1 # -*- coding: utf-8 -*-
2 # Description: web log netdata python.d module
5 from base import LogService
8 from os import access, R_OK
9 from os.path import getsize
10 from collections import namedtuple
11 from copy import deepcopy
16 ORDER = ['response_statuses', 'response_codes', 'bandwidth', 'response_time', 'requests_per_url', 'http_method',
17 'http_version', 'requests_per_ipproto', 'clients', 'clients_all']
20 'options': [None, 'Response Codes', 'requests/s', 'responses', 'web_log.response_codes', 'stacked'],
22 ['2xx', '2xx', 'incremental'],
23 ['5xx', '5xx', 'incremental'],
24 ['3xx', '3xx', 'incremental'],
25 ['4xx', '4xx', 'incremental'],
26 ['1xx', '1xx', 'incremental'],
27 ['0xx', 'other', 'incremental'],
28 ['unmatched', 'unmatched', 'incremental']
31 'options': [None, 'Bandwidth', 'KB/s', 'bandwidth', 'web_log.bandwidth', 'area'],
33 ['resp_length', 'received', 'incremental', 1, 1024],
34 ['bytes_sent', 'sent', 'incremental', -1, 1024]
37 'options': [None, 'Processing Time', 'milliseconds', 'timings', 'web_log.response_time', 'area'],
39 ['resp_time_min', 'min', 'incremental', 1, 1000],
40 ['resp_time_max', 'max', 'incremental', 1, 1000],
41 ['resp_time_avg', 'avg', 'incremental', 1, 1000]
44 'options': [None, 'Current Poll Unique Client IPs', 'unique ips', 'clients', 'web_log.clients', 'stacked'],
46 ['unique_cur_ipv4', 'ipv4', 'incremental', 1, 1],
47 ['unique_cur_ipv6', 'ipv6', 'incremental', 1, 1]
50 'options': [None, 'All Time Unique Client IPs', 'unique ips', 'clients', 'web_log.clients_all', 'stacked'],
52 ['unique_tot_ipv4', 'ipv4', 'absolute', 1, 1],
53 ['unique_tot_ipv6', 'ipv6', 'absolute', 1, 1]
56 'options': [None, 'Requests Per HTTP Method', 'requests/s', 'http methods', 'web_log.http_method', 'stacked'],
58 ['GET', 'GET', 'incremental', 1, 1]
61 'options': [None, 'Requests Per HTTP Version', 'requests/s', 'http versions',
62 'web_log.http_version', 'stacked'],
64 'requests_per_ipproto': {
65 'options': [None, 'Requests Per IP Protocol', 'requests/s', 'ip protocols', 'web_log.requests_per_ipproto',
68 ['req_ipv4', 'ipv4', 'incremental', 1, 1],
69 ['req_ipv6', 'ipv6', 'incremental', 1, 1]
71 'response_statuses': {
72 'options': [None, 'Response Statuses', 'requests/s', 'responses', 'web_log.response_statuses',
75 ['successful_requests', 'success', 'incremental', 1, 1],
76 ['server_errors', 'error', 'incremental', 1, 1],
77 ['redirects', 'redirect', 'incremental', 1, 1],
78 ['bad_requests', 'bad', 'incremental', 1, 1],
79 ['other_requests', 'other', 'incremental', 1, 1]
83 NAMED_URL_PATTERN = namedtuple('URL_PATTERN', ['description', 'pattern'])
85 DET_RESP_AGGR = ['', '_1xx', '_2xx', '_3xx', '_4xx', '_5xx', '_Other']
88 class Service(LogService):
89 def __init__(self, configuration=None, name=None):
93 # self._get_data = None # will be assigned in 'check' method.
94 # self.order = None # will be assigned in 'create_*_method' method.
95 # self.definitions = None # will be assigned in 'create_*_method' method.
