
*/
|
||||||
|
|
||||||
|
#include "Grid.h"
|
||||||
|
|
||||||
|
#include <iostream>
|
||||||
|
#include <string>
|
||||||
|
|
||||||
|
namespace Taiyan {
|
||||||
|
namespace Gramambular {
|
||||||
|
|
||||||
|
std::string Grid::dumpDOT() {
|
||||||
|
std::stringstream sst;
|
||||||
|
sst << "digraph {" << std::endl;
|
||||||
|
sst << "graph [ rankdir=LR ];" << std::endl;
|
||||||
|
sst << "BOS;" << std::endl;
|
||||||
|
|
||||||
|
for (size_t p = 0; p < m_spans.size(); p++) {
|
||||||
|
Span& span = m_spans[p];
|
||||||
|
for (size_t ni = 0; ni <= span.maximumLength(); ni++) {
|
||||||
|
Node* np = span.nodeOfLength(ni);
|
||||||
|
if (np) {
|
||||||
|
if (!p) {
|
||||||
|
sst << "BOS -> " << np->currentKeyValue().value << ";" << std::endl;
|
||||||
|
}
|
||||||
|
|
||||||
|
sst << np->currentKeyValue().value << ";" << std::endl;
|
||||||
|
|
||||||
|
if (p + ni < m_spans.size()) {
|
||||||
|
Span& dstSpan = m_spans[p + ni];
|
||||||
|
for (size_t q = 0; q <= dstSpan.maximumLength(); q++) {
|
||||||
|
Node* dn = dstSpan.nodeOfLength(q);
|
||||||
|
if (dn) {
|
||||||
|
sst << np->currentKeyValue().value << " -> "
|
||||||
|
<< dn->currentKeyValue().value << ";" << std::endl;
|
||||||
|
}
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
if (p + ni == m_spans.size()) {
|
||||||
|
sst << np->currentKeyValue().value << " -> "
|
||||||
|
<< "EOS;" << std::endl;
|
||||||
|
}
|
||||||
|
}
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
sst << "EOS;" << std::endl;
|
||||||
|
sst << "}";
|
||||||
|
return sst.str();
|
||||||
|
}
|
||||||
|
|
||||||
|
} // namespace Gramambular
|
||||||
|
} // namespace Taiyan
|
|
@ -17,47 +17,43 @@ THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABI
|
||||||
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
||||||
*/
|
*/
|
||||||
|
|
||||||
#ifndef KeyValuePair_h
|
#ifndef KEYVALUEPAIR_H_
|
||||||
#define KeyValuePair_h
|
#define KEYVALUEPAIR_H_
|
||||||
|
|
||||||
#include <ostream>
|
#include <ostream>
|
||||||
#include <string>
|
#include <string>
|
||||||
|
|
||||||
namespace Taiyan {
|
namespace Taiyan {
|
||||||
namespace Gramambular {
|
namespace Gramambular {
|
||||||
using namespace std;
|
|
||||||
|
|
||||||
class KeyValuePair {
|
class KeyValuePair {
|
||||||
public:
|
public:
|
||||||
string key;
|
std::string key;
|
||||||
string value;
|
std::string value;
|
||||||
|
|
||||||
bool operator==(const KeyValuePair& inAnother) const;
|
bool operator==(const KeyValuePair& another) const;
|
||||||
bool operator<(const KeyValuePair& inAnother) const;
|
bool operator<(const KeyValuePair& another) const;
|
||||||
};
|
};
|
||||||
|
|
||||||
inline ostream& operator<<(ostream& inStream, const KeyValuePair& inPair)
|
inline std::ostream& operator<<(std::ostream& stream,
|
||||||
{
|
const KeyValuePair& pair) {
|
||||||
inStream << "(" << inPair.key << "," << inPair.value << ")";
|
stream << "(" << pair.key << "," << pair.value << ")";
|
||||||
return inStream;
|
return stream;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline bool KeyValuePair::operator==(const KeyValuePair& inAnother) const
|
inline bool KeyValuePair::operator==(const KeyValuePair& another) const {
|
||||||
{
|
return key == another.key && value == another.value;
|
||||||
return key == inAnother.key && value == inAnother.value;
|
|
||||||
}
|
}
|
||||||
|
|
||||||
inline bool KeyValuePair::operator<(const KeyValuePair& inAnother) const
|
inline bool KeyValuePair::operator<(const KeyValuePair& another) const {
|
||||||
{
|
if (key < another.key) {
|
||||||
if (key < inAnother.key) {
|
|
||||||
return true;
|
return true;
|
||||||
}
|
} else if (key == another.key) {
|
||||||
else if (key == inAnother.key) {
|
return value < another.value;
|
||||||
return value < inAnother.value;
|
|
||||||
}
|
}
|
||||||
return false;
|
return false;
|
||||||
}
|
}
|
||||||
}
|
} // namespace Gramambular
|
||||||
}
|
} // namespace Taiyan
|
||||||
|
|
||||||
#endif
|
#endif
|
||||||
|
|
|
@ -17,28 +17,28 @@ THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABI
|
||||||
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
||||||
*/
|
*/
|
||||||
|
|
||||||
#ifndef LanguageModel_h
|
#ifndef LANGUAGEMODEL_H_
|
||||||
#define LanguageModel_h
|
#define LANGUAGEMODEL_H_
|
||||||
|
|
||||||
|
#include <string>
|
||||||
#include <vector>
|
#include <vector>
|
||||||
|
|
||||||
#include "Bigram.h"
|
#include "Bigram.h"
|
||||||
#include "Unigram.h"
|
#include "Unigram.h"
|
||||||
|
|
||||||
namespace Taiyan {
|
namespace Taiyan {
|
||||||
namespace Gramambular {
|
namespace Gramambular {
|
||||||
|
|
||||||
using namespace std;
|
|
||||||
|
|
||||||
class LanguageModel {
|
class LanguageModel {
|
||||||
public:
|
public:
|
||||||
virtual ~LanguageModel() {}
|
virtual ~LanguageModel() {}
|
||||||
|
|
||||||
virtual const vector<Bigram> bigramsForKeys(const string &preceedingKey, const string& key) = 0;
|
virtual const std::vector<Bigram> bigramsForKeys(
|
||||||
virtual const vector<Unigram> unigramsForKey(const string &key) = 0;
|
