diff --git a/maca_common/include/LimitedArray.hpp b/maca_common/include/LimitedArray.hpp
index c830423ee0aab50bd05ff2eaed5cca7d0f4ff86b..f7176f6ba6b25613c83d5cfe695a0f15555629d4 100644
--- a/maca_common/include/LimitedArray.hpp
+++ b/maca_common/include/LimitedArray.hpp
@@ -30,16 +30,17 @@ class LimitedArray
nbElements = 0;
lastElementDataIndex = -1;
lastElementRealIndex = -1;
+ data.clear();
}
void push(const T & elem)
{
nbElements++;
- if (nbElements > data.size())
+ if (nbElements > (int)data.size())
nbElements = data.size();
lastElementDataIndex++;
- if (lastElementDataIndex >= data.size())
+ if (lastElementDataIndex >= (int)data.size())
lastElementDataIndex = 0;
lastElementRealIndex++;
@@ -47,15 +48,25 @@ class LimitedArray
data[lastElementDataIndex] = elem;
}
- const T & get(unsigned int index)
+ const T & get(unsigned int index) const
{
return data[index % data.size()];
}
- int getLastIndex()
+ void set(unsigned int index, const T & value)
+ {
+ data[index % data.size()] = value;
+ }
+
+ int getLastIndex() const
{
return lastElementRealIndex;
}
+
+ void copy(LimitedArray<T> & other, unsigned int from, unsigned int to)
+ {
+ std::copy(other.data.begin()+from, other.data.begin()+to, std::back_inserter(data));
+ }
};
#endif
diff --git a/maca_common/include/LimitedStack.hpp b/maca_common/include/LimitedStack.hpp
index 9994451a6e81b1c1b63e608986a20320dd78d41f..32422abd60fdf1f344fb2c0d9588e8c763cc289d 100644
--- a/maca_common/include/LimitedStack.hpp
+++ b/maca_common/include/LimitedStack.hpp
@@ -33,10 +33,10 @@ class LimitedStack
void push(T elem)
{
nbElements++;
- if (nbElements > data.size())
+ if (nbElements > (int)data.size())
nbElements = data.size();
lastElementIndex++;
- if (lastElementIndex >= data.size())
+ if (lastElementIndex >= (int)data.size())
lastElementIndex = 0;
data[lastElementIndex] = elem;
@@ -80,7 +80,7 @@ class LimitedStack
bool contains(const T & element)
{
int currentIndex = lastElementIndex;
- for (unsigned int i = 0; i < nbElements; i++)
+ for (int i = 0; i < nbElements; i++)
{
if (data[currentIndex] == element)
return true;
diff --git a/neural_network/src/MLPBase.cpp b/neural_network/src/MLPBase.cpp
index 8fd98e450db028d18a69d785469409af1690d8ac..ed4149603b77154616a15f047be768f5a26a1768 100644
--- a/neural_network/src/MLPBase.cpp
+++ b/neural_network/src/MLPBase.cpp
@@ -312,9 +312,9 @@ void MLPBase::loadStruct(dynet::ParameterCollection & model, const std::string &
exit(1);
}
- do
- {
- } while (fscanf(fd, "#TOPOLOGY# # {1,1} 0 Layer : %d %d %s %f\n", &input, &output, activation, &dropout) == 4);
+ do
+ {
+ } while (fscanf(fd, "#TOPOLOGY# # {1,1} 0 Layer : %d %d %s %f\n", &input, &output, activation, &dropout) == 4);
}
while (fscanf(fd, "#TOPOLOGY# # {1,1} 0 Layer : %d %d %s %f\n", &input, &output, activation, &dropout) != 4)
@@ -346,7 +346,7 @@ void MLPBase::loadParameters(dynet::ParameterCollection & model, const std::stri
{
parameters[i][0] = loader.load_param(model, prefix + std::to_string(i) + "_W");
parameters[i][1] = loader.load_param(model, prefix + std::to_string(i) + "_b");
- } catch(const std::runtime_error e)
+ } catch(const std::runtime_error & e)
{
fprintf(stderr, "WARNING (%s) : Could not find parameter with key \'%s\' in the model. Ignore this if this model was trained with an older version of Macaon.\n", ERRINFO, (prefix+std::to_string(i) + "_W").c_str());
prefix = "Layer_";
diff --git a/trainer/src/Trainer.cpp b/trainer/src/Trainer.cpp
index 6784c3ccc369f90e81898d2e55447395eb85d435..63c07cb43838cf962c24c23e8b4e2a8ef1f2df9d 100644
--- a/trainer/src/Trainer.cpp
+++ b/trainer/src/Trainer.cpp
@@ -56,7 +56,7 @@ void Trainer::computeScoreOnDev()
// Print current iter advancement in percentage
if (ProgramParameters::interactive)
{
- int totalSize = devConfig->tapes[0].hyp.size();
+ int totalSize = ProgramParameters::tapeSize;
int steps = devConfig->getHead();
if (steps && (steps % 200 == 0 || totalSize-steps < 200))
{
@@ -208,7 +208,7 @@ void Trainer::train()
// Print current iter advancement in percentage
if (ProgramParameters::interactive)
{
- int totalSize = ProgramParameters::iterationSize == -1 ? trainConfig.tapes[0].hyp.size() : ProgramParameters::iterationSize;
+ int totalSize = ProgramParameters::iterationSize == -1 ? ProgramParameters::tapeSize : ProgramParameters::iterationSize;
int steps = ProgramParameters::iterationSize == -1 ? trainConfig.getHead() : nbSteps;
if (steps % 200 == 0 || totalSize-steps < 200)
{
@@ -226,7 +226,7 @@ void Trainer::train()
if (zeroCostActions.empty())
{
- if (trainConfig.getHead() >= (int)trainConfig.tapes[0].ref.size()-1)
+ if (trainConfig.endOfTapes())
{
while (!trainConfig.stackEmpty())
trainConfig.stackPop();
diff --git a/trainer/src/macaon_train.cpp b/trainer/src/macaon_train.cpp
index 02f251d6da8ac7c17d622d157ab17ba1f2c1934a..4f8e10d931a33fdea44c1e6421b01e5dda541d34 100644
--- a/trainer/src/macaon_train.cpp
+++ b/trainer/src/macaon_train.cpp
@@ -34,7 +34,9 @@ po::options_description getOptionsDescription()
("mcd", po::value<std::string>()->required(),
"MCD file that describes the input")
("train,T", po::value<std::string>()->required(),
- "Training corpus formated according to the MCD");
+ "Training corpus formated according to the MCD")
+ ("tapeSize", po::value<int>()->required(),
+ "Number of lines in the input file.");
po::options_description opt("Optional");
opt.add_options()
@@ -67,8 +69,6 @@ po::options_description getOptionsDescription()
"Remove identical training examples")
("showFeatureRepresentation", po::value<int>()->default_value(0),
"For each state of the Config, show its feature representation")
- ("tapeSize", po::value<int>()->default_value(100000),
- "Number of lines in the input file.")
