Total Complexity | 67 |
Total Lines | 424 |
Duplicated Lines | 0 % |
Complex classes like buildtimetrend.travis.TravisData often do a lot of different things. To break such a class down, we need to identify a cohesive component within that class. A common approach to find such a component is to look for fields/methods that share the same prefixes, or suffixes.
Once you have determined the fields that belong together, you can apply the Extract Class refactoring. If the component makes sense as a sub-class, Extract Subclass is also a candidate, and is often faster.
1 | # vim: set expandtab sw=4 ts=4: |
||
56 | class TravisData(object): |
||
57 | |||
58 | """Gather data from Travis CI using the API.""" |
||
59 | |||
60 | def __init__(self, repo, build_id, connector=None): |
||
61 | """ |
||
62 | Retrieve Travis CI build data using the API. |
||
63 | |||
64 | Parameters: |
||
65 | - repo : github repository slug (fe. buildtimetrend/python-lib) |
||
66 | - build_id : Travis CI build id (fe. 158) |
||
67 | - connector : Travis Connector instance |
||
68 | """ |
||
69 | self.builds_data = {} |
||
70 | self.build_jobs = {} |
||
71 | self.current_build_data = {} |
||
72 | self.current_job = BuildJob() |
||
73 | self.travis_substage = None |
||
74 | self.repo = repo |
||
75 | self.build_id = str(build_id) |
||
76 | # set TravisConnector if it is defined |
||
77 | if isinstance(connector, TravisConnector): |
||
78 | self.connector = connector |
||
79 | # use Travis Org connector by default |
||
80 | else: |
||
81 | self.connector = TravisOrgConnector() |
||
82 | |||
83 | def get_build_data(self): |
||
84 | """ |
||
85 | Retrieve Travis CI build data. |
||
86 | |||
87 | Returns true if retrieving data was succesful, false on error. |
||
88 | """ |
||
89 | request = 'repos/{repo}/builds?number={build_id}'.format( |
||
90 | repo=self.repo, build_id=self.build_id |
||
91 | ) |
||
92 | try: |
||
93 | self.builds_data = self.connector.json_request(request) |
||
94 | except (HTTPError, URLError) as msg: |
||
95 | logger.error("Error getting build data from Travis CI: %s", msg) |
||
96 | return False |
||
97 | |||
98 | # log builds_data |
||
99 | logger.debug( |
||
100 | "Build #%s data : %s", |
||
101 | str(self.build_id), |
||
102 | json.dumps(self.builds_data, sort_keys=True, indent=2) |
||
103 | ) |
||
104 | |||
105 | return True |
||
106 | |||
107 | def get_substage_name(self, command): |
||
108 | """ |
||
109 | Resolve Travis CI substage name that corresponds to a cli command. |
||
110 | |||
111 | Parameters: |
||
112 | - command : cli command |
||
113 | """ |
||
114 | if not is_string(command): |
||
115 | return "" |
||
116 | |||
117 | if len(self.current_build_data) > 0 and \ |
||
118 | "config" in self.current_build_data: |
||
119 | build_config = self.current_build_data["config"] |
||
120 | else: |
||
121 | logger.warning( |
||
122 | "Travis CI build config is not set" |
||
123 | ) |
||
124 | return "" |
||
125 | |||
126 | # check if build_config collection is empty |
||
127 | if build_config: |
||
128 | for stage_name, commands in build_config.items(): |
||
129 | if type(commands) is list and command in commands: |
||
130 | substage_number = commands.index(command) + 1 |
||
131 | substage_name = "{stage}.{substage:d}".format( |
||
132 | stage=stage_name, substage=substage_number |
||
133 | ) |
||
134 | logger.debug( |
||
135 | "Substage %s corresponds to '%s'", |
||
136 | substage_name, command |
||
137 | ) |
||
138 | return substage_name |
||
139 | |||
140 | return "" |
||
141 | |||
142 | def process_build_jobs(self): |
||
143 | """ |
||
144 | Retrieve Travis CI build job data. |
||
145 | |||
146 | Method is a generator, iterate result to get each processed build job. |
||
147 | """ |
||
148 | if len(self.builds_data) > 0 and "builds" in self.builds_data: |
||
149 | for build in self.builds_data['builds']: |
||
150 | self.current_build_data = build |
||
151 | |||
152 | if "job_ids" in build: |
||
153 | for job_id in build['job_ids']: |
||
154 | yield self.process_build_job(job_id) |
