Visual Studio Code snippet: convert 'PascalCase' to lowercase with spaces

Question

I'm trying to write a Visual Studio Code snippet that converts Pascal case to lowercase with spaces. I'm almost there:

${1/([A-Z]*)([A-Z][a-z]+)/$1 ${2:/downcase} /g}

Group 1 matches acronyms and group 2 matches capitalized words.

So if the placeholder were MyUIDTest, the expected result would be my UID test. But currently I get my UID test (notice the spaces on either side).

How do I only add a space after group 1 if group 1 has a match? And how do I remove the space at the end of the line?

Answer 1

According to the docs, ${1:+ } means "insert a space iff group 1 is captured":

${
  TM_SELECTED_TEXT
  /
    (?<=([A-Za-z])?)      # Lookbehind for and capture ($1) a preceding letter, if any,
    (?:                   # then match a "word" – either
      ([A-Z]+)(?![a-z])   # ($2) 1+ uppercase not followed by a lowercase
    |                     # or
      ([A-Z][a-z]+)       # ($3) an uppercase followed by 1+ lowercase.
    )                     #
  /                       # Replace each match with
    ${1:+\u0020}          # a space, iff group 1 is captured,
    ${2:/upcase}          # the uppercase version of group 2
    ${3:/downcase}        # and the lowercase version of group 3.
  /g                      # for all matches found.
}

Since $2 and $3 are mutually exclusive, using them both always results in only one being actually inserted.

Before	After
PascalCase	pascal case
MyUIDTest	my UID test
FoOBaR	fo O ba R

A more precise (and verbose) version would be (note that \b needs to be escaped again when written in a JSON string: "\\b"):

${
  TM_SELECTED_TEXT
  /
    (?<=                                    # Lookbehind for
      \b                                    # a word boundary followed by
      ([A-Z]+(?![a-z])|[A-Z][a-z]+)*        # 0+ "words"
                                            # and capture the last word, if any.
    )
    (?:([A-Z]+)(?![a-z])|([A-Z][a-z]+))
  /
    ${1:+\u0020}${2:/upcase}${3:/downcase}
  /g
}

I'm using TM_SELECTED_TEXT here, but you might want to change that depending on your needs.