Different behavior of regexp() on linux and windows in matlab-CodePudding

I am encountering a different behavior of regexp() on linux and windows using MatLab. I am trying to separate a string based on a separator. Here is a minimal example:

Linux

test_string = '<some_path>/tool/test/unit_test'
seperator = sprintf('%stest%s',filesep,filesep)
regexp(test_string, seperator,'split')

Output:

1×2 cell array
{'<some_path>/tool'}    {'unit_test'}

Windows

test_string = '<some_path>\tool\test\unit_test'
seperator = sprintf('%stest%s',filesep,filesep)
regexp(test_string, seperator,'split')

Output

1×1 cell array
{'<some_path>\src\tool\test\unit_test'}

The output of this code snippet on Linux represents the behavior I want. Could anyone explain or point towards resources to understand what is going on?

CodePudding user response：

The path separators are different in Linux (/) and Windows (\).

The \ character is a special regex metacharacter, it is used to form "regex escapes", like \d to match digits, etc. To match a literal backslash, it must be doubled, or escaped.

To escape any special regex metacharacters, you can use regexptranslate(op, str) with op set to escape:

seperator = sprintf('%stest%s',regexptranslate('escape',filesep), regexptranslate('escape',filesep))

Other op possible values are:

Type of Translation	Description
`'escape'`	Translate all special characters in str, such as `'$'`, `'.'`, `'?'`,`'['`, so that they are treated as literal characters when used in `regexp`, `regexpi`, and `regexprep`. The translation inserts a backslash, or escape, character, `'\'`, before each special character in `str`.
`'wildcard'`	Translate all wildcard and `'.'` characters in `str` so that they are treated as literal wildcard characters and periods when used in `regexp`, `regexpi`, and `regexprep`. The translation replaces all instances of `''` with `'.'`, all instances of `'?'` with `'.'`, and all instances of `'.'` with `'\.'`.
`'flexible'`	Replace text in `str` with a regular expression that matches the text. If you specify `'flexible'`, then also specify a regular expression to use as a replacement: `newStr = regexptranslate('flexible',str,expression)`. The expression input can be a character vector or string scalar. This syntax is equivalent to `newStr = regexprep(str,expression,regexptranslate('escape',expression))`.