Extract Text and Values from String

Dear Forum,

I am trying to extract the following information from this string.

str = 'AB(16.7)CD[20.6]EF[.1] 864.4 Round'

The first thing I did is the following.

vals1 = strsplit(str)

I get as expected.

{'AB(16.7)CD[20.6]EF[.1]'} {'864.4'} {'Roundl'}

What I am having trouble is trying to get the next part.

type1 = AB

val1 = 16.7

type2 = CD

val2 = 20.6

type3 = EF

val3 = .1

If I set

str1 = 'AB(16.7)CD[20.6]EF[.1]'

I can only get this to work

val1 = regexp(str1, '(?<=[)\d+\.\d(?=)','match','once')

val1 = '20.6'

What I don't understand is that is if str1 = 'AB(16)CD[20.6]EF[.1]'

Where there is no decimal I get an error.

I am unsure how to achieve the the text [type1,type2,type3] and three values [val1,val2,val3]

Answer 1

[Matt J]

str = 'AB(16.7)CD[20.6]EF[.1] 864.4 Round';

S=strsplit(str,{'(',')','[',']',' '});

Types=S([1,3,5])

Types = 1x3 cell array
    {'AB'}    {'CD'}    {'EF'}

Vals=str2double(S([2,4,6]))

Vals = 1×3

16.7000 20.6000 0.1000

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

Answer 2

[Image Analyst]

How much does the string vary? Are all the things at fixed, specific locations? If so just use indexing

str = 'AB(16.7)CD[20.6]EF[.1] 864.4 Round';
type1 = str(1:2);
val1 = str2double(str(4:7))

val1 = 16.7000

type2 = str(9:10)

type2 = 'CD'

val2 = str2double(str(12:15))

val2 = 20.6000

type3 = str(17:18)

type3 = 'EF'

val3 = str2double(str(20:21))

val3 = 0.1000

Sure, it's not as compact as some regexp but way less cryptic and far easier to understand.

If the values and types are not in fixed index locations, you should look at using the more modern patterns instead of the older (and more cryptic/harder to use) regexp function. Also see extract as shown in the examples for digitsPattern and lettersPattern

patNum = digitsPattern;
numbers = extract(str,patNum)

numbers = 7x1 cell array
    {'16' }
    {'7'  }
    {'20' }
    {'6'  }
    {'1'  }
    {'864'}
    {'4'  }

You would then have to use str2double and combine cells that you know belong as parts of the same number, like

val1 = str2double(numbers{1}) + str2double(numbers{2}) / 10

val1 = 16.7000

val2 = str2double(numbers{3}) + str2double(numbers{4}) / 10

val2 = 20.6000

val3 = str2double(numbers{5}) / 10

val3 = 0.1000

patLetters = lettersPattern;
letters = extract(str,patLetters)

letters = 4x1 cell array
    {'AB'   }
    {'CD'   }
    {'EF'   }
    {'Round'}

type1 = letters{1}

type1 = 'AB'

type2 = letters{2}

type2 = 'CD'

type3 = letters{3}

type3 = 'EF'

Documentation for lettersPattern

Documentation for digitsPattern

Again, perhaps not as compact as regexp but I'd go for understandability, readability, and intuitiveness over compactness every time.

Answer 3

[Walter Roberson]

strs = {'AB(16.7)CD[20.6]EF[.1]', 'AB(16)CD[20.6]EF[.1]'};
for idx = 1 : length(strs)
    str1 = strs{idx};
    info = regexp(str1, '^(?<type1>\w+)\((?<val1>(\d+(\.\d*)?|\.\d+))\)(?<type2>\w+)\[(?<val2>(\d+(\.\d*)?|\.\d+))\](?<type3>\w+)\[(?<val3>(\d+(\.\d*)?|\.\d+))\]', 'names', 'once')
end

info = struct with fields:
    type1: 'AB'
     val1: '16.7'
    type2: 'CD'
     val2: '20.6'
    type3: 'EF'
     val3: '.1'

info = struct with fields:
    type1: 'AB'
     val1: '16'
    type2: 'CD'
     val2: '20.6'
    type3: 'EF'
     val3: '.1'