# Copyright Aleksey Gurtovoy 2001-2004
#
# Distributed under the Boost Software License, Version 1.0. 
# (See accompanying file LICENSE_1_0.txt or copy at 
# http://www.boost.org/LICENSE_1_0.txt)
#
# See http://www.boost.org/libs/mpl for documentation.

# $Id: pp.py 49270 2008-10-11 06:35:10Z agurtovoy $
# $Date: 2008-10-10 23:35:10 -0700 (Fri, 10 Oct 2008) $
# $Revision: 49270 $

import fileinput
import os
import re
import string
import sys

if_else = lambda a,b,c:(a and [b] or [c])[0]
max_len = 79
ident = 4

def nearest_ident_pos(text):
    return (len(text)/ident) * ident
    
def block_format(limits, text, first_sep='  ', sep=',', need_last_ident=1 ):
    if sep == ',' and string.find( text, '<' ) != -1:
        sep = '%s ' % sep
    
    words = string.split(
          string.join( string.split( text ), ' ' )
        , sep
        )

    s = ' ' * limits[0]
    max_len = limits[1]
    return '%s\n%s' \
        % (
         reduce(
            lambda t,w,max_len=max_len,s=s,sep=sep:
                if_else(t[1] + len(w) < max_len
                    , ('%s%s%s'% (t[0],t[2],w), t[1]+len(w)+len(t[2]), sep)
                    , ('%s\n%s%s%s'% (t[0],s,sep,w), len(s)+len(w)+len(sep), sep)
                    )
            , words
            , (s,len(s)+len(first_sep),first_sep)
            )[0]
        , if_else(need_last_ident,s,'')
        )

def handle_args( match ):
    if re.compile('^\s*(typedef|struct|static)\s+.*?$').match(match.group(0)):
        return match.group(0)
    
    return '%s'\
        % block_format(
              (nearest_ident_pos(match.group(1)),max_len)
            , match.group(3)
            , match.group(2)
            , ','
            , 0
            )


def handle_inline_args(match):
    if len(match.group(0)) < max_len:
        return match.group(0)

    if match.group(9) == None:
        return '%s%s<\n%s>\n'\
            % (
                  match.group(1)
                , match.group(3)
                , block_format(
                      (nearest_ident_pos(match.group(1))+ident,max_len)
                    , match.group(4)
                    )
            )
        
    return '%s%s<\n%s>\n%s%s'\
        % (
              match.group(1)
            , match.group(3)
            , block_format(
                 (nearest_ident_pos(match.group(1))+ident,max_len-len(match.group(9)))
                , match.group(4)
                )
            , string.replace(match.group(1),',',' ')
            , match.group(9)
          )

def handle_simple_list(match):
    if match.group(1) == 'template':
        return match.group(0)

    single_arg = re.compile('^\s*(\w|\d)+\s*$').match(match.group(2))
    return if_else(single_arg,'%s<%s>','%s< %s >') %\
        (
          match.group(1)
        , string.join(string.split(match.group(2)), '')
        )

def handle_static(match):
    if len(match.group(0)) < max_len:
        return match.group(0)

    (first_sep,sep) = if_else(string.find(match.group(0),'+') == -1, (' ',' '),('  ','+'))
    return '%s%s\n%s%s' %\
        (
          match.group(1)
        , string.join(string.split(match.group(2)), ' ')
        , block_format(
              (nearest_ident_pos(match.group(1))+ident,max_len)
            , match.group(4)
            , first_sep
            , sep
            )
        , match.group(5)
        )

def handle_typedefs(match):
    if string.count(match.group(2), ';') == 1:
        return match.group(0)

    join_sep = ';\n%s' % match.group(1)

    return '%s%s\n' \
        % (
            match.group(1)
          , string.join(map(string.strip, string.split(match.group(2), ';')), join_sep)
          )

def fix_angle_brackets( match ):
    return ' '.join( ''.join( match.group(1).split( ' ' ) ) ) + match.group(3)
    
    
class pretty:
    def __init__(self, name):
        self.output = open(name, "w")
        self.prev_line = ''

        self.re_copyright_start = re.compile( r'^// Copyright .*$' )
        self.re_copyright_end = re.compile( r'^// See .* for documentation.$' )
        self.reading_copyright = 0
        self.copyright = None
        
        self.re_header_name_comment = re.compile( 
              r'^\s*//\s+\$[I]d:\s+(.*?%s\.hpp)\s+[^$]+[$]$'
                % os.path.splitext( name )[0]
            )
        
        self.header_was_written = 0

        self.re_junk = re.compile(r'^\s*(#|//[^/]|////).*$')
        self.re_c_comment_start = re.compile(r'^\s*/\*.*')
        self.re_c_comment_end = re.compile(r'^.*\*/\s*$')
        self.inside_c_comment = 0