97 LogService.__init__(self, configuration=configuration, name=name)
98 # Variables from module configuration file
99 self.log_type = self.configuration.get('type', 'web_access')
100 self.log_path = self.configuration.get('path')
101 self.url_pattern = self.configuration.get('categories') # dict
102 self.custom_log_format = self.configuration.get('custom_log_format') # dict
104 self.regex = None # will be assigned in 'find_regex' or 'find_regex_custom' method
105 self.data = {'bytes_sent': 0, 'resp_length': 0, 'resp_time_min': 0, 'resp_time_max': 0,
106 'resp_time_avg': 0, 'unique_cur_ipv4': 0, 'unique_cur_ipv6': 0, '2xx': 0,
107 '5xx': 0, '3xx': 0, '4xx': 0, '1xx': 0, '0xx': 0, 'unmatched': 0, 'req_ipv4': 0,
108 'req_ipv6': 0, 'unique_tot_ipv4': 0, 'unique_tot_ipv6': 0, 'successful_requests': 0,
109 'redirects': 0, 'bad_requests': 0, 'server_errors': 0, 'other_requests': 0, 'GET': 0}
115 1. "log_path" is specified in the module configuration file
116 2. "log_path" must be readable by netdata user and must exist
117 3. "log_path' must not be empty. We need at least 1 line to find appropriate pattern to parse
118 4. other checks depends on log "type"
120 if not self.log_path:
121 self.error('log path is not specified')
124 if not access(self.log_path, R_OK):
125 self.error('%s not readable or not exist' % self.log_path)
128 if not getsize(self.log_path):
129 self.error('%s is empty' % self.log_path)
132 # Read last line (or first if there is only one line)
133 with open(self.log_path, 'rb') as logs:
135 while logs.read(1) != b'\n':
139 last_line = logs.readline()
142 last_line = last_line.decode()
143 except UnicodeDecodeError:
145 last_line = last_line.decode(encoding='utf-8')
146 except (TypeError, UnicodeDecodeError) as error:
147 self.error(str(error))
150 if self.log_type == 'web_access':
151 self.unique_all_time = list() # sorted list of unique IPs
152 self.detailed_response_codes = self.configuration.get('detailed_response_codes', True)
153 self.detailed_response_aggregate = self.configuration.get('detailed_response_aggregate', True)
154 self.all_time = self.configuration.get('all_time', True)
156 # Custom_log_format or predefined log format.
157 if self.custom_log_format:
158 match_dict, error = self.find_regex_custom(last_line)
160 match_dict, error = self.find_regex(last_line)
162 # "match_dict" is None if there are any problems
163 if match_dict is None:
164 self.error(str(error))
167 # self.url_pattern check
169 self.url_pattern = check_req_per_url_pattern('rpu', self.url_pattern)
171 self.create_access_charts(match_dict) # Create charts
172 self._get_data = self._get_access_data # _get_data assignment
174 self.error('Not implemented')
179 self.error('That can not happen, but it happened. "regex" is None')
181 self.info('Collected data: %s' % list(match_dict.keys()))
184 def find_regex_custom(self, last_line):
186 :param last_line: str: literally last line from log file
187 :return: tuple where:
188 [0]: dict or None: match_dict or None
189 [1]: str: error description
191 We are here only if "custom_log_format" is in logs. We need to make sure:
192 1. "custom_log_format" is a dict
193 2. "pattern" in "custom_log_format" and pattern is <str> instance
194 3. if "time_multiplier" is in "custom_log_format" it must be <int> instance
196 If all parameters is ok we need to make sure:
197 1. Pattern search is success
198 2. Pattern search contains named subgroups (?P<subgroup_name>) (= "match_dict")
200 If pattern search is success we need to make sure:
201 1. All mandatory keys ['address', 'code', 'bytes_sent', 'method', 'url'] are in "match_dict"
203 If this is True we need to make sure:
204 1. All mandatory key values from "match_dict" have the correct format
205 ("code" is integer, "method" is uppercase word, etc)
207 If non mandatory keys in "match_dict" we need to make sure:
208 1. All non mandatory key values from match_dict ['resp_length', 'resp_time'] have the correct format
209 ("resp_length" is integer or "-", "resp_time" is integer or float)
212 if not hasattr(self.custom_log_format, 'keys'):
213 return find_regex_return(msg='Custom log: "custom_log_format" is not a <dict>')
215 pattern = self.custom_log_format.get('pattern')
216 if not (pattern and isinstance(pattern, str)):
217 return find_regex_return(msg='Custom log: "pattern" option is not specified or type is not <str>')
219 resp_time_func = self.custom_log_format.get('time_multiplier') or 0
221 if not isinstance(resp_time_func, int):
222 return find_regex_return(msg='Custom log: "time_multiplier" is not an integer')
225 regex = re.compile(pattern)
226 except re.error as error:
227 return find_regex_return(msg='Pattern compile error: %s' % str(error))
229 match = regex.search(last_line)
231 match_dict = match.groupdict() or None
233 return find_regex_return(msg='Custom log: pattern search FAILED')
235 if match_dict is None:
236 find_regex_return(msg='Custom log: search OK but contains no named subgroups'
237 ' (you need to use ?P<subgroup_name>)')
239 mandatory_dict = {'address': r'[\da-f.:]+',
240 'code': r'[1-9]\d{2}',
242 'bytes_sent': r'\d+|-'}
243 optional_dict = {'resp_length': r'\d+',
244 'resp_time': r'[\d.]+',
245 'http_version': r'\d\.\d'}
247 mandatory_values = set(mandatory_dict) - set(match_dict)
249 return find_regex_return(msg='Custom log: search OK but some mandatory keys (%s) are missing'
250 % list(mandatory_values))
252 for key in mandatory_dict:
253 if not re.search(mandatory_dict[key], match_dict[key]):
254 return find_regex_return(msg='Custom log: can\'t parse "%s": %s'
255 % (key, match_dict[key]))
257 optional_values = set(optional_dict) & set(match_dict)
258 for key in optional_values:
259 if not re.search(optional_dict[key], match_dict[key]):
260 return find_regex_return(msg='Custom log: can\'t parse "%s": %s'
261 % (key, match_dict[key]))
263 dot_in_time = '.' in match_dict.get('resp_time', '')
265 self.resp_time_func = lambda time: time * (resp_time_func or 1000000)
267 self.resp_time_func = lambda time: time * (resp_time_func or 1)
270 return find_regex_return(match_dict=match_dict)
272 def find_regex(self, last_line):
274 :param last_line: str: literally last line from log file
275 :return: tuple where:
276 [0]: dict or None: match_dict or None
277 [1]: str: error description
278 We need to find appropriate pattern for current log file
279 All logic is do a regex search through the string for all predefined patterns
280 until we find something or fail.
282 # REGEX: 1.IPv4 address 2.HTTP method 3. URL 4. Response code
283 # 5. Bytes sent 6. Response length 7. Response process time
284 acs_default = re.compile(r'(?P<address>[\da-f.:]+)'
285 r' -.*?"(?P<method>[A-Z]+)'
287 r' [A-Z]+/(?P<http_version>\d\.\d)"'
288 r' (?P<code>[1-9]\d{2})'
289 r' (?P<bytes_sent>\d+|-)')
291 acs_apache_ext_insert = re.compile(r'(?P<address>[\da-f.:]+)'
292 r' -.*?"(?P<method>[A-Z]+)'
294 r' [A-Z]+/(?P<http_version>\d\.\d)"'
295 r' (?P<code>[1-9]\d{2})'
296 r' (?P<bytes_sent>\d+|-)'
297 r' (?P<resp_length>\d+)'
298 r' (?P<resp_time>\d+) ')
300 acs_apache_ext_append = re.compile(r'(?P<address>[\da-f.:]+)'
301 r' -.*?"(?P<method>[A-Z]+)'
303 r' [A-Z]+/(?P<http_version>\d\.\d)"'
304 r' (?P<code>[1-9]\d{2})'