const std::string& preceedingKey, const std::string& key) = 0;
|
||||||
virtual bool hasUnigramsForKey(const string& key) = 0;
|
virtual const std::vector<Unigram> unigramsForKey(const std::string& key) = 0;
|
||||||
|
virtual bool hasUnigramsForKey(const std::string& key) = 0;
|
||||||
};
|
};
|
||||||
}
|
} // namespace Gramambular
|
||||||
}
|
} // namespace Taiyan
|
||||||
|
|
||||||
|
|
||||||
#endif
|
#endif
|
||||||
|
|
|
@ -17,75 +17,75 @@ THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABI
|
||||||
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
||||||
*/
|
*/
|
||||||
|
|
||||||
#ifndef Node_h
|
#ifndef NODE_H_
|
||||||
#define Node_h
|
#define NODE_H_
|
||||||
|
|
||||||
#include <limits>
|
#include <limits>
|
||||||
|
#include <map>
|
||||||
|
#include <string>
|
||||||
#include <vector>
|
#include <vector>
|
||||||
|
|
||||||
#include "LanguageModel.h"
|
#include "LanguageModel.h"
|
||||||
|
|
||||||
namespace Taiyan {
|
namespace Taiyan {
|
||||||
namespace Gramambular {
|
namespace Gramambular {
|
||||||
using namespace std;
|
|
||||||
|
|
||||||
class Node {
|
class Node {
|
||||||
public:
|
public:
|
||||||
Node();
|
Node();
|
||||||
Node(const string& inKey, const vector<Unigram>& inUnigrams, const vector<Bigram>& inBigrams);
|
Node(const std::string& key, const std::vector<Unigram>& unigrams,
|
||||||
|
const std::vector<Bigram>& bigrams);
|
||||||
|
|
||||||
void primeNodeWithPreceedingKeyValues(const vector<KeyValuePair>& inKeyValues);
|
void primeNodeWithPreceedingKeyValues(
|
||||||
|
const std::vector<KeyValuePair>& keyValues);
|
||||||
|
|
||||||
bool isCandidateFixed() const;
|
bool isCandidateFixed() const;
|
||||||
const vector<KeyValuePair>& candidates() const;
|
const std::vector<KeyValuePair>& candidates() const;
|
||||||
void selectCandidateAtIndex(size_t inIndex = 0, bool inFix = true);
|
void selectCandidateAtIndex(size_t index = 0, bool fix = true);
|
||||||
void resetCandidate();
|
void resetCandidate();
|
||||||
void selectFloatingCandidateAtIndex(size_t index, double score);
|
void selectFloatingCandidateAtIndex(size_t index, double score);
|
||||||
|
|
||||||
const string& key() const;
|
const std::string& key() const;
|
||||||
double score() const;
|
double score() const;
|
||||||
// double scoreForCandidate(string &candidate) const; // Prevents the override model to remember symbols with scode -X or lower.
|
double scoreForCandidate(const std::string& candidate) const;
|
||||||
const KeyValuePair currentKeyValue() const;
|
const KeyValuePair currentKeyValue() const;
|
||||||
double highestUnigramScore() const;
|
double highestUnigramScore() const;
|
||||||
|
|
||||||
protected:
|
protected:
|
||||||
const LanguageModel* m_LM;
|
const LanguageModel* m_LM;
|
||||||
|
|
||||||
string m_key;
|
std::string m_key;
|
||||||
double m_score;
|
double m_score;
|
||||||
|
|
||||||
vector<Unigram> m_unigrams;
|
std::vector<Unigram> m_unigrams;
|
||||||
vector<KeyValuePair> m_candidates;
|
std::vector<KeyValuePair> m_candidates;
|
||||||
map<string, size_t> m_valueUnigramIndexMap;
|
std::map<std::string, size_t> m_valueUnigramIndexMap;
|
||||||
map<KeyValuePair, vector<Bigram> > m_preceedingGramBigramMap;
|
std::map<KeyValuePair, std::vector<Bigram> > m_preceedingGramBigramMap;
|
||||||
|
|
||||||
bool m_candidateFixed;
|
bool m_candidateFixed;
|
||||||
size_t m_selectedUnigramIndex;
|
size_t m_selectedUnigramIndex;
|
||||||
|
|
||||||
friend ostream& operator<<(ostream& inStream, const Node& inNode);
|
friend std::ostream& operator<<(std::ostream& stream, const Node& node);
|
||||||
};
|
};
|
||||||
|
|
||||||
inline ostream& operator<<(ostream& inStream, const Node& inNode)
|
inline std::ostream& operator<<(std::ostream& stream, const Node& node) {
|
||||||
{
|
stream << "(node,key:" << node.m_key
|
||||||
inStream << "(node,key:" << inNode.m_key << ",fixed:" << (inNode.m_candidateFixed ? "true" : "false")
|
<< ",fixed:" << (node.m_candidateFixed ? "true" : "false")
|
||||||
<< ",selected:" << inNode.m_selectedUnigramIndex
|
<< ",selected:" << node.m_selectedUnigramIndex << ","
|
||||||
<< "," << inNode.m_unigrams << ")";
|
<< node.m_unigrams << ")";
|
||||||
return inStream;
|
return stream;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline Node::Node()
|
inline Node::Node()
|
||||||
: m_candidateFixed(false)
|
: m_candidateFixed(false), m_selectedUnigramIndex(0), m_score(0.0) {}
|
||||||
, m_selectedUnigramIndex(0)
|
|
||||||
, m_score(0.0)
|
|
||||||
{
|
|
||||||
}
|
|
||||||
|
|
||||||
inline Node::Node(const string& inKey, const vector<Unigram>& inUnigrams, const vector<Bigram>& inBigrams)
|
inline Node::Node(const std::string& key, const std::vector<Unigram>& unigrams,
|
||||||
: m_key(inKey)
|
const std::vector<Bigram>& bigrams)
|
||||||
, m_unigrams(inUnigrams)
|
: m_key(key),
|
||||||
, m_candidateFixed(false)
|
m_unigrams(unigrams),
|
||||||
, m_selectedUnigramIndex(0)
|
m_candidateFixed(false),
|
||||||
, m_score(0.0)
|
m_selectedUnigramIndex(0),
|
||||||
{
|
m_score(0.0) {
|
||||||
stable_sort(m_unigrams.begin(), m_unigrams.end(), Unigram::ScoreCompare);
|
stable_sort(m_unigrams.begin(), m_unigrams.end(), Unigram::ScoreCompare);
|
||||||
|
|
||||||
if (m_unigrams.size()) {
|
if (m_unigrams.size()) {
|
||||||
|
@ -93,33 +93,39 @@ namespace Taiyan {
|
||||||
}
|
}
|
||||||
|
|
||||||
size_t i = 0;
|
size_t i = 0;
|
||||||
for (vector<Unigram>::const_iterator ui = m_unigrams.begin() ; ui != m_unigrams.end() ; ++ui) {
|
for (std::vector<Unigram>::const_iterator ui = m_unigrams.begin();
|
||||||
|
ui != m_unigrams.end(); ++ui) {
|
||||||
m_valueUnigramIndexMap[(*ui).keyValue.value] = i;
|
m_valueUnigramIndexMap[(*ui).keyValue.value] = i;
|
||||||
i++;
|
i++;
|
||||||
|
|
||||||
m_candidates.push_back((*ui).keyValue);
|
m_candidates.push_back((*ui).keyValue);
|
||||||
}
|
}
|
||||||
|
|
||||||
for (vector<Bigram>::const_iterator bi = inBigrams.begin() ; bi != inBigrams.end() ; ++bi) {
|
for (std::vector<Bigram>::const_iterator bi = bigrams.begin();
|
||||||
|
bi != bigrams.end(); ++bi) {
|
||||||
m_preceedingGramBigramMap[(*bi).preceedingKeyValue].push_back(*bi);
|
m_preceedingGramBigramMap[(*bi).preceedingKeyValue].push_back(*bi);
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
inline void Node::primeNodeWithPreceedingKeyValues(const vector<KeyValuePair>& inKeyValues)
|
inline void Node::primeNodeWithPreceedingKeyValues(
|
||||||
{
|
const std::vector<KeyValuePair>& keyValues) {
|
||||||
size_t newIndex = m_selectedUnigramIndex;
|
size_t newIndex = m_selectedUnigramIndex;
|
||||||
double max = m_score;
|
double max = m_score;
|
||||||
|
|
||||||
if (!isCandidateFixed()) {
|
if (!isCandidateFixed()) {
|
||||||
for (vector<KeyValuePair>::const_iterator kvi = inKeyValues.begin() ; kvi != inKeyValues.end() ; ++kvi) {
|
for (std::vector<KeyValuePair>::const_iterator kvi = keyValues.begin();
|
||||||
map<KeyValuePair, vector<Bigram> >::const_iterator f = m_preceedingGramBigramMap.find(*kvi);
|
kvi != keyValues.end(); ++kvi) {
|
||||||
|
std::map<KeyValuePair, std::vector<Bigram> >::const_iterator f =
|
||||||
|
m_preceedingGramBigramMap.find(*kvi);
|
||||||
if (f != m_preceedingGramBigramMap.end()) {
|
if (f != m_preceedingGramBigramMap.end()) {
|
||||||
const vector<Bigram>& bigrams = (*f).second;
|
const std::vector<Bigram>& bigrams = (*f).second;
|
||||||
|
|
||||||
for (vector<Bigram>::const_iterator bi = bigrams.begin() ; bi != bigrams.end() ; ++bi) {
|
for (std::vector<Bigram>::const_iterator bi = bigrams.begin();
|
||||||
|
bi != bigrams.end(); ++bi) {
|
||||||
const Bigram& bigram = *bi;
|
const Bigram& bigram = *bi;
|
||||||
if (bigram.score > max) {
|
if (bigram.score > max) {
|
||||||
map<string, size_t>::const_iterator uf = m_valueUnigramIndexMap.find((*bi).keyValue.value);
|
std::map<std::string, size_t>::const_iterator uf =
|
||||||
|
m_valueUnigramIndexMap.find((*bi).keyValue.value);
|
||||||
if (uf != m_valueUnigramIndexMap.end()) {
|
if (uf != m_valueUnigramIndexMap.end()) {
|
||||||
newIndex = (*uf).second;
|
newIndex = (*uf).second;
|
||||||
max = bigram.score;
|
max = bigram.score;
|
||||||
|
@ -139,31 +145,24 @@ namespace Taiyan {
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
inline bool Node::isCandidateFixed() const
|
inline bool Node::isCandidateFixed() const { return m_candidateFixed; }
|
||||||
{
|
|
||||||
return m_candidateFixed;
|
|
||||||
}
|
|
||||||
|
|
||||||
inline const vector<KeyValuePair>& Node::candidates() const
|
inline const std::vector<KeyValuePair>& Node::candidates() const {
|
||||||
{
|
|
||||||
return m_candidates;
|
return m_candidates;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline void Node::selectCandidateAtIndex(size_t inIndex, bool inFix)
|
inline void Node::selectCandidateAtIndex(size_t index, bool fix) {
|
||||||
{
|
if (index >= m_unigrams.size()) {
|
||||||
if (inIndex >= m_unigrams.size()) {
|
|
||||||
m_selectedUnigramIndex = 0;
|
m_selectedUnigramIndex = 0;
|
||||||
}
|
} else {
|
||||||
else {
|
m_selectedUnigramIndex = index;
|
||||||
m_selectedUnigramIndex = inIndex;
|
|
||||||
}
|
}
|
||||||
|
|
||||||
m_candidateFixed = inFix;
|
m_candidateFixed = fix;
|
||||||
m_score = 99;
|
m_score = 99;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline void Node::resetCandidate()
|
inline void Node::resetCandidate() {
|
||||||
{
|
|
||||||
m_selectedUnigramIndex = 0;
|
m_selectedUnigramIndex = 0;
|
||||||
m_candidateFixed = 0;
|
m_candidateFixed = 0;
|
||||||
if (m_unigrams.size()) {
|
if (m_unigrams.size()) {
|
||||||
|
@ -181,19 +180,12 @@ namespace Taiyan {
|
||||||
m_score = score;
|
m_score = score;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline const string& Node::key() const
|
inline const std::string& Node::key() const { return m_key; }
|
||||||
{
|
|
||||||
return m_key;
|
|
||||||
}
|
|
||||||
|
|
||||||
inline double Node::score() const
|
inline double Node::score() const { return m_score; }
|
||||||
{
|
|
||||||
return m_score;
|
|
||||||
}
|
|
||||||
|
|
||||||
// Prevents the override model to remember symbols with scode -X or lower.
|
// Prevents the override model to remember symbols with scode -X or lower.
|
||||||
// inline double Node::scoreForCandidate(string &candidate) const
|
//inline double Node::scoreForCandidate(const std::string& candidate) const {
|
||||||
// {
|
|
||||||
// for (auto unigram : m_unigrams) {
|
// for (auto unigram : m_unigrams) {
|
||||||
// if (unigram.keyValue.value == candidate) {
|
// if (unigram.keyValue.value == candidate) {
|
||||||
// return unigram.score;
|
// return unigram.score;
|
||||||
|
@ -209,16 +201,14 @@ namespace Taiyan {
|
||||||
return m_unigrams[0].score;
|
return m_unigrams[0].score;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline const KeyValuePair Node::currentKeyValue() const
|
inline const KeyValuePair Node::currentKeyValue() const {
|
||||||
{
|
|
||||||
if (m_selectedUnigramIndex >= m_unigrams.size()) {
|
if (m_selectedUnigramIndex >= m_unigrams.size()) {
|
||||||
return KeyValuePair();
|
return KeyValuePair();
|
||||||
}
|
} else {
|
||||||
else {
|
|
||||||
return m_candidates[m_selectedUnigramIndex];
|
return m_candidates[m_selectedUnigramIndex];
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
} // namespace Gramambular
|
||||||
}
|
} // namespace Taiyan
|
||||||
|
|
||||||
#endif
|
#endif
|
||||||
|
|
|
@ -17,55 +17,48 @@ THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABI
|
||||||
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
||||||
*/
|
*/
|
||||||
|
|
||||||
#ifndef NodeAnchor_h
|
#ifndef NODEANCHOR_H_
|
||||||
#define NodeAnchor_h
|
#define NODEANCHOR_H_
|
||||||
|
|
||||||
|
#include <vector>
|
||||||
|
|
||||||
#include "Node.h"
|
#include "Node.h"
|
||||||
|
|
||||||
namespace Taiyan {
|
namespace Taiyan {
|
||||||
namespace Gramambular {
|
namespace Gramambular {
|
||||||
class NodeAnchor {
|
|
||||||
public:
|
struct NodeAnchor {
|
||||||
NodeAnchor();
|
const Node* node = nullptr;
|
||||||
const Node *node;
|
size_t location = 0;
|
||||||
size_t location;
|
size_t spanningLength = 0;
|
||||||
size_t spanningLength;
|
double accumulatedScore = 0.0;
|
||||||
double accumulatedScore;
|
|
||||||
};
|
};
|
||||||
|
|
||||||
inline NodeAnchor::NodeAnchor()
|
inline std::ostream& operator<<(std::ostream& stream,
|
||||||
: node(0)
|
const NodeAnchor& anchor) {
|
||||||
, location(0)
|
stream << "{@(" << anchor.location << "," << anchor.spanningLength << "),";
|
||||||
, spanningLength(0)
|
if (anchor.node) {
|
||||||
, accumulatedScore(0.0)
|
stream << *(anchor.node);
|
||||||
{
|
} else {
|
||||||
|
stream << "null";
|
||||||
|
}
|
||||||
|
stream << "}";
|
||||||
|
return stream;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline ostream& operator<<(ostream& inStream, const NodeAnchor& inAnchor)
|
inline std::ostream& operator<<(std::ostream& stream,
|
||||||
{
|
const std::vector<NodeAnchor>& anchor) {
|
||||||
inStream << "{@(" << inAnchor.location << "," << inAnchor.spanningLength << "),";
|
for (std::vector<NodeAnchor>::const_iterator i = anchor.begin();
|
||||||
if (inAnchor.node) {
|
i != anchor.end(); ++i) {
|
||||||
inStream << *(inAnchor.node);
|
stream << *i;
|
||||||
}
|
if (i + 1 != anchor.end()) {
|
||||||
else {
|
stream << "<-";
|
||||||
inStream << "null";
|
|
||||||
}
|
|
||||||
inStream << "}";
|
|
||||||
return inStream;
|
|
||||||
}
|
|
||||||
|
|
||||||
inline ostream& operator<<(ostream& inStream, const vector<NodeAnchor>& inAnchor)
|
|
||||||
{
|
|
||||||
for (vector<NodeAnchor>::const_iterator i = inAnchor.begin() ; i != inAnchor.end() ; ++i) {
|
|
||||||
inStream << *i;
|
|
||||||
if (i + 1 != inAnchor.end()) {
|
|
||||||
inStream << "<-";
|
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
return inStream;
|
return stream;
|
||||||
}
|
|
||||||
}
|
|
||||||
}
|
}
|
||||||
|
} // namespace Gramambular
|
||||||
|
} // namespace Taiyan
|
||||||
|
|
||||||
#endif
|
#endif
|
||||||
|
|
|
@ -17,88 +17,77 @@ THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABI
|
||||||
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
||||||
*/
|
*/
|
||||||
|
|
||||||
#ifndef Span_h
|
#ifndef SPAN_H_
|
||||||
#define Span_h
|
#define SPAN_H_
|
||||||
|
|
||||||
#include <map>
|
#include <map>
|
||||||
#include <set>
|
#include <set>
|
||||||
#include <sstream>
|
#include <sstream>
|
||||||
|
|
||||||
#include "Node.h"
|
#include "Node.h"
|
||||||
|
|
||||||
namespace Taiyan {
|
namespace Taiyan {
|
||||||
namespace Gramambular {
