proto/Source/init.cpp

Go to the documentation of this file.

#include <fstream>

// Settings
#include "config.h"
// Static members initialization
#include "evaluation.h"
// Self header file
#include "init.h"
// Static members initialization
#include "ngram.h"
// Hash function tables check.
#include "ngrams.h"
// Errors, warnings
#include "notifier.h"
// Global vars "set up"
#include "persistent.h"

namespace ace {

/*  Interface item implementation - for info see description in header file.
 */
bool read_and_check_input_datafiles(filenames_t& files_to_process, size_t& files_total_size) {
    
    std::ifstream file_list(settings::input_file);
    
    if ( !file_list ) {
        notifier::error.fatal_error("Cannot open the file: " + std::string(settings::input_file) + "!");
        return false;
    }
    
    bool status = true;
    
    std::ifstream tmp;

    std::string dir = (settings::dir == 0) ? std::string() : std::string(settings::dir);
    
    while ( file_list ) {
        std::string str;
        std::getline(file_list, str);
        // Ignore empty lines.
        if ( !str.empty() ) {
            // Check, if file can be opened for reading.
            tmp.open((dir + str).c_str());
            if ( !tmp ) {
                // What a pity...
                notifier::error.fatal_error("Input datafile : " + dir + str + " cannot be opened for reading!");
                status = false;
            } else {
                // Count file length.
                tmp.seekg(0, std::ios::end);
                std::streampos pos = tmp.tellg();
                tmp.seekg(0, std::ios::beg);
                if ( pos > 0 ) { // tellg() returns -1 in case of failure.
                    files_total_size += pos;
                }
                // Close and store.
                tmp.close();
                files_to_process.push_back(dir + str);
            }
        }
    }
    
    file_list.close();

    if ( files_to_process.empty() ) {
        notifier::error.fatal_error("Input data files list is empty!");
        return false;
    }

    // Otherwise.
    return status;
}
    
/*  Interface item implementation - for info see description in header file.
 */
bool startup_init(void) {
    
    NGram::init();
    
    NGramToken::init();

    if ( !NGramHashCounter::check(settings::n) ) {
        // TODO: This should be compile time error.
        notifier::error.fatal_error("Predefined # counter tables are to small!");
    }

    EvaluationTables::init();
    
    persistent::ngrams.init(constants::ngram_within_block);

    // TODO: Literals -m!
    
    // Read morphological filter rules?
    if ( settings::morphologic_filter_file != 0 ) {
        // Morphological filter is set on.
        if ( settings::morphologic_tag_mask == 0 ) {
            // No tag info to use with filter!
            notifier::error.warning("Extraction of morphologic information is not set on (parameter -m not specified). Morphologic filter will not be applied...");
        }
        // Otherwise read the rules.
        else if ( !persistent::morphologic_filter.read_rules(settings::morphologic_filter_file, settings::n) ) {
            // Error, rules cannot be read.
            notifier::error.fatal_error("Cannot read rules from file: " + std::string(settings::morphologic_filter_file) + "!");
            return false;
        }
    }

    if ( settings::morphologic_filter_stats_file != 0 ) {
        // 
        if ( settings::morphologic_filter_file == 0 ) {
            // Sorry man, but stats just cannot be sucked out from the toe...:)
            notifier::error.warning("Morphologic filter is not turn on (parameter -f not specified). Morphologic filter stats cannot be retrieved...");
        }
    }
    
    // Read context filter rules?
    if ( settings::context_filter_file != 0 ) {
                
        if ( settings::context_output_file == 0 ) {
            // No context output file specified!
            notifier::error.warning("Context output file is not specified (parameter -c). Morphologic filter for N-gram context will not be applied...");
        } else if ( settings::morphologic_tag_mask == 0 ) {
            // No tag info to use with filter!
            notifier::error.warning("Extraction of morphologic information is not set on (parameter -m is not specified). Morphologic filter for N-gram context will not be applied...");
        }
        // Otherwise read the rules.
        else if ( !persistent::context_filter.read_rules(settings::context_filter_file) ) {
            // Error, rules cannot be read.
            notifier::error.fatal_error("Cannot read rules from file: " + std::string(settings::context_filter_file) + "!");
            return false;
        }
    }
    
    // Some statistics can be obtained only for bigrams...
    if ( settings::n != 2 ) {
        if ( settings::thresholds.mutual_information > 0.0 ) {
            notifier::error.warning("Mutual information evaluation only available for N equal 2. Threshold ignored...");
        }
        if ( settings::thresholds.pearsons_coefficient > 0.0 ) {
            notifier::error.warning("Pearson's coefficient evaluation only available for N equal 2. Threshold ignored...");
        }
        if ( settings::thresholds.t_test > 0.0 ) {
            notifier::error.warning("t test evaluation only available for N equal 2. Threshold ignored...");
        }
        if ( settings::thresholds.z_score > 0.0 ) {
            notifier::error.warning("z score evaluation only available for N equal 2. Threshold ignored...");
        }
    }

    return true;
}

} // namespace ace

---