("interactive", po::value<bool>()->default_value(true),
"Is the shell interactive ? Display advancement informations")
("randomEmbeddings", po::value<bool>()->default_value(false),
diff --git a/transition_machine/include/Config.hpp b/transition_machine/include/Config.hpp
index b0c0814b67c85a9672ecc5742075c26e30d1659e..922f642354749a3d1960d1a3eecdef1461e606b0 100644
--- a/transition_machine/include/Config.hpp
+++ b/transition_machine/include/Config.hpp
@@ -60,11 +60,16 @@ class Config
/// @param relativeIndex The index of the cell relatively to the head.
///
/// @return The content of the cell.
- const std::string & gethyp(int relativeIndex);
+ const std::string & getHyp(int relativeIndex);
+ /// @brief Set the value of a cell of the hyp.
+ ///
+ /// @param relativeIndex The index of the cell relatively to the head.
+ /// @param elem The new content of the cell.
+ void setHyp(int relativeIndex, const std::string & elem);
/// @brief Return true if the head of this tape is on the last cell.
///
/// @return True if the head of this tape is on the last cell.
- bool headIsAtEnd();
+ bool headIsAtEnd() const;
public :
@@ -74,18 +79,38 @@ class Config
///
/// @return The name of this Tape.
const std::string & getName();
- /// @brief Set the name of this Tape.
- ///
- /// @param name The desired name.
- void setName(const std::string & name);
- /// @brief Set if the content of this tape is known or predicted.
- ///
- /// @param known The value to set.
void setKnown(bool known);
/// @brief Move the head of this tape.
///
/// @param mvt The relative movement to apply to the head.
void moveHead(int mvt);
+ /// @brief Return the current size of the Tape, in number of cells.
+ ///
+ /// @return The current number of cells in this Tape.
+ int size();
+ /// @brief Return the current size of the ref Tape, in number of cells.
+ ///
+ /// @return The current number of cells in this Tape.
+ int refSize();
+ /// @brief Return the current size of the hyp Tape, in number of cells.
+ ///
+ /// @return The current number of cells in this Tape.
+ int hypSize();
+ /// @brief Add cell to ref.
+ ///
+ /// @param elem The content of the new cell.
+ void addToRef(const std::string & elem);
+ /// @brief Add cell to hyp.
+ ///
+ /// @param elem The content of the new cell.
+ void addToHyp(const std::string & elem);
+ /// @brief Empty the Tape.
+ void clear();
+ /// @brief Copy a chunk of an other Tape inside this one.
+ ///
+ /// @param from first cell index of the chunk to copy.
+ /// @param to last cell index of the chunk to copy.
+ void copyPart(Tape & other, unsigned int from, unsigned int to);
};
private :
diff --git a/transition_machine/src/ActionBank.cpp b/transition_machine/src/ActionBank.cpp
index 2de43a8c4e744c3695452a39657eff12d06926df..84b063d5505782693ac81e74387aada204ed97a7 100644
--- a/transition_machine/src/ActionBank.cpp
+++ b/transition_machine/src/ActionBank.cpp
@@ -394,7 +394,7 @@ std::vector<Action::BasicAction> ActionBank::str2sequence(const std::string & na
for (int i = c.stackSize()-1; i >= 0; i--)
{
auto s = c.stackGetElem(i);
- if (govs.getHyp[s-b0] == "0")
+ if (govs.getHyp(s-b0) == "0")
{
simpleBufferWrite(c, "GOV", "", s-b0);
break;
@@ -578,9 +578,7 @@ void ActionBank::simpleBufferWrite(Config & config, const std::string & tapeName
{
auto & tape = config.getTape(tapeName);
- int index = config.getHead() + relativeIndex;
-
- tape.setHyp(index, value);
+ tape.setHyp(relativeIndex, value);
}
bool ActionBank::simpleBufferWriteAppliable(Config & config,
@@ -590,10 +588,10 @@ bool ActionBank::simpleBufferWriteAppliable(Config & config,
int index = config.getHead() + relativeIndex;
- if (c.endOfTapes())
+ if (config.endOfTapes())
return true;
- return !(index < 0);
+ return !(index < 0) && index < tape.size();
}
bool ActionBank::isRuleAppliable(Config & config,
@@ -611,7 +609,7 @@ void ActionBank::writeRuleResult(Config & config, const std::string & fromTapeNa
auto & from = fromTape.getRef(relativeIndex);
- toTape.setHyp(config.getHead() + relativeIndex, applyRule(from, rule));
+ toTape.setHyp(relativeIndex, applyRule(from, rule));
}
int ActionBank::getLinkLength(const Config & c, const std::string & action)
diff --git a/transition_machine/src/Config.cpp b/transition_machine/src/Config.cpp
index 2a9b04707c706f0deef4018b592397ea3c066a0d..1be439008ef66969edc73033cd2ddf37d7b282fc 100644
--- a/transition_machine/src/Config.cpp
+++ b/transition_machine/src/Config.cpp
@@ -36,7 +36,7 @@ void Config::readInput()
{
if (!file.get())
file.reset(new File(inputFilename, "r"));
- FILE * fd = file.getDescriptor();
+ FILE * fd = file->getDescriptor();
char buffer[10000];
std::vector<std::string> cols;
@@ -45,8 +45,6 @@ void Config::readInput()
int toRead = 100;
int haveRead = 0;
- bool finishedFile = false;
-
while(fscanf(fd, "%[^\n]\n", buffer) == 1 && haveRead < toRead)
{
