function [words,counts]=parseEmail(fname,dictionary)
% [words,counts]=parseEmail(fname,dictionary)
% Reads in file FNAME, collects unique words and counts their occurences.
% A word is defined as any contiguous sequence of letters, _ and @
% (surrounded by anything that is not a letter, _ or @).
% If non-empty cell-array DICTIONARY is given, only words in it are considered
% (i.e., any word not in DICTIONARY is ignored).
% Returns:
%  WORDS{I} is the I-th distinct word;
%  COUNTS(I) is the number of times WORDS{I} appeared in this document.
% Note: words that do not appear in the file are not included in WORDS.

if (nargin < 2)
  dictionary = [];
end

f=fopen(fname,'rb');
if (f < 0)
  error(['Can not open ' fname]);
end

% read in characters (fastest to read a stream of bytes), 
% convert to char type, 
% convert all letters to lower case and finally enclose with spaces on
% both ends (to make regexp's life easier)
email=[32 lower(char(fread(f,'char')')) 32];

% this will find all words (non unique i.e. some words will appear
% multiple times)
matches=regexp(email,'([a-z_@]*)','match');

% collect unique words
[words,ignore,ind]=unique(matches);
if (~isempty(dictionary))
  % only keep words in the dictionary
  [words,keep]=intersect(words,dictionary);
else
  % keep all of them
  keep=1:length(words);
end

% count them using the output of unique
counts=zeros(1,length(words));
for i=1:length(keep)
  counts(i)=sum(ind==keep(i));
end

fclose(f);