terom@50: """
terom@50:     Format LogLines into some other representation
terom@50: """
terom@50: 
terom@69: import re, xml.sax.saxutils
terom@50: 
terom@50: from log_line import LogTypes
terom@79: from log_formatter_pil import PILImageFormatter
terom@80: from log_formatter_rss import RSSFormatter
terom@50: 
terom@50: class LogFormatter (object) :
terom@50:     """
terom@65:         Provides a method to format series of LogLines into various output formats, with varying themes.
terom@50:     """
terom@51: 
terom@72:     # machine-readable name
terom@51:     name = None
terom@72: 
terom@72:     # human-readable name
terom@72:     title = None
terom@72: 
terom@72:     ## parameters
terom@72:     # use a fixed-width font for HTML output
terom@72:     html_fixedwidth = True
terom@72: 
terom@79:     def __init__ (self, tz, timestamp_fmt, img_ttf_path, img_font_size) :
terom@50:         """
terom@79:             Initialize to format timestamps with the given timezone and timestamp.
terom@79: 
terom@79:             Use the given TTF font to render image text with the given size, if given, otherwise, a default one.
terom@50:         """
terom@97:         
terom@97:         # store
terom@50:         self.tz = tz
terom@50:         self.timestamp_fmt = timestamp_fmt
terom@79:         self.img_ttf_path = img_ttf_path
terom@79:         self.img_font_size = img_font_size
terom@97:         
terom@97:         # XXX: harcoded
terom@97:         self.date_fmt = '%Y-%m-%d'
terom@50:     
terom@97:     def _format_line_text (self, line, template_dict, type=None, full_timestamp=False, **extra) :
terom@50:         """
terom@92:             Format the given line as text, using the given { type: string template } dict.
terom@92:             
terom@92:             If type is given, then it overrides line.type
terom@97: 
terom@97:             Any additional keyword args will also be available for the template to use
terom@50:         """
terom@92: 
terom@92:         # default type?
terom@92:         if type is None :
terom@92:             type = line.type
terom@65:             
terom@50:         # look up the template
terom@97:         if type in template_dict :
terom@97:             template = template_dict[type]
terom@97: 
terom@97:         else :
terom@97:             raise Exception("Format template not defined for type: %s" % LogTypes.name_from_code(type))
terom@65:         
terom@65:         # convert timestamp into display timezone
terom@65:         dtz = line.timestamp.astimezone(self.tz)
terom@65:         
terom@65:         # full timestamps?
terom@65:         if full_timestamp :
terom@97:             # XXX: let the user define a 'datetime' format instead?
terom@97:             timestamp_fmt = self.date_fmt + self.timestamp_fmt
terom@65: 
terom@65:         else :
terom@65:             timestamp_fmt = self.timestamp_fmt
terom@86:         
terom@86:         # breakdown source
terom@86:         source_nickname, source_username, source_hostname, source_chanflag = line.source
terom@86:         target_nickname = line.target
terom@50:         
terom@50:         # format with dict
terom@50:         return template % dict(
terom@86:             channel_name    = line.channel.name,
terom@86:             datetime        = dtz.strftime('%a %b %d %H:%M:%S %Y'),
terom@97:             date            = dtz.strftime(self.date_fmt),
terom@86:             timestamp       = dtz.strftime(timestamp_fmt),
terom@86:             source_nickname = source_nickname,
terom@86:             source_username = source_username,
terom@86:             source_hostname = source_hostname,
terom@86:             source_chanflag = source_chanflag,
terom@86:             target_nickname = target_nickname,
terom@86:             message         = line.data,
terom@97:             **extra
terom@50:         )
terom@50:     
terom@65:     def format_txt (self, lines, full_timestamps=False) :
terom@50:         """
terom@65:             Format given lines as plaintext.
terom@65: 
terom@65:             If full_timestamps is given, the output will contain full timestamps with both date and time.
terom@65: 
terom@65:             No trailing newlines.
terom@50:         """
terom@50: 
terom@50:         abstract
terom@50: 
terom@65:     def format_html (self, lines, full_timestamps=False) :
terom@50:         """
terom@65:             Format as HTML.
terom@65:             
terom@65:             See format_txt for information about arguments
terom@50:         """
terom@50: 
terom@50:         abstract
terom@79:     
terom@80:     def format_png (self, lines, full_timestamps=False) :
terom@79:         """
terom@79:             Format as a PNG image, returning the binary PNG data
terom@79:         """
terom@50: 
terom@80:         abstract
terom@80:     
terom@80:     def format_rss (self, lines, full_timestamps=False) :
terom@80:         """
terom@80:             Format as an XML RSS document
terom@80:         """
terom@80:         
terom@80:         abstract
terom@80: 
terom@86: class BaseHTMLFormatter (LogFormatter) :
terom@69:     """
terom@69:         Implements some HTML-formatting utils
terom@69:     """
terom@86:     
terom@86:     # parameters
terom@86:     html_fixedwidth = True
terom@69: 
terom@86:     # regexp to match URLs
terom@69:     URL_REGEXP = re.compile(r"http://\S+")
terom@69: 
terom@69:     def _process_links (self, line) :
terom@69:         """
terom@69:             Processed the rendered line, adding in <a href>'s for things that look like URLs, returning the new line.