cols = split(buffer, '\t');
@@ -55,7 +53,7 @@ void Config::readInput()
if (cols.size() != usualColsSize)
{
- fprintf(stderr, "ERROR (%s) : input (%s) line %lu has %lu columns instead of %u. Aborting.\n", ERRINFO, filename.c_str(), tapes[0].ref.size(), cols.size(), usualColsSize);
+ fprintf(stderr, "ERROR (%s) : input (%s) line %d has %lu columns instead of %u. Aborting.\n", ERRINFO, inputFilename.c_str(), tapes[0].size(), cols.size(), usualColsSize);
exit(1);
}
@@ -72,7 +70,7 @@ void Config::readInput()
}
// Making all tapes the same size
- unsigned int maxTapeSize = 0;
+ int maxTapeSize = 0;
for(auto & tape : tapes)
maxTapeSize = std::max<unsigned int>(maxTapeSize, tape.refSize());
@@ -128,7 +126,7 @@ void Config::printForDebug(FILE * output)
{
cols[0].emplace_back(tape.getName());
- for(int i = std::max(0, head-window); i < std::min((int)tape.hyp.size(), head+window); i++)
+ for(int i = std::max(0, head-window); i < std::min(tape.hypSize(), head+window); i++)
{
unsigned int colIndex = i - std::max(0, head-window)+1;
while(cols.size() <= colIndex)
@@ -142,7 +140,6 @@ void Config::printForDebug(FILE * output)
cols[colIndex].emplace_back(shrink(tape[i]));
}
-
}
fprintf(output, "Configuration :\n");
@@ -167,8 +164,11 @@ void Config::printAsExample(FILE *)
exit(1);
}
-void Config::printAsOutput(FILE * output)
+void Config::printAsOutput(FILE *)
{
+ //TODO : Output should be done when reading, and at the end of the program.
+
+ /*
unsigned int lastToPrint = 0;
for (unsigned int j = 0; j < tapes.size(); j++)
if(bd.mustPrintLine(j))
@@ -180,11 +180,12 @@ void Config::printAsOutput(FILE * output)
if(bd.mustPrintLine(j))
fprintf(output, "%s%s", tapes[j][i].empty() ? "0" : tapes[j][i].c_str(), j == lastToPrint ? "\n" : "\t");
}
+ */
}
void Config::moveHead(int mvt)
{
- if (head + mvt < (int)tapes[0].ref.size())
+ if (head + mvt < tapes[0].size())
{
head += mvt;
@@ -195,16 +196,13 @@ void Config::moveHead(int mvt)
bool Config::isFinal()
{
- return head >= (int)getTapeByInputCol(0).hyp.size()-1 && stack.empty();
+ return endOfTapes() && stack.empty();
}
void Config::reset()
{
for (auto & tape : tapes)
- {
- tape.ref.clear();
- tape.hyp.clear();
- }
+ tape.clear();
actionHistory.clear();
pastActions.clear();
@@ -215,25 +213,31 @@ void Config::reset()
head = 0;
- readInput(inputFilename);
+ file.reset();
+ readInput();
}
const std::string & Config::Tape::operator[](int relativeIndex)
{
if(isKnown)
- return getRef[relativeIndex];
+ return getRef(relativeIndex);
- return getHyp[relativeIndex];
+ return getHyp(relativeIndex);
}
const std::string & Config::Tape::getRef(int relativeIndex)
{
- return ref[head + relativeIndex];
+ return ref.get(head + relativeIndex);
}
-const std::string & Config::Tape::gethyp(int relativeIndex)
+const std::string & Config::Tape::getHyp(int relativeIndex)
{
- return hyp[head + relativeIndex];
+ return hyp.get(head + relativeIndex);
+}
+
+void Config::Tape::setHyp(int relativeIndex, const std::string & elem)
+{
+ hyp.set(head + relativeIndex, elem);
}
std::string & Config::getCurrentStateName()
@@ -282,8 +286,8 @@ void Config::shuffle(const std::string & delimiterTape, const std::string & deli
std::vector< std::pair<unsigned int, unsigned int> > delimiters;
unsigned int previousIndex = 0;
- for (unsigned int i = 0; i < tape.ref.size(); i++)
- if (tape.ref[i] == delimiter)
+ for (int i = 0; i < tape.refSize(); i++)
+ if (tape.getRef(i-head) == delimiter)
{
delimiters.emplace_back(previousIndex, i);
previousIndex = i+1;
@@ -295,28 +299,21 @@ void Config::shuffle(const std::string & delimiterTape, const std::string & deli
exit(1);
}
- std::pair<unsigned int, unsigned int> suffix = {delimiters.back().second+1, tape.ref.size()-1};
+ std::pair<unsigned int, unsigned int> suffix = {delimiters.back().second+1, tape.refSize()-1};
std::random_shuffle(delimiters.begin(), delimiters.end());
- std::vector<Tape> newTapes = tapes;
+ auto newTapes = tapes;
for (unsigned int tape = 0; tape < tapes.size(); tape++)
{
- newTapes[tape].ref.clear();
- newTapes[tape].hyp.clear();
+ newTapes[tape].clear();
for (auto & delimiter : delimiters)
- {
- std::copy(tapes[tape].ref.begin()+delimiter.first, tapes[tape].ref.begin()+delimiter.second+1, std::back_inserter(newTapes[tape].ref));
- std::copy(tapes[tape].hyp.begin()+delimiter.first, tapes[tape].hyp.begin()+delimiter.second+1, std::back_inserter(newTapes[tape].hyp));
- }
+ newTapes[tape].copyPart(tapes[tape], delimiter.first, delimiter.second+1);
if (suffix.first <= suffix.second)
- {
- std::copy(tapes[tape].ref.begin()+suffix.first, tapes[tape].ref.begin()+suffix.second+1, std::back_inserter(newTapes[tape].ref));
- std::copy(tapes[tape].hyp.begin()+suffix.first, tapes[tape].hyp.begin()+suffix.second+1, std::back_inserter(newTapes[tape].hyp));
- }
+ newTapes[tape].copyPart(tapes[tape], suffix.first, suffix.second+1);
}
tapes = newTapes;
@@ -397,15 +394,15 @@ std::size_t Config::computeHash()
{
static int window = 3;
- unsigned int start = std::max(0, head-window);
- unsigned int end = std::min((unsigned int)tapes[0].ref.size()-1, (unsigned int)head+window);
+ int start = std::max(0, head-window);
+ int end = std::min(tapes[0].refSize()-1, head+window);
std::hash<std::string> hasher;
std::size_t result = 0;
- for (unsigned int i = start; i < end; i++)
+ for (int i = start; i < end; i++)
for (auto & tape : tapes)
- result ^= (hasher(tape[i])*0x9e3779b9+(result << 6)+(result >>2));
+ result ^= (hasher(tape[i-head])*0x9e3779b9+(result << 6)+(result >>2));
return result;
}
@@ -435,28 +432,56 @@ const std::string & Config::Tape::getName()
return name;
}
-void Config::Tape::setName(const std::string & name)
+void Config::Tape::moveHead(int mvt)
{
- this->name = name;
+ head += mvt;
}
-void Config::Tape::setKnown(bool known)
+bool Config::endOfTapes() const
{
- this->isKnown = known;
+ return inputAllRead && tapes[0].headIsAtEnd();
}
-void Config::Tape::moveHead(int mvt)
+bool Config::Tape::headIsAtEnd() const
{
- head += mvt;
+ return head == ref.getLastIndex();
}
-bool Config::endOfTapes() const
+int Config::Tape::size()
{
- return inputAllRead && tapes[0].headIsAtEnd();
+ return refSize();
}
-bool Config::Tape::headIsAtEnd()
+int Config::Tape::refSize()
{
- return head == ref.getLastIndex();
+ return ref.getLastIndex();
+}
+
+int Config::Tape::hypSize()
+{
+ return hyp.getLastIndex();
+}
+
+void Config::Tape::addToHyp(const std::string & elem)
+{
+ hyp.push(elem);
+}
+
+void Config::Tape::addToRef(const std::string & elem)
+{
+ ref.push(elem);
+}
+void Config::Tape::clear()
+{
+ head = 0;
+ ref.clear();
+ hyp.clear();
+}
+
+void Config::Tape::copyPart(Tape & other, unsigned int from, unsigned int to)
+{
+ ref.copy(other.ref, from, to);
+ hyp.copy(other.hyp, from, to);
+
}
diff --git a/transition_machine/src/FeatureBank.cpp b/transition_machine/src/FeatureBank.cpp
index 4d81ff699e18fd5930c25a0f5298eca37fbd1827..bf8f67f9629567c5404731ada458d60b41b2a968 100644
--- a/transition_machine/src/FeatureBank.cpp
+++ b/transition_machine/src/FeatureBank.cpp
@@ -190,6 +190,7 @@ FeatureModel::FeatureValue FeatureBank::ldep(Config & config, int index, const s
auto & eos = config.getTape(ProgramParameters::sequenceDelimiterTape);
Dict * dict = config.getDictOfLine(tapeName);
auto policy = dictPolicy2FeaturePolicy(dict->policy);
+ int b0 = config.getHead();
if(object == "s")
{
@@ -203,26 +204,26 @@ FeatureModel::FeatureValue FeatureBank::ldep(Config & config, int index, const s
index += config.getHead();
}
- if(index < 0 || index >= (int)tape.hyp.size())
+ if(index < 0 || index >= tape.hypSize())
return {dict, featName, Dict::nullValueStr, policy};
int candidate = -1;
unsigned int maxDist = 10;
- for (int i = std::max<int>(index-1, 0); index - i <= (int)maxDist && i >= 0 && eos[i] != ProgramParameters::sequenceDelimiter; i--)
+ for (int i = std::max<int>(index-1, 0); index - i <= (int)maxDist && i >= 0 && eos[i-b0] != ProgramParameters::sequenceDelimiter; i--)
{
int dist = index - i;
- if(govs[i] == std::to_string(dist))
+ if(govs[i-b0] == std::to_string(dist))
candidate = i;
}
if(candidate == -1)
return {dict, featName, Dict::nullValueStr, policy};
- if(tape[candidate].empty())
+ if(tape[candidate-b0].empty())
return {dict, featName, Dict::nullValueStr, policy};
- return {dict, featName, tape[candidate], policy};
+ return {dict, featName, tape[candidate-b0], policy};
}
FeatureModel::FeatureValue FeatureBank::dist(Config & config, const std::string & object1, int index1, const std::string & object2, int index2, const std::string & featName)
@@ -234,7 +235,7 @@ FeatureModel::FeatureValue FeatureBank::dist(Config & config, const std::string
if (object1 == "b")
{
- if(index1 < 0 || index1 >= (int)config.tapes[0].hyp.size())
+ if(index1 < 0 || index1+config.getHead() >= config.tapes[0].hypSize())
return {dict, featName, Dict::nullValueStr, policy};
elem1 = config.getHead() + index1;
@@ -249,7 +250,7 @@ FeatureModel::FeatureValue FeatureBank::dist(Config & config, const std::string
if (object2 == "b")
{
- if(index2 < 0 || index2 >= (int)config.tapes[0].hyp.size())
+ if(index2 < 0 || index2+config.getHead() >= config.tapes[0].hypSize())
return {dict, featName, Dict::nullValueStr, policy};
elem2 = config.getHead() + index2;
@@ -285,26 +286,26 @@ FeatureModel::FeatureValue FeatureBank::rdep(Config & config, int index, const s
index += config.getHead();
}
- if(index < 0 || index >= (int)tape.hyp.size())
+ if(index < 0 || index >= tape.hypSize())
return {dict, featName, Dict::nullValueStr, policy};
int candidate = -1;
- unsigned int maxDist = 10;
+ int maxDist = 10;
- for (unsigned int i = index+1; i - index <= maxDist && i < eos.hyp.size() && eos[i-1] != ProgramParameters::sequenceDelimiter; i++)
