def afterActionsRun(self, testingStep, executionSessions, traces):
trace = [trace for trace in traces if trace is not None][0]
self.listOfTimesForScreenshot.append(trace.timeForScreenshot)
self.listOfTimesForActionMapRetrieval.append(
trace.timeForActionMapRetrieval)
self.listOfTimesForActionDecision.append(trace.timeForActionDecision)
self.listOfTimesForActionExecution.append(trace.timeForActionExecution)
self.listOfTimesForMiscellaneous.append(trace.timeForMiscellaneous)
totalTime = trace.timeForScreenshot + \
trace.timeForActionMapRetrieval + \
trace.timeForActionDecision + \
trace.timeForActionExecution + \
trace.timeForMiscellaneous
self.listOfTotalLoopTimes.append(totalTime)
if trace.traceNumber % self.config['testing_print_every'] == (
self.config['testing_print_every'] -
1) or trace.traceNumber == 0:
msg = f"Finished {trace.traceNumber + 1} testing actions."
if len(self.listOfTimesForScreenshot):
msg += f"\n Avg Screenshot time: {numpy.average(self.listOfTimesForScreenshot[-self.config['testing_print_every']:]):.4f}"
msg += f"\n Avg Action Map Retrieval Time: {numpy.average(self.listOfTimesForActionMapRetrieval[-self.config['testing_print_every']:]):.4f}"
msg += f"\n Avg Action Decision Time: {numpy.average(self.listOfTimesForActionDecision[-self.config['testing_print_every']:]):.4f}"
msg += f"\n Avg Action Execution Time: {numpy.average(self.listOfTimesForActionExecution[-self.config['testing_print_every']:]):.4f}"
msg += f"\n Avg Miscellaneous Time: {numpy.average(self.listOfTimesForMiscellaneous[-self.config['testing_print_every']:]):.4f}"
msg += f"\n Avg Total Loop Time: {numpy.average(self.listOfTotalLoopTimes[-self.config['testing_print_every']:]):.4f}"
getLogger().info(msg)
def testingStepFinished(self, testingStep, executionSessions):
if self.config[
'precompute_sample_cache_single_threaded'] or self.config[
'precompute_sample_cache_num_workers'] == 1:
for session in executionSessions:
getLogger().info(
f"Preparing samples for {session.id} and adding them to the sample cache."
)
TrainingManager.addExecutionSessionToSampleCache(
session.id, self.config)
else:
# For some reason, we are getting frequent errors with the multi-process
# based version of this code shown below in our cloud environment.
with concurrent.futures.ProcessPoolExecutor(
max_workers=self.
config['precompute_sample_cache_num_workers']) as executor:
futures = []
for session in executionSessions:
getLogger().info(
f"Preparing samples for {session.id} and adding them to the sample cache."
)
futures.append(
executor.submit(
PrecomputeSessionsForSampleCache.
addExecutionSessionToSampleCache, session.id,
self.config))
for future in futures:
future.result()
def getResourceLink(resourceURL,
baseURL,
resourceDataModifyFunc=None):
resourceURL = urlWithoutFragment(
urllib.parse.urljoin(baseURL, resourceURL))
versionId, data = proxy.getResourceVersion(resourceURL)
if versionId is None:
try:
proxies = {
'http': f'http://127.0.0.1:{proxy.port}',
'https': f'http://127.0.0.1:{proxy.port}',
}
response = requests.get(resourceURL,
proxies=proxies,
verify=False)
versionId, data = proxy.getResourceVersion(resourceURL)
except requests.exceptions.RequestException:
pass
if data is None:
if resourceURL not in self.failedResourceUrls:
getLogger().warning(
f"WARNING! Failed to grab the resource at URL: {resourceURL}"
)
self.failedResourceUrls.add(resourceURL)
return None
else:
