Here is how you can parse Pd patches into rows of atoms in three languages using regular expressions:

Javascript

        var lines_re = new RegExp("(#((.|\r|\n)*?)[^\\\\])\r{0,1}\n{0,1};\r{0,1}\n", "gi");
        for (pdline = lines_re.exec(patchtext)) {
                var atoms = pdline[1].split(/ |\r\n?|\n/);
        }

Python

        lines_re = re.compile("(#(.*?)[^\\\])\r{0,1}\n{0,1};\r{0,1}\n", re.MULTILINE | re.DOTALL)
        split_re = re.compile(" |\r\n?|\n", re.MULTILINE)
        for found in lines_re.finditer(patch):
                line = found.group(1)
                atoms = split_re.split(line)

Java

        private static final String line_re = "(#((.|\r|\n)*?)[^\\\\])\r{0,1}\n{0,1};\r{0,1}\n";
        private static final String token_re = " |\r\n?|\n";

        Pattern pattern = Pattern.compile(line_re, Pattern.MULTILINE);
        Pattern token_pattern = Pattern.compile(token_re, Pattern.MULTILINE);
        Matcher matcher = pattern.matcher(patchtext);
        ArrayList<String[]> atomlines = new ArrayList<String[]>();
        while (matcher.find()) {
                String[] s = token_pattern.split(matcher.group(1));
                atomlines.add(token_pattern.split(matcher.group(1)));
        }

Regexp

Also here is a regular expression for matching dollar args:

        /(?:\\{0,1}\$)(\d+)/g;