+ for (int i = index+1; i - index <= maxDist && i < eos.hypSize() && eos[i-1-config.getHead()] != ProgramParameters::sequenceDelimiter; i++)
{
int dist = index - i;
- if(govs[i] == std::to_string(dist))
+ if(govs[i-config.getHead()] == std::to_string(dist))
candidate = i;
}
if(candidate == -1)
return {dict, featName, Dict::nullValueStr, policy};
- if(tape[candidate].empty())
+ if(tape[candidate-config.getHead()].empty())
return {dict, featName, Dict::nullValueStr, policy};
- return {dict, featName, tape[candidate], policy};
+ return {dict, featName, tape[candidate-config.getHead()], policy};
}
FeatureModel::FeatureValue FeatureBank::simpleBufferAccess(Config & config, int relativeIndex, const std::string & tapeName, const std::string & featName)
@@ -315,13 +316,13 @@ FeatureModel::FeatureValue FeatureBank::simpleBufferAccess(Config & config, int
int index = config.getHead() + relativeIndex;
- if(index < 0 || index >= (int)tape.hyp.size())
+ if(index < 0 || index >= tape.hypSize())
return {dict, featName, Dict::nullValueStr, policy};
- if(tape[index].empty())
+ if(tape[relativeIndex].empty())
return {dict, featName, Dict::nullValueStr, policy};
- return {dict, featName, tape[index], policy};
+ return {dict, featName, tape[relativeIndex], policy};
}
FeatureModel::FeatureValue FeatureBank::simpleStackAccess(Config & config, int relativeIndex, const std::string & tapeName, const std::string & featName)
@@ -335,13 +336,13 @@ FeatureModel::FeatureValue FeatureBank::simpleStackAccess(Config & config, int r
int index = config.stackGetElem(relativeIndex);
- if(index < 0 || index >= (int)tape.hyp.size())
+ if(index < 0 || index >= tape.hypSize())
return {dict, featName, Dict::nullValueStr, policy};
- if(tape[index].empty())
+ if(tape[index-config.getHead()].empty())
return {dict, featName, Dict::nullValueStr, policy};
- return {dict, featName, tape[index], policy};
+ return {dict, featName, tape[index-config.getHead()], policy};
}
FeatureModel::FeatureValue FeatureBank::getUppercase(Config &, const FeatureModel::FeatureValue & fv)
@@ -435,7 +436,7 @@ FeatureModel::FeatureValue FeatureBank::aggregateBuffer(Config & c, int from, in
{
int index = c.getHead() + i;
std::string featName = "b."+std::to_string(i)+"."+tape.getName();
- if(index < 0 || index >= (int)tape.hyp.size())
+ if(index < 0 || index >= tape.hypSize())
{
result.dicts.emplace_back(dict);
result.names.emplace_back(featName);
@@ -443,7 +444,7 @@ FeatureModel::FeatureValue FeatureBank::aggregateBuffer(Config & c, int from, in
result.policies.emplace_back(policy);
continue;
}
- if(tape[index].empty())
+ if(tape[i].empty())
{
result.dicts.emplace_back(dict);
result.names.emplace_back(featName);
@@ -454,7 +455,7 @@ FeatureModel::FeatureValue FeatureBank::aggregateBuffer(Config & c, int from, in
result.dicts.emplace_back(dict);
result.names.emplace_back(featName);
- result.values.emplace_back(tape[index]);
+ result.values.emplace_back(tape[i]);
result.policies.emplace_back(policy);
}
}
@@ -492,7 +493,7 @@ FeatureModel::FeatureValue FeatureBank::aggregateStack(Config & c, int from, con
continue;
}
int index = c.stackGetElem(i);
- if((index < 0 || index >= (int)tape.hyp.size())||tape[index].empty())
+ if( (index < 0 || index >= tape.hypSize()) || tape[index-c.getHead()].empty() )
{
result.dicts.emplace_back(dict);
result.names.emplace_back(featName);
@@ -503,7 +504,7 @@ FeatureModel::FeatureValue FeatureBank::aggregateStack(Config & c, int from, con
result.dicts.emplace_back(dict);
result.names.emplace_back(featName);
- result.values.emplace_back(tape[index]);
+ result.values.emplace_back(tape[index-c.getHead()]);
result.policies.emplace_back(policy);
}
}
diff --git a/transition_machine/src/Oracle.cpp b/transition_machine/src/Oracle.cpp
index aab0be882f7dfcb2dd983099adce5957d6ee6380..2ca022dbca69fce22ce2b78a2be24d608ef6665a 100644
--- a/transition_machine/src/Oracle.cpp
+++ b/transition_machine/src/Oracle.cpp
@@ -204,7 +204,7 @@ void Oracle::createDatabase()
},
[](Config & c, Oracle *, const std::string & action)
{
- return action == "WRITE b.0 POS " + c.getTape("POS").getRef[0] || c.endOfTapes() ? 0 : 1;
+ return (action == "WRITE b.0 POS " + c.getTape("POS").getRef(0) || c.endOfTapes()) ? 0 : 1;
})));
str2oracle.emplace("tokenizer", std::unique_ptr<Oracle>(new Oracle(
@@ -220,7 +220,7 @@ void Oracle::createDatabase()
},
[](Config & c, Oracle *, const std::string & action)
{
- return action == "WRITE b.0 BIO " + c.getTape("BIO").ref[c.getHead()] || c.getHead() >= (int)c.tapes[0].ref.size()-1 ? 0 : 1;
+ return (action == "WRITE b.0 BIO " + c.getTape("BIO").getRef(0) || c.endOfTapes()) ? 0 : 1;
})));
str2oracle.emplace("eos", std::unique_ptr<Oracle>(new Oracle(
@@ -236,7 +236,7 @@ void Oracle::createDatabase()
},
[](Config & c, Oracle *, const std::string & action)
{
