%%%%%%%%%%%%%%%%%%%%%%%% TSAPadj2Hmat.m %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% Created by Amy Langville (edited by Rebecca Wills)
%% reading link data provided by Panayiotis Tsaparas
%% available at 
%% http://www.cs.toronto.edu/~tsap/experiments/datasets/index.html
%% Loading: Expanded Datasets (instead of Refined or Regular)

function  [H,numberofnodes,a] = TSAPadj2Hmat(filename);

% EXAMPLE: [H,numberofnodes,a] = TSAPadj2Hmat('TSAP_abortion.txt');

% OUTPUT: 
% H = hyperlink matrix
% numberofnodes = n (where H = n x n matrix)
% a = dangling node vector, where a_i=1 if node i is dangling, 0 o.w.



fid=fopen(filename,'r');

% The adjacency list data sets label the nodes from 0 up.
% This function makes the adjustment so that the nodes run from 1 up.
% NOTE: An entire E matrix is not added.  H is sparse and contains 0
% rows for dangling nodes.

dline = fgetl(fid);
if (dline == -1 )
  error('Empty file.')
end

row=0;
rowindex=[];
colindex=[];
elements=[];
while length(dline) > 0 & dline ~= -1,
  [lstr,dline]=strtok(dline);
  dline=dline(1:length(dline)-3);
  adjlist=str2num(dline);
  adjlist=adjlist+1;
  row=row+1;
  rowindex(length(rowindex)+1:length(rowindex)+length(adjlist))=row;
  colindex(length(colindex)+1:length(colindex)+length(adjlist))=adjlist;
  if length(adjlist) >0
    elements(length(elements)+1:length(elements)+length(adjlist))= ...
         (1/length(adjlist))*ones(length(adjlist),1);
  end
  dline = fgetl(fid);
end  

H=spconvert([rowindex' colindex' elements']);
% H is not necessarily a square matrix at this point.  
numberofnodes=row;
H(numberofnodes,numberofnodes)=0;
% The above step makes H square (if it isn't already).
rowsum=H*ones(size(H,2),1); 
a=ones(numberofnodes,1)-rowsum;

fclose(fid);
% Done.

 
save ('bb','H','numberofnodes','a') 
% creates filename.mat file containing H, numberofnodes, a


% then to load data for filename.mat, do run short 'load' m-file