selfile.cpp File Reference

#include <fstream>
#include <stdio.h>
#include <string.h>
#include <algorithm>
#include <core/xmipp_funcs.h>
#include "selfile.h"
#include <core/xmipp_image.h>

Include dependency graph for selfile.cpp:

Go to the source code of this file.

Functions
bool	operator< (const SelLine &l1, const SelLine &l2)
std::ostream &	operator<< (std::ostream &o, const SelLine &SFL)
std::istream &	operator>> (std::istream &o, SelLine &SFL)
std::ostream &	operator<< (std::ostream &o, const SelFile &SF)
std::vector< SelLine >::iterator	find (std::vector< SelLine > &text, const std::string &img_name)
SelFile	compare (SelFile &SF1, SelFile &SF2, const int mode)

Function Documentation

compare()

SelFile compare	(	SelFile &	SF1,
		SelFile &	SF2,
		const int	mode
	)

Compare two selection files.

The result is another selection file. At the beginning of it there is information about the number of active and discarded images on both input selection files, about the number of matching files (a file is said to match if it is active in both selection files), the number of active files which are only in the first selection file, and the number of active files which are only in the second. Then goes the list of matching files, the list of files only in SF1 and the list of files only in SF2. There are comments enough to know where things start and finish, and what the numbers are at the beginning. If a file is active in a file and discarded in the other, then it is said to match and it is kept as active, a preceding comment warns of this situation.

if mode<0, output file will contain all information if mode=0, output will be selfile with overlapping images if mode=1, output will be selfile with images only in file 1 if mode=2, output will be selfile with images only in file 2

sel3 = compare(sel1, sel2);

Definition at line 872 of file selfile.cpp.

{
    std::vector<SelLine>     only_in_SF1;
    std::vector<SelLine>     only_in_SF2;
    std::vector<SelLine>     in_both;
    SelFile           result;
    SelLine           temp;
    int               SF1_discarded = 0, SF2_discarded = 0;
    int maxLen = 15;
    char              str[maxLen + 1]; // + terminating null character

    // Search in File 1
    std::vector<SelLine>::iterator current = SF1.text_line.begin();
    std::vector<SelLine>::iterator last    = SF1.text_line.end();
    std::vector<SelLine>::iterator last_SF = SF2.text_line.end();
    std::vector<SelLine>::iterator found;

    while (current != last)
    {
        // Skip if not active
        if ((*current).line_type != SelLine::DATALINE)
        {
            current++;
            continue;
        }
        if ((*current).label == SelLine::DISCARDED)
        {
            SF1_discarded++;
            current++;
            continue;
        }

        // Try to find this archive into Sel File 2
        found = SF2.find((*current).text);
        if (found == last_SF)
            only_in_SF1.push_back(*current);
        else
            if ((*found).label == SelLine::DISCARDED)
                only_in_SF1.push_back(*current);
            else
                in_both.push_back(*current);
        current++;
    }

    // Search in File 2
    current = SF2.text_line.begin();
    last    = SF2.text_line.end();

    while (current != last)
    {
        // Skip if not active
        if ((*current).line_type != SelLine::DATALINE)
        {
            current++;
            continue;
        }
        if ((*current).label == SelLine::DISCARDED)
        {
            SF2_discarded++;
            current++;
            continue;
        }

        // Try to find this archive into Sel File 2
        found = find(in_both, (*current).text);
        if (found != in_both.end())
        {
            current++;
            continue;
        }
        only_in_SF2.push_back(*current);
        current++;
    }

    // Write Statistics
    if (mode < 0)
    {
    temp.line_type = SelLine::COMMENT;
    temp.label = SelLine::DISCARDED;
    temp.text = "# Statistics of comparison";
    result.text_line.push_back(temp);
    temp.text = "# -------------------------------------------------------------";
    result.text_line.push_back(temp);
    snprintf(str, maxLen, "%6d", SF1.no_imgs);
    temp.text = "# File 1: " + SF1.fn_sel + "(VALID: " + str;
    snprintf(str, maxLen, "%6d", SF1_discarded);
    temp.text += (std::string) " DISCARDED: " + str + ")";
    result.text_line.push_back(temp);
    snprintf(str, maxLen, "%6d", SF2.no_imgs);
    temp.text = "# File 2: " + SF2.fn_sel + "(VALID: " + str;
    snprintf(str, maxLen, "%6d", SF2_discarded);
    temp.text += (std::string) " DISCARDED: " + str + ")";
    result.text_line.push_back(temp);
    temp.text = "";
    result.text_line.push_back(temp);
    snprintf(str, maxLen, "%6lu", (unsigned long int)in_both.size());
    temp.text = (std::string)"# Matching Files: " + str;
    result.text_line.push_back(temp);
    snprintf(str, maxLen, "%6lu", (unsigned long int)only_in_SF1.size());
    temp.text = (std::string)"# Only in file 1: " + str;
    result.text_line.push_back(temp);
    snprintf(str, maxLen, "%6lu", (unsigned long int)only_in_SF2.size());
    temp.text = (std::string)"# Only in file 2: " + str;
    result.text_line.push_back(temp);
    temp.text = "# -------------------------------------------------------------";
    result.text_line.push_back(temp);

