diff options
Diffstat (limited to 'utils')
-rw-r--r-- | utils/Makefile.am | 2 | ||||
-rw-r--r-- | utils/alignment_io.cc (renamed from utils/alignment_pharaoh.cc) | 36 | ||||
-rw-r--r-- | utils/alignment_io.h | 42 | ||||
-rw-r--r-- | utils/alignment_pharaoh.h | 14 | ||||
-rw-r--r-- | utils/atools.cc | 11 | ||||
-rw-r--r-- | utils/ccrp.h | 10 | ||||
-rw-r--r-- | utils/mfcr.h | 4 |
7 files changed, 89 insertions, 30 deletions
diff --git a/utils/Makefile.am b/utils/Makefile.am index 3ea21835..2fc6ae21 100644 --- a/utils/Makefile.am +++ b/utils/Makefile.am @@ -23,7 +23,7 @@ atools_SOURCES = atools.cc noinst_LIBRARIES = libutils.a libutils_a_SOURCES = \ - alignment_pharaoh.cc \ + alignment_io.cc \ b64tools.cc \ corpus_tools.cc \ dict.cc \ diff --git a/utils/alignment_pharaoh.cc b/utils/alignment_io.cc index 890ff565..1d923f7f 100644 --- a/utils/alignment_pharaoh.cc +++ b/utils/alignment_io.cc @@ -1,12 +1,10 @@ -#include "utils/alignment_pharaoh.h" - -#include <set> +#include "utils/alignment_io.h" using namespace std; static bool is_digit(char x) { return x >= '0' && x <= '9'; } -boost::shared_ptr<Array2D<bool> > AlignmentPharaoh::ReadPharaohAlignmentGrid(const string& al) { +boost::shared_ptr<Array2D<bool> > AlignmentIO::ReadPharaohAlignmentGrid(const string& al) { int max_x = 0; int max_y = 0; int i = 0; @@ -64,14 +62,36 @@ boost::shared_ptr<Array2D<bool> > AlignmentPharaoh::ReadPharaohAlignmentGrid(con return grid; } -void AlignmentPharaoh::SerializePharaohFormat(const Array2D<bool>& alignment, ostream* out) { +void AlignmentIO::SerializePharaohFormat(const Array2D<bool>& alignment, ostream* o) { + ostream& out = *o; bool need_space = false; for (int i = 0; i < alignment.width(); ++i) for (int j = 0; j < alignment.height(); ++j) if (alignment(i,j)) { - if (need_space) (*out) << ' '; else need_space = true; - (*out) << i << '-' << j; + if (need_space) out << ' '; else need_space = true; + out << i << '-' << j; } - (*out) << endl; + out << endl; +} + +void AlignmentIO::SerializeTypedAlignment(const Array2D<AlignmentType>& alignment, ostream* o) { + ostream& out = *o; + bool need_space = false; + for (int i = 0; i < alignment.width(); ++i) + for (int j = 0; j < alignment.height(); ++j) { + const AlignmentType& aij = alignment(i,j); + if (aij != kNONE) { + if (need_space) out << ' '; else need_space = true; + if (aij == kTRANSLATION) {} + else if (aij == kTRANSLITERATION) { + out << 'T' << ':'; + } else { + cerr << "\nUnexpected alignment point type: " << static_cast<int>(aij) << endl; + abort(); + } + out << i << '-' << j; + } + } + out << endl; } diff --git a/utils/alignment_io.h b/utils/alignment_io.h new file mode 100644 index 00000000..36bcecd7 --- /dev/null +++ b/utils/alignment_io.h @@ -0,0 +1,42 @@ +#ifndef _ALIGNMENT_IO_H_ +#define _ALIGNMENT_IO_H_ + +#include <string> +#include <iostream> +#include <boost/shared_ptr.hpp> +#include "array2d.h" + +struct AlignmentIO { + enum AlignmentType { kNONE = 0, kTRANSLATION = 1, kTRANSLITERATION = 2 }; + + static boost::shared_ptr<Array2D<bool> > ReadPharaohAlignmentGrid(const std::string& al); + static void SerializePharaohFormat(const Array2D<bool>& alignment, std::ostream* out); + static void SerializeTypedAlignment(const Array2D<AlignmentType>& alignment, std::ostream* out); +}; + +inline std::ostream& operator<<(std::ostream& os, const Array2D<AlignmentIO::AlignmentType>& m) { + os << ' '; + for (int j=0; j<m.height(); ++j) + os << (j%10); + os << "\n"; + for (int i=0; i<m.width(); ++i) { + os << (i%10); + for (int j=0; j<m.height(); ++j) { + switch (m(i,j)) { + case AlignmentIO::kNONE: os << '.'; break; + case AlignmentIO::kTRANSLATION: os << '*'; break; + case AlignmentIO::kTRANSLITERATION: os << '#'; break; + default: os << '?'; break; + } + } + os << (i%10) << "\n"; + } + os << ' '; + for (int j=0; j<m.height(); ++j) + os << (j%10); + os << "\n"; + return os; +} + + +#endif diff --git a/utils/alignment_pharaoh.h b/utils/alignment_pharaoh.h deleted file mode 100644 index d111c8bf..00000000 --- a/utils/alignment_pharaoh.h +++ /dev/null @@ -1,14 +0,0 @@ -#ifndef _PHARAOH_ALIGNMENT_H_ -#define _PHARAOH_ALIGNMENT_H_ - -#include <string> -#include <iostream> -#include <boost/shared_ptr.hpp> -#include "array2d.h" - -struct AlignmentPharaoh { - static boost::shared_ptr<Array2D<bool> > ReadPharaohAlignmentGrid(const std::string& al); - static void SerializePharaohFormat(const Array2D<bool>& alignment, std::ostream* out); -}; - -#endif diff --git a/utils/atools.cc b/utils/atools.cc index c0a91731..bce7822e 100644 --- a/utils/atools.cc +++ b/utils/atools.cc @@ -8,11 +8,10 @@ #include <boost/shared_ptr.hpp> #include "filelib.h" -#include "alignment_pharaoh.h" +#include "alignment_io.h" namespace po = boost::program_options; using namespace std; -using boost::shared_ptr; struct Command { virtual ~Command() {} @@ -348,10 +347,10 @@ int main(int argc, char **argv) { } } if (line1.empty() && !*in1) break; - shared_ptr<Array2D<bool> > out(new Array2D<bool>); - shared_ptr<Array2D<bool> > a1 = AlignmentPharaoh::ReadPharaohAlignmentGrid(line1); + boost::shared_ptr<Array2D<bool> > out(new Array2D<bool>); + boost::shared_ptr<Array2D<bool> > a1 = AlignmentIO::ReadPharaohAlignmentGrid(line1); if (in2) { - shared_ptr<Array2D<bool> > a2 = AlignmentPharaoh::ReadPharaohAlignmentGrid(line2); + boost::shared_ptr<Array2D<bool> > a2 = AlignmentIO::ReadPharaohAlignmentGrid(line2); cmd.Apply(*a1, *a2, out.get()); } else { Array2D<bool> dummy; @@ -359,7 +358,7 @@ int main(int argc, char **argv) { } if (cmd.Result() == 1) { - AlignmentPharaoh::SerializePharaohFormat(*out, &cout); + AlignmentIO::SerializePharaohFormat(*out, &cout); } } if (cmd.Result() == 2) diff --git a/utils/ccrp.h b/utils/ccrp.h index 4a8b80e7..8635b422 100644 --- a/utils/ccrp.h +++ b/utils/ccrp.h @@ -55,6 +55,10 @@ class CCRP { double discount() const { return discount_; } double strength() const { return strength_; } + void set_hyperparameters(double d, double s) { + discount_ = d; strength_ = s; + check_hyperparameters(); + } void set_discount(double d) { discount_ = d; check_hyperparameters(); } void set_strength(double a) { strength_ = a; check_hyperparameters(); } @@ -93,8 +97,10 @@ class CCRP { } // returns +1 or 0 indicating whether a new table was opened + // p = probability with which the particular table was selected + // excluding p0 template <typename T> - int increment(const Dish& dish, const T& p0, MT19937* rng) { + int increment(const Dish& dish, const T& p0, MT19937* rng, T* p = NULL) { DishLocations& loc = dish_locs_[dish]; bool share_table = false; if (loc.total_dish_count_) { @@ -108,6 +114,7 @@ class CCRP { ti != loc.table_counts_.end(); ++ti) { r -= (*ti - discount_); if (r <= 0.0) { + if (p) { *p = T(*ti - discount_) / T(strength_ + num_customers_); } ++(*ti); break; } @@ -119,6 +126,7 @@ class CCRP { } } else { loc.table_counts_.push_back(1u); + if (p) { *p = T(strength_ + discount_ * num_tables_) / T(strength_ + num_customers_); } ++num_tables_; } ++loc.total_dish_count_; diff --git a/utils/mfcr.h b/utils/mfcr.h index 886f01ef..4aacb567 100644 --- a/utils/mfcr.h +++ b/utils/mfcr.h @@ -73,6 +73,10 @@ class MFCR { double discount() const { return discount_; } double strength() const { return strength_; } + void set_hyperparameters(double d, double s) { + discount_ = d; strength_ = s; + check_hyperparameters(); + } void set_discount(double d) { discount_ = d; check_hyperparameters(); } void set_strength(double a) { strength_ = a; check_hyperparameters(); } |