cpp to matlab converter using python and regular expressions (regex)

cpp_to_m.py

# Handles lots of cases but only for certain kinds of c++ coding

# Example conversion:
#     int count = 0;
#     for(int i = 0; i < 5; i++)
#         count++;
#
# changes to
# 
#     count = 0;
#     for i = 0: 5
#         count = count + 1;
#     end

# Use at your own risk!  Doesn't handle function headers, includes, struct types, etc, or most things besides the basics
# This conversion is only supposed to speed up some of the find and replace tasks when porting, but is not a replacement
# for real time and effort into doing the port from c++ to matlab and testing it.


import re

files = ['temp.cpp']

four_spaces = re.compile(r" {4}")

multi_line_comment = re.compile(r"/\*([\s\S]*?)\*/\s*", re.MULTILINE)
single_line_comment = re.compile(r"//")
square_brackets = re.compile(r"\[(.+?)\]")
log_statements = re.compile(r".*__LOG.*\n")
else_block = re.compile(r"\}\s*else\s*\{")
trailing_curly_brace = re.compile(r"\}")
leading_curly_brace = re.compile(r"\{")
casting_to_double = re.compile(r"\(double\)|\(float\)")
local_var_init_1 = re.compile(r"^(\s*)(double|int|float)\s+(\S+)\s*=\s*(\S+)\s*;", re.MULTILINE)
local_var_init_2 = re.compile(r"^(\s*)(double|int|float)\s+(\S+)\s*;", re.MULTILINE)

static_member_refs = re.compile(r"m_s_");
member_refs = re.compile(r"m_");

inc_for_loop = re.compile(r"for\s*\(\s*(\S*\s)?(\S+)\s*=\s*(\S+)\s*;\s*\2\s*<\s*(\S+)\s*;\s*((\+\+\2)|(\2\+\+))\s*\)")
generic_for_loop = re.compile(r"(\s*)for\s*\(([\s\S]*?);([\s\S]*?);([\s\S]*?)\)")

simple_inc_1 = re.compile(r"^(\s*)\+\+(\S*);\s*$", re.MULTILINE)
simple_inc_2 = re.compile(r"^(\s*)(\S*)\+\+;\s*$", re.MULTILINE)
simple_dec_1 = re.compile(r"^(\s*)\-\-(\S*);\s*$", re.MULTILINE)
simple_dec_2 = re.compile(r"^(\s*)(\S*)\-\-;\s*$", re.MULTILINE)
simple_arith = re.compile(r"^(\s*)(\S*)\s*([\*\+\/\-])=\s*(\S*);\s*$", re.MULTILINE)

# Not yet working!!!
if_with_no_curly_brackets = re.compile(r"^(\s*?)(if\s*\(.+?\)[^\n]*)$\s*([^\s{/%][^\n]*)$", re.MULTILINE)
for_with_no_curly_brackets = re.compile(r"^(\s*?)(for\s*\(.+?\)[^\n]*)$\s*([^\s{/%][^\n]*)$", re.MULTILINE)

inline_if_statement = re.compile(r"^(\s*?)(if\s*\(.+?\))\s*([^\s\/\)].*;)$", re.MULTILINE)

pointer_to_dot = re.compile(r"->")

for file in files:

	data = ""
	with open (file, "r") as cppfile:
		data = cppfile.read()
		data = four_spaces.sub("\t", data)
		data = multi_line_comment.sub('', data)
		
		data = inline_if_statement.sub(r'\1\2\n\1\t\3', data)
		
		data = if_with_no_curly_brackets.sub(r'\1\2\n\1\t\3\n\1end\n', data)
		data = for_with_no_curly_brackets.sub(r'\1\2\n\1\t\3\n\1end\n', data)
		
		data = single_line_comment.sub("% ", data)
		data = square_brackets.sub(r'(\1)', data)
		data = square_brackets.sub(r'(\1)', data) # catch any double nested square_brackets
		data = square_brackets.sub(r'(\1)', data) # catch any triple nested square_brackets
		data = log_statements.sub('', data)
		data = else_block.sub('else', data)
		data = trailing_curly_brace.sub('end', data)
		data = leading_curly_brace.sub('', data)
		data = inc_for_loop.sub(r'for \2 = \3 : \4 - 1', data)
		data = generic_for_loop.sub(r'\1%FIXME: for(\2;\3;\4)\1for \2:\3:\4', data)
		
		data = casting_to_double.sub('', data)
		
		data = static_member_refs.sub('', data)
		data = member_refs.sub('', data)
		
		data = local_var_init_1.sub(r'\1\3 = \4;', data)
		data = local_var_init_2.sub(r'\1\3 = 0;', data)
		
		data = simple_inc_1.sub(r'\1\2 = \2 + 1;', data)
		data = simple_inc_2.sub(r'\1\2 = \2 + 1;', data)
		
		data = simple_dec_1.sub(r'\1\2 = \2 - 1;', data)
		data = simple_dec_2.sub(r'\1\2 = \2 - 1;', data)
		
		data = simple_arith.sub(r'\1\2 = \2 \3 \4;', data)
		
		data = pointer_to_dot.sub(r'.', data)
		
		mfile = file.replace('.cpp', '.m')
		with open (mfile, "w") as outfile:
			outfile.write(data)
		
		print(data)