    // Write files in both
    temp.text = "";
    result.text_line.push_back(temp);
    temp.text = "# Files in both .sel files";
    result.text_line.push_back(temp);
    }
    if (mode<0 || mode==0) 
    {
    current = in_both.begin();
    last    = in_both.end();
    while (current != last)
        result.text_line.push_back(*current++);
    }

    if (mode<0)
    {
    // Write files only in Sel File 1
    temp.text = "";
    result.text_line.push_back(temp);
    temp.text = "# Files only in the first file";
    result.text_line.push_back(temp);
    }
    if (mode<0 || mode==1) 
    {
    current = only_in_SF1.begin();
    last    = only_in_SF1.end();
    while (current != last)
        result.text_line.push_back(*current++);
    }

    if (mode<0)
    {
    // Write files only in Sel File 2
    temp.text = "";
    result.text_line.push_back(temp);
    temp.text = "# Files only in the second file";
    result.text_line.push_back(temp);
    }
    if (mode<0 || mode==2) 
    {
    current = only_in_SF2.begin();
    last    = only_in_SF2.end();
    while (current != last)
        result.text_line.push_back(*current++);
    }
    // Adjust the remaining fields
    if (mode<0) 
    result.no_imgs = in_both.size() + only_in_SF1.size() + only_in_SF2.size();
    else if (mode==0) 
    result.no_imgs = in_both.size();
    else if (mode==1)
    result.no_imgs = only_in_SF1.size();
    else if (mode==2)
    result.no_imgs = only_in_SF2.size();
    result.current_line = result.text_line.begin();

    return result;
}

find()

std::vector<SelLine>::iterator find	(	std::vector< SelLine > &	text,
		const std::string &	img_name
	)

Definition at line 553 of file selfile.cpp.

{
    std::vector<SelLine>::iterator current = text.begin();
    std::vector<SelLine>::iterator last    = text.end();

    while (current != last)
    {
        if ((*current).line_type == SelLine::DATALINE &&
            (*current).text == img_name)
            return current;
        current++;
    }
    return current;
}

operator<()

bool operator<	(	const SelLine &	l1,
		const SelLine &	l2
	)

Lesser than. l1 is lesser than l2 if line_type(l1)<line_type(l2) or if they are equal if the text of l1 is lesser than the text of l2. The order of the line types are NOT_CONSIDERED, NOT_ASSIGNED, DATALINE, COMMENT

Definition at line 49 of file selfile.cpp.

{
    if (l1.line_type < l2.line_type)
        return 1;
    else if (l1.line_type > l2.line_type)
        return 0;
    else
        return l1.text < l2.text;
}

operator<<() [1/2]

std::ostream& operator<<	(	std::ostream &	o,
		const SelLine &	SFL
	)

Show a SelLine.

Definition at line 59 of file selfile.cpp.

{
    switch (SFL.line_type)
    {
    case SelLine::DATALINE:
                    o << SFL.text << " " << SFL.label << std::endl;
        break;
    case SelLine::COMMENT:
                    o << SFL.text << std::endl;
        break;
    default:
        break;
    }
    return o;
}

operator<<() [2/2]

std::ostream& operator<<	(	std::ostream &	o,
		const SelFile &	SF
	)

Show a selection file.

Shows all the lines either they are comments, active images or discarded images. A new line is printed at the end.

std::cout << sel;

Definition at line 145 of file selfile.cpp.

{
    std::vector<SelLine>::const_iterator current = SF.text_line.begin();
    std::vector<SelLine>::const_iterator last    = SF.text_line.end();
    while (current != last)
    {
        o << *current;
        current++;
    }
    return o;
}

operator>>()

std::istream& operator>>	(	std::istream &	o,
		SelLine &	SFL
	)

Read a Selection Line.

An exception is thrown if the line doesn't meet the SelFile specifications.

Definition at line 75 of file selfile.cpp.

{
    std::string line;
    char     img_name[1024];
    int      no_elements_read;
    int      label;

    // Get line
    getline(o, line);

    // Initialise target
    SFL.line_type = SelLine::NOT_ASSIGNED;
    SFL.text = "";
    SFL.label = SelLine::DISCARDED;
    if (line.length() == 0)
        return o;

    // Check if comment or empty line
    if (line[0] == '#' || line[0] == '\0' || line[0] == ';')
    {
        line[line.length()-1] = '\0';
        SFL.line_type = SelLine::COMMENT;
        SFL.text = line;

        // Check if a true "filename label" line
    }
    else
    {
        no_elements_read = sscanf(line.c_str(), "%s %d", img_name, &label);
        // *** THE SSCANF CAN BE REPLACED BY A STRING I/O OPERATION
        if (no_elements_read == 2)
        {
            SFL.line_type = SelLine::DATALINE;
            SFL.text = img_name;
            SFL.label = (label >= 0) ? SelLine::ACTIVE : SelLine::DISCARDED;
            SFL.number = label;
        }
        else
            REPORT_ERROR(ERR_SELFILE, "Format error when reading Selection line");
    }
    return o;
}