1

import datetime, time

     2

import re

     3

     4

import logging; log = logging.getLogger('pvl.syslog.parser')

     5

     6

class SyslogParser (object) :

     7

    """

     8

        Parse syslog lines in text format, as used in logfiles/fifos.

     9

    """

    49

    # default syslogd format

    50

    SYSLOG_RE = re.compile(

    51

        # the timestamp+hostname header

    52

        # XXX:  hostname may be missing

    53

        #       at least in Ubuntu 11.10 syslogd 'last message repeated 2 times'...

    56

    57

        # the message, including possible tag/pid

    58

        +   r"(?P<message>(?P<tag>(?P<program>[^:\]]+)(?:\[(?P<pid>\d+)\])?: )?(?P<text>.*))\n?"

    59

    )

    60

    61

    TIMESTAMP_FMT = '%b %d %H:%M:%S'

    62

    63

    def __init__ (self, raw=False) :

    64

        """

    65

            Using given underlying line source.

    66

        """

    67

    68

        self.raw = raw

    69

    89

    def parse_timestamp (self, match) :

    90

        """

    91

            Parse timstamp from line into datetime.

    92

        """

    93

    94

        timestamp = match.group('timestamp')

    95

    96

        # add missing year; assume current

    97

        timestamp = time.strftime('%Y') + ' ' + timestamp

    98

    99

        # k

   100

        timestamp = datetime.datetime.strptime(timestamp, '%Y ' + self.TIMESTAMP_FMT)

   101

   102

        return timestamp

   103

   104

    def parse_prog (self, match) :

   105

        """

   106

            Parse prog from line.

   107

        """

   108

   109

        prog = match.group('program')

   110

   111

        if prog :

   112

            return prog

   113

        else :

   114

            # no tag

   115

            return None

   116

   117

    def parse (self, line) :

   118

        """

   119

            Parse given input line into SyslogMessage.

   120

        """

   121

   122

        # ignore whitespace

   123

        line = line.strip()

   124

   125

        # debug

   126

        log.debug("%s", line)

   127

   128

        # timestamp?

   129

        if self.raw :

   130

            # from defaults

   131

            return dict(

   132

                timestamp   = datetime.datetime.now(), # XXX: None?

   133

                host        = None,

   134

                prog        = None,

   135

                pid         = None,

   136

                msg         = line,

   137

            )

   138

   139

        else :

   140

            # parse

   141

            match = self.SYSLOG_RE.match(line)

   142

   143

            if not match :

   144

                log.warn("Unparseable syslog message: %r", line)

   145

                return

   146

   147

            # parse

   149

                timestamp   = self.parse_timestamp(match),

   150

                host        = match.group('hostname'),

   151

                prog        = self.parse_prog(match),

   152

                pid         = match.group('pid'),

   153

                msg         = match.group('text'),

   154

            )

   161

   162

    def process (self, lines) :

   163

        """

   164

            Yield SyslogMessages from given series of lines.

   165

        """

   166

   167

        for line in lines :

   168

            item = self.parse(line)

   169

   170

            if item :

   171

                yield item

   172

   173

    __call__ = process

   174