# urlPath = urllib.parse.urlparse(resourceURL).path
# extension = os.path.splitext(urlPath)[1]
# if resourceDataModifyFunc is not None:
# data = resourceDataModifyFunc(data, resourceURL)
return f"kwolaResourceVersion://{versionId}"
def encodeAndSaveVideo(self, executionSession, folder, lossless):
codec = chooseBestFfmpegVideoCodec(losslessPreferred=lossless)
crfRating = 25
if lossless:
crfRating = 0
result = subprocess.run(
[
'ffmpeg', '-f', 'image2', "-r", "1", '-i',
'kwola-screenshot-%05d.png', '-vcodec', codec, '-crf',
str(crfRating), '-preset', 'veryslow',
self.movieFileName(executionSession)
],
cwd=self.screenshotDirectory[executionSession.id],
stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
if result.returncode != 0:
errorMsg = f"Error! Attempted to create a movie using ffmpeg and the process exited with exit-code {result.returncode}. The following output was observed:\n"
errorMsg += str(result.stdout, 'utf8') + "\n"
errorMsg += str(result.stderr, 'utf8') + "\n"
getLogger().error(errorMsg)
raise RuntimeError(errorMsg)
else:
localVideoPath = self.movieFilePath(executionSession)
with open(localVideoPath, 'rb') as f:
videoData = f.read()
fileName = f'{str(executionSession.id)}.mp4'
self.config.saveKwolaFileData(folder, fileName, videoData)
os.unlink(localVideoPath)
def testingStepFinished(self, testingStep, executionSessions):
pool = multiprocessing.Pool(
self.config['testing_video_generation_processes'],
maxtasksperchild=1)
futures = []
for session in executionSessions:
future = pool.apply_async(func=createDebugVideoSubProcess,
args=(self.config.serialize(),
str(session.id), "", False, False,
None, None, "annotated_videos"))
futures.append((session, future))
for session, future in futures:
localFuture = future
# for retry in range(5):
# try:
value = localFuture.get(
timeout=self.config['debug_video_generation_timeout'])
if value:
getLogger().error(value)
# break
# except billiard.exceptions.WorkerLostError:
# if retry == 4:
# raise
# localFuture = pool.apply_async(func=createDebugVideoSubProcess,
# args=(self.config.serialize(), str(session.id), "", False, False, None, None, "annotated_videos"))
# except BrokenPipeError:
# if retry == 4:
# raise
# localFuture = pool.apply_async(func=createDebugVideoSubProcess,
# args=(self.config.serialize(), str(session.id), "", False, False, None, None, "annotated_videos"))
pool.close()
pool.join()
def extractBranchTrace(self, webDriver):
# The JavaScript that we want to inject. This will extract out the Kwola debug information.
if isinstance(webDriver, Firefox):
injected_javascript = (
'const newCounters = {};'
'if (window.kwolaCounters)'
'{'
' for (const [key, value] of Object.entries(window.kwolaCounters))'
' {'
' newCounters[key] = Array.from(value);'
' }'
' return newCounters;'
'}'
'else'
'{'
' return null;'
'}')
elif isinstance(webDriver, Chrome) or isinstance(webDriver, Edge):
injected_javascript = ('return window.kwolaCounters;')
else:
raise RuntimeError("Unrecognized web driver class.")
result = webDriver.execute_script(injected_javascript)
# The JavaScript that we want to inject. This will extract out the Kwola debug information.
injected_javascript = (
'if (!window.kwolaCounters)'
'{'
' window.kwolaCounters = {};'
'}'
'Object.keys(window.kwolaCounters).forEach((fileName) => {'
' window.kwolaCounters[fileName].fill(0);'
'});')
try:
webDriver.execute_script(injected_javascript)
except selenium.common.exceptions.TimeoutException:
getLogger().warning(
f"Warning, timeout while running the script to reset the kwola line counters."
