2011-09-11 15:10:05 +00:00
|
|
|
|
# Copyright 2010-2011 Florent Le Coz <louiz@louiz.org>
|
2011-03-29 14:11:37 +00:00
|
|
|
|
#
|
|
|
|
|
# This file is part of Poezio.
|
|
|
|
|
#
|
|
|
|
|
# Poezio is free software: you can redistribute it and/or modify
|
2011-09-11 15:10:05 +00:00
|
|
|
|
# it under the terms of the zlib license. See the COPYING file.
|
2011-03-29 14:11:37 +00:00
|
|
|
|
|
|
|
|
|
"""
|
|
|
|
|
Various methods to convert
|
|
|
|
|
shell colors to poezio colors,
|
|
|
|
|
xhtml code to shell colors,
|
|
|
|
|
poezio colors to xhtml code
|
|
|
|
|
"""
|
|
|
|
|
|
|
|
|
|
import re
|
|
|
|
|
import subprocess
|
2011-04-10 14:44:26 +00:00
|
|
|
|
from sleekxmpp.xmlstream import ET
|
|
|
|
|
from xml.etree.ElementTree import ElementTree
|
2011-05-04 20:25:26 +00:00
|
|
|
|
from sys import version_info
|
2011-03-29 14:11:37 +00:00
|
|
|
|
|
2011-04-04 17:19:27 +00:00
|
|
|
|
from config import config
|
2011-03-29 18:36:02 +00:00
|
|
|
|
import logging
|
|
|
|
|
|
2011-06-04 18:15:18 +00:00
|
|
|
|
digits = '0123456789' # never trust the modules
|
|
|
|
|
|
2011-03-29 18:36:02 +00:00
|
|
|
|
log = logging.getLogger(__name__)
|
|
|
|
|
|
2011-03-29 14:11:37 +00:00
|
|
|
|
shell_colors_re = re.compile(r'(\[(?:\d+;)*(?:\d+m))')
|
2011-03-30 02:34:45 +00:00
|
|
|
|
start_indent_re = re.compile(r'\[0;30m\[0;37m ')
|
|
|
|
|
newline_indent_re = re.compile('\n\[0;37m ')
|
2011-03-29 14:11:37 +00:00
|
|
|
|
|
2011-03-29 18:36:02 +00:00
|
|
|
|
def get_body_from_message_stanza(message):
|
|
|
|
|
"""
|
|
|
|
|
Returns a string with xhtml markups converted to
|
|
|
|
|
poezio colors if there's an xhtml_im element, or
|
|
|
|
|
the body (without any color) otherwise
|
|
|
|
|
"""
|
2011-04-10 14:44:26 +00:00
|
|
|
|
if config.get('enable_xhtml_im', 'true') == 'true':
|
2011-04-04 17:19:27 +00:00
|
|
|
|
xhtml_body = message['xhtml_im']
|
|
|
|
|
if xhtml_body:
|
2011-04-10 14:44:26 +00:00
|
|
|
|
xhtml_body = convert_links_to_plaintext(xhtml_body)
|
2011-04-04 17:19:27 +00:00
|
|
|
|
try:
|
|
|
|
|
shell_body = xhtml_code_to_shell_colors(xhtml_body)
|
|
|
|
|
except OSError:
|
2011-08-16 12:53:29 +00:00
|
|
|
|
log.debug('html parsing failed')
|
2011-04-04 17:19:27 +00:00
|
|
|
|
else:
|
|
|
|
|
return shell_colors_to_poezio_colors(shell_body)
|
2011-03-29 18:36:02 +00:00
|
|
|
|
return message['body']
|
|
|
|
|
|
2011-04-10 14:44:26 +00:00
|
|
|
|
def convert_links_to_plaintext(text):
|
|
|
|
|
"""
|
|
|
|
|
Replace
|
|
|
|
|
<a href='URL'>click</a>
|
|
|
|
|
by
|
|
|
|
|
<url> (click)
|
|
|
|
|
in plain text
|
|
|
|
|
"""
|
|
|
|
|
log.debug(text)
|
|
|
|
|
xml = ElementTree(ET.fromstring(text))
|
|
|
|
|
for parent in xml.getiterator():
|
|
|
|
|
previous_child = None
|
|
|
|
|
for child in parent:
|
|
|
|
|
if child.tag == '{http://www.w3.org/1999/xhtml}a':
|
2011-05-27 21:36:20 +00:00
|
|
|
|
if child.attrib['href'] != child.text:
|
|
|
|
|
if child.text is None and 'title' in child.attrib:
|
2011-06-04 18:15:18 +00:00
|
|
|
|
link_text = '\n%s (%s)' % (child.attrib['href'], child.attrib['title'])
|
2011-05-27 21:36:20 +00:00
|
|
|
|
else:
|
