% File -> Import Data (gas_all_grades.csv)
% Number of header rows = 1

% constants to determine how data is filtered
FILTER_CITY = 1;
FILTER_STATE = 2;
FILTER_REGION = 3;
FILTER_COUNTRY = 4;
FILTER_ALL = 5;

% filter data using ...
filterData = FILTER_STATE;

% format numbers with at most 5 digits after decimal
format short g

% sort data and return index of sort
[dataSorted,sortIndex] = sortrows(data,[5 1 4]);

% first row of textdata is garbage
textdata2 = textdata(2:size(textdata,1),:);
% apply sort index to textdata
textdataSorted = textdata2(sortIndex,:);

% read geography info into vectors, which will act as lookup tables
%   read from file 'geography.csv'
%   read each line as '%q %q' = [string string]
%   read all lines (= -1)
%   as the 'delimiter', use a comma (',')
[geoCode,geoName] = textread('geography.csv', '%q %q', -1, 'delimiter', ',');
% first lines of each read column are garbage
geoCode = geoCode(2:length(geoCode));
geoName = geoName(2:length(geoName));

% extract data by iterating over every region code and extracting data for
% that region
dataByGeo = [];
dataByGeoEndIndex = [];
% text is represented in cells
textdataByGeo = cell(0);
textdataByGeoEndIndex = [];
for i=1:length(geoCode)
    % for any given string, the function 'letters' returns an array where
    % for each character of the string, a 1 denotes that the character is a
    % letter, and a 0 denotes that it is not.
    % Example: letters('Zoo100') = [1 1 1 0 0 0]
    % data type 'cell' is converted to 'char' using function 'char'
    letters = isletter(char(geoCode(i)));
    % filter: city=2nd letter of code is digit, e.g. N1=New York City
    if ((filterData == FILTER_CITY) && (letters(2) == 1))
        dataT=[];
        textdataT=[];
    % filter: region=1st letter of code is digit, e.g. 1A=New England Region
    elseif ((filterData == FILTER_REGION) && (letters(1) == 1))
        dataT = [];
        textdataT = [];
    % filter: only country considered: US
    elseif ((filterData == FILTER_COUNTRY) && (strcmp(geoCode(i), 'US') == 0))
        dataT = [];
        textdataT = [];
    % filter: both letters 
    elseif ((filterData == FILTER_STATE) && ((sum(letters) ~= 2) || (strcmp(geoCode(i), 'US') == 1)))
        dataT = [];
        textdataT = [];
    else
        [dataT, textdataT] = extractRegion(dataSorted, textdataSorted, geoCode(i));
    end

    % collect extracted data and textdata per geographical area in a matrix
    dataByGeo(1:size(dataT,1), 1:size(dataT,2), i) = dataT;
    dataByGeoEndIndex(i,:) = size(dataT);
    textdataByGeo(1:size(textdataT,1), 1:size(textdataT,2), i) = textdataT;
    textdataByGeoEndIndex(i,:) = size(textdataT);
end

% find minimum year to offset each data set
yearMin = min(data(:,5));

% make an empty plot with titles and labels
plot([], []), xlabel('Time'), ylabel('Price in cents/gallon'), title('Gasoline prices in various regions in U.S.');
% used to overlay next plots
hold on;

% line colors used
colors = {'b','g','r','c','m'};
ci = 1;
% line marks used
marks = {'o','x','+','*','d'};
mi = 1;

% collect legend labels to create legend after plotting
labels = cell(0);
li = 1;

% iterate over all geographical regions and make plot
for i = 1:length(geoCode)
    % since data is irregularly sized and data is located in 3D matrix
    % structure, ei denotes the "real" length of each data slice
    ei = dataByGeoEndIndex(i,1);
    % if data has been collected, length is greater than 0, so plot it
    if (ei > 0)
        % create timeline
        timeT = ((dataByGeo(1:ei, 5, i) - yearMin) * 52) + (dataByGeo(1:ei, 1, i) * 4) + dataByGeo(1:ei, 4, i);
        % plot data
        plot(timeT, dataByGeo(1:ei, 2, i), [char(colors(ci)) char(marks(mi)) '-']);
        % collect label
        labels(li) = geoName(indexOf(geoCode, geoCode(i)));
        li = li + 1;
        % rotate colors and marks
        if (mod(ci, length(colors)) == 0)
            ci = 1;
            mi = mi + 1;
        else
            ci = ci + 1;
        end
    end
end

% stop overlaying graphs
hold off;

% create legend
legend(labels, 'Location', 'BestOutside');