def setupLogging(logLevel, fileName = None, logToStd = True):
"""Logging for the server module. Default
level is INFO.
"""
#Message queue
taskLogger = logging.getLogger("Asrt")
taskLogger.setLevel(logLevel)
#Rendering engines
mediaparlFormatter = MultiLineFormatter("%(lineno)-4d : %(levelname)-10s %(name)-30s %(asctime)-25s %(message)s")
if fileName != None:
#Check and make directory
MyFile.checkDirExists(MyFile(fileName).getFileDir())
fileHandler = logging.handlers.RotatingFileHandler(filename=fileName,maxBytes=1024000, backupCount=5)
fileHandler.setLevel(logLevel)
fileHandler.setFormatter(mediaparlFormatter)
taskLogger.addHandler(fileHandler)
if logToStd:
streamHandler = logging.StreamHandler(sys.stdout)
streamHandler.setLevel(logLevel)
streamHandler.setFormatter(mediaparlFormatter)
taskLogger.addHandler(streamHandler)
return taskLogger
def setupLogging(logLevel, fileName=None, logToStd=True):
"""Logging for the server module. Default
level is INFO.
"""
#Message queue
taskLogger = logging.getLogger("Asrt")
taskLogger.setLevel(logLevel)
#Rendering engines
mediaparlFormatter = MultiLineFormatter(
"%(lineno)-4d : %(levelname)-10s %(name)-30s %(asctime)-25s %(message)s"
)
if fileName != None:
#Check and make directory
MyFile.checkDirExists(MyFile(fileName).getFileDir())
fileHandler = logging.handlers.RotatingFileHandler(filename=fileName,
maxBytes=1024000,
backupCount=5)
fileHandler.setLevel(logLevel)
fileHandler.setFormatter(mediaparlFormatter)
taskLogger.addHandler(fileHandler)
if logToStd:
streamHandler = logging.StreamHandler(sys.stdout)
streamHandler.setLevel(logLevel)
streamHandler.setFormatter(mediaparlFormatter)
taskLogger.addHandler(streamHandler)
return taskLogger
def execute(commandList, logPath, outFileName = None, errFileName = None):
"""Wrapper to execute a sub process.
"""
#Make sure the directory exists
MyFile.checkDirExists(logPath)
stdout, stderr, retCode = None, None, 0
try:
#Default to one log
p = subprocess.Popen(commandList, stdout=subprocess.PIPE,
stderr=subprocess.STDOUT)
if errFileName is not None:
p = subprocess.Popen(commandList, stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
#Run the subprocess
stdout, stderr = p.communicate()
retCode = p.poll()
except Exception, e:
AsrtSubprocess.logger.critical("Subprocess error: %s" % str(e))
errorMessage = str(commandList) + "\n" + \
"------------ Begin stack ------------\n" + \
traceback.format_exc().rstrip() + "\n" + \
"------------ End stack --------------"
print errorMessage
#Make sure the trace is logged
if stderr is None:
stderr = errorMessage
else:
stderr += errorMessage
retCode = 1
def _readDataList(self):
"""Read the only data list from 'target directory'.