||
155 | |||
156 | # reset current_build_data after builds are processed |
||
157 | self.current_build_data = {} |
||
158 | |||
159 | def process_build_job(self, job_id): |
||
160 | """ |
||
161 | Retrieve Travis CI build job data. |
||
162 | |||
163 | Parameters: |
||
164 | - job_id : ID of the job to process |
||
165 | """ |
||
166 | if job_id is None: |
||
167 | return None |
||
168 | |||
169 | # retrieve job data from Travis CI |
||
170 | job_data = self.get_job_data(job_id) |
||
171 | # process build/job data |
||
172 | self.process_job_data(job_data) |
||
173 | # parse Travis CI job log file |
||
174 | self.parse_job_log(job_id) |
||
175 | |||
176 | # store build job |
||
177 | self.build_jobs[str(job_id)] = self.current_job |
||
178 | # create new build job instance |
||
179 | self.current_job = BuildJob() |
||
180 | |||
181 | # return processed build job |
||
182 | return self.build_jobs[str(job_id)] |
||
183 | |||
184 | def get_job_data(self, job_id): |
||
185 | """ |
||
186 | Retrieve Travis CI job data. |
||
187 | |||
188 | Parameters: |
||
189 | - job_id : ID of the job to process |
||
190 | """ |
||
191 | request = 'jobs/{:s}'.format(str(job_id)) |
||
192 | job_data = self.connector.json_request(request) |
||
193 | |||
194 | # log job_data |
||
195 | logger.debug( |
||
196 | "Job #%s data : %s", |
||
197 | str(job_id), |
||
198 | json.dumps(job_data, sort_keys=True, indent=2) |
||
199 | ) |
||
200 | |||
201 | return job_data |
||
202 | |||
203 | def process_job_data(self, job_data): |
||
204 | """ |
||
205 | Process Job/build data. |
||
206 | |||
207 | Set build/job properties : |
||
208 | - Build/job ID |
||
209 | - build result : passed, failed, errored |
||
210 | - git repo |
||
211 | - git branch |
||
212 | - CI platform : Travis |
||
213 | - build matrix (language, language version, compiler, ...) |
||
214 | - build_trigger : push, pull_request |
||
215 | - pull_request (is_pull_request, title, number) |
||
216 | |||
217 | Parameters: |
||
218 | - job_data : dictionary with Travis CI job data |
||
219 | """ |
||
220 | self.current_job.add_property( |
||
221 | "build", |
||
222 | # buildnumber is part before "." of job number |
||
223 | job_data['job']['number'].split(".")[0] |
||
224 | ) |
||
225 | self.current_job.add_property("job", job_data['job']['number']) |
||
226 | self.current_job.add_property("branch", job_data['commit']['branch']) |
||
227 | self.current_job.add_property( |
||
228 | "repo", |
||
229 | job_data['job']['repository_slug'] |
||
230 | ) |
||
231 | self.current_job.add_property("ci_platform", 'travis') |
||
232 | self.current_job.add_property("result", job_data['job']['state']) |
||
233 | |||
234 | self.set_build_matrix(job_data) |
||
235 | |||
236 | self.process_pull_request_data() |
||
237 | |||
238 | self.current_job.set_started_at(job_data['job']['started_at']) |
||
239 | self.current_job.set_finished_at(job_data['job']['finished_at']) |
||
240 | |||
241 | # calculate job duration from start and finished timestamps |
||
242 | # if no timing tags are available |
||
243 | if not self.has_timing_tags(): |
||
244 | self.current_job.add_property("duration", self.get_job_duration()) |
||
245 | |||
246 | def set_build_matrix(self, job_data): |
||
247 | """ |
||
248 | Retrieve build matrix data from job data and store in properties. |
||
249 | |||
250 | Properties : |
||
251 | - language |
||
252 | - language version (if applicable) |
||
253 | - compiler (if applicable) |
||
254 | - operating system |
||
255 | - environment parameters |
||
256 | |||
257 | Parameters: |
||
258 | - job_data : dictionary with Travis CI job data |
||
259 | """ |
||
260 | build_matrix = Collection() |
||
261 | |||
262 | job_config = job_data['job']['config'] |
||
263 | |||
264 | language = job_config['language'] |
||
265 | build_matrix.add_item("language", language) |
||
266 | |||
267 | # set language version |
||
268 | # ('d', 'dart', 'go', 'perl', 'php', 'python', 'rust') |
||
269 | if language in job_config: |
||