)
if result is not None:
# Cast everything to a numpy array so we don't have to do it later
for fileName, vector in result.items():
result[fileName] = numpy.array(vector)
else:
getLogger().warning(
f"Warning, did not find the kwola line counter object in the browser. This usually "
"indicates that there was an error either in translating the javascript, an error "
"in loading the page, or that the page has absolutely no javascript. "
f"On page: {webDriver.current_url}")
result = {}
return result
def afterActionRuns(self, webDriver, proxy, executionSession,
executionTrace, actionExecuted):
exceptions = self.extractExceptions(webDriver)
for exception in exceptions:
msg, source, lineno, colno, stack = tuple(exception)
msg = str(msg)
source = str(source)
stack = str(stack)
combinedMessage = msg + source + stack
kwolaJSRewriteErrorFound = False
for detectionString in kwolaJSRewriteErrorDetectionStrings:
if detectionString in combinedMessage:
kwolaJSRewriteErrorFound = True
if kwolaJSRewriteErrorFound:
logMsgString = f"Error. There was a bug generated by the underlying javascript application, " \
f"but it appears to be a bug in Kwola's JS rewriting. Please notify the Kwola " \
f"developers that this url: {webDriver.current_url} gave you a js-code-rewriting " \
f"issue. \n"
logMsgString += f"{msg} at line {lineno} column {colno} in {source}\n"
logMsgString += f"{str(stack)}\n"
getLogger().error(logMsgString)
else:
error = ExceptionError(type="exception",
page=executionTrace.startURL,
stacktrace=stack,
message=msg,
source=source,
lineNumber=lineno,
columnNumber=colno)
executionTrace.errorsDetected.append(error)
errorHash = error.computeHash()
executionTrace.didErrorOccur = True
if errorHash not in self.errorHashes[executionSession.id]:
if errorHash not in self.allErrorHashes and not self.isDuplicate(
error):
logMsgString = f"An unhandled exception was detected in client application:\n"
logMsgString += f"{msg} at line {lineno} column {colno} in {source}\n"
logMsgString += f"{str(stack)}"
getLogger().info(logMsgString)
self.allErrorHashes.add(errorHash)
self.allErrors.append(error)
self.errorHashes[executionSession.id].add(errorHash)
executionTrace.didNewErrorOccur = True
def afterActionRuns(self, webDriver, proxy, executionSession, executionTrace, actionExecuted):
branchTrace = self.extractBranchTrace(webDriver)
newBranches = False
filteredBranchTrace = {}
for fileName in branchTrace.keys():
traceVector = branchTrace[fileName]
didExecuteFile = bool(numpy.sum(traceVector) > 0)
if didExecuteFile:
filteredBranchTrace[fileName] = traceVector
if fileName in self.cumulativeBranchTrace[executionSession.id]:
cumulativeTraceVector = self.cumulativeBranchTrace[executionSession.id][fileName]
if len(traceVector) == len(cumulativeTraceVector):
newBranchCount = numpy.sum(traceVector[cumulativeTraceVector == 0])
if newBranchCount > 0:
newBranches = True
else:
if didExecuteFile:
newBranches = True
else:
if didExecuteFile:
newBranches = True
executionTrace.branchTrace = {k: v.tolist() for k, v in filteredBranchTrace.items()}
executionTrace.didCodeExecute = bool(len(filteredBranchTrace) > 0)
executionTrace.didNewBranchesExecute = bool(newBranches)
total = 0
executedAtleastOnce = 0
for fileName in self.cumulativeBranchTrace[executionSession.id]:
total += len(self.cumulativeBranchTrace[executionSession.id][fileName])
executedAtleastOnce += numpy.count_nonzero(self.cumulativeBranchTrace[executionSession.id][fileName])
# Just an extra check here to cover our ass in case of division by zero
if total == 0:
total += 1
executionTrace.cumulativeBranchCoverage = float(executedAtleastOnce) / float(total)
for fileName in filteredBranchTrace.keys():
if fileName in self.cumulativeBranchTrace[executionSession.id]:
if len(branchTrace[fileName]) == len(self.cumulativeBranchTrace[executionSession.id][fileName]):
self.cumulativeBranchTrace[executionSession.id][fileName] += branchTrace[fileName]
else:
getLogger().warning(
f"Warning! The file with fileName {fileName} has changed the size of its trace vector. This "
f"is very unusual and could indicate some strange situation with dynamically loaded javascript")
else:
self.cumulativeBranchTrace[executionSession.id][fileName] = branchTrace[fileName]
def browserSessionFinished(self, webDriver, proxy, executionSession):
result = subprocess.run(['ffmpeg', '-f', 'image2', "-r", "3", '-i', 'kwola-screenshot-%05d.png', '-vcodec',
chooseBestFfmpegVideoCodec(), '-pix_fmt', 'yuv420p', '-crf', '15', '-preset',
'veryslow', self.movieFileName(executionSession)],
cwd=self.screenshotDirectory[executionSession.id], stdout=subprocess.PIPE, stderr=subprocess.PIPE)
if result.returncode != 0:
errorMsg = f"Error! Attempted to create a movie using ffmpeg and the process exited with exit-code {result.returncode}. The following output was observed:\n"
errorMsg += str(result.stdout, 'utf8') + "\n"
errorMsg += str(result.stderr, 'utf8') + "\n"
getLogger().error(errorMsg)
def testingStepFinished(self, testingStep, executionSessions):
totalRewards = []
for session in executionSessions:
getLogger().info(
f"Session {session.tabNumber} finished with total reward: {session.totalReward:.3f}"
)
totalRewards.append(session.totalReward)
if len(totalRewards) > 0:
getLogger().info(
f"Mean total reward of all sessions: {numpy.mean(totalRewards):.3f}"
)
def testingStepFinished(self, testingStep, executionSessions):
with concurrent.futures.ProcessPoolExecutor(max_workers=2) as executor:
futures = []
for session in executionSessions:
getLogger().info(
f"[{os.getpid()}] Preparing samples for {session.id} and adding them to the sample cache."
)
futures.append(
executor.submit(
TrainingManager.addExecutionSessionToSampleCache,
session.id, self.config))
for future in futures:
future.result()
def willRewriteFile(self, url, contentType, fileData):
jsMimeTypes = [
"application/x-javascript", "application/javascript",
"application/ecmascript", "text/javascript", "text/ecmascript"
]
cleanedFileName = self.getCleanedFileName(url)
if ('_js' in url and not "_json" in url and not "_jsp" in url
and not url.endswith("_css")
) or str(contentType).strip().lower() in jsMimeTypes:
kind = filetype.guess(fileData)
mime = ''
if kind is not None:
mime = kind.mime
# Next, check to see that we haven't gotten an image or something else that we should ignore. This happens, surprisingly.
if mime.startswith("image/") or mime.startswith(
"video/") or mime.startswith("audio/") or mime.startswith(
"application/"):
return False
# For some reason, some websites send JSON data in files labelled as javascript files.
# So we have to double check to make sure we aren't looking at JSON data
try:
json.loads(str(fileData, 'utf8').lower())
return False
except json.JSONDecodeError:
pass
except UnicodeDecodeError:
pass
if fileData.startswith(b"<html>"):
return False
ignoreKeyword = self.findMatchingJavascriptFilenameIgnoreKeyword(
cleanedFileName)
if ignoreKeyword is None:
return True
else:
getLogger().info(
f"[{os.getpid()}] Warning: Ignoring the javascript file '{cleanedFileName}' because it matches the javascript ignore keyword '{ignoreKeyword}'. "
f"This means that no learnings will take place on the code in this file. If this file is actually part of your "
f"application and should be learned on, then please modify your config file kwola.json and remove the ignore "
f"keyword '{ignoreKeyword}' from the variable 'web_session_ignored_javascript_file_keywords'. This file will be "
f"cached without Kwola line counting installed. Its faster to install line counting only in the files that need "
f"it.")
return False
else:
return False
def afterActionRuns(self, webDriver, proxy, executionSession,
executionTrace, actionExecuted):
logEntries = webDriver.get_log(
'browser')[self.startLogCounts[executionSession.id]:]
for log in logEntries:
if log['level'] == 'SEVERE':
message = str(log['message'])
message = message.replace("\\n", "\n")
# If it looks like a network error, then ignore it because those are handled separately
if RecordLogEntriesAndLogErrors.networkErrorRegex.search(
message) is not None:
continue
kwolaJSRewriteErrorFound = False
for detectionString in kwolaJSRewriteErrorDetectionStrings:
if detectionString in message:
kwolaJSRewriteErrorFound = True
if kwolaJSRewriteErrorFound:
logMsgString = f"[{os.getpid()}] Error. There was a bug generated by the underlying javascript application, " \
f"but it appears to be a bug in Kwola's JS rewriting. Please notify the Kwola " \
f"developers that this url: {webDriver.current_url} gave you a js-code-rewriting " \
f"issue.\n"
logMsgString += f"{message}\n"
getLogger().error(logMsgString)
else:
error = LogError(type="log",
page=executionTrace.startURL,
message=message,
logLevel=log['level'])
executionTrace.errorsDetected.append(error)
errorHash = error.computeHash()
executionTrace.didErrorOccur = True
if errorHash not in self.errorHashes[executionSession.id]:
logMsgString = f"[{os.getpid()}] A log error was detected in client application:\n"
logMsgString += f"{message}\n"
getLogger().info(logMsgString)
self.errorHashes[executionSession.id].add(errorHash)
executionTrace.didNewErrorOccur = True
executionTrace.logOutput = "\n".join([str(log) for log in logEntries])
executionTrace.hadLogOutput = bool(executionTrace.logOutput)
def afterActionRuns(self, webDriver, proxy, executionSession, executionTrace, actionExecuted):
for networkError in proxy.getNetworkErrors():
networkError.page = executionTrace.startURL
executionTrace.errorsDetected.append(networkError)
errorHash = networkError.computeHash()
if errorHash not in self.errorHashes[executionSession.id]:
networkErrorMsgString = f"[{os.getpid()}] A network error was detected in client application:\n"
networkErrorMsgString += f"Path: {networkError.path}\n"