"""
self._log(logging.INFO,
"Gather input lists from %s" % self.getTargetDirectory())
#Input list
dataListFiles = MyFile.dirContent(self.getTargetDirectory(),
"*" + Task.OUTPUTLISTEXTENSION)
#Input data or representations
if len(dataListFiles) == 0:
raise Exception("No data list found in %s!" %
self.getTargetDirectory())
elif len(dataListFiles) > 1:
raise Exception("One input list max, %d found!" %
len(dataListFiles))
self._log(logging.INFO, "Found data list: %s!" % dataListFiles[0])
#Copy from target directory
dataListSrcPath = self.getTargetDirectory() + os.sep + dataListFiles[0]
dataListDestPath = self.getInputDirectory() + os.sep +\
MyFile.removeExtension(dataListFiles[0])[0] + Task.INPUTLISTEXTENSION
MyFile.copyFile(dataListSrcPath, dataListDestPath)
#Read content
self.inputList = DataList()
self.inputList.readFile(dataListDestPath)
def testConvertToText(self): rep = TextRepresentation(TestTextRepresentation.pdfFile, TEMPDIRUNITTEST, LOGDIR) rep.convertToText() MyFile.checkFileExists(TestTextRepresentation.tmpPdfFile) rep = TextRepresentation(TestTextRepresentation.textFile, TEMPDIRUNITTEST, LOGDIR) rep.convertToText() MyFile.checkFileExists(TestTextRepresentation.tmpTextFile)
def testConvertToText(self):
rep = TextRepresentation(TestTextRepresentation.pdfFile,
TEMPDIRUNITTEST, LOGDIR)
rep.convertToText()
MyFile.checkFileExists(TestTextRepresentation.tmpPdfFile)
rep = TextRepresentation(TestTextRepresentation.textFile,
TEMPDIRUNITTEST, LOGDIR)
rep.convertToText()
MyFile.checkFileExists(TestTextRepresentation.tmpTextFile)
def execute(commandList, logPath, outFileName=None, errFileName=None):
"""Wrapper to execute a sub process.
"""
#Make sure the directory exists
MyFile.checkDirExists(logPath)
stdout, stderr, retCode = None, None, 0
try:
#Default to one log
p = subprocess.Popen(commandList,
stdout=subprocess.PIPE,
stderr=subprocess.STDOUT)
if errFileName is not None:
p = subprocess.Popen(commandList,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
#Run the subprocess
stdout, stderr = p.communicate()
retCode = p.poll()
except Exception as e:
AsrtSubprocess.logger.critical("Subprocess error: %s" % str(e))
errorMessage = str(commandList) + "\n" + \
"------------ Begin stack ------------\n" + \
traceback.format_exc().rstrip() + "\n" + \
"------------ End stack --------------"
print(errorMessage)
#Make sure the trace is logged
if stderr is None:
stderr = errorMessage
else:
stderr += errorMessage
retCode = 1
#Now log results
#It is important to be ouside exception management as we
#still want to log what happened
io = Ioread()
if stdout != None and len(stdout) > 0 and outFileName != None:
io.writeFileContent("%s/%s" % (logPath, outFileName),
str(stdout, 'utf-8'))
if stderr != None and len(stderr) > 0 and errFileName != None:
io.writeFileContent("%s/%s" % (logPath, errFileName),
str(stderr, 'utf-8'))
return retCode, stdout, stderr
def loadTextFile(self):
"""Load converted text file.
"""
if self.tempFilePath is None or not MyFile.checkFileExists(self.tempFilePath):
raise Exception("Temporary text file does not exist!")
io = Ioread()
self.sentencesList = io.readFileContentList(self.tempFilePath)
def loadTextFile(self):
"""Load converted text file.
"""
if self.tempFilePath is None or not MyFile.checkFileExists(
self.tempFilePath):
raise Exception("Temporary text file does not exist!")
io = Ioread()
self.sentencesList = io.readFileContentList(self.tempFilePath)
def prepareOutputData(self):
"""Copy results, old lists and build new input
and map lists.
"""
self._log(logging.INFO, "Copy results files to output folder:%s" % self.getOutputDirectory())
# Data maps
dataMapFiles = MyFile.dirContent(self.getTempDirectory(), "*sentences_*.txt")
for sentenceFile in dataMapFiles:
srcFile = self.getTempDirectory() + os.sep + sentenceFile
shutil.copy(srcFile, self.getOutputDirectory())
def prepareOutputData(self):
"""Copy results, old lists and build new input
and map lists.
