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@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@51:     # the formatter's code name
terom@51:     name = None
terom@50:     
terom@50:     def __init__ (self, tz, timestamp_fmt="%H:%M:%S") :
terom@50:         """
terom@50:             Initialize to format timestamps with the given timezone and timestamp
terom@50:         """
terom@50: 
terom@50:         self.tz = tz
terom@50:         self.timestamp_fmt = timestamp_fmt
terom@50:     
terom@65:     def _format_line_text (self, line, template_dict, full_timestamp=False) :
terom@50:         """
terom@50:             Format the given line as text, using the given { type: string template } dict
terom@50:         """
terom@65:             
terom@50:         # look up the template
terom@50:         template = template_dict[line.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@65:             # XXX: ugly
terom@65:             timestamp_fmt = '%Y-%m-%d ' + self.timestamp_fmt
terom@65: 
terom@65:         else :
terom@65:             timestamp_fmt = self.timestamp_fmt
terom@50: 
terom@50:         # build timestamp
terom@65:         timestamp = dtz.strftime(timestamp_fmt)
terom@50:         
terom@50:         # format with dict
terom@50:         return template % dict(
terom@50:             timestamp       = timestamp,
terom@50:             source          = line.source,
terom@50:             data            = line.data,
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@50: 
terom@69: class BaseHTMLFormatter (object) :
terom@69:     """
terom@69:         Implements some HTML-formatting utils
terom@69:     """
terom@69: 
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@69: 
terom@50: class IrssiTextFormatter (LogFormatter) :
terom@50:     """
terom@50:         Implements format_txt for irssi-style output
terom@50:     """
terom@50: 
terom@50:     # format definitions by type
terom@50:     __FMT = {
terom@50:         LogTypes.RAW        : "%(timestamp)s %(data)s",
terom@50:     }
terom@50: 
terom@65:     def format_txt (self, lines, full_timestamps=False) :
terom@50:         # ...handle each line
terom@50:         for line in lines :
terom@50:             # using __TYPES
terom@65:             yield self._format_line_text(line, self.__FMT, full_timestamps)
terom@50: 
terom@69: class IrssiFormatter (IrssiTextFormatter, BaseHTMLFormatter) :
terom@50:     """
terom@50:         Implements plain black-and-white irssi-style formatting
terom@50:     """
terom@50: 
terom@51:     name = 'irssi'
terom@59:     title = "Irssi (plain)"
terom@51: 
terom@65:     def format_html (self, lines, full_timestamps=False) :
terom@50:         """
terom@50:             Just uses format_txt, but wraps in <pre></pre>
terom@50:         """
terom@50:         
terom@50:         # open pre
terom@50:         yield "<pre>"
terom@50:         
terom@50:         # format using IrssiTextFormatter
terom@65:         for line in self.format_txt(lines, full_timestamps) :
terom@50:             # escape HTML
terom@69:             line = xml.sax.saxutils.escape(line)
terom@69: 
terom@69:             # process links
terom@69:             line = self._process_links(line)
terom@69: 
terom@69:             # yield
terom@69:             yield line
terom@50: 
terom@50:         # close pre
terom@50:         yield "</pre>"
terom@50: 
terom@50: # define formatters by name
terom@50: FORMATTERS = {
terom@50:     'irssi':        IrssiFormatter,
terom@50: }
terom@50: 
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: