#!/usr/bin/python

from HTMLParser import HTMLParser
import sys

class Html2Breve ( HTMLParser ):
    __depth = 0
    __indent = 4
    __injs = False
    __jsdepth = 0
    __out = sys.stdout
    __prev = None
    __needscomma = ( None, 'data', 'entity' )
    
    def fmt_attrs ( self, attrs ):
        if attrs:
            attrs = ' ( %s )' % ', '.join ( [ '%s_ = "%s"' % ( a, v )
                                              for ( a, v ) in attrs ] )
        return attrs or ''
    
    def handle_starttag ( self, tag, attrs ):
        if tag == 'script':
            self.handle_startjs ( tag, attrs )
        else:
            self.__out.write (  '%s\n%s%s%s [' % (
                    [ '', ',' ][ self.__depth > 0
                                 and self.__prev in self.__needscomma ],
                    ' ' * self.__depth,
                    tag,
                    self.fmt_attrs ( attrs )
                )
            )
            self.__depth += self.__indent
        self.__prev = tag
        
    def handle_endtag ( self, tag ):
        if tag == 'script':
            self.handle_endjs ( tag )
        else:
            self.__depth -= self.__indent
            self.__out.write ( '\n%s]' % (
                    # [ '', '#,#' ][ self.__prev is not None ],
                    ' ' * self.__depth,
                    # [ '\n', ',' ] [ self.__depth > 0 ]
                )
            )
        self.__prev = None

    def handle_startendtag ( self, tag, attrs ):
        self.__out.write (  '%s\n%s%s%s' % (
                [ '', ',' ][ self.__depth > 0
                             and self.__prev in self.__needscomma ],
                ' ' * self.__depth,
                tag,
                self.fmt_attrs ( attrs )
            )
        )
        self.__prev = None

    def handle_data ( self, data ):
        data = data.strip ( )
        if not data:
            return

        if self.__injs:
            if data.startswith ( '}' ):
                self.__jsdepth -= self.__indent
                
            self.__out.write (  '\n%s%s' % ( ' ' * ( self.__depth + self.__jsdepth ), data ) )
            
            if data.startswith ( '{' ):
                self.__jsdepth += self.__indent
        else:
            bquote = equote = '"'
            if '\n' in data:
                bquote = "'''\n%s" % ( ' ' * self.__depth )
                equote = "\n%s'''" % ( ' ' * self.__depth )
                data = data.replace ( '\n', '\n' + ' ' * self.__depth ) 

            self.__out.write (  '%s\n%s%s%s%s' % (
                    [ '', ',' ][ self.__prev in self.__needscomma ],
                    ' ' * self.__depth,
                    bquote,
                    data,
                    equote
                )
           )
        self.__prev = 'data'

    def handle_startjs ( self, tag, attrs ):
        self.__injs = True
        self.__jsdepth = self.__indent
        self.__out.write (  "\n%sinlineJS ( '''" % ( ' ' * self.__depth ) )
        self.__prev = tag

    def handle_endjs ( self, tag ):
        self.__injs = False
        self.__jsdepth = 0
        self.__out.write (  "%s''' )," % ( ' ' * self.__depth ) )
        self.__prev = tag

    def handle_entityref ( self, name ):
        sys.stdout.write ( '%s\n%s%s' % (
                (
                    [ '', ',' ][ self.__depth > 0
                                 and self.__prev in self.__needscomma ],
                    ' ' * self.__depth,
                    name
                )
            )
        )
        self.__prev = 'entity'

def usage ( ):
    print '''
        Usage:
            %s <htmlfile>
    ''' % sys.argv [ 0 ]
    
if __name__ == '__main__':
    if len ( sys.argv ) < 2:
        usage ( )
        raise SystemExit
    
    p = Html2Breve ( )
    html = '\n'.join ( [ l.strip ( )
                         for l in file ( sys.argv [ 1 ], 'rU' ).readlines ( ) ] )
    p.feed ( html )
    p.close ( )