270 | if language == "android": |
||
271 | build_matrix.add_item( |
||
272 | "language_components", |
||
273 | " ".join(job_config[language]["components"]) |
||
274 | ) |
||
275 | else: |
||
276 | build_matrix.add_item( |
||
277 | "language_version", |
||
278 | str(job_config[language]) |
||
279 | ) |
||
280 | |||
281 | # language specific build matrix parameters |
||
282 | parameters = { |
||
283 | 'ghc': 'ghc', # Haskell |
||
284 | 'jdk': 'jdk', # Java, Android, Groovy, Ruby, Scala |
||
285 | 'lein': 'lein', # Clojure |
||
286 | 'mono': 'mono', # C#, F#, Visual Basic |
||
287 | 'node_js': 'node_js', # Javascript |
||
288 | 'otp_release': 'otp_release', # Erlang |
||
289 | 'rvm': 'rvm', # Ruby, Objective-C |
||
290 | 'gemfile': 'gemfile', # Ruby, Objective-C |
||
291 | 'xcode_sdk': 'xcode_sdk', # Objective-C |
||
292 | 'xcode_scheme': 'xcode_scheme', # Objective-C |
||
293 | 'compiler': 'compiler', # C, C++ |
||
294 | 'os': 'os', |
||
295 | 'env': 'parameters' |
||
296 | } |
||
297 | for parameter, name in parameters.items(): |
||
298 | if parameter in job_config: |
||
299 | build_matrix.add_item(name, str(job_config[parameter])) |
||
300 | |||
301 | self.current_job.add_property( |
||
302 | "build_matrix", |
||
303 | build_matrix.get_items_with_summary() |
||
304 | ) |
||
305 | |||
306 | def process_pull_request_data(self): |
||
307 | """Retrieve pull request data from Travis CI API.""" |
||
308 | # check if collection is empty |
||
309 | if self.current_build_data: |
||
310 | if "event_type" in self.current_build_data: |
||
311 | # build trigger (push or pull_request) |
||
312 | self.current_job.add_property( |
||
313 | "build_trigger", |
||
314 | self.current_build_data["event_type"] |
||
315 | ) |
||
316 | |||
317 | # pull_request |
||
318 | pull_request_data = {} |
||
319 | if "pull_request" in self.current_build_data: |
||
320 | pull_request_data["is_pull_request"] = \ |
||
321 | self.current_build_data["pull_request"] |
||
322 | else: |
||
323 | pull_request_data["is_pull_request"] = False |
||
324 | |||
325 | if "pull_request_title" in self.current_build_data: |
||
326 | pull_request_data["title"] = \ |
||
327 | self.current_build_data["pull_request_title"] |
||
328 | |||
329 | if "pull_request_number" in self.current_build_data: |
||
330 | pull_request_data["number"] = \ |
||
331 | self.current_build_data["pull_request_number"] |
||
332 | |||
333 | self.current_job.add_property("pull_request", pull_request_data) |
||
334 | |||
335 | def parse_job_log(self, job_id): |
||
336 | """ |
||
337 | Parse Travis CI job log. |
||
338 | |||
339 | Parameters: |
||
340 | - job_id : ID of the job to process |
||
341 | """ |
||
342 | self.parse_job_log_stream(self.connector.download_job_log(job_id)) |
||
343 | |||
344 | def parse_job_log_file(self, filename): |
||
345 | """ |
||
346 | Open a Travis CI log file and parse it. |
||
347 | |||
348 | Parameters : |
||
349 | - filename : filename of Travis CI log |
||
350 | Returns false if file doesn't exist, true if it was read successfully. |
||
351 | """ |
||
352 | # load timestamps file |
||
353 | if not check_file(filename): |
||
354 | return False |
||
355 | |||
356 | # read timestamps, calculate stage duration |
||
357 | with open(filename, 'rb') as file_stream: |
||
358 | self.parse_job_log_stream(file_stream) |
||
359 | |||
360 | return True |
||
361 | |||
362 | def parse_job_log_stream(self, stream): |
||
363 | """ |
||
364 | Parse Travis CI job log stream. |
||
365 | |||
366 | Parameters: |
||
367 | - stream : stream of job log file |
||
368 | """ |
||
369 | self.travis_substage = TravisSubstage() |
||
370 | check_timing_tags = self.has_timing_tags() |
||
371 | |||
372 | for line in stream: |
||
373 | # convert to str if line is bytes type |
||
374 | if isinstance(line, bytes): |
||
375 | line = line.decode('utf-8') |
||
376 | # parse Travis CI timing tags |
||
377 | if check_timing_tags and 'travis_' in line: |
||
378 | self.parse_travis_time_tag(line) |
||
379 | # parse Travis CI worker tag |
||
380 | if 'Using worker:' in line: |
