stepan
/
dynare
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Improvements in load_csv_file_data 

* autodetect whether there are variable names and dates in the file
 * optimize under Octave, using csv2cell and parcecell from io package
time-shift
Sébastien Villemot 2013-11-26 17:45:19 +01:00
parent eb75397e0f
commit 93416a93da
2 changed files with 43 additions and 168 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
doc/dynare.texi
View File

@ -4334,12 +4334,14 @@ univariate convergence diagnostic.
@item datafile = @var{FILENAME}
@anchor{datafile}
The datafile: a @file{.m} file, a @file{.mat} file, a @file{.csv} file, or a
@file{.xls}/@file{.xlsx} file (the latter format is supported under Octave
if the @uref{http://octave.sourceforge.net/io/,io} and
@uref{http://octave.sourceforge.net/java/,java} packages from
Octave-Forge are installed, along with a
@uref{http://www.java.com/download,Java Runtime Environment})
The datafile: a @file{.m} file, a @file{.mat} file, a @file{.csv}
file, or a @file{.xls}/@file{.xlsx} file (under Octave, the
@uref{http://octave.sourceforge.net/io/,io} from Octave-Forge is
required for the @file{.csv}, @file{.xls} and @file{.xlsx} formats; in
addition, for the @file{.xls} and @file{.xlsx} formats, the
@uref{http://octave.sourceforge.net/java/,java} package is required,
along with a @uref{http://www.java.com/download,Java Runtime
Environment})
@item xls_sheet = @var{NAME}
@anchor{xls_sheet}

197
matlab/load_csv_file_data.m
View File

@ -1,18 +1,20 @@
function [freq, init, data, varlist] = load_csv_file_data(file, withtime, withnames, noemptycell)
function [freq, init, data, varlist] = load_csv_file_data(file)
% Loads data in a csv file.
%
% INPUTS
% o file string, name of the csv file (with path).
% o withtime integer scalar, nonzero iff the first column is for the dates of the observations.
% o withnames integer scalar, nonzero iff the first row is for the names of the variables.
% o noemptycell integer scalar, nonzero the csv file does not have empty cells.
%
% OUTPUTS
% o freq integer scalar equal to 1, 4, 12 or 52 (for annual, quaterly, monthly or weekly frequencies).
% o init dates object, initial date in the dataset.
% o data matrix of doubles, the data.
% o varlist cell of strings, names of the variables.
%
% REMARKS
% The varlist output will be set only if the first line contains variable
% names. Similarly, if the first column does not contain dates, then
% freq will be 1 and init will be year 1.
% Copyright (C) 2012-2013 Dynare Team
%
@ -31,174 +33,45 @@ function [freq, init, data, varlist] = load_csv_file_data(file, withtime, withna
% You should have received a copy of the GNU General Public License
% along with Dynare. If not, see <http://www.gnu.org/licenses/>.
% Set defaults.
if nargin<4
noemptycell = 1;
if nargin<3
withnames = 1;
if nargin <2
withtime = 1;
if nargin<1
error('load_csv_file_data:: I need at least one input (name of the csv file)!')
end
end
end
end
if ~withtime && ~withnames && noemptycell
% Use matlab builtin routine!
data = csvread(file);
end
if ~( isequal(withtime,0) || isequal(withtime,1) )
error('load_csv_file_data:: Second input argument has to be equal to 1 or 0!')
end
if ~( isequal(withnames,0) || isequal(withnames,1) )
error('load_csv_file_data:: Third input argument has to be equal to 1 or 0!')
end
% Output initialization
freq = 1; % Default frequency is annual.
init = dates(1,1); % Default initial date is year one.
varlist = [];
if ~isoctave
% Under Matlab, save time by using importdata
assert(exist(file, 'file') == 2, ['load_csv_file_data: I can''t find file ' file '!']);
A = importdata(file, ',', withnames);
if withnames && withtime
if size(A.textdata, 1) == 1
% year dates confused for data
varlist = A.textdata(1, 2:end);
init = dates([num2str(A.data(1, 1)) 'Y']);
data = A.data(:, 2:end);
else
varlist = A.textdata(1, 2:end);
init = dates(A.textdata{2, 1});
data = A.data;
end
elseif withnames && ~withtime
varlist = A.textdata;
data = A.data;
elseif ~withnames && withtime
if ~isstruct(A)
% year dates confused for data
init = dates([num2str(A.data(1, 1)) 'Y']);
data = A(:, 2:end);
else
init = dates(A.textdata{1, 1});
data = A.data;
end
assert(exist(file, 'file') == 2, ['load_csv_file_data: I can''t find file ' file '!']);
if isoctave
if ~user_has_octave_forge_package('io')
error('The io package is required to read CSV files from Octave')
end
A = csv2cell(file);
[data, T, L] = parsecell(A);
withvars = L.numlimits(2,1) > L.txtlimits(2,1);
withtime = L.numlimits(1,1) > L.txtlimits(1,1);
else
A = importdata(file);
if ~isstruct(A)
data = A;
T = {};
withvars = 0;
withtime = 0;
else
error('load_csv_file_data:: Shouldn''t arrive here');
data = A.data;
T = A.textdata;
% importdata() allows text only at the top and the left, so the following
% tests are sufficient.
withvars = size(T, 2) >= size(data, 2);
withtime = size(T, 1) >= size(data, 1);
end
freq = init.freq;
varlist = transpose(varlist);
return
end
% Check if file exists.
if check_file_extension(file,'csv')
try
fid = fopen(file,'r');
catch
error(['load_csv_file_data: I can''t find file ' file '!'])
end
else
error('load_csv_file_data: Wrong file extension!')
if withvars
varlist = T(1, 2:end);
T = T(2:end, :);
end
% bfile contains a vector of ascii codes.
bfile = fread(fid);
% Close (csv) file.
fclose(fid);
% Set newline code (ok for *nix, check for mac and windows)
if isunix
newline_code = 10;
elseif ispc
newline_code = 13;
elseif ismac
newline_code = 10;
else
error('load_csv_file_data:: Not implemented for your OS!')
end
% Get the positions of the end-of-line code;
end_of_line_locations = find(bfile==newline_code);
if ispc && isempty(end_of_line_locations)
newline_code=10;
end_of_line_locations = find(bfile==newline_code);
end;
tmp = find(bfile==newline_code);
% Get the number of lines in the file.
ndx = length(tmp);
% Create a cell of indices for each line.
b = [1; end_of_line_locations+1];
c = [end_of_line_locations-1; length(bfile)+1];
b = b(1:end-1);
c = c(1:end-1);
linea = 1;
if withnames
% Get the first line of the csv file (names of the variables).
linee = char(transpose(bfile(b(linea):c(linea))));
% Get the content of the first line and determine the number of variables and their names.
[B,C] = get_cells_id(linee,',');
if withtime
B = B(2:end);
C = C(2:end);
end
varlist = cell(length(B),1);
number_of_variables = length(varlist);
for i=1:number_of_variables
varlist(i) = {linee(B(i):C(i))};
end
varlist = strtrim(varlist);
linea = linea+1;
end
% Get following line (number 1 or 2 depending on withnames flag)
linee = char(transpose(bfile(b(linea):c(linea))));
comma_locations = transpose(strfind(linee,','));
B = 1;
C = comma_locations(1)-1;
if withtime
tmp = linee(B:C);
% Check the dates formatting
if isnumeric(tmp) && isint(tmp)
tmp = [num2str(tmp) 'Y'];
end
if ~isdate(tmp)
error('load_csv_file_data:: Formatting error. I can''t read the dates!')
end
init = dates(tmp);
init = dates(T{1, 1});
freq = init.freq;
first = 2;
else
first = 1;
end
if ~withnames
number_of_variables = length(tmp)-withtime;
end
% Initialization of matrix data.
data = zeros(ndx,number_of_variables);
% Populate data.
for linea = 1+withnames:ndx
linee = char(transpose(bfile(b(linea):c(linea))));
[B,C] = get_cells_id(linee,',');
for i=first:length(B)
data(linea,i-withtime) = str2double(linee(B(i):C(i)));
end
end
% Remove first line if withnames
data = data(1+withnames:ndx,:);
varlist = transpose(varlist);