Removing commas between columns in text data

I have a txt file which is the ouput of a lemmatizer, in the form

Sometimes, ,, I, use, commas, .
I, like, writing, ,, I, like, reading

How can I read it into a tokenizedDocument deleting the unneccessary commas between tokens? A simple approach would be

test=readlines('/path/to/file.txt')
test=strrep(test,',','')
test=tokenizedDocument(test)

but it would remove even the commas already present in the original text, while I'd like to preserve punctuation-

Answer 1

[Walter Roberson]

test = {'Sometimes, ,, I, use, commas, .'
    'I, like, writing, ,, I, like, reading'};

test = regexprep(test, {'(?<=[^,]),\s', '\s*,,', '\s+\.'}, {' ', ',', '.'})

test = 2×1 cell array
    {'Sometimes, I use commas.'      }
    {'I like writing, I like reading'}

Notice we had to have a special rule for periods. You have 'use, commas' which should almost certainly translate to 'use commas' (so comma space becomes space), but after that 'commas, .' should not become 'commas .' .

To put it another way, we cannot use the rule that comma space pair is to be deleted: that works for the comma space between the word 'commas' and the period, but it does not work for the comma space pair between 'use' and 'commas': if you tried to apply that rule then 'use, commas' would merge together to 'usecommas' .

[Kim Maria Damiani]

Thank you!

Answer 2

[Chunru]

test = {'Sometimes, ,, I, use, commas, .'
    'I, like, writing, ,, I, like, reading'};

test = regexprep(test, ',\s', ' ')

test = 2×1 cell array
    {'Sometimes , I use commas .'     }
    {'I like writing , I like reading'}