2011-06-04 18:15:18 +00:00
|
|
|
|
link_text = '\n%s (%s)' % (child.attrib['href'], child.text)
|
2011-05-27 21:36:20 +00:00
|
|
|
|
else:
|
|
|
|
|
link_text = child.text
|
2011-04-10 14:44:26 +00:00
|
|
|
|
if previous_child is not None:
|
|
|
|
|
if previous_child.tail is None:
|
|
|
|
|
previous_child.tail = link_text
|
|
|
|
|
else:
|
|
|
|
|
previous_child.tail += link_text
|
|
|
|
|
else:
|
|
|
|
|
if parent.text is None:
|
|
|
|
|
parent.text = link_text
|
|
|
|
|
else:
|
|
|
|
|
parent.text += link_text
|
|
|
|
|
parent.remove(child)
|
|
|
|
|
previous_child = child
|
2011-05-04 20:25:26 +00:00
|
|
|
|
if version_info.minor <= 1:
|
|
|
|
|
return ET.tostring(xml.getroot())
|
2011-05-04 20:05:02 +00:00
|
|
|
|
return ET.tostring(xml.getroot(), encoding=str)
|
2011-04-10 14:44:26 +00:00
|
|
|
|
|
2011-05-04 20:25:26 +00:00
|
|
|
|
|
2011-03-30 02:34:45 +00:00
|
|
|
|
def clean_text(string):
|
|
|
|
|
"""
|
|
|
|
|
Remove all \x19 from the string
|
|
|
|
|
"""
|
|
|
|
|
pos = string.find('\x19')
|
|
|
|
|
while pos != -1:
|
|
|
|
|
string = string[:pos] + string[pos+2:]
|
|
|
|
|
pos = string.find('\x19')
|
|
|
|
|
return string
|
|
|
|
|
|
2011-03-29 18:56:11 +00:00
|
|
|
|
number_to_color_names = {
|
|
|
|
|
1: 'red',
|
|
|
|
|
2: 'green',
|
2011-03-30 02:34:45 +00:00
|
|
|
|
3: 'yellow',
|
|
|
|
|
4: 'blue',
|
|
|
|
|
5: 'violet',
|
|
|
|
|
6: 'turquoise',
|
|
|
|
|
7: 'white'
|
2011-03-29 18:56:11 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
def poezio_colors_to_html(string):
|
|
|
|
|
"""
|
|
|
|
|
Convert poezio colors to html makups
|
|
|
|
|
(e.g. \x191: <span style='color: red'>)
|
|
|
|
|
"""
|
|
|
|
|
# TODO underlined
|
|
|
|
|
|
|
|
|
|
# a list of all opened elements, e.g. ['strong', 'span']
|
|
|
|
|
# So that we know what we need to close
|
|
|
|
|
opened_elements = []
|
2011-03-30 02:34:45 +00:00
|
|
|
|
res = "<body xmlns='http://www.w3.org/1999/xhtml'><p>"
|
2011-03-29 18:56:11 +00:00
|
|
|
|
next_attr_char = string.find('\x19')
|
|
|
|
|
while next_attr_char != -1:
|
|
|
|
|
attr_char = string[next_attr_char+1].lower()
|
|
|
|
|
if next_attr_char != 0:
|
|
|
|
|
res += string[:next_attr_char]
|
|
|
|
|
string = string[next_attr_char+2:]
|
|
|
|
|
if attr_char == 'o':
|
|
|
|
|
for elem in opened_elements[::-1]:
|
|
|
|
|
res += '</%s>' % (elem,)
|
|
|
|
|
opened_elements = []
|
|
|
|
|
elif attr_char == 'b':
|
|
|
|
|
if 'strong' not in opened_elements:
|
|
|
|
|
opened_elements.append('strong')
|
|
|
|
|
res += '<strong>'
|
2011-05-13 09:46:50 +00:00
|
|
|
|
elif attr_char in digits:
|
2011-03-29 18:56:11 +00:00
|
|
|
|
number = int(attr_char)
|
|
|
|
|
if number in number_to_color_names:
|
2011-05-11 12:33:22 +00:00
|
|
|
|
if 'strong' in opened_elements:
|
|
|
|
|
res += '</strong>'
|
|
|
|
|
opened_elements.remove('strong')
|
2011-03-29 18:56:11 +00:00
|
|
|
|
if 'span' in opened_elements:
|
|
|
|
|
res += '</span>'
|
2011-03-30 02:48:02 +00:00
|
|
|
|
else:
|
|
|
|
|
opened_elements.append('span')
|
2011-03-29 18:56:11 +00:00
|
|
|
|
res += "<span style='color: %s'>" % (number_to_color_names[number])
|
|
|
|
|
next_attr_char = string.find('\x19')