networkErrorMsgString += f"Status Code: {networkError.statusCode}\n"
networkErrorMsgString += f"Message: {networkError.message}\n"
getLogger().info(networkErrorMsgString)
self.errorHashes[executionSession.id].add(errorHash)
executionTrace.didNewErrorOccur = True
def generateVideoFilesForBugs(self, testingStep, bugObjects):
pool = multiprocessing.Pool(
self.config['testing_video_generation_processes'],
maxtasksperchild=1)
futures = []
for bugIndex, bug in enumerate(bugObjects):
future = pool.apply_async(func=createDebugVideoSubProcess,
args=(self.config.serialize(),
str(bug.executionSessionId),
f"{bug.id}_bug", False, False,
bug.stepNumber, bug.stepNumber + 3,
"bugs"))
futures.append((bugIndex, bug, future))
for bugIndex, bug, future in futures:
localFuture = future
# for retry in range(5):
# try:
value = localFuture.get(
timeout=self.config['debug_video_generation_timeout'])
if value:
getLogger().error(value)
# break
# except billiard.exceptions.WorkerLostError:
# if retry == 4:
# raise
# localFuture = pool.apply_async(func=createDebugVideoSubProcess, args=(
# self.config.serialize(), str(bug.executionSessionId), f"{bug.id}_bug", False, False, bug.stepNumber,
# bug.stepNumber + 3, "bugs"))
# except BrokenPipeError:
# if retry == 4:
# raise
# localFuture = pool.apply_async(func=createDebugVideoSubProcess, args=(
# self.config.serialize(), str(bug.executionSessionId), f"{bug.id}_bug", False, False, bug.stepNumber,
# bug.stepNumber + 3, "bugs"))
pool.close()
pool.join()
def extractBranchTrace(self, webDriver):
# The JavaScript that we want to inject. This will extract out the Kwola debug information.
injected_javascript = (
'return window.kwolaCounters;'
)
result = webDriver.execute_script(injected_javascript)
# The JavaScript that we want to inject. This will extract out the Kwola debug information.
injected_javascript = (
'if (!window.kwolaCounters)'
'{'
' window.kwolaCounters = {};'
'}'
'Object.keys(window.kwolaCounters).forEach((fileName) => {'
' window.kwolaCounters[fileName].fill(0);'
'});'
)
try:
webDriver.execute_script(injected_javascript)
except selenium.common.exceptions.TimeoutException:
getLogger().warning(f"[{os.getpid()}] Warning, timeout while running the script to reset the kwola line counters.")
if result is not None:
# Cast everything to a numpy array so we don't have to do it later
for fileName, vector in result.items():
result[fileName] = numpy.array(vector)
else:
getLogger().warning(f"[{os.getpid()}] Warning, did not find the kwola line counter object in the browser. This usually "
"indicates that there was an error either in translating the javascript, an error "
"in loading the page, or that the page has absolutely no javascript. "
f"On page: {webDriver.current_url}")
result = {}
return result
def rewriteFile(self, url, contentType, fileData):
jsFileContents = fileData.strip()
strictMode = False
if jsFileContents.startswith(
b"'use strict';") or jsFileContents.startswith(
b'"use strict";'):
strictMode = True
jsFileContents = jsFileContents.replace(b"'use strict';", b"")
jsFileContents = jsFileContents.replace(b'"use strict";', b"")
wrapperStart = b""
wrapperEnd = b""
for wrapper in JSRewriter.knownResponseWrappers:
if jsFileContents.startswith(
wrapper[0]) and jsFileContents.endswith(wrapper[1]):
jsFileContents = jsFileContents[len(wrapper[0]
):-len(wrapper[1])]
wrapperStart = wrapper[0]
wrapperEnd = wrapper[1]
cleanedFileName = self.getCleanedFileName(url)
longFileHash, shortFileHash = ProxyPluginBase.computeHashes(
bytes(fileData))
fileNameForBabel = shortFileHash + "_" + cleanedFileName
environment = dict(os.environ)
environment['KWOLA_ENABLE_LINE_COUNTING'] = 'true'
environment['KWOLA_ENABLE_EVENT_HANDLER_TRACKING'] = 'true'
noLineCountingKeyword = self.findMatchingJavascriptFilenameNoLineCountingKeyword(
cleanedFileName)
if noLineCountingKeyword is not None:
environment['KWOLA_ENABLE_LINE_COUNTING'] = 'false'
getLogger().info(
f"Warning: Not installing line counting in the javascript file '{cleanedFileName}' because it matches the "
f"javascript no line counting keyword '{noLineCountingKeyword}'. Event handler tracking will still be installed."
