2012-06-30 20:58:06 -03:00
|
|
|
#!/usr/bin/env python3
|
2012-07-03 18:11:40 -03:00
|
|
|
'Add syntax highlighting to Python source code'
|
2012-06-30 20:58:06 -03:00
|
|
|
|
2012-07-03 18:11:40 -03:00
|
|
|
__all__ = ['colorize', 'build_page', 'default_css', 'default_html',
|
|
|
|
|
'ansi_colorize', 'default_ansi']
|
2012-07-01 03:19:30 -03:00
|
|
|
__author__ = 'Raymond Hettinger'
|
2012-06-30 20:58:06 -03:00
|
|
|
|
|
|
|
|
import keyword, tokenize, cgi, functools
|
|
|
|
|
|
|
|
|
|
def is_builtin(s):
|
|
|
|
|
'Return True if s is the name of a builtin'
|
|
|
|
|
return s in vars(__builtins__)
|
|
|
|
|
|
2012-07-03 17:13:52 -03:00
|
|
|
def combine_range(lines, start, end):
|
|
|
|
|
'Join content from a range of lines between start and end'
|
2012-07-02 17:29:57 -03:00
|
|
|
(srow, scol), (erow, ecol) = start, end
|
|
|
|
|
if srow == erow:
|
|
|
|
|
rows = [lines[srow-1][scol:ecol]]
|
|
|
|
|
else:
|
|
|
|
|
rows = [lines[srow-1][scol:]] + lines[srow: erow-1] + [lines[erow-1][:ecol]]
|
2012-07-03 17:13:52 -03:00
|
|
|
return ''.join(rows), end
|
2012-07-02 17:29:57 -03:00
|
|
|
|
2012-07-03 17:13:52 -03:00
|
|
|
def isolate_tokens(source):
|
|
|
|
|
'Generate chunks of source and indentify chunks to be highlighted'
|
2012-07-03 04:15:59 -03:00
|
|
|
lines = source.splitlines(True)
|
2012-07-02 17:29:57 -03:00
|
|
|
lines.append('')
|
2012-06-30 20:58:06 -03:00
|
|
|
readline = functools.partial(next, iter(lines), '')
|
|
|
|
|
kind = tok_str = ''
|
|
|
|
|
tok_type = tokenize.COMMENT
|
2012-07-02 17:29:57 -03:00
|
|
|
written = (1, 0)
|
2012-06-30 20:58:06 -03:00
|
|
|
for tok in tokenize.generate_tokens(readline):
|
|
|
|
|
prev_tok_type, prev_tok_str = tok_type, tok_str
|
|
|
|
|
tok_type, tok_str, (srow, scol), (erow, ecol), logical_lineno = tok
|
2012-07-03 04:12:27 -03:00
|
|
|
kind = ''
|
2012-06-30 20:58:06 -03:00
|
|
|
if tok_type == tokenize.COMMENT:
|
|
|
|
|
kind = 'comment'
|
2012-07-01 04:37:05 -03:00
|
|
|
elif tok_type == tokenize.OP and tok_str[:1] not in '{}[](),.:;':
|
2012-06-30 20:58:06 -03:00
|
|
|
kind = 'operator'
|
|
|
|
|
elif tok_type == tokenize.STRING:
|
|
|
|
|
kind = 'string'
|
|
|
|
|
if prev_tok_type == tokenize.INDENT or scol==0:
|
|
|
|
|
kind = 'docstring'
|
|
|
|
|
elif tok_type == tokenize.NAME:
|
|
|
|
|
if tok_str in ('def', 'class', 'import', 'from'):
|
|
|
|
|
kind = 'definition'
|
|
|
|
|
elif prev_tok_str in ('def', 'class'):
|
|
|
|
|
kind = 'defname'
|
|
|
|
|
elif keyword.iskeyword(tok_str):
|
|
|
|
|
kind = 'keyword'
|
|
|
|
|
elif is_builtin(tok_str) and prev_tok_str != '.':
|
|
|
|
|
kind = 'builtin'
|
2012-07-03 17:13:52 -03:00
|
|
|
line_upto_token, written = combine_range(lines, written, (srow, scol))
|
|
|
|
|
line_thru_token, written = combine_range(lines, written, (erow, ecol))
|
|
|
|
|
yield kind, line_upto_token, line_thru_token
|
|
|
|
|
|
2012-07-03 18:11:40 -03:00
|
|
|
default_ansi = {
|
|
|
|
|
'comment': '\033[0;31m',
|
|
|
|
|
'string': '\033[0;32m',
|
|
|
|
|
'docstring': '\033[0;32m',
|
|
|
|
|
'keyword': '\033[0;33m',
|
|
|
|
|
'builtin': '\033[0;35m',
|
|
|
|
|
'definition': '\033[0;33m',
|
|
|
|
|
'defname': '\033[0;34m',
|
|
|
|
|
'operator': '\033[0;33m',
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
def colorize_ansi(source, colors=default_ansi):
|
|
|
|
|
'Add syntax highlighting to Python source code using ANSI escape sequences'
|
|
|
|
|
# http://en.wikipedia.org/wiki/ANSI_escape_code
|
|
|
|
|
result = []
|
|
|
|
|
for kind, line_upto_token, line_thru_token in isolate_tokens(source):
|
|
|
|
|
if kind:
|
|
|
|
|
result += [line_upto_token, colors[kind], line_thru_token, '\033[0m']
|
|
|
|
|
else:
|
|
|
|
|
result += [line_upto_token, line_thru_token]
|
|
|
|
|
return ''.join(result)
|
|
|
|
|
|
|
|
|
|
def colorize_html(source):
|
2012-07-03 17:13:52 -03:00
|
|
|
'Convert Python source code to an HTML fragment with colorized markup'
|
|
|
|
|
result = ['<pre class="python">\n']
|
|
|
|
|
for kind, line_upto_token, line_thru_token in isolate_tokens(source):
|
2012-06-30 20:58:06 -03:00
|
|
|
if kind:
|
2012-07-03 17:13:52 -03:00
|
|
|
result += [cgi.escape(line_upto_token),
|
|
|
|
|
'<span class="%s">' % kind,
|
|
|
|
|
cgi.escape(line_thru_token),
|
|
|
|
|
'</span>']
|
2012-07-02 17:29:57 -03:00
|
|
|
else:
|
2012-07-03 17:13:52 -03:00
|
|
|
result += [cgi.escape(line_upto_token),
|
|
|
|
|
cgi.escape(line_thru_token)]
|
|
|
|
|
result += ['</pre>\n']
|
2012-07-02 17:29:57 -03:00
|
|
|
return ''.join(result)
|
2012-06-30 20:58:06 -03:00
|
|
|
|
|
|
|
|
default_css = {
|
|
|
|
|
'.comment': '{color: crimson;}',
|
|
|
|
|
'.string': '{color: forestgreen;}',
|
2012-07-03 17:13:52 -03:00
|
|
|
'.docstring': '{color: forestgreen; font-style:italic;}',
|
2012-06-30 20:58:06 -03:00
|
|
|
'.keyword': '{color: darkorange;}',
|
|
|
|
|
'.builtin': '{color: purple;}',
|
|
|
|
|
'.definition': '{color: darkorange; font-weight:bold;}',
|
|
|
|
|
'.defname': '{color: blue;}',
|
|
|
|
|
'.operator': '{color: brown;}',
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
default_html = '''\
|
2012-07-01 02:19:04 -03:00
|
|
|
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
|
|
|
|
|
"http://www.w3.org/TR/html4/strict.dtd">
|
|
|
|
|
<html>
|
|
|
|
|
<head>
|
|
|
|
|
<meta http-equiv="Content-type" content="text/html;charset=UTF-8">
|
2012-07-02 21:17:16 -03:00
|
|
|
<title> {title} </title>
|
2012-07-01 02:19:04 -03:00
|
|
|
<style type="text/css">
|
2012-07-02 21:17:16 -03:00
|
|
|
{css}
|
2012-07-01 02:19:04 -03:00
|
|
|
</style>
|
|
|
|
|
</head>
|
2012-06-30 20:58:06 -03:00
|
|
|
<body>
|
2012-07-02 21:17:16 -03:00
|
|
|
{body}
|
2012-07-01 02:19:04 -03:00
|
|
|
</body>
|
|
|
|
|
</html>
|
