terom@44: import logging; log = logging.getLogger('pvl.syslog.filter')
terom@44: 
terom@52: import re # XXX
terom@52: import os.path, fnmatch
terom@52: 
terom@44: class SyslogFilter (object) :
terom@44:     """
terom@52:         Match syslog messages fields against given patterns.
terom@44:     """
terom@44: 
terom@52:     def __init__ (self, **filters) :
terom@44:         """
terom@52:             Match using given { field: regex }.
terom@44:         """
terom@45:         
terom@52:         self.filters = filters
terom@44: 
terom@52:     def match_prog (self, attr, filter, prog=None) :
terom@44:         """
terom@52:             XXX: Match given prog as glob?
terom@44:         """
terom@44: 
terom@52:         if not filter :
terom@52:             # ignore
terom@52:             return None
terom@45: 
terom@44:         # normalize
terom@52:         prog = prog.strip().lower()
terom@52:         _, prog = os.path.split(prog)
terom@44: 
terom@44:         # match
terom@52:         if fnmatch.fnmatch(prog, filter) :
terom@52:             return { attr: prog }
terom@52:         else :
terom@44:             return False
terom@44: 
terom@52:     def match (self, attr, regex, value=None) :
terom@52:         """
terom@52:             Match given value against given pattern.
terom@52:         """
terom@44: 
terom@52:         if not filter :
terom@52:             # ignore
terom@52:             return None
terom@52: 
terom@52:         if not value :
terom@52:             # XXX: optional = match empty string?
terom@52:             value = ''
terom@44:         else :
terom@52:             # normalize; XXX: unicode?
terom@52:             value = str(value).strip()
terom@52:         
terom@52:         # match
terom@52:         match = regex.match(value)
terom@52: 
terom@52:         if match :
terom@52:             # as matche-values
terom@52:             match = match.groupdict()
terom@52:             match[attr] = value
terom@52: 
terom@52:         return match
terom@44: 
terom@44:     def filter (self, item) :
terom@44:         """
terom@52:             Match given item. Returns any matched values (including regexp capture groups) across all fields.
terom@44:         """
terom@44: 
terom@52:         matches = {}
terom@52: 
terom@52:         for attr in self.filters :
terom@52:             # lookup match-func
terom@52:             match = getattr(self, 'match_{attr}'.format(attr=attr), self.match)
terom@52: 
terom@52:             # filter
terom@52:             filter = self.filters[attr]
terom@52:             
terom@52:             # apply match
terom@52:             if attr in item :
terom@52:                 match = match(attr, filter, item[attr])
terom@52:             else :
terom@52:                 match = match(attr, filter)
terom@52: 
terom@52:             log.debug("%s: %s", attr, match)
terom@44: 
terom@45:             if match :
terom@52:                 # match
terom@52:                 matches.update(match)
terom@45:             
terom@45:             elif match is None :
terom@52:                 # ignore
terom@52:                 continue
terom@45: 
terom@45:             else :
terom@52:                 # reject
terom@52:                 return
terom@45:         
terom@52:         # XXX: test last match, in case they were all None
terom@45:         if match is None :
terom@45:             # XXX: empty filter!?
terom@45:             return True
terom@44:         else :
terom@52:             return matches
terom@44: 
terom@52:     def process (self, items) :
terom@44:         for item in items:
terom@52:             match = self.filter(item)
terom@52: 
terom@52:             if match :
terom@44:                 yield item
terom@52:     
terom@52:     __call__ = process