% table2admb(catchfile,surveyfiles,firstyear,lastyear, maxage, num_cohorts, outfilename)
%
% Conversion of ICES table data to the .dat format used by AD Model builder.
%
% Input:
%
% catchfile   - string with path to ascii file containing catch data
% surveyfiles - either string with path to ascii file containing
%               survey data, or cell array of such strings.
% firstyear   - first year of observation/estimation (e.g 1995)
% lastyear    - last year of observation (e.g 2001)
% maxage      - maximum age for cohort (after maxage is reached no more
%               computation is done for the cohort in question).
% num_cohorts - number of cohorts to follow  (e.g. 4)
%
% NOTE: first year, last year, num_cohorts and maxage should have
%       consistent values. We always include the cohort born in 
%       firstyear, and at the moment always have cohorts born in
%       consecutive years.
%
% Output:
%
% outfilename  - (Optional) string with filename of output file without suffix,
%                that is, outfilename = '../cod' will create ../cod.dat and
%                ../cod.pin. .pin-file generation may be omitted in the future.
%                If outfilename is not defined, then output is written to the 
%                screen.
%
%     
% Written by Lennart Frimannslund, April 2010 

% Test input variables
if (in_maxage > in_lastyear-firstyear),
  fprintf(1,'Maxage = %i, but lastyear-firstyear = %i. Setting maxage to %i.\n',maxage,in_lastyear-firstyear,in_lastyear-firstyear);
  maxage = lastyear-firstyear;
else,
  maxage = in_maxage;
end;

if (in_lastyear > firstyear + (num_cohorts-1) + maxage),
  fprintf(1,'Lastyear = %i, but firstyear = %i, num_cohorts = %i and maxage = %i.\nSetting lastyear to %i.\n', ...
          in_lastyear,firstyear,num_cohorts,maxage,firstyear + (num_cohorts-1) + maxage);
  lastyear = firstyear + (num_cohorts-1) + maxage;
else,
 lastyear = in_lastyear;
end;


if (num_cohorts > lastyear-firstyear),
  fprintf(2,'Error: firstyear = %i, lastyear = %i, maxage = %i and num_cohorts = %i not consistent.\nExiting...\n', ...
          firstyear, lastyear, maxage,num_cohorts);

end;

% Read input files
C = load(catchfile,'-ascii');
if (iscell(surveyfiles)),
  S = cell(length(surveyfiles),1);
  for (i=1:length(surveyfiles)),
    S{i} = load(surveyfiles{i},'-ascii');
  end;
else,
  S = load(surveyfiles,'-ascii');
end;

% Create output file, or OVERWRITE it if it exists.
if (exist('outfilename','var') == 0),
 fid = 1;
else,
 try,
  fid = fopen([outfilename '.dat'],'w');
 catch,
  disp(lasterr);
  fid = 1;
 end;
end;

% Print data to file, item by item

% Headeer and number of cohorts
fprintf(fid,'#Generated .dat-file with catch and acoustic/trawl data from ICES report.\n#\n#Number of cohorts\n');
fprintf(fid,'%i\n',num_cohorts);

% Birth years
fprintf(fid,'\n# Cohort birth years (Startaar)\n');
for (i=1:num_cohorts),
 fprintf(fid,'%4i ',firstyear+i-1);
end;

% Last year for each cohorst
fprintf(fid,'\n\n# Last year of observation (Sluttaar)\n');
for (i=1:num_cohorts),
 fprintf(fid,'%4i ',min(firstyear+i-1+maxage,lastyear));
end;

% Catch. Note that catch in the table is in thousands, but here we divide the volume by 1000 so that it is in millions
fprintf(fid,'\n\n# Catch data IN MILLIONS. One cohort per row, one year per column.\n# First column is the birth year. Last column is not used by .tpl file\n# but must be present.\n');

youngest_age_caught = C(1,2);
max_age_in_catch_table = max(C(1,:));

% Populate catch matrix, one cohort at a time
for (i=1:num_cohorts),
  [this_birthyear_index] = find(C(:,1) == firstyear + i - 1);

  % Set catch to zero for age groups too young to be in the table
  for (j=0:youngest_age_caught - 1),
    fprintf(fid,'0.0 ');
  end;

  % Go diagonally in the catch table until either the end is reached or maxage is reached
  for (j=youngest_age_caught:min(max_age_in_catch_table,min(maxage,lastyear-firstyear - i + 1))),
    % Make sure we don't exceed size of table
    desired_row = this_birthyear_index+j-youngest_age_caught;
    if (desired_row <= size(C,1)),
      % Divide by 1000 since catch table data is in thousands, not millions
      fprintf(fid,'%f ',C(desired_row,j-youngest_age_caught+2)/1000);
    end;
  end;
  fprintf(fid,'\n');
end;

% Surveys
fprintf(fid,'\n# Survey section. Surveys included:\n#\n');
if (iscell(S)),
  for (i=1:length(S)),
    fprintf(fid,'# %s\n',surveyfiles{i});
  end;
  fprintf(fid,'#\n\n# Number of surveys per year\n%i\n',length(S));
else,
  fprintf(fid,'# %s\n',surveyfiles);
  fprintf(fid,'#\n\n# Number of surveys per year\n1\n');
end;

% For the surveys we don't yet know how many survey indices there will be,
% even though the data format requires this number to come first.
% We first write the survey data to a matrix of maximum possible size, and then to file
if (iscell(S)),
  num_surveys = length(S);
else,
  num_surveys = 1;
end;


I = zeros(num_cohorts*num_surveys*(lastyear-firstyear+1),4);


num_survey_indices = 0;

for (k=1:num_surveys),
  
   if (num_surveys == 1),
     this_survey = S;
   else, 
     this_survey = S{k};
   end;

  % Populate one cohort at a time
  for (i=1:num_cohorts),
    % Find the row in column 2 which corresponds to the current cohort
    this_cohort_row_in_col_2 = find( (this_survey(:,1) - this_survey(1,2) - (firstyear + (i-1)) ) == 0);

    last_legal_row = find(this_survey(:,1) == lastyear);
    last_legal_col = find(this_survey(1,:) == lastyear-firstyear - i + 1);
    max_j = min(last_legal_row-this_cohort_row_in_col_2,last_legal_col-2);

    % go diagonally, same as for catch.
    for (j=0:max_j),
      try,
       candidate_volume = this_survey(this_cohort_row_in_col_2+j,2+j);
        if (isfinite(candidate_volume)),
          % If we encounter an observation of zero (which will make the 
          % objective function infinity), don't include it
          if (candidate_volume > 0.0),
             num_survey_indices = num_survey_indices + 1;
             I(num_survey_indices,:) = [ i this_survey(1,2+j) k candidate_volume];
          end;
        end;

      % For debug purposes, report if our indices are out of bounds. Shouldn't happen anymore.
      catch,
        fprintf(1,'Survey table index exceeded. (i,j) = (%i,%i), but size S{%i} = (%i,%i).\n', ...
           this_cohort_row_in_col_2+j,this_cohort_row_in_col_2+1+j,k,size(this_survey,1),size(this_survey,2));
      end;
    end;
   
  end;
 
end;

% Print everything to file
fprintf(fid,'\n# Number of survey indices\n');
fprintf(fid,'%i \n',num_survey_indices);
fprintf(fid,'\n# Survey table: (Cohort, age, survey_number, observed_volume)\n');
for (i=1:num_survey_indices),
  fprintf(fid,'%3i %3i %3i %f\n',I(i,1),I(i,2),I(i,3),I(i,4));
end;

fprintf(fid,'# End of file');

% Done, close file if applicable
if (fid ~= 1),
  fclose(fid);
end;

%return;
% Make .pin-file too
if (exist('outfilename','var') == 0),
 fid = 1;
else,
 try,
  fid = fopen([outfilename '.pin'],'w');
 catch,
  disp(lasterr);
  fid = 1;
 end;
end;

fprintf(fid,'\n\n# Autogenerated .pin file\n# N0\n');
for (i=1:num_cohorts),

   fprintf(fid,'%f  ',1e3);

end;
fprintf(fid,'\n\n# q\n');

for (i=1:num_surveys),

  fprintf(fid,'%4.2f ',0.4);

end;

fprintf(fid,'\n\n# log(s)\n0.1\n\n# M\n0.2\n');

if (fid ~= 1),
  fclose(fid);
end;