proto/Source/config.h

Go to the documentation of this file.

// TODO: Consider putting implementation constants (such as bits_per_char) somewhere else.

#ifndef _CONFIG_H
#define _CONFIG_H

#include <cstddef>

// "32-bit or 64-bit?", that is the question, my friend.
#if defined(_M_X64) || defined(_M_IA64) || defined(__x86_64__) || defined(__IA64__)
// _M_X64, _M_IA64 - Visual Studio
// __x86_64__, __IA64__ - GCC
#define _64_BIT

#elif defined(_M_IX86) || defined(__i386__)
// _M_IX86 - Visual Studio
// __i386__ - GCC
#define _32_BIT
#else
#error "CPU word size not detected!"
#endif

// Surface or dependency mode?
#if !defined(_DEPENDENCY_MODE) && !defined(_SURFACE_MODE)
#define _DEPENDENCY_MODE // Default mode.
#elif defined(_DEPENDENCY_MODE) && defined(_SURFACE_MODE)
#undef _SURFACE_MODE // Only one mode allowed.
#endif


namespace ace {

namespace constants {

extern const size_t max_grammity; // If you change this, you should probably change typedef for ngram_type_t
// and static-initialized tables used to hashing (see ngrams.h and ngrams.cpp)!

extern const size_t bits_per_char;
const size_t max_string_length = 65535; // Not declared as extern - must be visible for other compilation units.
extern const size_t max_collocation_window_size;
extern const size_t ngram_within_block;
extern const size_t default_precision;
const char null_char = '\0';
const char eol = '\n';
extern const size_t max_sentence_length;
extern const char significant_position_mark;
extern const char insignificant_position_mark;
extern const size_t sentences_radius_max_size;
extern const size_t words_radius_max_size;

} // namespace constants



typedef unsigned short word_order_t;
typedef unsigned int string_index_t;
typedef unsigned char ngram_size_t;
typedef unsigned char dependency_index_t;
typedef unsigned char tag_index_t;
typedef unsigned char ngram_type_t;

class TaggedLemma {
    union {
        string_index_t _lemma;
        tag_index_t _bits[sizeof(string_index_t) / sizeof(tag_index_t)];
    };
    static const size_t _last_bit_index;
    static const string_index_t _lemma_bits_mask;
public:
    TaggedLemma(string_index_t lemma, tag_index_t tag): _lemma(lemma) {
        _bits[_last_bit_index] = tag;
    }
    inline string_index_t lemma(void) const { return _lemma & _lemma_bits_mask; }
    inline tag_index_t tag(void) const { return _bits[_last_bit_index]; }
};

struct ContextWindow {
    enum Mode {
        SENTENCE_MODE,
        WORD_MODE
    } mode;
    size_t radius;
    ContextWindow(Mode m = SENTENCE_MODE, size_t r = 0): mode(m), radius(r) {}
    inline bool check(void) { 
        return ((mode == SENTENCE_MODE) && (radius <= constants::sentences_radius_max_size)) || ((mode == WORD_MODE) && (1 <= radius) && (radius <= constants::words_radius_max_size));
    }
};

struct Sieves {
    double expected_frequency;
    size_t frequency;
    Sieves(void): expected_frequency(0.0), frequency(0) {}
};

struct Thresholds {
    double chi_square_test;
    double log_likelihood_ratio;
    double mutual_information;
    double pearsons_coefficient;
    double t_test;
    double z_score;
    Thresholds(void): chi_square_test(0.0), log_likelihood_ratio(0.0), mutual_information(0.0), pearsons_coefficient(0.0), t_test(0.0), z_score(0.0) {}
};



namespace settings {
    extern size_t bucket_size;
#ifdef _SURFACE_MODE

    extern size_t collocation_window_size;
#endif

    extern ContextWindow context_window;
    extern const char *context_filter_file;
    extern const char *context_output_file;
//  TODO: Remove?
//  extern const char *context_tag_mask;
    extern size_t crop_ratio;
    extern const char *dir;
    extern const char *morphologic_filter_file;
    extern const char *morphologic_filter_stats_file;
    extern const char *input_file;
    extern ngram_size_t n;
    extern const char *output_file;
    extern size_t precision;
    extern bool sort;
    extern const char *stats_file;
    extern Sieves sieves;
    extern Thresholds thresholds;
    extern bool all_thresholds_together;
    extern const char *morphologic_tag_mask;
    extern bool verbose;

    bool check_mask(const char *mask);
    bool context_tracing_on(void);
    bool correct(void);
    bool ready(void);

} // namespace settings

} // namespace ace

#endif

---