diff options
| author | Emmanuel Gil Peyrot <linkmauve@linkmauve.fr> | 2011-09-23 23:47:25 +0200 |
|---|---|---|
| committer | Emmanuel Gil Peyrot <linkmauve@linkmauve.fr> | 2011-09-23 23:47:25 +0200 |
| commit | 8f675044b1fb9d102b7a2732682faa00d812ab19 (patch) | |
| tree | 545067e30b0099391b6435795f8957dca7a0edd0 | |
| parent | 8845cdce8e84004e5e4878bfad84ec59ea2f68ac (diff) | |
| download | poezio-8f675044b1fb9d102b7a2732682faa00d812ab19.tar.gz poezio-8f675044b1fb9d102b7a2732682faa00d812ab19.tar.bz2 poezio-8f675044b1fb9d102b7a2732682faa00d812ab19.tar.xz poezio-8f675044b1fb9d102b7a2732682faa00d812ab19.zip | |
Really fix whitespace handling, and malformed CSS.
| -rw-r--r-- | src/xhtml.py | 14 |
1 files changed, 10 insertions, 4 deletions
diff --git a/src/xhtml.py b/src/xhtml.py index fd50bb11..517d9b80 100644 --- a/src/xhtml.py +++ b/src/xhtml.py @@ -174,6 +174,7 @@ colors = { log = logging.getLogger(__name__) +whitespace_re = re.compile(r'\s') shell_colors_re = re.compile(r'(\[(?:\d+;)*(?:\d+m))') start_indent_re = re.compile(r'\[0;30m\[0;37m ') newline_indent_re = re.compile('\n\[0;37m ') @@ -221,6 +222,8 @@ def xhtml_to_poezio_colors(text): shell = '' rules = css.split(';') for rule in rules: + if ':' not in rule: + continue key, value = rule.split(':', 1) key = key.strip() value = value.strip() @@ -244,13 +247,16 @@ def xhtml_to_poezio_colors(text): shell += '\x19a' return shell + def trim(string): + return re.sub(whitespace_re, ' ', string) + log.debug(text) xml = ET.fromstring(text) message = '' for elem in xml.iter(): if elem.tag == '{http://www.w3.org/1999/xhtml}a': if 'href' in elem.attrib and elem.attrib['href'] != elem.text: - message += '\x19u%s\x19o (%s)' % (elem.attrib['href'].strip(), elem.text.strip()) + message += '\x19u%s\x19o (%s)' % (trim(elem.attrib['href']), trim(elem.text)) else: message += '\x19u' + elem.text + '\x19o' elif elem.tag == '{http://www.w3.org/1999/xhtml}blockquote': @@ -265,7 +271,7 @@ def xhtml_to_poezio_colors(text): message += '\x19i' elif elem.tag == '{http://www.w3.org/1999/xhtml}img' and 'src' in elem.attrib: if 'alt' in elem.attrib: - message += '%s (%s)' % (elem.attrib['src'].strip(), elem.attrib['alt'].strip()) + message += '%s (%s)' % (trim(elem.attrib['src']), trim(elem.attrib['alt'])) else: message += elem.attrib['src'] elif elem.tag == '{http://www.w3.org/1999/xhtml}li': @@ -289,7 +295,7 @@ def xhtml_to_poezio_colors(text): if (elem.text and elem.tag != '{http://www.w3.org/1999/xhtml}a' and elem.tag != '{http://www.w3.org/1999/xhtml}br' and elem.tag != '{http://www.w3.org/1999/xhtml}img'): - message += elem.text.strip() + message += trim(elem.text) if ('style' in elem.attrib and elem.tag != '{http://www.w3.org/1999/xhtml}br' and elem.tag != '{http://www.w3.org/1999/xhtml}em' @@ -311,7 +317,7 @@ def xhtml_to_poezio_colors(text): message += ' [' + elem.attrib['title'] + ']' if elem.tail: - message += elem.tail.strip() + message += trim(elem.tail) return message |