Why doesn't the "textscan" function display the empty character after the delimiter?

I found that the "textscan" function returns a different number of tokens than expected when a string ends with a delimiter. For example:

'a,b,c' --> 3 tokens
'a,,c' --> 3 tokens
'a,b,' --> 2 tokens, but expected 3 tokens

I would expect n+1 tokens for n delimiters assuming multiple delimiters are not being combined together.

Is this expected behavior? And is there a way to display the empty character at the end after the delimiter?

Answer 1

[MathWorks Support Team]

This is the expected behavior. When textscan reaches the end of an array (or file), it only returns data for the specifiers which actually had data:

'a,b,c' % parses as |a,|b,|c|<end>
'a,,c' % parses as |a,|,|c|<end>
'a,b,' % parses as |a,|b,|<end>

Since there's only one specifier, each time it consumes a delimiter, it considers the next piece of data as the start of a new line. So, given the input in "textscan":

'a,b,c' is equivalent to sprintf('a\nb\nc\n')

'a,b,' is equivalent to sprintf('a\nb\n')

As a workaround, concatenate the string with another delimiter that would make "textscan" treat the empty trailing element as a field, and not as the end of the string.