finding string in between

Hi, so I have a cell string with 100 X 1 like:

18WABO1-12345-0X

18WABO2-12345-0N

18WACE3-12345-00

18WACE4-12345-0R

18WAGUG-12345-0G

18WDUER-12345-0N

I would like to find the string sequence that is always between 18W and first - so the result is:

ABO1

AB02

ACE3

ACE4

AGUG

DUER etc

my example of a code:

 %
somestring(:)= eic_p; 
underscore_indices= strfind(somestring,'18W');
underscore_indices=cell2mat(underscore_indices);
fs_indices = strfind(somestring,'-');
fs_indices=fs_indices';
your_number=cellfun(@(v)v(1),fs_indices);
somestring(:)= somestring'; 
for i=1:length(fs_indices)
    yourNumber= somestring{i}(underscore_indices(i)+2:your_number(i)-1);
    %HOW i can save every iteration? thanks
end

in the last for loop somehow I am getting the weird output and can not save all results so I can have all those 205 abbreviations in one variable (yourNumber).

Thanks a lot,

Answer 1

[per isakson]

yourNumber is overwritten in the loop and only the last value is saved. The first step to fix your code is

yourNumber = cell( length(fs_indices), 1 );
for i = 1 : length(fs_indices)
    yourNumber{i} = somestring{i}(underscore_indices(i)+2:your_number(i)-1);
end

There are other ways, e.g. with regular expressions

>> str = '18WABO1-12345-0X';
>> regexp( str, '(?<=18W)[^\-]+(?=\-)', 'match' )
ans = 
    'ABO1'

and

    cac = { 
            '18WABO1-12345-0X'
            '18WABO2-12345-0N'
            '18WACE3-12345-00'
            '18WACE4-12345-0R'
            '18WAGUG-12345-0G'
            '18WDUER-12345-0N' };
    %    
    out = regexp( cac, '(?<=18W).+?(?=\-)', 'match' ); 
    out = cat( 1, out{:} );

and

>> out
out = 
    'ABO1'
    'ABO2'
    'ACE3'
    'ACE4'
    'AGUG'
    'DUER'

and with indexing

>> str = char( cac );
>> str = str( :, 4:7 )
str =
ABO1
ABO2
ACE3
ACE4
AGUG
DUER
>>

[Cedric]

No need for the look forward/around with your first pattern, and you can add the option 'once' to avoid the CAT. And if you need to debug it .. well I'm not really sure .. yet ;)

 >> out2 = regexp( cac, '(?<=18W)[^-]+', 'match', 'once' )
 out2 =
  6×1 cell array
    {'ABO1'}
    {'ABO2'}
    {'ACE3'}
    {'ACE4'}
    {'AGUG'}
    {'DUER'}

[Stephen23]

I tried this myself, and came up with the almost the same reg exp, just with the ^ to match the start:

regexp(C,'(?<=^18W)[^-]+','once','match')

[per isakson]

Yes, the look-ahead is overkill and 'once' will save a microsecond. With long strings 'once' makes a significant difference.

However, regexp with or without 'once' returns a cell array of scalar cell arrays, which in turn contain the strings. cat "flattens" the cell array.

[Cedric]

There is always one more level without the 'once':

 >> out2 = regexp( cac, '(?<=18W)[^-]+', 'match' )
 out2 =
  6×1 cell array
    {1×1 cell}
    {1×1 cell}
    {1×1 cell}
    {1×1 cell}
    {1×1 cell}
    {1×1 cell}

 >> out2 = regexp( cac, '(?<=18W)[^-]+', 'match', 'once' )
 out2 =
  6×1 cell array
    {'ABO1'}
    {'ABO2'}
    {'ACE3'}
    {'ACE4'}
    {'AGUG'}
    {'DUER'}

.. you probably forgot to copy one of the lines (call to CAT) when you copy-pasted your example from the command window.

[per isakson]

"There is always one more level without the 'once':" Yes, that's correct. Now, I'll remember. A pity there isn't a strike-out feature.

One thing still puzzles me

out2 =
  6×1 cell array
    {'ABO1'}

why the braces around 'AB01'. Here on R2016a I get

>> out2
out2 = 
    'ABO1'

Have The MathWorks changed the display format?

[Cedric]

Wow, you're right, I had never realized, or already forgotten(!) My output is from 2017b, but I was on 2016b until very recently .. I'm wondering if I didn't pay attention or if the update was between 2016a/b (?)

[sensation]

Thanks a lot guys for your answers! One quick question: can you just briefly eleborate (?<=18W)[^-]+ ?, or where I can find those expressions when I should use ? ^ or/and +. Thanks!

[Cedric]

Look at my comment starting with "Not far" here for a brief summary.

Understanding this, you will understand that

• (?<=..) is a look-behind and (?<=18W) imposes that what is matched (by the rest of the pattern) is preceded by 18W
• [^..] defines a set of elements not to match, so [^-] matches all characters but the dash.
• + is a quantifier that means one or more times the expression that precedes directly (which is [^-])

So the whole thing reads: match one or more character that is not a dash (which translates into "read all until a dash"), preceded by the literal 18W.

[Stephen23]

"where I can find those expressions when I should use ? ^ or/and +."

By reading the documentation ten times:

https://www.mathworks.com/help/matlab/matlab_prog/regular-expressions.html

And then read it another ten times. And practice lots.

Regular expressions are powerful and very useful, but they require practice and attention to detail. Study that page I linked to, and the other pages that it links to as well.

[sensation]

Thanks!