"""
self._log(logging.INFO, "Copy results files to output folder:%s" %
self.getOutputDirectory())
#Data maps
dataMapFiles = MyFile.dirContent(self.getTempDirectory(),
"*sentences_*.txt")
for sentenceFile in dataMapFiles:
srcFile = self.getTempDirectory() + os.sep + sentenceFile
shutil.copy(srcFile,self.getOutputDirectory())
def loadDocumentAsSentences(self, tempDir):
"""Convert to text, remove new lines and
segment into sentences using NLTK
toolkit.
"""
#Pdf to text
tempFileName = self.convertToText(self.sourceFileName, tempDir, self.logDir)
#Segment into sentences using NLTK toolkit
self._loadTextDocumentAsSentences(tempFileName)
#Delete temporary file
MyFile(tempFileName).removeFile(tempFileName)
def gatherInputData(self):
"""Prepare task directories and load data list
and map lists.
"""
workingDirectory = self.getWorkingDirectory()
targetDirectory = self.getTargetDirectory()
self.taskDirectory = "%s%s%s" % (workingDirectory, os.sep,
self.taskInstanceName)
#Don't want to keep old results
MyFile.forceRemoveDir(self.taskDirectory)
#Make task working directory
MyFile.makeDir(self.taskDirectory)
#Sub folders
MyFile.makeDir(self.getInputDirectory())
MyFile.makeDir(self.getTempDirectory())
MyFile.makeDir(self.getOutputDirectory())
#Read data list and data maps
self._copyLists()
def _readMapLists(self):
"""Read the data maps from 'target directory'.
"""
#Data maps
dataMapFiles = MyFile.dirContent(self.getTargetDirectory(),
"*" + Task.OUTPUTMAPEXTENSION)
#Map of representations for data
if len(dataMapFiles) == 0:
raise Exception("No data map found in %s!" %
self.getTargetDirectory())
self._log(logging.INFO,
"Found %d input map list(s)!" % len(dataMapFiles))
for dataMapFile in dataMapFiles:
self._log(logging.INFO, "Found map list: %s!" % dataMapFile)
#Copy from target directory
dataMapSrcPath = self.getTargetDirectory() + os.sep + dataMapFile
dataMapDestPath = self.getInputDirectory() + os.sep +\
MyFile.removeExtension(dataMapFile)[0] + Task.INPUTMAPEXTENSION
MyFile.copyFile(dataMapSrcPath, dataMapDestPath)
#Read content
tempDataMap = DataMap()
tempDataMap.readFile(dataMapDestPath)
self.mapLists.append(tempDataMap)
self._log(logging.INFO,
"Lists have been copied to %s" % self.getInputDirectory())
#Debug information
dataListFiles = MyFile.dirContent(self.getInputDirectory(), "*")
self._log(logging.INFO,
"Files in input directory: '%s'." % ", ".join(dataListFiles))
def execute(commandList, logPath, outFileName=None, errFileName=None):
"""Wrapper to execute a sub process.
