MATLAB help always showing true to if condition

So, I wanted to write a program to classify text files based on their topic. The code below shows a very crude implementation of the same, but it isn't working properly. It is always showing a true value to the if condition, so I am getting Greetings four times and then Colloquial four times. What's up?

if true
File1 = fopen('Hello.xml','r');
File2 = fopen('Hello2.xml','r');
File3 = fopen('Colloquial1.xml','r');
File4 = fopen('Colloquial2.xml','r');
A = fscanf(File1, '%s');
B = fscanf(File2, '%s');
C = fscanf(File3, '%s');
D = fscanf(File4, '%s');
if ~( strcmp(A,'Hi') || strcmp(A,'Hello') || strcmp(A,'how'))
     disp('Greetings.');
end
if ~( strcmp(B,'Hi') || strcmp(B,'Hello') || strcmp(B,'how'))
      disp('Greetings');
end    
if ~( strcmp(C,'Hi') || strcmp(C,'Hello') || strcmp(C,'how'))
      disp('Greetings');
end    
if ~( strcmp(D,'Hi') || strcmp(D,'Hello') || strcmp(D,'how'))
      disp('Greetings');
end
if ~( strcmp(A,'lite'))
     disp('Colloquial.');
end
if ~( strcmp(B, 'lite'))
      disp('Colloquial');
end    
if ~( strcmp(C, 'lite'))
      disp('Colloquial');
end    
if ~( strcmp(D, 'lite'))
      disp('Colloquial');
end  
end

[Daniel Shub]

Have you thought about using REGEXP for this? It might be a little cleaner.

[Samyukta Ramnath]

But REGEXP is mainly for performing operations on the found substring, right? I just need to see if the word is there.

[Daniel Shub]

No, REGEXPREP Does replacements, REGEXP just does the search.

Answer 1

[Daniel Shub]

The strcmp function requires an exact match. We can define a simple test function to see if a string is either exactly 'Hi' or exactly 'lite'

test = @(A)([strcmp(A,'Hi'), strcmp(A,'lite')])

Then

 test('Hi')
     1     0
 test('lite')
     0     1
 test('Hilite')
     0     0

In words: Hi is Hi, but it is not lite. lite is not Hi but is lite, and Hilite is neither Hi nor lite.

Are you potentially looking for STRFIND

 test = @(A)([~isempty(strfind(A,'Hi')), ~isempty(strfind(A,'lite'))])

 test('Hi')
     1     0
 test('lite')
     0     1
 test('Hilite')
     1     1

[Samyukta Ramnath]

The file Hello has the content - 'Hello, Mr Anderson, how are you doing?' Hello1 has : 'Hello, my name is Samyukta.' Colloquial1 has : 'Arre no, take lite, ob you can’t do anything.' Colloquial2 has : 'Sam, gimme one donut, or else take lite!'

So now we know the exact contents, and strcmp should be able to find these correctly, right?

[Matt Kindig]

Actually, if you look at Daniel's posting, you'll see that strcmp() won't do what you want to do, since the strings are not identically 'hi', 'hello', 'lite', etc. Use the strfind() method that he describes above.

[Daniel Shub]

No, STRCMP requires an exact match of the entire string. You want strfind or regexp.

[Samyukta Ramnath]

Thank you! That works. :)

[Daniel Shub]

Please consider accepting an answer if your problem is solved. Also, please consider voting for other answers that were helpful.

Answer 2

[Ken Atwell]

It is not clear what the contents of the files are and what the expected ("correct") output ought to be. Be aware that strcmp returns different results than a C programmer might expect:

<http://www.mathworks.com/help/matlab/ref/strcmp.html>

strcmp might be returning 1 where you are expecting 0.

If that is not the problem, you will need to provide more context -- namely, the first line of the four files you are parsing.

[Samyukta Ramnath]

The Files which say Hello have Hi and Hello in them, and the files which have Colloquial in their name have the word 'lite' in them. So, it should not be returning a true for everything. And I realized that one error was that the strcmp returns a value different from that in C.

[Ken Atwell]

Take out the not operators (the difference between MATLAB and C) and it seems to work okay for me:

File1 = fopen('Hello.xml','r');
File2 = fopen('Hello2.xml','r');
File3 = fopen('Colloquial1.xml','r');
File4 = fopen('Colloquial2.xml','r');
A = fscanf(File1, '%s');
B = fscanf(File2, '%s');
C = fscanf(File3, '%s');
D = fscanf(File4, '%s');
if strcmp(A,'Hi') || strcmp(A,'Hello') || strcmp(A,'how')
   disp('Greetings A');
end
if strcmp(B,'Hi') || strcmp(B,'Hello') || strcmp(B,'how')
    disp('Greetings B');
end    
if strcmp(C,'Hi') || strcmp(C,'Hello') || strcmp(C,'how')
    disp('Greetings C');
end    
if strcmp(D,'Hi') || strcmp(D,'Hello') || strcmp(D,'how')
    disp('Greetings D');
end
if strcmp(A,'lite')
   disp('Colloquial A');
end
if strcmp(B, 'lite')
    disp('Colloquial B');
end    
if strcmp(C, 'lite')
    disp('Colloquial C');
end    
if strcmp(D, 'lite')
    disp('Colloquial D');
end  
if true
  % code
end

Output is:

 Greetings A
 Greetings B
 Colloquial C
 Colloquial D

[Samyukta Ramnath]

Oh My. I copied and pasted this exact code into my MATLAB script and ran it, and nothing happened.

Answer 3

[Iain]

Do A, B, C, and D actually have those strings in them, EXACTLY like what you're comparing them to?

I suspect they have a different case. Try strcmpi instead of strcmp.

What do you get when you get matlab to print out A, B, C & D?

[Daniel Shub]

For STRCMP it is not about having the strings IN them, it is about BEING them.