terom@69: 
terom@69:             The line should already be escaped
terom@69:         """
terom@69: 
terom@69:         def _encode_url (match) :
terom@69:             # encode URL
terom@69:             url_html = match.group(0)
terom@69:             url_link = xml.sax.saxutils.unescape(url_html)
terom@69: 
terom@69:             return '<a href="%(url_link)s">%(url_html)s</a>' % dict(url_link=url_link, url_html=url_html)
terom@69: 
terom@69:         return self.URL_REGEXP.sub(_encode_url, line)
terom@86:  
terom@79:     def format_html (self, lines, **kwargs) :
terom@50:         """
terom@72:             Just uses format_txt, but processes links, etc
terom@50:         """
terom@50:         
terom@50:         # format using IrssiTextFormatter
terom@79:         for line, txt in self.format_txt(lines, **kwargs) :
terom@50:             # escape HTML
terom@72:             html = xml.sax.saxutils.escape(txt)
terom@69: 
terom@69:             # process links
terom@72:             html = self._process_links(html)
terom@69: 
terom@69:             # yield
terom@72:             yield line, html
terom@50: 
terom@86:    
terom@86: class IrssiTextFormatter (RSSFormatter, PILImageFormatter, LogFormatter) :
terom@86:     """
terom@86:         Implements format_txt for irssi-style output
terom@86:     """
terom@86: 
terom@86:     # format definitions by type
terom@86:     __FMT = {
terom@86:         LogTypes.RAW        : "%(timestamp)s %(data)s",
terom@86:         LogTypes.LOG_OPEN   : "--- Log opened %(datetime)s",
terom@86:         LogTypes.LOG_CLOSE  : "--- Log closed %(datetime)s",
terom@97:         'DAY_CHANGED'       : "--- Day changed %(date)s",
terom@86: 
terom@86:         LogTypes.MSG        : "%(timestamp)s <%(source_chanflag)s%(source_nickname)s> %(message)s",
terom@86:         LogTypes.NOTICE     : "%(timestamp)s -%(source_nickname)s- %(message)s",
terom@86:         LogTypes.ACTION     : "%(timestamp)s  * %(source_nickname)s %(message)s",
terom@86: 
terom@86:         LogTypes.JOIN       : "%(timestamp)s -!- %(source_nickname)s [%(source_username)s@%(source_hostname)s] has joined %(channel_name)s",
terom@86:         LogTypes.PART       : "%(timestamp)s -!- %(source_nickname)s [%(source_username)s@%(source_hostname)s] has left %(channel_name)s [%(message)s]",
terom@86:         LogTypes.KICK       : "%(timestamp)s -!- %(target_nickname)s was kicked from %(channel_name)s by %(source_nickname)s [%(message)s]",
terom@86:         LogTypes.MODE       : "%(timestamp)s -!- mode/%(channel_name)s [%(message)s] by %(source_nickname)s",
terom@86: 
terom@86:         LogTypes.NICK       : "%(timestamp)s -!- %(source_nickname)s is now known as %(target_nickname)s",
terom@86:         LogTypes.QUIT       : "%(timestamp)s -!- %(source_nickname)s [%(source_username)s@%(source_hostname)s] has quit [%(message)s]",
terom@86: 
terom@86:         LogTypes.TOPIC      : "%(timestamp)s -!- %(source_nickname)s changed the topic of %(channel_name)s to: %(message)s",
terom@92:         'TOPIC_UNSET'       : "%(timestamp)s -!- Topic unset by %(source_nickname)s on %(channel_name)s",
terom@86: 
terom@86:         LogTypes.SELF_NOTICE: "%(timestamp)s -%(source_nickname)s- %(message)s",
terom@86:         LogTypes.SELF_NICK  : "%(timestamp)s -!- %(source_nickname)s is now known as %(target_nickname)s",
terom@97: 
terom@97:         LogTypes.NETSPLIT_START : 
terom@97:                               "%(timestamp)s -!- Netsplit %(source_hostname)s <-> %(target_nickname)s quits: %(_netsplit_targets)s",
terom@97:         LogTypes.NETSPLIT_END :
terom@97:                               "%(timestamp)s -!- Netsplit over, joins: %(_netsplit_targets)s",
terom@86:     }
terom@86: 
terom@86:     def format_txt (self, lines, full_timestamps=False) :
terom@86:         # ...handle each line
terom@86:         for line in lines :
terom@97:             # extra args
terom@97:             extra = {}
terom@97: 
terom@97:             # specialcase type?
terom@92:             if line.type == LogTypes.TOPIC and line.data is None :
terom@92:                 type = 'TOPIC_UNSET'
terom@92:             
terom@92:             else :
terom@92:                 type = line.type
terom@92: 
terom@97:             # format netsplit stuff
terom@97:             if line.type & LogTypes._NETSPLIT_MASK :
terom@97:                 # format the netsplit-targets stuff
terom@97:                 extra['_netsplit_targets'] = line.data
terom@97: 
terom@86:             # using __TYPES
terom@97:             yield line, self._format_line_text(line, self.__FMT, type, full_timestamps, **extra)
terom@86: 
terom@86: class IrssiFormatter (BaseHTMLFormatter, IrssiTextFormatter) :
terom@86:     """
terom@86:         Implements plain black-and-white irssi-style formatting
terom@86:     """
terom@86:     
terom@86:     # name
terom@86:     name = 'irssi'
terom@86:     title = "Irssi (plain)"
terom@86: 
terom@86: class DebugFormatter (BaseHTMLFormatter) :
terom@86:     """
terom@86:         Implements a raw debug-style formatting of LogLines
terom@86:     """
terom@86: 
terom@86:     # name
terom@86:     name = 'debug'
terom@86:     title = "Raw debugging format"
terom@86:     
terom@86:     def format_txt (self, lines, full_timestamps=False) :
terom@86:         # iterate
terom@86:         for line in lines :
terom@86:             # just dump
terom@97:             yield line, unicode(line)
terom@86: 
terom@50: def by_name (name) :
terom@50:     """
terom@64:         Lookup and return a class LogFormatter by name
terom@50:     """
terom@50: 
terom@64:     return FORMATTERS[name]
terom@50: