Switch statement reverse using sed

Question

I have code that looks like this:

switch (argument0) {
    case ("Goblin"):
        return 0;
    break;
    case ("Fang"):
        return 1;
    break;
    ...
}

How can I, using sed or another command line tool, switch around the returns and cases so it looks like this:

switch (argument0) {
    case (0):
        return "Goblin";
    break;
    case (1):
        return "Fang";
    break;
    ...
}

Answer 1

Something along the lines of

 sed '/^[[:space:]]*case/ { N; s/case (\(.*\)):\(.* return \)\(.*\);/case (\3):\2\1;/; }' filename

That is:

/^[[:space:]]*case/ {   # in lines that bein with "case" (optionally preceded by
               # whitespace)
  N            # fetch the next line

  # Then split, reassemble.
  s/case (\(.*\)):\(.* return \)\(.*\);/case (\3):\2\1;/
}

Note that this will only work for code that is formatted fairly strictly like the one you showed, with the return directly in the line after the case label and parentheses just the right way.

By the way, I can't think of a reason to have break; directly after return statements in C.

Answer 2

With GNU awk for multi-char RS:

$ gawk -vRS="^$" -vORS= '{$0=gensub(/(case \()([^)]+)(\):\s*return )(\S+);/,"\\1\\4\\3\\2;","g")}1' file  
switch (argument0) {
    case (0):
        return "Goblin";
    break;
    case (1):
        return "Fang";
    break;
    ...
}

Liberally sprinkle \\s* s in the regexp if/when white space can occur.

Answer 3

sed '/^[[:space:]]*case/ {N;s/\(.*\)\("[^"]*"\)\(.*return \)\([^;]*\);/\1\4\3\2;/;}' YourFile

• same concept as @Wintermute but with other separator.
• This kind of sed could never be secure (so wide possibility of coding) but do most of the job if source file is like the sample.