        self.re_empty_line = re.compile(r'^\s*$')
        self.re_comma = re.compile(r'(\S+)\s*,\s*')
        self.re_assign = re.compile(r'(\S+[^<|^!|^>])\s*(=+)\s*(\S+)')
        self.re_marked_empty_comment = re.compile(r'^\s*//\s*$')
        self.re_typedef = re.compile(r'^\s+typedef\s+.*?;$')
        self.re_nsl = re.compile(r'^(\s+typedef\s+.*?;|\s*(private|public):\s*|\s*{\s*|\s*(\w|\d|,)+\s*)$')
        self.re_templ_decl = re.compile(r'^(\s*template\s*<\s*.*?|\s*(private|public):\s*)$')
        self.re_type_const = re.compile(r'(const)\s+((unsigned|signed)?(bool|char|short|int|long))')
        #self.re_templ_args = re.compile(r'^(\s*)(, | {2})((.*::.*?,?)+)\s*$')
        self.re_templ_args = re.compile(r'^(\s*)(, | {2})((\s*(\w+)(\s+|::)\w+\s*.*?,?)+)\s*$')
        self.re_inline_templ_args = re.compile(
            r'^(\s+(,|:\s+)?|struct\s+)(\w+)\s*<((\s*(typename\s+)?\w+\s*(=\s*.*|<(\s*\w+\s*,?)+>\s*)?,?)+)\s*>\s+((struct|class).*?)?$'
            )

        self.re_simple_list = re.compile(r'(\w+)\s*<((\w|,| |-)+)>')
        self.re_static_const = re.compile(r'(\s*)((BOOST_STATIC_CONSTANT\(\s*\w+,\s*|enum\s*\w*\s*{\s*)value\s*=)(.*?)([}|\)];)$')
        self.re_typedefs = re.compile(r'(\s*)((\s*typedef\s*.*?;)+)\s*$')
        self.re_fix_angle_brackets = re.compile( r'(>(\s*>)+)(,|\n$)' )
        self.re_closing_curly_brace = re.compile(r'^(}|struct\s+\w+);\s*$')
        self.re_namespace_scope_templ = re.compile(r'^template\s*<\s*$')
        self.re_namespace = re.compile(r'^\n?namespace\s+\w+\s*{\s*\n?$')

    def process(self, line):
        if self.reading_copyright:
            if not self.re_copyright_end.match( line ):
                self.copyright += line
                return

            self.reading_copyright = 0
            
        if not self.header_was_written and self.re_copyright_start.match( line ):
            self.copyright = line
            self.reading_copyright = 1
            return
    
        # searching for header line
        if not self.header_was_written:
            if self.re_header_name_comment.match( line ):
                self.header_was_written = 1
                match = self.re_header_name_comment.match( line )
                self.output.write( \
                    '\n%s\n' \
                    '// *Preprocessed* version of the main "%s" header\n' \
                    '// -- DO NOT modify by hand!\n\n' \
                    % ( self.copyright, match.group(1) )
                    )
            return
        
        # skipping preprocessor directives, comments, etc.
        if self.re_junk.match(line):
            return

        if self.inside_c_comment or self.re_c_comment_start.match(line):
            self.inside_c_comment = not self.re_c_comment_end.match(line)
            return

        # restoring some empty lines
        if self.re_templ_decl.match(line) and self.re_typedef.match(self.prev_line) \
           or not self.re_empty_line.match(line) and self.re_closing_curly_brace.match(self.prev_line) \
           or not self.re_empty_line.match(self.prev_line) \
              and ( self.re_namespace_scope_templ.match(line) \
                    or self.re_namespace.match(line) and not self.re_namespace.match(self.prev_line) \
                    ):
            line = '\n%s' % line

        # removing excessive empty lines
        if self.re_empty_line.match(line):
            if self.re_empty_line.match(self.prev_line) or not self.header_was_written:
                return

            # skip empty line after typedef
            if self.re_nsl.match(self.prev_line):
                return

        # formatting
        line = self.re_comma.sub( r'\1, ', line )
        line = self.re_assign.sub( r'\1 \2 \3', line )
        line = self.re_marked_empty_comment.sub( r'\n', line )
        line = self.re_type_const.sub( r'\2 \1', line )
        line = self.re_templ_args.sub( handle_args, line )
        line = self.re_inline_templ_args.sub( handle_inline_args, line )
        line = self.re_simple_list.sub( handle_simple_list, line)
        line = self.re_static_const.sub( handle_static, line )
        line = self.re_typedefs.sub( handle_typedefs, line )        
        line = self.re_fix_angle_brackets.sub( fix_angle_brackets, line )

        # write the output
        self.output.write(line)
        self.prev_line = line

def main( src, dest ):
    p = pretty( os.path.basename( dest ) )
    for line in fileinput.input( src ):
        p.process(line)

if __name__ == '__main__':    
    main( sys.argv[1], sys.argv[2] )