|
|
|
|
|
res += string
|
|
|
|
|
for elem in opened_elements[::-1]:
|
|
|
|
|
res += '</%s>' % (elem,)
|
2011-03-29 19:07:53 +00:00
|
|
|
|
res += "</p></body>"
|
|
|
|
|
return res.replace('\n', '<br />')
|
2011-03-29 18:56:11 +00:00
|
|
|
|
|
2011-03-29 14:11:37 +00:00
|
|
|
|
def shell_colors_to_poezio_colors(string):
|
|
|
|
|
"""
|
|
|
|
|
'shell colors' means something like:
|
|
|
|
|
|
|
|
|
|
Bonjour ^[[0;32msalut^[[0m
|
|
|
|
|
|
|
|
|
|
The current understanding of this syntax is:
|
|
|
|
|
n = 0: reset all attributes to defaults
|
2011-03-30 02:34:45 +00:00
|
|
|
|
n = 1: activate bold
|
2011-03-29 14:11:37 +00:00
|
|
|
|
n >= 30 and n <= 37: set the foreground to n-30
|
|
|
|
|
|
|
|
|
|
"""
|
|
|
|
|
def repl(matchobj):
|
|
|
|
|
exp = matchobj.group(0)[2:-1]
|
|
|
|
|
numbers = [int(nb) for nb in exp.split(';')]
|
|
|
|
|
res = ''
|
|
|
|
|
for num in numbers:
|
|
|
|
|
if num == 0:
|
2011-03-30 02:34:45 +00:00
|
|
|
|
res += '\x19o'
|
|
|
|
|
elif num == 1:
|
|
|
|
|
res += '\x19b'
|
|
|
|
|
elif num >= 31 and num <= 37:
|
2011-04-04 17:19:27 +00:00
|
|
|
|
res += '\x19%d' % ((num-30)%7,)
|
2011-03-29 14:11:37 +00:00
|
|
|
|
return res
|
2011-03-30 02:34:45 +00:00
|
|
|
|
def remove_elinks_indent(string):
|
|
|
|
|
lines = string.split('\n')
|
|
|
|
|
for i, line in enumerate(lines):
|
|
|
|
|
lines[i] = re.sub(' ', '', line, 1)
|
|
|
|
|
return '\n'.join(lines)
|
|
|
|
|
res = shell_colors_re.sub(repl, string).strip()
|
|
|
|
|
res = remove_elinks_indent(res)
|
|
|
|
|
return res
|
2011-03-29 14:11:37 +00:00
|
|
|
|
|
|
|
|
|
def xhtml_code_to_shell_colors(string):
|
|
|
|
|
"""
|
|
|
|
|
Use a console browser to parse the xhtml and
|
|
|
|
|
make it return a shell-colored string
|
|
|
|
|
"""
|
|
|
|
|
process = subprocess.Popen(["elinks", "-dump", "-dump-color-mode", "2"], stdout=subprocess.PIPE, stdin=subprocess.PIPE)
|
|
|
|
|
result = process.communicate(input=string.encode('utf-8'))[0]
|
|
|
|
|
return result.decode('utf-8').strip()
|
|
|
|
|
|
2011-03-30 02:34:45 +00:00
|
|
|
|
def poezio_colors_to_xhtml(string):
|
|
|
|
|
"""
|
|
|
|
|
Generate a valid xhtml string from
|
|
|
|
|
the poezio colors in the given string
|
|
|
|
|
"""
|
|
|
|
|
res = "<body xmlns='http://www.w3.org/1999/xhtml'>"
|
|
|
|
|
next_attr_char = string.find('\x19')
|
|
|
|
|
open_elements = []
|
|
|
|
|
while next_attr_char != -1:
|
|
|
|
|
attr_char = string[next_attr_char+1].lower()
|
|
|
|
|
if next_attr_char != 0:
|
|
|
|
|
res += string[:next_attr_char]
|
|
|
|
|
string = string[next_attr_char+2:]
|
|
|
|
|
if attr_char == 'o':
|
|
|
|
|
# close all opened elements
|
|
|
|
|
for elem in open_elements:
|
|
|
|
|
res += '</%s>'
|
|
|
|
|
open_elements = []
|
|
|
|
|
elif attr_char == 'b':
|
|
|
|
|
if 'strong' not in open_elements:
|
|
|
|
|
res += '<strong>'
|
|
|
|
|
open_elements.append('strong')
|
2011-05-13 09:46:50 +00:00
|
|
|
|
elif attr_char in digits:
|
2011-03-30 02:34:45 +00:00
|
|
|
|
self._win.attron(common.curses_color_pair(int(attr_char)))
|
|
|
|
|
next_attr_char = string.find('\x19')
|
|
|
|
|
|
2011-04-10 14:44:26 +00:00
|
|
|
|
|