f"This means that no learnings will take place on the code in this file. If this file is actually part of your "
f"application and should be learned on, then please modify your config file kwola.json and remove the ignore "
f"keyword '{noLineCountingKeyword}' from the variable 'web_session_no_line_counting_javascript_file_keywords'. This file will be "
f"cached without Kwola line counting installed. Its faster to install line counting only in the files that need "
f"it.")
babelCmd = 'babel'
if sys.platform == "win32" or sys.platform == "win64":
babelCmd = 'babel.cmd'
result = subprocess.run([
babelCmd, '-f', fileNameForBabel, '--plugins',
'babel-plugin-kwola', '--retain-lines', '--source-type', "script"
],
input=jsFileContents,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
env=environment)
if result.returncode != 0 and "'import' and 'export' may appear only with" in str(
result.stderr, 'utf8'):
result = subprocess.run([
babelCmd, '-f', fileNameForBabel, '--plugins',
'babel-plugin-kwola', '--retain-lines', '--source-type',
"module"
],
input=jsFileContents,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
env=environment)
if result.returncode != 0:
cutoffLength = 250
kind = filetype.guess(fileData)
mime = ''
if kind is not None:
mime = kind.mime
getLogger().warning(
f"Unable to install Kwola line-counting in the Javascript file {url}. Most"
f" likely this is because Babel thinks your javascript has invalid syntax, or that"
f" babel is not working / not able to find the babel-plugin-kwola / unable to"
f" transpile the javascript for some other reason. See the following truncated"
f" output:")
if len(result.stdout) > 0:
getLogger().warning(result.stdout[:cutoffLength])
else:
getLogger().warning("No data in standard output")
if len(result.stderr) > 0:
getLogger().warning(result.stderr[:cutoffLength])
else:
getLogger().warning("No data in standard error output")
return fileData
else:
# Check to see if the resulting file object had multiple branches
if noLineCountingKeyword is None and not self.checkIfRewrittenJSFileHasMultipleBranches(
result.stdout):
getLogger().warning(
f"Ignoring the javascript file {url} because it looks like a JSONP-style request, or some other javascript "
f"file without a significant number of code branches.")
return fileData
getLogger().info(
f"Successfully translated {url} with Kwola modifications.")
transformed = wrapperStart + result.stdout + wrapperEnd
if strictMode:
transformed = b'"use strict";\n' + transformed
return transformed
def browserSessionFinished(self, webDriver, proxy, executionSession):
if executionSession.bestApplicationProvidedCumulativeFitness is not None:
getLogger().info(f"Session {executionSession.id} finished with fitness: {executionSession.bestApplicationProvidedCumulativeFitness:.0f}")
def rewriteFile(self, url, contentType, fileData):
jsFileContents = fileData.strip()
strictMode = False
if jsFileContents.startswith(
b"'use strict';") or jsFileContents.startswith(
b'"use strict";'):
strictMode = True
jsFileContents = jsFileContents.replace(b"'use strict';", b"")
jsFileContents = jsFileContents.replace(b'"use strict";', b"")
wrapperStart = b""
wrapperEnd = b""
for wrapper in JSRewriter.knownResponseWrappers:
if jsFileContents.startswith(
wrapper[0]) and jsFileContents.endswith(wrapper[1]):
jsFileContents = jsFileContents[len(wrapper[0]
):-len(wrapper[1])]
wrapperStart = wrapper[0]
wrapperEnd = wrapper[1]
cleanedFileName = self.getCleanedFileName(url)
longFileHash, shortFileHash = ProxyPluginBase.computeHashes(
bytes(fileData))
fileNameForBabel = shortFileHash + "_" + cleanedFileName
result = subprocess.run([
'babel', '-f', fileNameForBabel, '--plugins', 'babel-plugin-kwola',
'--retain-lines', '--source-type', "script"
],
input=jsFileContents,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
if result.returncode != 0 and "'import' and 'export' may appear only with" in str(
result.stderr, 'utf8'):
result = subprocess.run([
'babel', '-f', fileNameForBabel, '--plugins',
'babel-plugin-kwola', '--retain-lines', '--source-type',
"module"
],
input=jsFileContents,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
if result.returncode != 0:
cutoffLength = 250
kind = filetype.guess(fileData)
mime = ''
if kind is not None:
mime = kind.mime
getLogger().warning(
f"[{os.getpid()}] Unable to install Kwola line-counting in the Javascript file {url}. Most"
f" likely this is because Babel thinks your javascript has invalid syntax, or that"
f" babel is not working / not able to find the babel-plugin-kwola / unable to"
f" transpile the javascript for some other reason. See the following truncated"
f" output:")
if len(result.stdout) > 0:
getLogger().warning(result.stdout[:cutoffLength])
else:
getLogger().warning("No data in standard output")
if len(result.stderr) > 0:
getLogger().warning(result.stderr[:cutoffLength])
else:
getLogger().warning("No data in standard error output")
return fileData
else:
getLogger().info(
f"[{os.getpid()}] Successfully translated {url} with Kwola branch counting and event tracing."