305 r' (?P<bytes_sent>\d+|-)'
307 r' (?P<resp_length>\d+)'
308 r' (?P<resp_time>\d+)'
311 acs_nginx_ext_insert = re.compile(r'(?P<address>[\da-f.:]+)'
312 r' -.*?"(?P<method>[A-Z]+)'
314 r' [A-Z]+/(?P<http_version>\d\.\d)"'
315 r' (?P<code>[1-9]\d{2})'
316 r' (?P<bytes_sent>\d+)'
317 r' (?P<resp_length>\d+)'
318 r' (?P<resp_time>\d+\.\d+) ')
320 acs_nginx_ext_append = re.compile(r'(?P<address>[\da-f.:]+)'
321 r' -.*?"(?P<method>[A-Z]+)'
323 r' [A-Z]+/(?P<http_version>\d\.\d)"'
324 r' (?P<code>[1-9]\d{2})'
325 r' (?P<bytes_sent>\d+)'
327 r' (?P<resp_length>\d+)'
328 r' (?P<resp_time>\d+\.\d+)')
334 return time * 1000000
336 r_regex = [acs_apache_ext_insert, acs_apache_ext_append, acs_nginx_ext_insert,
337 acs_nginx_ext_append, acs_default]
338 r_function = [func_usec, func_usec, func_sec, func_sec, func_usec]
339 regex_function = zip(r_regex, r_function)
342 for regex, function in regex_function:
343 match = regex.search(last_line)
346 self.resp_time_func = function
347 match_dict = match.groupdict()
350 return find_regex_return(match_dict=match_dict or None,
351 msg='Unknown log format. You need to use "custom_log_format" feature.')
353 def create_access_charts(self, match_dict):
355 :param match_dict: dict: regex.search.groupdict(). Ex. {'address': '127.0.0.1', 'code': '200', 'method': 'GET'}
357 Create additional charts depending on the 'match_dict' keys and configuration file options
358 1. 'time_response' chart is removed if there is no 'resp_time' in match_dict.
359 2. Other stuff is just remove/add chart depending on yes/no in conf
362 def find_job_name(override_name, name):
364 :param override_name: str: 'name' var from configuration file
365 :param name: str: 'job_name' from configuration file
366 :return: str: new job name
367 We need this for dynamic charts. Actually same logic as in python.d.plugin.
369 add_to_name = override_name or name
371 return '_'.join(['web_log', re.sub('\s+', '_', add_to_name)])
375 self.order = ORDER[:]
376 self.definitions = deepcopy(CHARTS)
378 job_name = find_job_name(self.override_name, self.name)
380 self.http_method_chart = 'CHART %s.http_method' \
381 ' "" "Requests Per HTTP Method" requests/s "http methods"' \
382 ' web_log.http_method stacked 11 %s\n' \
383 'DIMENSION GET GET incremental\n' % (job_name, self.update_every)
384 self.http_version_chart = 'CHART %s.http_version' \
385 ' "" "Requests Per HTTP Version" requests/s "http versions"' \
386 ' web_log.http_version stacked 12 %s\n' % (job_name, self.update_every)
388 # Remove 'request_time' chart from ORDER if resp_time not in match_dict
389 if 'resp_time' not in match_dict:
390 self.order.remove('response_time')
391 # Remove 'clients_all' chart from ORDER if specified in the configuration
392 if not self.all_time:
393 self.order.remove('clients_all')
394 # Add 'detailed_response_codes' chart if specified in the configuration
395 if self.detailed_response_codes:
396 self.detailed_chart = list()
397 for prio, add_to_dim in enumerate(DET_RESP_AGGR):
398 self.detailed_chart.append('CHART %s.detailed_response_codes%s ""'
399 ' "Detailed Response Codes %s" requests/s responses'
400 ' web_log.detailed_response_codes%s stacked %s %s\n'
401 % (job_name, add_to_dim, add_to_dim[1:], add_to_dim,
402 str(prio), self.update_every))
404 codes = DET_RESP_AGGR[:1] if self.detailed_response_aggregate else DET_RESP_AGGR[1:]
406 self.order.append('detailed_response_codes%s' % code)
407 self.definitions['detailed_response_codes%s' % code] = {'options':
409 'Detailed Response Codes %s' % code[1:],
412 'web_log.detailed_response_codes%s' % code,
