formatHTML

#!/usr/bin/env python
'''
    Copyright (c) 2015, 2016, 2017, 2018 Timothy Savannah All Rights Reserved

    Licensed under terms of the Lesser GNU General Purpose License Version 2.1

    This is part of AdvancedHTMLParser ( https://github.com/kata198/AdvancedHTMLParser )

    See LICENSE for more information
'''

import os
import sys

from AdvancedHTMLParser import AdvancedHTMLFormatter, AdvancedHTMLMiniFormatter, AdvancedHTMLSlimTagFormatter, AdvancedHTMLSlimTagMiniFormatter

from AdvancedHTMLParser import __version__ as VERSION

def printVersion():
    sys.stdout.write ( 'AdvancedHTMLParser and formatHTML by Timothy Savannah (c) 2015-2018 Version %s\n\n' %(VERSION, ))

def printUsage():
    sys.stderr.write('''Usage: %s (Optional Arguments) (optional: /path/to/in.html) (optional: [/path/to/output.html])
  Formats HTML on input and writes to output.

 Optional Arguments:
 -------------------

    -e [encoding]        - Specify an encoding to use. Default is utf-8

    -m  or --mini        - Output "mini" HTML (only retain functional whitespace,
                            strip the rest and no indentation)

    -p  or --pretty      - Output "pretty" HTML [This is the defualt mode]

    -s or --slim         - Use the "slim" formatter to strip trailing-space in
                            start tags <span id="x"> vs <span id="x" >


    --indent='    '      - Use the provided string [default 4-spaces] to represent each
                            level of nesting. Use --indent="\t" for 1 tab insead, for example.
                           Affects pretty printing mode only

    --version            - Print version info on AdvancedHTMLParser suite and exit

    --help               - Print this message and exit


 If output filename is not specified or is empty string, output will be to stdout.
 If input filename is not specified or is empty string, input will be from stdin
 If -e is provided, will use that as the encoding. Defaults to utf-8

''' %(os.path.basename(sys.argv[0])))
    printVersion()

if __name__ == '__main__':


    indent = '  '
    inFilename  = None
    outFilename = None

    isMini = False
    isSlim = False

    args = sys.argv[1:]
    tmpArgs = args[:]
    encoding = 'utf-8'

    for i in range(len(args)):
        arg = args[i]
        if arg.startswith('-e'):
            if len(arg) > 2:
                encoding = arg[2:]
                args = args[:i] + args[i+1:]
                break
            else:
                if i == len(args) - 1:
                    sys.stderr.write('-e takes an argument, encoding.\n')
                    sys.exit(1)
                encoding = args[i+1]
                args = args[:i] + args[i+2:]
                break


    for arg in args[:]:
        if arg.startswith('--indent='):
            indent = arg[len('--indent='):].replace('\\t', '\t').replace('\\r', '\r').replace('\\n', '\n')
            if indent.isdigit():
                indent = ' ' * indent
            elif len(indent.replace(' ', '').replace('\t', '').replace('\r', '').replace('\n', '')) > 0:
                sys.stderr.write('Supported values for indent are: # of spaces, or a string of tabs, spaces, newlines which represent one level of indentation.\n')
                sys.exit(1)
            args.remove(arg)
        elif arg in ('-m', '--mini'):
            isMini = True
        elif arg in ('-p', '--pretty'):
            isMini = False
        elif arg in ('-s', '--slim'):
            isSlim = True
        elif arg == '--help':
            printUsage()
            sys.exit(0)
        elif arg == '--version':
            printVersion()
            sys.exit(0)
        elif inFilename is None:
            inFilename = arg
        elif outFilename is None:
            outFilename = arg
        else:
            sys.stderr.write('Too many arguments\n\n')
            printUsage()
            sys.exit(1)

    if inFilename:
        if not os.path.isfile(inFilename):
            sys.stderr.write('Input file "%s" does not exist.\n' %(inFilename,))
            sys.exit(1)
    else:
        try:
            inData = sys.stdin.read()
        except KeyboardInterrupt:
            # Intercept control+c and exit cleanly
            sys.exit(1)

    if outFilename:
        try:
            outFile = open(outFilename, 'w')
        except IOError:
            sys.stderr.write('Cannot open output file %s\n' %(outFilename,))
        
    
    if not isMini:
        if not isSlim:
            formatter = AdvancedHTMLFormatter(indent=indent, encoding=encoding)
        else:
            formatter = AdvancedHTMLSlimTagFormatter(indent=indent, encoding=encoding)
    else:
        if not isSlim:
            formatter = AdvancedHTMLMiniFormatter(encoding=encoding)
        else:
            formatter = AdvancedHTMLSlimTagMiniFormatter(encoding=encoding)

    if inFilename:
        formatter.parseFile(inFilename)
    else:
        formatter.parseStr(inData)

    if outFilename:
        outFile.write(formatter.getHTML())
        outFile.write('\n')
    else:
        sys.stdout.write(formatter.getHTML())
        sys.stdout.write('\n')