)
transformed = wrapperStart + result.stdout + wrapperEnd
if strictMode:
transformed = b'"use strict";\n' + transformed
return transformed
def afterActionRuns(self, webDriver, proxy, executionSession,
executionTrace, actionExecuted):
if isinstance(webDriver, Firefox):
logEntries = []
rawLogs = webDriver.execute_script(
"""const logs = window.kwolaLogs; window.kwolaLogs = []; return logs;"""
)
if rawLogs is not None:
for entry in rawLogs:
logEntries.append({
"level":
entry[0],
"message":
" ".join([str(m) for m in entry[1]])
})
elif isinstance(webDriver, Chrome) or isinstance(webDriver, Edge):
logEntries = webDriver.get_log(
'browser')[self.startLogCounts[executionSession.id]:]
else:
raise RuntimeError("Unrecognized web driver class")
for log in logEntries:
if log['level'] == 'SEVERE' or log['level'] == 'error':
message = str(log['message'])
message = message.replace("\\n", "\n")
# If it looks like a network error, then ignore it because those are handled separately
if RecordLogEntriesAndLogErrors.networkErrorRegex.search(
message) is not None:
continue
kwolaJSRewriteErrorFound = False
for detectionString in kwolaJSRewriteErrorDetectionStrings:
if detectionString in message:
kwolaJSRewriteErrorFound = True
break
ignoreErrorKeywordFound = False
matchingIgnoreKeyword = None
for ignoreKeyword in self.config[
'web_session_ignored_log_error_keywords']:
if ignoreKeyword in message:
ignoreErrorKeywordFound = True
matchingIgnoreKeyword = ignoreKeyword
break
if kwolaJSRewriteErrorFound:
logMsgString = f"Error. There was a bug generated by the underlying javascript application, " \
f"but it appears to be a bug in Kwola's JS rewriting. Please notify the Kwola " \
f"developers that this url: {webDriver.current_url} gave you a js-code-rewriting " \
f"issue.\n"
logMsgString += f"{message}\n"
getLogger().error(logMsgString)
elif ignoreErrorKeywordFound:
logMsgString = f"Suppressed an error message because it matched the " \
f"error ignore keyword {matchingIgnoreKeyword}. "
logMsgString += f"{message}\n"
getLogger().info(logMsgString)
else:
error = LogError(type="log",
page=executionTrace.startURL,
message=message,
logLevel=log['level'])
executionTrace.errorsDetected.append(error)
errorHash = error.computeHash()
executionTrace.didErrorOccur = True
if errorHash not in self.errorHashes[executionSession.id]:
if errorHash not in self.allErrorHashes and not self.isDuplicate(
error):
logMsgString = f"A log error was detected in client application:\n"
logMsgString += f"{message}\n"
getLogger().info(logMsgString)
self.allErrorHashes.add(errorHash)
self.allErrors.append(error)
self.errorHashes[executionSession.id].add(errorHash)
executionTrace.didNewErrorOccur = True
executionTrace.logOutput = "\n".join([str(log) for log in logEntries])
executionTrace.hadLogOutput = bool(executionTrace.logOutput)
def browserSessionFinished(self, webDriver, proxy, executionSession):
getLogger().info(
f"Creating movie file for the execution session {executionSession.id}"
)
self.encodeAndSaveVideo(executionSession, "videos", False)
self.encodeAndSaveVideo(executionSession, "videos_lossless", True)
def testingStepFinished(self, testingStep, executionSessions):
kwolaVideoDirectory = self.config.getKwolaUserDataDirectory("videos")
existingBugs = self.loadAllBugs(testingStep)
bugObjects = []
executionSessionsById = {}
for session in executionSessions:
executionSessionsById[session.id] = session
for errorIndex, error, executionSessionId, stepNumber in zip(
range(len(self.newErrorsThisTestingStep[testingStep.id])),
self.newErrorsThisTestingStep[testingStep.id],
self.newErrorOriginalExecutionSessionIds[testingStep.id],
self.newErrorOriginalStepNumbers[testingStep.id]):
if error.type == "http":
if error.statusCode == 400 and not self.config[
'enable_400_error']:
continue # Skip this error
if error.statusCode == 401 and not self.config[
'enable_401_error']:
continue # Skip this error
if error.statusCode == 403 and not self.config[
'enable_403_error']:
continue # Skip this error
if error.statusCode == 404 and not self.config[
'enable_404_error']:
continue # Skip this error
if error.statusCode >= 500 and not self.config[
'enable_5xx_error']:
continue # Skip this error
elif error.type == "log":
if not self.config['enable_javascript_console_error']:
continue # Skip this error
elif error.type == "exception":
if not self.config['enable_unhandled_exception_error']:
continue # Skip this error
bug = BugModel()
bug.owner = testingStep.owner
bug.applicationId = testingStep.applicationId
bug.testingStepId = testingStep.id
bug.executionSessionId = executionSessionId
bug.creationDate = datetime.now()
bug.stepNumber = stepNumber
bug.error = error
bug.testingRunId = testingStep.testingRunId
bug.actionsPerformed = [
trace.actionPerformed
for trace in self.executionSessionTraces[executionSessionId]
][:(bug.stepNumber + 2)]
bug.browser = executionSessionsById[executionSessionId].browser
bug.userAgent = executionSessionsById[executionSessionId].userAgent
bug.windowSize = executionSessionsById[
executionSessionId].windowSize
tracesForScore = [
trace
for trace in self.executionSessionTraces[executionSessionId]
[max(0, stepNumber - 5):(stepNumber + 1)]
if trace.codePrevalenceScore is not None
]
if len(tracesForScore) > 0:
bug.codePrevalenceScore = numpy.mean(
[trace.codePrevalenceScore for trace in tracesForScore])
else:
bug.codePrevalenceScore = None
bug.isBugNew = True
bug.recomputeBugQualitativeFeatures()
duplicate = False
for existingBug in existingBugs:
if bug.isDuplicateOf(existingBug):
duplicate = True
break
if not duplicate:
bug.id = CustomIDField.generateNewUUID(BugModel, self.config)
bug.saveToDisk(self.config,
overrideSaveFormat="json",
overrideCompression=0)
bug.saveToDisk(self.config)
bugTextFile = os.path.join(
self.config.getKwolaUserDataDirectory("bugs"),
bug.id + ".txt")
saveKwolaFileData(bugTextFile,
bytes(bug.generateBugText(), "utf8"),
self.config)
bugVideoFilePath = os.path.join(
self.config.getKwolaUserDataDirectory("bugs"),
bug.id + ".mp4")
origVideoFilePath = os.path.join(
kwolaVideoDirectory, f'{str(executionSessionId)}.mp4')
origVideoFileData = loadKwolaFileData(origVideoFilePath,
self.config)
saveKwolaFileData(bugVideoFilePath, origVideoFileData,
self.config)
getLogger().info(
f"\n\nBug #{errorIndex + 1}:\n{bug.generateBugText()}\n")
existingBugs.append(bug)
bugObjects.append(bug)
getLogger().info(
f"\n\nBug #{errorIndex + 1}:\n{bug.generateBugText()}\n")
getLogger().info(
f"Found {len(self.newErrorsThisTestingStep[testingStep.id])} new unique errors this session."
)
testingStep.bugsFound = len(
self.newErrorsThisTestingStep[testingStep.id])
testingStep.errors = self.newErrorsThisTestingStep[testingStep.id]
self.generateVideoFilesForBugs(testingStep, bugObjects)