||
381 | self.parse_travis_worker_tag(line) |
||
382 | |||
383 | def parse_travis_time_tag(self, line): |
||
384 | """ |
||
385 | Parse and process Travis CI timing tags. |
||
386 | |||
387 | Parameters: |
||
388 | - line : line from logfile containing Travis CI tags |
||
389 | """ |
||
390 | if self.travis_substage is None: |
||
391 | self.travis_substage = TravisSubstage() |
||
392 | |||
393 | escaped_line = line.replace('\x0d', '*').replace('\x1b', 'ESC') |
||
394 | logger.debug('line : %s', escaped_line) |
||
395 | |||
396 | # parse Travis CI timing tags |
||
397 | for parse_string in TRAVIS_LOG_PARSE_TIMING_STRINGS: |
||
398 | result = re.search(parse_string, line) |
||
399 | if result: |
||
400 | self.travis_substage.process_parsed_tags(result.groupdict()) |
||
401 | |||
402 | # when finished : log stage and create a new instance |
||
403 | if self.travis_substage.has_finished(): |
||
404 | # set substage name, if it is not set |
||
405 | if not self.travis_substage.has_name() and \ |
||
406 | self.travis_substage.has_command(): |
||
407 | self.travis_substage.set_name( |
||
408 | self.get_substage_name( |
||
409 | self.travis_substage.get_command() |
||
410 | ) |
||
411 | ) |
||
412 | |||
413 | # only log complete substages |
||
414 | if not self.travis_substage.finished_incomplete: |
||
415 | self.current_job.add_stage(self.travis_substage.stage) |
||
416 | self.travis_substage = TravisSubstage() |
||
417 | |||
418 | def parse_travis_worker_tag(self, line): |
||
419 | """ |
||
420 | Parse and process Travis CI worker tag. |
||
421 | |||
422 | Parameters: |
||
423 | - line : line from logfile containing Travis CI tags |
||
424 | """ |
||
425 | logger.debug('line : %s', line) |
||
426 | |||
427 | # parse Travis CI worker tags |
||
428 | result = re.search(TRAVIS_LOG_PARSE_WORKER_STRING, line) |
||
429 | if not result: |
||
430 | return |
||
431 | |||
432 | worker_tags = result.groupdict() |
||
433 | |||
434 | # check if parameter worker_tags is a dictionary and |
||
435 | # if it contains all required tags |
||
436 | tag_list = list({'hostname', 'os'}) |
||
437 | if check_dict(worker_tags, "worker_tags", tag_list): |
||
438 | logger.debug("Worker tags : %s", worker_tags) |
||
439 | self.current_job.add_property("worker", worker_tags) |
||
440 | |||
441 | def has_timing_tags(self): |
||
442 | """ |
||
443 | Check if Travis CI job log has timing tags. |
||
444 | |||
445 | Timing tags were introduced on Travis CI starting 2014-08-07, |
||
446 | check if started_at is more recent. |
||
447 | """ |
||
448 | started_at = self.current_job.get_property("started_at") |
||
449 | if started_at is None or "timestamp_seconds" not in started_at: |
||
450 | return False |
||
451 | |||
452 | # 1407369600 is epoch timestamp of 2014-08-07T00:00:00Z |
||
453 | return started_at["timestamp_seconds"] > 1407369600 |
||
454 | |||
455 | def get_job_duration(self): |
||
456 | """Calculate build job duration.""" |
||
457 | started_at = self.current_job.get_property("started_at") |
||
458 | finished_at = self.current_job.get_property("finished_at") |
||
459 | if started_at is None or "timestamp_seconds" not in started_at or \ |
||
460 | finished_at is None or "timestamp_seconds" not in finished_at: |
||
461 | return 0.0 |
||
462 | |||
463 | timestamp_start = float(started_at["timestamp_seconds"]) |
||
464 | timestamp_end = float(finished_at["timestamp_seconds"]) |
||
465 | return timestamp_end - timestamp_start |
||
466 | |||
467 | def get_started_at(self): |
||
468 | """Retrieve timestamp when build was started.""" |
||
469 | if check_dict(self.current_build_data, key_list=["started_at"]): |
||
470 | return self.current_build_data['started_at'] |
||
471 | else: |
||
472 | return None |
||
473 | |||
474 | def get_finished_at(self): |
||
475 | """Retrieve timestamp when build finished.""" |
||
476 | if check_dict(self.current_build_data, key_list=["finished_at"]): |
||
477 | return self.current_build_data['finished_at'] |
||
478 | else: |
||
479 | return None |
||
480 |