clear;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% Paramenters Section
% Group folders contain sample subfolers in which subclass-specific XIC is named as consecutive numbers as a form of text file
dirName = '.\Late stages';
% Set the start point of each subclass
dividePoint = [1,22,31]; 
% Set the export path
statFile = '.\stat_result_.txt';
% Annotation
note = {'IgG 1', 'IgG 4', 'IgG 2'};

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
if dirName(end) ~= '\'
    dirName = [dirName, '\'];
end
grpNum = length(dividePoint);
if length(note) ~= length(dividePoint)
    fprintf('The length of note should be the same as dividePoint.\n');
    return
end
dirList = dir(dirName);
pattern = '^(\D+)(\d+)'; 
% Fold selection '^(\D+)(\d+)'
tokens_ = regexp({dirList.name}, pattern, 'tokens') ; 
% Keep the part of letter uniform
strPart = cell(1,length(tokens_));
for d = 2:length(tokens_)
    if ~isempty(tokens_{1,d})
        stdStrPart = tokens_{1,d}{1,1}{1,1};
        break;
    end
end
flag = 0;
for dcnt = (d+1):length(tokens_)
    if stdStrPart ~= tokens_{1,dcnt}{1,1}{1,1}
        fprintf('The folder name in %s is not the same as the %s.\n', dirList(dcnt).name, tokens_{1,dcnt}{1,1}{1,1});
        fprintf('The string part of subfolders'' name should be the same.\n');
        flag = 1;
    end
end
if flag == 1
	flag = 0;
	return; 
end
% Sorted by number
numPart = -1.23*ones(1,length(tokens_));
numStrPart = cell(1,length(tokens_));
for dcnt = 1:length(tokens_)
    if ~isempty(tokens_{1,dcnt})
        numStrPart{dcnt} = tokens_{1,dcnt}{1,1}{1,2};
        numPart(dcnt) = str2double(tokens_{1,dcnt}{1,1}{1,2});
    end
end
numStrPart(numPart==-1.23)=[];
numPart(numPart==-1.23)=[]; % This two lines can't be reverse
[dirNameNum, dirNameIdx] = sort(numPart, 2,'ascend'); 
numStrPart = numStrPart(dirNameIdx);
% Number of valid folder
dirNum = length(dirNameNum); 

% The file name and amount of first subfold is set as the standard for variable initialization 
dcnt = 1;
stdFolderName = [stdStrPart, numStrPart{dcnt}];
subDirName = [dirName, stdFolderName, '\'];
stdFileList = dir([subDirName, '*.txt']);
stdFileNum = length(stdFileList); % Collecting the amout of text files in the first subfold

flag = 0;
stdFileNameList = cell(1, stdFileNum);
for i = 1:stdFileNum
    stdFileNameList{1, i} = [num2str(i), '.txt'];
end
stdFileNameList = sort(stdFileNameList);
% File names must be consecutive numbers
if ~isequal(stdFileNameList, {stdFileList.name})
    fprintf('The file name in %s is not consist of continuous numbers.\n', stdFolderName);
    flag = 1;
    return
end
fprintf('Prepare to use the file names in %s as the standard names.\n', stdFolderName);


for dcnt = 2:dirNum
    iterDirName = [stdStrPart, numStrPart{dcnt}];
    subDirName = [dirName, iterDirName, '\'];
    fileList = dir([subDirName, '*.txt']);
    fileNum = length(fileList); 
    if fileNum ~= stdFileNum
        fprintf('The file number in %s is not the same as the %s.\n', iterDirName, stdFolderName);
        flag = 1;
        continue;
    end
    if ~isequal({fileList.name}, {stdFileList.name})
        fprintf('The file names in %s is not the same as the %s.\n', iterDirName, stdFolderName);
        flag = 1;
        continue;
    end
end
if flag == 1
	flag = 0;
	return; 
end

% Building the format reading the string of txt files
readFormat = '%f';
% Five columns in a txt file is the default
colsNum = 5;
rowsNum = 1;
for cnt = 1:colsNum-1
    readFormat = strcat(readFormat, '\t%f');
end
% sumStats = zeros(dirNum, stdFileNum); % Data storage


% Start reading data
dividePoint = [dividePoint,stdFileNum+1]; 
headSet = dividePoint(1:end-1);
footSet = dividePoint(2:end);
memSpace = footSet - headSet;
summary = cell(grpNum, dirNum);
records = cell(grpNum, dirNum);
for dcnt = 1:dirNum
    subDirName = [dirName, stdStrPart, numStrPart{dcnt}, '\'];
    %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
    for gcnt = 1:numel(dividePoint)-1
        % Obtaining the break point for each subclass
        timeList = zeros(memSpace(gcnt),1);
        areaList = zeros(memSpace(gcnt),1);
        for mcnt = 1:memSpace(gcnt)
            fname = sprintf('%d.txt',dividePoint(gcnt)+mcnt-1);
            fullName = fullfile(subDirName,fname);
            fid = fopen(fullName, 'r');
            if fid == -1
                fsprintf('Cannot open the file %s.\n',fullName);
                return;
            end
            inMat = textscan(fid,readFormat,'headerlines',1);
            % Read the file
            time = inMat{1,1};
            area = inMat{1,2};
            fclose(fid);
            if size(time ,1)~=rowsNum
                fprintf('The data format of time in %s is incorrect.\n',fullName);
                return;
            end
            if size(area ,1)~=rowsNum
                fprintf('The data format of area in %s is incorrect.\n',fullName);
                return;
            end
            timeList(mcnt) = time;
            areaList(mcnt) = area;
        end
        areaSum = sum(areaList);
        proportion = areaList / areaSum * 100;
        records{gcnt, dcnt} = [timeList, areaList, proportion];
        summary{gcnt, dcnt}= proportion;
    end
end

% Data exported in a standard form
basicDist = '\t\t\t\t';
dcnt = 1;
header = [stdStrPart, numStrPart{dcnt}];
exNote = note;
for i = 1:length(note)
    exNote{i} = [note{i}, basicDist];
end
% Processing record part
dataFormat = '%f\t%f\t%f\t\t';
for dcnt = 2:dirNum
    subDirName = [stdStrPart, numStrPart{dcnt}];
    header = strcat(header, basicDist, subDirName);
    for i = 1:length(note)
        exNote{i} = [exNote{i}, note{i}, basicDist];
    end
    dataFormat = strcat(dataFormat, '%f\t%f\t%f\t\t');
end
% SummaryÁÐ
header = [header, basicDist, 'Summary'];
for dcnt = 1:dirNum
    subDirName = [stdStrPart, numStrPart{dcnt}];
    header = strcat(header, '\t', subDirName);
    dataFormat = strcat(dataFormat, '\t%f');
end
header = [header, '\n'];
dataFormat = [dataFormat, '\n'];
for i = 1:length(note)
    exNote{i} = [exNote{i}, '\t', note{i}, '\t\t\n'];
end


% Writing the data
fID = fopen(statFile,'w');
if fID == -1
    fprintf('Cannot open the file %s.\n',statFile);
    return;
end
fprintf(fID,header);
for gcnt = 1:grpNum
    fprintf(fID,exNote{gcnt});
    grpRecordData = cell2mat(records(gcnt, :));
    grpSummaryData = cell2mat(summary(gcnt, :));
    matData = [grpRecordData, grpSummaryData];
    for h = 1:size(matData, 1)
        fprintf(fID, dataFormat, matData(h, :));
    end
end
fclose(fID);
fclose all;