- return action == "WRITE b.0 " + ProgramParameters::sequenceDelimiterTape + " " + (c.getTape(ProgramParameters::sequenceDelimiterTape).ref[c.getHead()] == std::string(ProgramParameters::sequenceDelimiter) ? std::string(ProgramParameters::sequenceDelimiter) : std::string("0")) || c.getHead() >= (int)c.tapes[0].ref.size()-1 ? 0 : 1;
+ return (action == "WRITE b.0 " + ProgramParameters::sequenceDelimiterTape + " " + (c.getTape(ProgramParameters::sequenceDelimiterTape).getRef(0) == std::string(ProgramParameters::sequenceDelimiter) ? std::string(ProgramParameters::sequenceDelimiter) : std::string("0")) || c.endOfTapes()) ? 0 : 1;
})));
str2oracle.emplace("morpho", std::unique_ptr<Oracle>(new Oracle(
@@ -252,7 +252,7 @@ void Oracle::createDatabase()
},
[](Config & c, Oracle *, const std::string & action)
{
- return action == "WRITE b.0 MORPHO " + c.getTape("MORPHO").ref[c.getHead()] || c.getHead() >= (int)c.tapes[0].ref.size()-1 ? 0 : 1;
+ return (action == "WRITE b.0 MORPHO " + c.getTape("MORPHO").getRef(0) || c.endOfTapes()) ? 0 : 1;
})));
str2oracle.emplace("signature", std::unique_ptr<Oracle>(new Oracle(
@@ -270,20 +270,20 @@ void Oracle::createDatabase()
[](Config & c, Oracle * oracle)
{
int window = 3;
- int start = std::max<int>(c.getHead()-window, 0);
- int end = std::min<int>(c.getHead()+window, c.getTape("SGN").hyp.size()-1);
+ int start = std::max<int>(c.getHead()-window, 0) - c.getHead();
+ int end = std::min<int>(c.getHead()+window, c.getTape("SGN").size()-1) - c.getHead();
- while (start < (int)c.getTape("SGN").hyp.size() && !c.getTape("SGN").hyp[start].empty())
+ while (start+c.getHead() < c.getTape("SGN").size() && !c.getTape("SGN").getHyp(start).empty())
start++;
if (start > end)
return std::string("NOTHING");
- std::string action("MULTIWRITE " + std::to_string(start-c.getHead()) + " " + std::to_string(end-c.getHead()) + " " + std::string("SGN"));
+ std::string action("MULTIWRITE " + std::to_string(start) + " " + std::to_string(end) + " " + std::string("SGN"));
for(int i = start; i <= end; i++)
{
- const std::string & form = c.getTape("FORM").ref[i];
+ const std::string & form = c.getTape("FORM").getRef(i);
std::string & signature = oracle->data[form];
if(signature.empty())
@@ -357,11 +357,11 @@ void Oracle::createDatabase()
},
[](Config & c, Oracle *, const std::string & action)
{
- const std::string & form = c.getTape("FORM").ref[c.getHead()];
- const std::string & lemma = c.getTape("LEMMA").ref[c.getHead()];
+ const std::string & form = c.getTape("FORM").getRef(0);
+ const std::string & lemma = c.getTape("LEMMA").getRef(0);
std::string rule = getRule(form, lemma);
- return action == std::string("RULE LEMMA ON FORM ") + rule || c.getHead() >= (int)c.tapes[0].ref.size()-1 ? 0 : 1;
+ return (action == std::string("RULE LEMMA ON FORM ") + rule || c.endOfTapes()) ? 0 : 1;
})));
str2oracle.emplace("parser", std::unique_ptr<Oracle>(new Oracle(
@@ -383,20 +383,20 @@ void Oracle::createDatabase()
int head = c.getHead();
int stackHead = c.stackEmpty() ? 0 : c.stackTop();
- int stackGov = stackHead + std::stoi(govs.ref[stackHead]);
- int headGov = head + std::stoi(govs.ref[head]);
+ int stackGov = stackHead + std::stoi(govs.getRef(stackHead-head));
+ int headGov = head + std::stoi(govs.getRef(0));
int sentenceStart = c.getHead()-1 < 0 ? 0 : c.getHead()-1;
int sentenceEnd = c.getHead();
int cost = 0;
- while(sentenceStart >= 0 && eos.ref[sentenceStart] != ProgramParameters::sequenceDelimiter)
+ while(sentenceStart >= 0 && eos.getRef(sentenceStart-head) != ProgramParameters::sequenceDelimiter)
sentenceStart--;
if (sentenceStart != 0)
sentenceStart++;
- while(sentenceEnd < (int)eos.ref.size() && eos.ref[sentenceEnd] != ProgramParameters::sequenceDelimiter)
+ while(sentenceEnd < eos.refSize() && eos.getRef(sentenceEnd-head) != ProgramParameters::sequenceDelimiter)
sentenceEnd++;
- if (sentenceEnd == (int)eos.ref.size())
+ if (sentenceEnd == eos.refSize())
sentenceEnd--;
auto parts = split(action);
@@ -405,13 +405,13 @@ void Oracle::createDatabase()
{
for (int i = sentenceStart; i <= sentenceEnd; i++)
{
- if (!isNum(govs.ref[i]))
+ if (!isNum(govs.getRef(i-head)))
{
- fprintf(stderr, "ERROR (%s) : govs.ref[%d] = <%s>. Aborting.\n", ERRINFO, i, govs.ref[i].c_str());
+ fprintf(stderr, "ERROR (%s) : govs.ref[%d] = <%s>. Aborting.\n", ERRINFO, i, govs.getRef(i-head).c_str());
exit(1);
}
- int otherGov = i + std::stoi(govs.ref[i]);
+ int otherGov = i + std::stoi(govs.getRef(i-head));
for (int j = 0; j < c.stackSize(); j++)
{
@@ -422,7 +422,7 @@ void Oracle::createDatabase()
}
}
- return eos.ref[stackHead] != ProgramParameters::sequenceDelimiter ? cost : cost+1;
+ return eos.getRef(stackHead-head) != ProgramParameters::sequenceDelimiter ? cost : cost+1;
}
else if (parts[0] == "WRITE" && parts.size() == 4)
{
@@ -430,17 +430,17 @@ void Oracle::createDatabase()
if (object[0] == "b")
{
if (parts[2] == "LABEL")
- return action == "WRITE b.0 LABEL " + c.getTape("LABEL").ref[c.getHead()] || c.getHead() >= (int)c.tapes[0].ref.size()-1 || c.getTape("LABEL").ref[c.getHead()] == "root" ? 0 : 1;
+ return (action == "WRITE b.0 LABEL " + c.getTape("LABEL").getRef(0) || c.endOfTapes() || c.getTape("LABEL").getRef(0) == "root") ? 0 : 1;
else if (parts[2] == "GOV")
- return action == "WRITE b.0 GOV " + c.getTape("GOV").ref[c.getHead()] || c.getHead() >= (int)c.tapes[0].ref.size()-1 ? 0 : 1;
+ return (action == "WRITE b.0 GOV " + c.getTape("GOV").getRef(0) || c.endOfTapes()) ? 0 : 1;
}
else if (object[0] == "s")
{
int index = c.stackGetElem(-1);
if (parts[2] == "LABEL")
- return action == "WRITE s.-1 LABEL " + c.getTape("LABEL").ref[index] || c.getTape("LABEL").ref[index] == "root" ? 0 : 1;
+ return (action == "WRITE s.-1 LABEL " + c.getTape("LABEL").getRef(index-head) || c.getTape("LABEL").getRef(index-head) == "root") ? 0 : 1;
else if (parts[2] == "GOV")
- return action == "WRITE s.-1 GOV " + c.getTape("GOV").ref[index] ? 0 : 1;
+ return (action == "WRITE s.-1 GOV " + c.getTape("GOV").getRef(index-head)) ? 0 : 1;
}
return 1;
@@ -449,21 +449,21 @@ void Oracle::createDatabase()
{
for (int i = head; i <= sentenceEnd; i++)
{
- int otherGov = i + std::stoi(govs.ref[i]);
+ int otherGov = i + std::stoi(govs.getRef(i-head));
if (otherGov == stackHead)
cost++;
}
- return eos.ref[stackHead] != ProgramParameters::sequenceDelimiter ? cost : cost+1;
+ return eos.getRef(stackHead-head) != ProgramParameters::sequenceDelimiter ? cost : cost+1;
}
else if (parts[0] == "LEFT")
{
- if (eos.ref[stackHead] == ProgramParameters::sequenceDelimiter)
+ if (eos.getRef(stackHead-head) == ProgramParameters::sequenceDelimiter)
cost++;
for (int i = head+1; i <= sentenceEnd; i++)
{
- int otherGov = i + std::stoi(govs.ref[i]);
+ int otherGov = i + std::stoi(govs.getRef(i-head));
if (otherGov == stackHead || stackGov == i)
cost++;
}
@@ -471,7 +471,7 @@ void Oracle::createDatabase()
if (stackGov != head)
cost++;
- return parts.size() == 1 || labels.ref[stackHead] == parts[1] ? cost : cost+1;
+ return parts.size() == 1 || labels.getRef(stackHead-head) == parts[1] ? cost : cost+1;
}
else if (parts[0] == "RIGHT")
{
@@ -482,7 +482,7 @@ void Oracle::createDatabase()
if (s == c.stackTop())
continue;
- int otherGov = s + std::stoi(govs.ref[s]);
+ int otherGov = s + std::stoi(govs.getRef(s-head));
if (otherGov == head || headGov == s)
cost++;
}
@@ -491,11 +491,11 @@ void Oracle::createDatabase()
if (headGov == i)
cost++;
- return parts.size() == 1 || labels.ref[head] == parts[1] ? cost : cost+1;
+ return parts.size() == 1 || labels.getRef(0) == parts[1] ? cost : cost+1;
}
else if (parts[0] == ProgramParameters::sequenceDelimiterTape)
{
- return eos.ref[stackHead] == ProgramParameters::sequenceDelimiter ? cost : cost+1;
+ return eos.getRef(stackHead-head) == ProgramParameters::sequenceDelimiter ? cost : cost+1;
}
return cost;
@@ -558,13 +558,13 @@ void Oracle::explainCostOfAction(FILE * output, Config & c, const std::string &
if (object[0] == "b")
{
int index = c.getHead() + std::stoi(object[1]);
- expected = c.getTape(tape).ref[index];
+ expected = c.getTape(tape).getRef(index-c.getHead());
}
else if (object[0] == "s")
{
int stackIndex = std::stoi(object[1]);
int bufferIndex = c.stackGetElem(stackIndex) + c.getHead();
- expected = c.getTape(tape).ref[bufferIndex];
+ expected = c.getTape(tape).getRef(bufferIndex-c.getHead());
}
else
{
@@ -581,25 +581,25 @@ void Oracle::explainCostOfAction(FILE * output, Config & c, const std::string &
int head = c.getHead();
int stackHead = c.stackEmpty() ? 0 : c.stackTop();
- int stackGov = stackHead + std::stoi(govs.ref[stackHead]);
- int headGov = head + std::stoi(govs.ref[head]);
+ int stackGov = stackHead + std::stoi(govs.getRef(stackHead-head));
+ int headGov = head + std::stoi(govs.getRef(0));
int sentenceStart = c.getHead()-1 < 0 ? 0 : c.getHead()-1;
int sentenceEnd = c.getHead();
- while(sentenceStart >= 0 && eos.ref[sentenceStart] != ProgramParameters::sequenceDelimiter)
+ while(sentenceStart >= 0 && eos.getRef(sentenceStart-head) != ProgramParameters::sequenceDelimiter)
sentenceStart--;
if (sentenceStart != 0)
sentenceStart++;
- while(sentenceEnd < (int)eos.ref.size() && eos.ref[sentenceEnd] != ProgramParameters::sequenceDelimiter)
+ while(sentenceEnd < eos.refSize() && eos.getRef(sentenceEnd-head) != ProgramParameters::sequenceDelimiter)
sentenceEnd++;
- if (sentenceEnd == (int)eos.ref.size())
+ if (sentenceEnd == eos.refSize())
sentenceEnd--;
if (parts[0] == "SHIFT")
{
for (int i = sentenceStart; i <= sentenceEnd; i++)
{
- int otherGov = i + std::stoi(govs.ref[i]);
+ int otherGov = i + std::stoi(govs.getRef(i-head));
for (int j = 0; j < c.stackSize(); j++)
{
@@ -608,19 +608,19 @@ void Oracle::explainCostOfAction(FILE * output, Config & c, const std::string &
{
if (otherGov == head)
{
- fprintf(output, "Word on stack %d(%s)\'s governor is the current getHead()\n", s, c.getTape("FORM").ref[s].c_str());
+ fprintf(output, "Word on stack %d(%s)\'s governor is the current getHead()\n", s, c.getTape("FORM").getRef(s-head).c_str());
return;
}
else if (headGov == s)
{
- fprintf(output, "The current getHead()\'s governor is on the stack %d(%s)\n", s, c.getTape("FORM").ref[s].c_str());
+ fprintf(output, "The current getHead()\'s governor is on the stack %d(%s)\n", s, c.getTape("FORM").getRef(s-head).c_str());
return;
}
}
}
}
- if (eos.ref[stackHead] != ProgramParameters::sequenceDelimiter)
+ if (eos.getRef(0) != ProgramParameters::sequenceDelimiter)
{
fprintf(output, "Zero cost\n");
return;
@@ -634,15 +634,15 @@ void Oracle::explainCostOfAction(FILE * output, Config & c, const std::string &
{
for (int i = head; i <= sentenceEnd; i++)
{
- int otherGov = i + std::stoi(govs.ref[i]);
+ int otherGov = i + std::stoi(govs.getRef(i-head));
if (otherGov == stackHead)
{
- fprintf(output, "Stack getHead() is the governor of %d(%s)\n", i, c.getTape("FORM").ref[i].c_str());
+ fprintf(output, "Stack getHead() is the governor of %d(%s)\n", i, c.getTape("FORM").getRef(i-head).c_str());
return;
}
}
- if (eos.ref[stackHead] != ProgramParameters::sequenceDelimiter)
+ if (eos.getRef(stackHead-head) != ProgramParameters::sequenceDelimiter)
{
fprintf(output, "Zero cost\n");
return;
@@ -654,7 +654,7 @@ void Oracle::explainCostOfAction(FILE * output, Config & c, const std::string &
}
else if (parts[0] == "LEFT")
{
- if (parts.size() == 2 && stackGov == head && labels.ref[stackHead] == parts[1])
+ if (parts.size() == 2 && stackGov == head && labels.getRef(stackHead-head) == parts[1])
{
fprintf(output, "Zero cost\n");
return;
@@ -665,23 +665,23 @@ void Oracle::explainCostOfAction(FILE * output, Config & c, const std::string &
return;
}
- if (labels.ref[stackHead] != parts[1])
+ if (labels.getRef(stackHead-head) != parts[1])
{
- fprintf(output, "Stack getHead() label %s mismatch with action label %s\n", labels.ref[stackHead].c_str(), parts[1].c_str());
+ fprintf(output, "Stack getHead() label %s mismatch with action label %s\n", labels.getRef(stackHead-head).c_str(), parts[1].c_str());
return;
}
for (int i = head; i <= sentenceEnd; i++)
{
- int otherGov = i + std::stoi(govs.ref[i]);
+ int otherGov = i + std::stoi(govs.getRef(i-head));
if (otherGov == stackHead)
{
- fprintf(output, "Word %d(%s)\'s governor is the stack getHead()\n", i, c.getTape("FORM").ref[i].c_str());
+ fprintf(output, "Word %d(%s)\'s governor is the stack getHead()\n", i, c.getTape("FORM").getRef(i-head).c_str());
return;
}
else if (stackGov == i)
{
- fprintf(output, "Stack getHead()\'s governor is the word %d(%s)\n", i, c.getTape("FORM").ref[i].c_str());
+ fprintf(output, "Stack getHead()\'s governor is the word %d(%s)\n", i, c.getTape("FORM").getRef(i-head).c_str());
return;
}
}
@@ -703,15 +703,15 @@ void Oracle::explainCostOfAction(FILE * output, Config & c, const std::string &
if (s == c.stackTop())
continue;
- int otherGov = s + std::stoi(govs.ref[s]);
+ int otherGov = s + std::stoi(govs.getRef(s-head));
if (otherGov == head)
{
- fprintf(output, "The governor of %d(%s) in the stack, is the current head\n", s, c.getTape("FORM").ref[s].c_str());
+ fprintf(output, "The governor of %d(%s) in the stack, is the current head\n", s, c.getTape("FORM").getRef(s-head).c_str());
return;
}
else if (headGov == s)
{
- fprintf(output, "The current head's governor is the stack element %d(%s)\n", s, c.getTape("FORM").ref[s].c_str());
+ fprintf(output, "The current head's governor is the stack element %d(%s)\n", s, c.getTape("FORM").getRef(s-head).c_str());
return;
}
}
@@ -719,7 +719,7 @@ void Oracle::explainCostOfAction(FILE * output, Config & c, const std::string &
for (int i = head; i <= sentenceEnd; i++)
if (headGov == i)
{
- fprintf(output, "The current head's governor is the future word %d(%s)\n", i, c.getTape("FORM").ref[i].c_str());
+ fprintf(output, "The current head's governor is the future word %d(%s)\n", i, c.getTape("FORM").getRef(i-head).c_str());
return;
}
@@ -729,20 +729,20 @@ void Oracle::explainCostOfAction(FILE * output, Config & c, const std::string &
return;
}
- if (labels.ref[head] == parts[1])
+ if (labels.getRef(0) == parts[1])
{
fprintf(output, "Zero cost\n");
return;
}
else
{
- fprintf(output, "Current head's label %s mismatch action label %s\n", labels.ref[head].c_str(), parts[1].c_str());
+ fprintf(output, "Current head's label %s mismatch action label %s\n", labels.getRef(0).c_str(), parts[1].c_str());
return;
}
}
else if (parts[0] == ProgramParameters::sequenceDelimiterTape)
{
- if (eos.ref[stackHead] == ProgramParameters::sequenceDelimiter)
+ if (eos.getRef(stackHead-head) == ProgramParameters::sequenceDelimiter)
{
fprintf(output, "Zero cost\n");
return;