416 # Add 'requests_per_url' chart if specified in the configuration
418 self.definitions['requests_per_url'] = {'options': [None, 'Requests Per Url', 'requests/s',
419 'urls', 'web_log.requests_per_url', 'stacked'],
420 'lines': [['rpu_other', 'other', 'incremental']]}
421 for elem in self.url_pattern:
422 self.definitions['requests_per_url']['lines'].append([elem.description, elem.description[4:],
424 self.data.update({elem.description: 0})
425 self.data.update({'rpu_other': 0})
427 self.order.remove('requests_per_url')
429 def add_new_dimension(self, dimension, line_list, chart_string, key):
431 :param dimension: str: response status code. Ex.: '202', '499'
432 :param line_list: list: Ex.: ['202', '202', 'incremental']
433 :param chart_string: Current string we need to pass to netdata to rebuild the chart
434 :param key: str: CHARTS dict key (chart name). Ex.: 'response_time'
435 :return: str: new chart string = previous + new dimensions
437 self.data.update({dimension: 0})
438 # SET method check if dim in _dimensions
439 self._dimensions.append(dimension)
440 # UPDATE method do SET only if dim in definitions
441 self.definitions[key]['lines'].append(line_list)
443 chart += "%s %s\n" % ('DIMENSION', ' '.join(line_list))
447 def _get_access_data(self):
450 :return: dict OR None
451 None if _get_raw_data method fails.
452 In all other cases - dict.
454 raw = self._get_raw_data()
458 request_time, unique_current = list(), list()
459 request_counter = {'count': 0, 'sum': 0}
460 ip_address_counter = {'unique_cur_ip': 0}
462 match = self.regex.search(line)
464 match_dict = match.groupdict()
466 code = ''.join([match_dict['code'][0], 'xx'])
469 self.data['0xx'] += 1
470 # detailed response code
471 if self.detailed_response_codes:
472 self._get_data_detailed_response_codes(match_dict['code'])
474 self._get_data_statuses(match_dict['code'])
477 self._get_data_per_url(match_dict['url'])
478 # requests per http method
479 self._get_data_http_method(match_dict['method'])
480 # requests per http version
481 if 'http_version' in match_dict:
482 self._get_data_http_version(match_dict['http_version'])
484 bytes_sent = match_dict['bytes_sent'] if '-' not in match_dict['bytes_sent'] else 0
485 self.data['bytes_sent'] += int(bytes_sent)
486 # request processing time and bandwidth received
487 if 'resp_length' in match_dict:
488 self.data['resp_length'] += int(match_dict['resp_length'])
489 if 'resp_time' in match_dict:
490 resp_time = self.resp_time_func(float(match_dict['resp_time']))
491 bisect.insort_left(request_time, resp_time)
492 request_counter['count'] += 1
493 request_counter['sum'] += resp_time
494 # requests per ip proto
495 proto = 'ipv4' if '.' in match_dict['address'] else 'ipv6'
496 self.data['req_' + proto] += 1
498 if address_not_in_pool(self.unique_all_time, match_dict['address'],
499 self.data['unique_tot_ipv4'] + self.data['unique_tot_ipv6']):
500 self.data['unique_tot_' + proto] += 1
501 if address_not_in_pool(unique_current, match_dict['address'], ip_address_counter['unique_cur_ip']):
502 self.data['unique_cur_' + proto] += 1
503 ip_address_counter['unique_cur_ip'] += 1
505 self.data['unmatched'] += 1
509 self.data['resp_time_min'] += int(request_time[0])
510 self.data['resp_time_avg'] += int(round(float(request_counter['sum']) / request_counter['count']))
511 self.data['resp_time_max'] += int(request_time[-1])
514 def _get_data_detailed_response_codes(self, code):
516 :param code: str: CODE from parsed line. Ex.: '202, '499'
518 Calls add_new_dimension method If the value is found for the first time
520 if code not in self.data:
521 if self.detailed_response_aggregate:
522 chart_string_copy = self.detailed_chart[0]
523 self.detailed_chart[0] = self.add_new_dimension(code, [code, code, 'incremental'],
524 chart_string_copy, 'detailed_response_codes')
526 code_index = int(code[0]) if int(code[0]) < 6 else 6
527 chart_string_copy = self.detailed_chart[code_index]
528 chart_name = 'detailed_response_codes' + DET_RESP_AGGR[code_index]
529 self.detailed_chart[code_index] = self.add_new_dimension(code, [code, code, 'incremental'],
530 chart_string_copy, chart_name)
533 def _get_data_http_method(self, method):
535 :param method: str: METHOD from parsed line. Ex.: 'GET', 'POST'
537 Calls add_new_dimension method If the value is found for the first time
539 if method not in self.data:
540 chart_string_copy = self.http_method_chart
541 self.http_method_chart = self.add_new_dimension(method, [method, method, 'incremental'],
542 chart_string_copy, 'http_method')
543 self.data[method] += 1
545 def _get_data_http_version(self, http_version):
547 :param http_version: str: METHOD from parsed line. Ex.: '1.1', '1.0'
549 Calls add_new_dimension method If the value is found for the first time
551 http_version_dim_id = http_version.replace('.', '_')
552 if http_version_dim_id not in self.data:
553 chart_string_copy = self.http_version_chart
554 self.http_version_chart = self.add_new_dimension(http_version_dim_id,
555 [http_version_dim_id, http_version, 'incremental'],
556 chart_string_copy, 'http_version')
557 self.data[http_version_dim_id] += 1
559 def _get_data_per_url(self, url):
561 :param url: str: URL from parsed line
563 Scan through string looking for the first location where patterns produce a match for all user
567 for elem in self.url_pattern:
568 if elem.pattern.search(url):
569 self.data[elem.description] += 1
573 self.data['rpu_other'] += 1
575 def _get_data_statuses(self, code):
577 :param code: str: response status code. Ex.: '202', '499'
581 if code_class == '2' or code == '304' or code_class == '1':
582 self.data['successful_requests'] += 1
583 elif code_class == '3':
584 self.data['redirects'] += 1
585 elif code_class == '4':
586 self.data['bad_requests'] += 1
587 elif code_class == '5':
588 self.data['server_errors'] += 1
590 self.data['other_requests'] += 1
593 def address_not_in_pool(pool, address, pool_size):
595 :param pool: list of ip addresses
596 :param address: ip address
597 :param pool_size: current pool size
598 :return: True if address not in pool. False if address in pool.
600 index = bisect.bisect_left(pool, address)
601 if index < pool_size:
602 if pool[index] == address:
605 bisect.insort_left(pool, address)
608 bisect.insort_left(pool, address)
612 def find_regex_return(match_dict=None, msg='Generic error message'):
614 :param match_dict: dict: re.search.groupdict() or None
615 :param msg: str: error description
618 return match_dict, msg
621 def check_req_per_url_pattern(string, url_pattern):
624 :param url_pattern: dict: ex. {'dim1': 'pattern1>', 'dim2': '<pattern2>'}
625 :return: list of named tuples or None:
626 We need to make sure all patterns are valid regular expressions
628 if not hasattr(url_pattern, 'keys'):
633 def is_valid_pattern(pattern):
636 :return: re.compile(pattern) or None
638 if not isinstance(pattern, str):
642 compile_pattern = re.compile(pattern)
646 return compile_pattern
648 for dimension, regex in url_pattern.items():
649 valid_pattern = is_valid_pattern(regex)
650 if isinstance(dimension, str) and valid_pattern:
651 result.append(NAMED_URL_PATTERN(description='_'.join([string, dimension]), pattern=valid_pattern))
653 return result or None