2012-06-30 20:58:06 -03:00
|
|
|
'''
|
|
|
|
|
|
2012-07-01 03:19:30 -03:00
|
|
|
def build_page(source, title='python', css=default_css, html=default_html):
|
2012-06-30 20:58:06 -03:00
|
|
|
'Create a complete HTML page with colorized Python source code'
|
2012-07-01 02:19:04 -03:00
|
|
|
css_str = '\n'.join(['%s %s' % item for item in css.items()])
|
2012-07-03 18:11:40 -03:00
|
|
|
result = colorize_html(source)
|
2012-07-01 03:19:30 -03:00
|
|
|
title = cgi.escape(title)
|
2012-07-02 21:17:16 -03:00
|
|
|
return html.format(title=title, css=css_str, body=result)
|
2012-06-30 20:58:06 -03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
if __name__ == '__main__':
|
|
|
|
|
import sys, argparse, webbrowser, os
|
|
|
|
|
|
|
|
|
|
parser = argparse.ArgumentParser(
|
2012-07-03 18:11:40 -03:00
|
|
|
description = 'Add syntax highlighting to Python source')
|
2012-07-03 04:12:27 -03:00
|
|
|
parser.add_argument('sourcefile', metavar = 'SOURCEFILE',
|
2012-06-30 20:58:06 -03:00
|
|
|
help = 'File containing Python sourcecode')
|
2012-07-03 18:11:40 -03:00
|
|
|
parser.add_argument('-a', '--ansi', action = 'store_true',
|
|
|
|
|
help = 'emit ANSI escape highlighted source')
|
2012-06-30 20:58:06 -03:00
|
|
|
parser.add_argument('-b', '--browser', action = 'store_true',
|
|
|
|
|
help = 'launch a browser to show results')
|
2012-07-03 04:12:27 -03:00
|
|
|
parser.add_argument('-s', '--section', action = 'store_true',
|
|
|
|
|
help = 'show an HTML section rather than a complete webpage')
|
2012-06-30 20:58:06 -03:00
|
|
|
args = parser.parse_args()
|
2012-07-03 18:11:40 -03:00
|
|
|
|
2012-07-03 04:12:27 -03:00
|
|
|
if args.browser and args.section:
|
|
|
|
|
parser.error('The -s/--section option is incompatible with '
|
2012-06-30 20:58:06 -03:00
|
|
|
'the -b/--browser option')
|
2012-07-03 18:11:40 -03:00
|
|
|
if args.ansi and (args.browser or args.section):
|
|
|
|
|
parser.error('The -a/--ansi option is incompatible with '
|
|
|
|
|
'the -b/--browser and -s/--section options')
|
2012-06-30 20:58:06 -03:00
|
|
|
|
2012-07-03 04:12:27 -03:00
|
|
|
sourcefile = args.sourcefile
|
2012-06-30 20:58:06 -03:00
|
|
|
with open(sourcefile) as f:
|
|
|
|
|
page = f.read()
|
2012-07-03 18:11:40 -03:00
|
|
|
|
|
|
|
|
if args.ansi:
|
|
|
|
|
encoded = colorize_ansi(page)
|
|
|
|
|
elif args.section:
|
|
|
|
|
encoded = colorize_html(page)
|
|
|
|
|
else:
|
|
|
|
|
encoded = build_page(page, title=sourcefile)
|
|
|
|
|
|
2012-06-30 20:58:06 -03:00
|
|
|
if args.browser:
|
|
|
|
|
htmlfile = os.path.splitext(os.path.basename(sourcefile))[0] + '.html'
|
|
|
|
|
with open(htmlfile, 'w') as f:
|
2012-07-03 18:11:40 -03:00
|
|
|
f.write(encoded)
|
2012-06-30 20:58:06 -03:00
|
|
|
webbrowser.open('file://' + os.path.abspath(htmlfile))
|
|
|
|
|
else:
|
2012-07-03 18:11:40 -03:00
|
|
|
sys.stdout.write(encoded)
|