|
namespace Gramambular {
|
||||||
class Span {
|
class Span {
|
||||||
public:
|
public:
|
||||||
Span();
|
|
||||||
|
|
||||||
void clear();
|
void clear();
|
||||||
void insertNodeOfLength(const Node& inNode, size_t inLength);
|
void insertNodeOfLength(const Node& node, size_t length);
|
||||||
void removeNodeOfLengthGreaterThan(size_t inLength);
|
void removeNodeOfLengthGreaterThan(size_t length);
|
||||||
|
|
||||||
Node* nodeOfLength(size_t inLength);
|
Node* nodeOfLength(size_t length);
|
||||||
size_t maximumLength() const;
|
size_t maximumLength() const;
|
||||||
|
|
||||||
protected:
|
protected:
|
||||||
map<size_t, Node> m_lengthNodeMap;
|
std::map<size_t, Node> m_lengthNodeMap;
|
||||||
size_t m_maximumLength;
|
size_t m_maximumLength = 0;
|
||||||
};
|
};
|
||||||
|
|
||||||
inline Span::Span()
|
inline void Span::clear() {
|
||||||
: m_maximumLength(0)
|
|
||||||
{
|
|
||||||
}
|
|
||||||
|
|
||||||
inline void Span::clear()
|
|
||||||
{
|
|
||||||
m_lengthNodeMap.clear();
|
m_lengthNodeMap.clear();
|
||||||
m_maximumLength = 0;
|
m_maximumLength = 0;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline void Span::insertNodeOfLength(const Node& inNode, size_t inLength)
|
inline void Span::insertNodeOfLength(const Node& node, size_t length) {
|
||||||
{
|
m_lengthNodeMap[length] = node;
|
||||||
m_lengthNodeMap[inLength] = inNode;
|
if (length > m_maximumLength) {
|
||||||
if (inLength > m_maximumLength) {
|
m_maximumLength = length;
|
||||||
m_maximumLength = inLength;
|
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
inline void Span::removeNodeOfLengthGreaterThan(size_t inLength)
|
inline void Span::removeNodeOfLengthGreaterThan(size_t length) {
|
||||||
{
|
if (length > m_maximumLength) {
|
||||||
if (inLength > m_maximumLength) {
|
|
||||||
return;
|
return;
|
||||||
}
|
}
|
||||||
|
|
||||||
size_t max = 0;
|
size_t max = 0;
|
||||||
set<size_t> removeSet;
|
std::set<size_t> removeSet;
|
||||||
for (map<size_t, Node>::iterator i = m_lengthNodeMap.begin(), e = m_lengthNodeMap.end() ; i != e ; ++i) {
|
for (std::map<size_t, Node>::iterator i = m_lengthNodeMap.begin(),
|
||||||
if ((*i).first > inLength) {
|
e = m_lengthNodeMap.end();
|
||||||
|
i != e; ++i) {
|
||||||
|
if ((*i).first > length) {
|
||||||
removeSet.insert((*i).first);
|
removeSet.insert((*i).first);
|
||||||
}
|
} else {
|
||||||
else {
|
|
||||||
if ((*i).first > max) {
|
if ((*i).first > max) {
|
||||||
max = (*i).first;
|
max = (*i).first;
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
for (set<size_t>::iterator i = removeSet.begin(), e = removeSet.end(); i != e; ++i) {
|
for (std::set<size_t>::iterator i = removeSet.begin(), e = removeSet.end();
|
||||||
|
i != e; ++i) {
|
||||||
m_lengthNodeMap.erase(*i);
|
m_lengthNodeMap.erase(*i);
|
||||||
}
|
}
|
||||||
|
|
||||||
m_maximumLength = max;
|
m_maximumLength = max;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline Node* Span::nodeOfLength(size_t inLength)
|
inline Node* Span::nodeOfLength(size_t length) {
|
||||||
{
|
std::map<size_t, Node>::iterator f = m_lengthNodeMap.find(length);
|
||||||
map<size_t, Node>::iterator f = m_lengthNodeMap.find(inLength);
|
|
||||||
return f == m_lengthNodeMap.end() ? 0 : &(*f).second;
|
return f == m_lengthNodeMap.end() ? 0 : &(*f).second;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline size_t Span::maximumLength() const
|
inline size_t Span::maximumLength() const { return m_maximumLength; }
|
||||||
{
|
} // namespace Gramambular
|
||||||
return m_maximumLength;
|
} // namespace Taiyan
|
||||||
}
|
|
||||||
}
|
|
||||||
}
|
|
||||||
|
|
||||||
#endif
|
#endif
|
||||||
|
|
|
@ -17,14 +17,16 @@ THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABI
|
||||||
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
||||||
*/
|
*/
|
||||||
|
|
||||||
#ifndef Unigram_h
|
#ifndef UNIGRAM_H_
|
||||||
#define Unigram_h
|
#define UNIGRAM_H_
|
||||||
|
|
||||||
#include <vector>
|
#include <vector>
|
||||||
|
|
||||||
#include "KeyValuePair.h"
|
#include "KeyValuePair.h"
|
||||||
|
|
||||||
namespace Taiyan {
|
namespace Taiyan {
|
||||||
namespace Gramambular {
|
namespace Gramambular {
|
||||||
|
|
||||||
class Unigram {
|
class Unigram {
|
||||||
public:
|
public:
|
||||||
Unigram();
|
Unigram();
|
||||||
|
@ -32,65 +34,58 @@ namespace Taiyan {
|
||||||
KeyValuePair keyValue;
|
KeyValuePair keyValue;
|
||||||
double score;
|
double score;
|
||||||
|
|
||||||
bool operator==(const Unigram& inAnother) const;
|
bool operator==(const Unigram& another) const;
|
||||||
bool operator<(const Unigram& inAnother) const;
|
bool operator<(const Unigram& another) const;
|
||||||
|
|
||||||
static bool ScoreCompare(const Unigram& a, const Unigram& b);
|
static bool ScoreCompare(const Unigram& a, const Unigram& b);
|
||||||
};
|
};
|
||||||
|
|
||||||
inline ostream& operator<<(ostream& inStream, const Unigram& inGram)
|
inline std::ostream& operator<<(std::ostream& stream, const Unigram& gram) {
|
||||||
{
|
std::streamsize p = stream.precision();
|
||||||
streamsize p = inStream.precision();
|
stream.precision(6);
|
||||||
inStream.precision(6);
|
stream << "(" << gram.keyValue << "," << gram.score << ")";
|
||||||
inStream << "(" << inGram.keyValue << "," << inGram.score << ")";
|
stream.precision(p);
|
||||||
inStream.precision(p);
|
return stream;
|
||||||
return inStream;
|
|
||||||
}
|
}
|
||||||
|
|
||||||
inline ostream& operator<<(ostream& inStream, const vector<Unigram>& inGrams)
|
inline std::ostream& operator<<(std::ostream& stream,
|
||||||
{
|
const std::vector<Unigram>& grams) {
|
||||||
inStream << "[" << inGrams.size() << "]=>{";
|
stream << "[" << grams.size() << "]=>{";
|
||||||
|
|
||||||
size_t index = 0;
|
size_t index = 0;
|
||||||
|
|
||||||
for (vector<Unigram>::const_iterator gi = inGrams.begin() ; gi != inGrams.end() ; ++gi, ++index) {
|
for (std::vector<Unigram>::const_iterator gi = grams.begin();
|
||||||
inStream << index << "=>";
|
gi != grams.end(); ++gi, ++index) {
|
||||||
inStream << *gi;
|
stream << index << "=>";
|
||||||
if (gi + 1 != inGrams.end()) {
|
stream << *gi;
|
||||||
inStream << ",";
|
if (gi + 1 != grams.end()) {
|
||||||
|
stream << ",";
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
inStream << "}";
|
stream << "}";
|
||||||
return inStream;
|
return stream;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline Unigram::Unigram()
|
inline Unigram::Unigram() : score(0.0) {}
|
||||||
: score(0.0)
|
|
||||||
{
|
inline bool Unigram::operator==(const Unigram& another) const {
|
||||||
|
return keyValue == another.keyValue && score == another.score;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline bool Unigram::operator==(const Unigram& inAnother) const
|
inline bool Unigram::operator<(const Unigram& another) const {
|
||||||
{
|
if (keyValue < another.keyValue) {
|
||||||
return keyValue == inAnother.keyValue && score == inAnother.score;
|
|
||||||
}
|
|
||||||
|
|
||||||
inline bool Unigram::operator<(const Unigram& inAnother) const
|
|
||||||
{
|
|
||||||
if (keyValue < inAnother.keyValue) {
|
|
||||||
return true;
|
return true;
|
||||||
}
|
} else if (keyValue == another.keyValue) {
|
||||||
else if (keyValue == inAnother.keyValue) {
|
return score < another.score;
|
||||||
return score < inAnother.score;
|
|
||||||
}
|
}
|
||||||
return false;
|
return false;
|
||||||
}
|
}
|
||||||
|
|
||||||
inline bool Unigram::ScoreCompare(const Unigram& a, const Unigram& b)
|
inline bool Unigram::ScoreCompare(const Unigram& a, const Unigram& b) {
|
||||||
{
|
|
||||||
return a.score > b.score;
|
return a.score > b.score;
|
||||||
}
|
}
|
||||||
}
|
} // namespace Gramambular
|
||||||
}
|
} // namespace Taiyan
|
||||||
|
|
||||||
#endif
|
#endif
|
||||||
|
|
|
@ -17,59 +17,61 @@ THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABI
|
||||||
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
||||||
*/
|
*/
|
||||||
|
|
||||||
#ifndef Walker_h
|
#ifndef WALKER_H_
|
||||||
#define Walker_h
|
#define WALKER_H_
|
||||||
|
|
||||||
#include <algorithm>
|
#include <algorithm>
|
||||||
|
#include <vector>
|
||||||
|
|
||||||
#include "Grid.h"
|
#include "Grid.h"
|
||||||
|
|
||||||
namespace Taiyan {
|
namespace Taiyan {
|
||||||
namespace Gramambular {
|
namespace Gramambular {
|
||||||
using namespace std;
|
|
||||||
|
|
||||||
class Walker {
|
class Walker {
|
||||||
public:
|
public:
|
||||||
Walker(Grid* inGrid);
|
explicit Walker(Grid* inGrid);
|
||||||
const vector<NodeAnchor> reverseWalk(size_t inLocation, double inAccumulatedScore = 0.0);
|
const std::vector<NodeAnchor> reverseWalk(size_t location,
|
||||||
|
double accumulatedScore = 0.0);
|
||||||
|
|
||||||
protected:
|
protected:
|
||||||
Grid* m_grid;
|
Grid* m_grid;
|
||||||
};
|
};
|
||||||
|
|
||||||
