author | Tero Marttila <terom@fixme.fi> |
Fri, 05 Feb 2010 22:56:14 +0200 | |
changeset 65 | 1de3c0ca2baf |
parent 53 | 21ab25ffa1e8 |
permissions | -rw-r--r-- |
50 | 1 |
""" |
2 |
Log messages (e.g. syslog) |
|
3 |
""" |
|
4 |
||
5 |
# stdlib syslog, for facilities and levels |
|
6 |
import syslog |
|
7 |
import time, datetime |
|
8 |
import re |
|
65
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
9 |
import os.path |
50 | 10 |
|
11 |
PRIORITY_NAMES = { |
|
12 |
syslog.LOG_EMERG: 'emerg', |
|
13 |
syslog.LOG_ALERT: 'alert', |
|
14 |
syslog.LOG_CRIT: 'crit', |
|
15 |
syslog.LOG_ERR: 'err', |
|
16 |
syslog.LOG_WARNING: 'warning', |
|
17 |
syslog.LOG_NOTICE: 'notice', |
|
18 |
syslog.LOG_INFO: 'info', |
|
19 |
syslog.LOG_DEBUG: 'debug', |
|
20 |
} |
|
21 |
||
22 |
FACILITY_NAMES = { |
|
23 |
syslog.LOG_KERN: 'kern', # 0 |
|
24 |
syslog.LOG_USER: 'user', # 1 |
|
25 |
syslog.LOG_MAIL: 'mail', # 2 |
|
26 |
syslog.LOG_DAEMON: 'daemon', # 3 |
|
27 |
syslog.LOG_AUTH: 'auth', # 4 |
|
28 |
5: 'syslog', # 5 |
|
29 |
syslog.LOG_LPR: 'lpr', # 6 |
|
30 |
syslog.LOG_NEWS: 'news', # 7 |
|
31 |
syslog.LOG_UUCP: 'uucp', # 8 |
|
32 |
syslog.LOG_CRON: 'cron', # 9 |
|
33 |
10: 'authpriv', # 10 |
|
34 |
11: 'ftp', # 11 |
|
35 |
12: 'ntp', # 12 |
|
36 |
13: 'audit', # 13 |
|
37 |
14: 'alert', # 14 |
|
38 |
15: 'clock', # 15 |
|
39 |
syslog.LOG_LOCAL0: 'local0', # 16 |
|
40 |
syslog.LOG_LOCAL1: 'local1', # 17 |
|
41 |
syslog.LOG_LOCAL2: 'local2', # 18 |
|
42 |
syslog.LOG_LOCAL3: 'local3', # 19 |
|
43 |
syslog.LOG_LOCAL4: 'local4', # 20 |
|
44 |
syslog.LOG_LOCAL5: 'local5', # 21 |
|
45 |
syslog.LOG_LOCAL6: 'local6', # 22 |
|
46 |
syslog.LOG_LOCAL7: 'local7', # 23 |
|
47 |
} |
|
48 |
||
49 |
class SyslogMessage (object) : |
|
50 |
""" |
|
51 |
A message from syslog with the following fields: |
|
52 |
||
53 |
pri - (optional) raw integer priority field |
|
54 |
priority - (optional) message priority as text |
|
55 |
facility - (optional) message facility as text |
|
56 |
timestamp - message timestamp as a datetime |
|
57 |
hostname - originating hostname |
|
58 |
tag - (optional) message tag including process name and id |
|
59 |
program - (optional) process name part of message tag |
|
60 |
pid - (optional) process ID part of message tag |
|
61 |
text - the log message following the tag |
|
62 |
message - the full log message including the tag |
|
63 |
raw - the full syslog-format message |
|
64 |
""" |
|
65 |
||
66 |
# the regular expression used to parse the lines |
|
67 |
LINE_RE = re.compile( |
|
68 |
# the priority field as in raw syslog messages |
|
69 |
r"(?:<(?P<pri>\d+)>)?" |
|
70 |
||
71 |
# the timestamp+hostname header |
|
72 |
+ r"(?P<timestamp>\w{3} [0-9 ]\d \d{2}:\d{2}:\d{2}) (?P<hostname>\S+)" |
|
73 |
||
74 |
# the message, including possible tag/pid |
|
53 | 75 |
+ r" (?P<message>(?P<tag>(?P<program>[^:\]]+)(?:\[(?P<pid>\d+)\])?: )?(?P<text>.*))\n?" |
50 | 76 |
) |
77 |
||
78 |
# strptime format of timestamp |
|
79 |
TIMESTAMP_FMT = "%b %d %H:%M:%S" |
|
80 |
||
81 |
def _parse_pri (self, match) : |
|
82 |
""" |
|
83 |
Parse the priority/facility from the given match object |
|
84 |
""" |
|
85 |
||
86 |
# raw integer |
|
87 |
self.pri = int(match.group('pri')) if match.group('pri') else None |
|
88 |
||
89 |
if self.pri : |
|
90 |
# unpack |
|
91 |
priority = self.pri % 8 |
|
92 |
facility = self.pri // 8 |
|
93 |
||
94 |
# translate to names |
|
95 |
self.priority = PRIORITY_NAMES.get(priority, str(priority)) |
|
96 |
self.facility = FACILITY_NAMES.get(facility, str(facility)) |
|
97 |
||
98 |
else : |
|
99 |
self.priority = self.facility = None |
|
100 |
||
101 |
def _parse_timestamp (self, match) : |
|
102 |
""" |
|
103 |
Parse the timestamp field into a datetime.datetime from the given match object |
|
104 |
""" |
|
105 |
||
106 |
timestamp = match.group('timestamp') |
|
107 |
||
108 |
try : |
|
109 |
ts = time.strptime(timestamp, self.TIMESTAMP_FMT) |
|
110 |
||
111 |
except Exception, ex : |
|
112 |
raise ValueError("Invalid timestamp: %s: %s" % (timestamp, ex)) |
|
113 |
||
114 |
# build timestamp |
|
115 |
self.timestamp = datetime.datetime( |
|
116 |
# fix year - strptime default is 1900 |
|
117 |
ts.tm_year if ts.tm_year != 1900 else time.localtime().tm_year, |
|
118 |
||
119 |
# month, day, hour, minute, second |
|
120 |
*ts[1:6] |
|
121 |
) |
|
122 |
||
123 |
def _parse_hostname (self, match) : |
|
124 |
""" |
|
125 |
Parse the hostname from the given match object |
|
126 |
""" |
|
127 |
||
128 |
# nothing much needed.. |
|
129 |
self.hostname = match.group('hostname') |
|
130 |
||
131 |
def _parse_message (self, match) : |
|
132 |
""" |
|
133 |
Parse the message with tag from the given match object |
|
134 |
""" |
|
135 |
||
136 |
self.message = match.group('message') |
|
137 |
self.tag = match.group('tag') |
|
138 |
self.program = match.group('program') |
|
139 |
self.pid = int(match.group('pid')) if match.group('pid') else None |
|
140 |
self.text = match.group('text') |
|
141 |
||
142 |
def __init__ (self, line) : |
|
143 |
""" |
|
144 |
Construct this message from the given line |
|
145 |
""" |
|
146 |
||
147 |
# apply regexp |
|
148 |
match = self.LINE_RE.match(line) |
|
149 |
||
150 |
if not match : |
|
151 |
# fail |
|
152 |
raise ValueError("Invalid syslog data format") |
|
153 |
||
154 |