inline Walker::Walker(Grid* inGrid)
|
inline Walker::Walker(Grid* inGrid) : m_grid(inGrid) {}
|
||||||
: m_grid(inGrid)
|
|
||||||
{
|
inline const std::vector<NodeAnchor> Walker::reverseWalk(
|
||||||
|
size_t location, double accumulatedScore) {
|
||||||
|
if (!location || location > m_grid->width()) {
|
||||||
|
return std::vector<NodeAnchor>();
|
||||||
}
|
}
|
||||||
|
|
||||||
inline const vector<NodeAnchor> Walker::reverseWalk(size_t inLocation, double inAccumulatedScore)
|
std::vector<std::vector<NodeAnchor> > paths;
|
||||||
{
|
|
||||||
if (!inLocation || inLocation > m_grid->width()) {
|
|
||||||
return vector<NodeAnchor>();
|
|
||||||
}
|
|
||||||
|
|
||||||
vector<vector<NodeAnchor> > paths;
|
std::vector<NodeAnchor> nodes = m_grid->nodesEndingAt(location);
|
||||||
|
|
||||||
vector<NodeAnchor> nodes = m_grid->nodesEndingAt(inLocation);
|
for (std::vector<NodeAnchor>::iterator ni = nodes.begin(); ni != nodes.end();
|
||||||
|
++ni) {
|
||||||
for (vector<NodeAnchor>::iterator ni = nodes.begin() ; ni != nodes.end() ; ++ni) {
|
|
||||||
if (!(*ni).node) {
|
if (!(*ni).node) {
|
||||||
continue;
|
continue;
|
||||||
}
|
}
|
||||||
|
|
||||||
(*ni).accumulatedScore = inAccumulatedScore + (*ni).node->score();
|
(*ni).accumulatedScore = accumulatedScore + (*ni).node->score();
|
||||||
|
|
||||||
vector<NodeAnchor> path = reverseWalk(inLocation - (*ni).spanningLength, (*ni).accumulatedScore);
|
std::vector<NodeAnchor> path =
|
||||||
|
reverseWalk(location - (*ni).spanningLength, (*ni).accumulatedScore);
|
||||||
path.insert(path.begin(), *ni);
|
path.insert(path.begin(), *ni);
|
||||||
|
|
||||||
paths.push_back(path);
|
paths.push_back(path);
|
||||||
}
|
}
|
||||||
|
|
||||||
if (!paths.size()) {
|
if (!paths.size()) {
|
||||||
return vector<NodeAnchor>();
|
return std::vector<NodeAnchor>();
|
||||||
}
|
}
|
||||||
|
|
||||||
vector<NodeAnchor>* result = &*(paths.begin());
|
std::vector<NodeAnchor>* result = &*(paths.begin());
|
||||||
for (vector<vector<NodeAnchor> >::iterator pi = paths.begin() ; pi != paths.end() ; ++pi) {
|
for (std::vector<std::vector<NodeAnchor> >::iterator pi = paths.begin();
|
||||||
|
pi != paths.end(); ++pi) {
|
||||||
if ((*pi).back().accumulatedScore > result->back().accumulatedScore) {
|
if ((*pi).back().accumulatedScore > result->back().accumulatedScore) {
|
||||||
result = &*pi;
|
result = &*pi;
|
||||||
}
|
}
|
||||||
|
@ -77,7 +79,7 @@ namespace Taiyan {
|
||||||
|
|
||||||
return *result;
|
return *result;
|
||||||
}
|
}
|
||||||
}
|
} // namespace Gramambular
|
||||||
}
|
} // namespace Taiyan
|
||||||
|
|
||||||
#endif
|
#endif
|
||||||
|
|
|
@ -50,6 +50,7 @@
|
||||||
5BD05C6827B2BBEF004C4F1D /* Content.swift in Sources */ = {isa = PBXBuildFile; fileRef = 5BD05C6327B2BBEF004C4F1D /* Content.swift */; };
|
5BD05C6827B2BBEF004C4F1D /* Content.swift in Sources */ = {isa = PBXBuildFile; fileRef = 5BD05C6327B2BBEF004C4F1D /* Content.swift */; };
|
||||||
5BD05C6927B2BBEF004C4F1D /* WindowController.swift in Sources */ = {isa = PBXBuildFile; fileRef = 5BD05C6427B2BBEF004C4F1D /* WindowController.swift */; };
|
5BD05C6927B2BBEF004C4F1D /* WindowController.swift in Sources */ = {isa = PBXBuildFile; fileRef = 5BD05C6427B2BBEF004C4F1D /* WindowController.swift */; };
|
||||||
5BD05C6A27B2BBEF004C4F1D /* ViewController.swift in Sources */ = {isa = PBXBuildFile; fileRef = 5BD05C6527B2BBEF004C4F1D /* ViewController.swift */; };
|
5BD05C6A27B2BBEF004C4F1D /* ViewController.swift in Sources */ = {isa = PBXBuildFile; fileRef = 5BD05C6527B2BBEF004C4F1D /* ViewController.swift */; };
|
||||||
|
5BDC5CAB27C2873D00E1CCE2 /* Grid.mm in Sources */ = {isa = PBXBuildFile; fileRef = 5BDC5CAA27C2873D00E1CCE2 /* Grid.mm */; };
|
||||||
5BDCBB2E27B4E67A00D0CC59 /* vChewingPhraseEditor.app in Resources */ = {isa = PBXBuildFile; fileRef = 5BD05BB827B2A429004C4F1D /* vChewingPhraseEditor.app */; };
|
5BDCBB2E27B4E67A00D0CC59 /* vChewingPhraseEditor.app in Resources */ = {isa = PBXBuildFile; fileRef = 5BD05BB827B2A429004C4F1D /* vChewingPhraseEditor.app */; };
|
||||||
5BE78BD927B3775B005EA1BE /* ctlAboutWindow.swift in Sources */ = {isa = PBXBuildFile; fileRef = 5BE78BD827B37750005EA1BE /* ctlAboutWindow.swift */; };
|
5BE78BD927B3775B005EA1BE /* ctlAboutWindow.swift in Sources */ = {isa = PBXBuildFile; fileRef = 5BE78BD827B37750005EA1BE /* ctlAboutWindow.swift */; };
|
||||||