# unpack the various portions |
|
155 |
self._parse_pri(match) |
|
156 |
self._parse_timestamp(match) |
|
157 |
self._parse_hostname(match) |
|
158 |
self._parse_message(match) |
|
159 |
||
160 |
# the raw line as matched |
|
161 |
self.raw = match.group(0) |
|
53 | 162 |
|
163 |
def properties (self) : |
|
164 |
""" |
|
165 |
Return a dict containing the attributes specified for this message |
|
166 |
""" |
|
167 |
||
168 |
# XXX: ugh... doesn't do @property |
|
169 |
return dict( |
|
170 |
facility = self.facility, |
|
171 |
priority = self.priority, |
|
172 |
timestamp = self.timestamp, |
|
173 |
hostname = self.hostname, |
|
174 |
tag = self.tag, |
|
175 |
program = self.program, |
|
176 |
pid = self.pid, |
|
177 |
text = self.text, |
|
178 |
message = self.message, |
|
179 |
) |
|
50 | 180 |
|
65
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
181 |
@property |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
182 |
def normalized_program (self) : |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
183 |
""" |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
184 |
Returns a normalized version of the program name: |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
185 |
- lowercase |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
186 |
- no path components |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
187 |
|
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
188 |
Returns None if there is no program tag |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
189 |
""" |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
190 |
|
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
191 |
if self.program is None : |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
192 |
return None |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
193 |
|
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
194 |
else : |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
195 |
return os.path.basename(self.program).lower() |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
196 |
|
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
197 |
def match_program (self, program) : |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
198 |
""" |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
199 |
If the given value is False, checks that this message does *not* have a program tag. |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
200 |
|
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
201 |
Otherwise, does a case-insensitive basename() compare of the given value against the message tag's program |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
202 |
name portion. |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
203 |
""" |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
204 |
|
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
205 |
if program is False : |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
206 |
# only match if no tag |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
207 |
return self.program is None |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
208 |
|
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
209 |
elif self.program is None : |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
210 |
# no program to match against... |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
211 |
return False |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
212 |
|
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
213 |
else : |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
214 |
# match against normalized name |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
215 |
return self.normalized_program == program.lower() |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
216 |
|
53 | 217 |
def match (self, regexp=None, program=None) : |
218 |
""" |
|
219 |
Evaluate a match against this message using the given criteria |
|
220 |
||
221 |
regexp - match regexp against the message's contents |
|
222 |
program - match program component of tag against given value, may also be False |
|
223 |
""" |
|
224 |
||
225 |
params = {} |
|
226 |
||
65
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
227 |
if program is not None : |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
228 |
if not self.match_program(program) : |
1de3c0ca2baf
improved SyslogMessage.match_program
Tero Marttila <terom@fixme.fi>
parents:
53
diff
changeset
|
229 |
# program did not match |
53 | 230 |
return False |
231 |
||
232 |
## pattern |
|
233 |
if regexp is not None : |
|
234 |
# apply regexp |
|
235 |
match = regexp.match(self.text) |
|
236 |
||
237 |
if not match : |
|
238 |
# no match |
|
239 |
return False |
|
240 |
||
241 |
# good, params |
|
242 |
params.update(match.groupdict()) |
|
243 |
||
244 |
# we appear to have a match |
|
245 |
return params |
|
246 |
||
50 | 247 |
def __str__ (self) : |
248 |
""" |
|
249 |
Format to default format |
|
250 |
""" |
|
251 |
||
252 |
return "%s %s %s" % (self.timestamp.strftime(self.TIMESTAMP_FMT), self.hostname, self.message) |
|
253 |
||
254 |
def __repr__ (self) : |
|
255 |
return "pri=%s:%s, timestamp=%s, hostname=%s, tag=%s[%s], text=%r" % ( |
|
256 |
self.facility, self.priority, self.timestamp.isoformat(), self.hostname, self.program, self.pid, self.text |
|
257 |
) |
|
258 |