5BE78BDD27B3776D005EA1BE /* frmAboutWindow.xib in Resources */ = {isa = PBXBuildFile; fileRef = 5BE78BDA27B37764005EA1BE /* frmAboutWindow.xib */; };
|
5BE78BDD27B3776D005EA1BE /* frmAboutWindow.xib in Resources */ = {isa = PBXBuildFile; fileRef = 5BE78BDA27B37764005EA1BE /* frmAboutWindow.xib */; };
|
||||||
|
@ -195,6 +196,7 @@
|
||||||
5BD05C6327B2BBEF004C4F1D /* Content.swift */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.swift; path = Content.swift; sourceTree = "<group>"; };
|
5BD05C6327B2BBEF004C4F1D /* Content.swift */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.swift; path = Content.swift; sourceTree = "<group>"; };
|
||||||
5BD05C6427B2BBEF004C4F1D /* WindowController.swift */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.swift; path = WindowController.swift; sourceTree = "<group>"; };
|
5BD05C6427B2BBEF004C4F1D /* WindowController.swift */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.swift; path = WindowController.swift; sourceTree = "<group>"; };
|
||||||
5BD05C6527B2BBEF004C4F1D /* ViewController.swift */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.swift; path = ViewController.swift; sourceTree = "<group>"; };
|
5BD05C6527B2BBEF004C4F1D /* ViewController.swift */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.swift; path = ViewController.swift; sourceTree = "<group>"; };
|
||||||
|
5BDC5CAA27C2873D00E1CCE2 /* Grid.mm */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.cpp.objcpp; path = Grid.mm; sourceTree = "<group>"; };
|
||||||
5BDCBB4227B4F6C600D0CC59 /* zh-Hant */ = {isa = PBXFileReference; lastKnownFileType = text.plist.strings; name = "zh-Hant"; path = "zh-Hant.lproj/MainMenu.strings"; sourceTree = "<group>"; };
|
5BDCBB4227B4F6C600D0CC59 /* zh-Hant */ = {isa = PBXFileReference; lastKnownFileType = text.plist.strings; name = "zh-Hant"; path = "zh-Hant.lproj/MainMenu.strings"; sourceTree = "<group>"; };
|
||||||
5BDCBB4327B4F6C600D0CC59 /* zh-Hant */ = {isa = PBXFileReference; lastKnownFileType = text.plist.strings; name = "zh-Hant"; path = "zh-Hant.lproj/frmAboutWindow.strings"; sourceTree = "<group>"; };
|
5BDCBB4327B4F6C600D0CC59 /* zh-Hant */ = {isa = PBXFileReference; lastKnownFileType = text.plist.strings; name = "zh-Hant"; path = "zh-Hant.lproj/frmAboutWindow.strings"; sourceTree = "<group>"; };
|
||||||
5BDCBB4527B4F6C600D0CC59 /* zh-Hant */ = {isa = PBXFileReference; lastKnownFileType = text.plist.strings; name = "zh-Hant"; path = "Source/WindowNIBs/zh-Hant.lproj/frmPrefWindow.strings"; sourceTree = "<group>"; };
|
5BDCBB4527B4F6C600D0CC59 /* zh-Hant */ = {isa = PBXFileReference; lastKnownFileType = text.plist.strings; name = "zh-Hant"; path = "Source/WindowNIBs/zh-Hant.lproj/frmPrefWindow.strings"; sourceTree = "<group>"; };
|
||||||
|
@ -663,6 +665,7 @@
|
||||||
6A0D4F1515FC0EB100ABF4B3 /* BlockReadingBuilder.h */,
|
6A0D4F1515FC0EB100ABF4B3 /* BlockReadingBuilder.h */,
|
||||||
6A0D4F1615FC0EB100ABF4B3 /* Gramambular.h */,
|
6A0D4F1615FC0EB100ABF4B3 /* Gramambular.h */,
|
||||||
6A0D4F1715FC0EB100ABF4B3 /* Grid.h */,
|
6A0D4F1715FC0EB100ABF4B3 /* Grid.h */,
|
||||||
|
5BDC5CAA27C2873D00E1CCE2 /* Grid.mm */,
|
||||||
6A0D4F1815FC0EB100ABF4B3 /* KeyValuePair.h */,
|
6A0D4F1815FC0EB100ABF4B3 /* KeyValuePair.h */,
|
||||||
6A0D4F1915FC0EB100ABF4B3 /* LanguageModel.h */,
|
6A0D4F1915FC0EB100ABF4B3 /* LanguageModel.h */,
|
||||||
6A0D4F1A15FC0EB100ABF4B3 /* Node.h */,
|
6A0D4F1A15FC0EB100ABF4B3 /* Node.h */,
|
||||||
|
@ -943,6 +946,7 @@
|
||||||
D47F7DD3278C1263002F9DD7 /* UserOverrideModel.cpp in Sources */,
|
D47F7DD3278C1263002F9DD7 /* UserOverrideModel.cpp in Sources */,
|
||||||
5B62A33627AE795800A19448 /* PreferencesModule.swift in Sources */,
|
5B62A33627AE795800A19448 /* PreferencesModule.swift in Sources */,
|
||||||
5B62A33827AE79CD00A19448 /* NSStringUtils.swift in Sources */,
|
5B62A33827AE79CD00A19448 /* NSStringUtils.swift in Sources */,
|
||||||
|
5BDC5CAB27C2873D00E1CCE2 /* Grid.mm in Sources */,
|
||||||
5B62A33227AE792F00A19448 /* InputSourceHelper.swift in Sources */,
|
5B62A33227AE792F00A19448 /* InputSourceHelper.swift in Sources */,
|
||||||
5B62A34927AE7CD900A19448 /* TooltipController.swift in Sources */,
|
5B62A34927AE7CD900A19448 /* TooltipController.swift in Sources */,
|
||||||
6A0D4F4515FC0EB100ABF4B3 /* Mandarin.cpp in Sources */,
|
6A0D4F4515FC0EB100ABF4B3 /* Mandarin.cpp in Sources */,
|
||||||
|
|
Loading…
Reference in New Issue