"""
#Make sure the directory exists
MyFile.checkDirExists(logPath)
stdout, stderr, retCode = None, None, 0
try:
#Default to one log
p = subprocess.Popen(commandList,
stdout=subprocess.PIPE,
stderr=subprocess.STDOUT)
if errFileName is not None:
p = subprocess.Popen(commandList,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
#Run the subprocess
stdout, stderr = p.communicate()
retCode = p.poll()
except Exception, e:
AsrtSubprocess.logger.critical("Subprocess error: %s" % str(e))
errorMessage = str(commandList) + "\n" + \
"------------ Begin stack ------------\n" + \
traceback.format_exc().rstrip() + "\n" + \
"------------ End stack --------------"
print errorMessage
#Make sure the trace is logged
if stderr is None:
stderr = errorMessage
else:
stderr += errorMessage
retCode = 1
def testGatherInputData(self):
#No data olist
task1 = Task(
TaskInfo("", TestTask.workingDirectory, TestTask.targetFolderErr1))
with self.assertRaises(Exception):
task1.gatherInputData()
#No map .omap
task1 = Task(
TaskInfo("", TestTask.workingDirectory, TestTask.targetFolderErr2))
with self.assertRaises(Exception):
task1.gatherInputData()
task1 = Task(
TaskInfo("", TestTask.workingDirectory, TestTask.targetFolder1))
task1.gatherInputData()
self.assertTrue(MyFile.checkFileExists(task1.getTaskDirectory()))
self.assertTrue(MyFile.checkFileExists(task1.getInputDirectory()))
self.assertTrue(MyFile.checkFileExists(task1.getTempDirectory()))
self.assertTrue(MyFile.checkFileExists(task1.getOutputDirectory()))
dataListPath = "%s%s%s" % (task1.getInputDirectory(), os.sep,
'data.ilist')
dataMap1Path = "%s%s%s" % (task1.getInputDirectory(), os.sep,
'audio.imap')
dataMap2Path = "%s%s%s" % (task1.getInputDirectory(), os.sep,
'model.imap')
self.assertTrue(MyFile.checkFileExists(dataListPath))
self.assertTrue(MyFile.checkFileExists(dataMap1Path))
self.assertTrue(MyFile.checkFileExists(dataMap2Path))
self.assertEqual(15, task1.inputList.getCount())
self.assertEqual(2, len(task1.mapLists))
for dataMap in task1.mapLists:
self.assertTrue(dataMap.getCount() in [2, 1])
task1 = Task(
TaskInfo("", TestTask.workingDirectory, TestTask.targetFolderErr))
#Two input lists
with self.assertRaises(Exception):
task1.gatherInputData()
def testGatherInputData(self):
#No data olist
task1 = Task(TaskInfo("",TestTask.workingDirectory,
TestTask.targetFolderErr1))
with self.assertRaises(Exception):
task1.gatherInputData()
#No map .omap
task1 = Task(TaskInfo("",TestTask.workingDirectory,
TestTask.targetFolderErr2))
with self.assertRaises(Exception):
task1.gatherInputData()
task1 = Task(TaskInfo("",TestTask.workingDirectory,
TestTask.targetFolder1))
task1.gatherInputData()
self.assertTrue(MyFile.checkFileExists(task1.getTaskDirectory()))
self.assertTrue(MyFile.checkFileExists(task1.getInputDirectory()))
self.assertTrue(MyFile.checkFileExists(task1.getTempDirectory()))
self.assertTrue(MyFile.checkFileExists(task1.getOutputDirectory()))
dataListPath = "%s%s%s" % (task1.getInputDirectory(),os.sep,'data.ilist')
dataMap1Path = "%s%s%s" % (task1.getInputDirectory(),os.sep,'audio.imap')
dataMap2Path = "%s%s%s" % (task1.getInputDirectory(),os.sep,'model.imap')
self.assertTrue(MyFile.checkFileExists(dataListPath))
self.assertTrue(MyFile.checkFileExists(dataMap1Path))
self.assertTrue(MyFile.checkFileExists(dataMap2Path))
self.assertEquals(15,task1.inputList.getCount())
self.assertEquals(2,len(task1.mapLists))
for dataMap in task1.mapLists:
self.assertTrue(dataMap.getCount() in [2,1])
task1 = Task(TaskInfo("",TestTask.workingDirectory,
TestTask.targetFolderErr))
#Two input lists
with self.assertRaises(Exception):
task1.gatherInputData()
def getTempFilePath(self):
"""Temporary version of the source file.
"""
return self.tempDir + os.sep + MyFile(
self.sourceFileName).getCurrentFileName() + ".tmp"
setupLogging(logging.INFO, outputDir + "/data_preparation_log.txt")
# Api setup
api = DataPreparationAPI(None, outputDir)
api.setRegexFile(regexFile)
api.setFilterSentences(filterSentences)
api.setFilterSentences2ndStage(filterSentences2ndStage)
api.setLMModeling(lmModeling)
api.setRemovePunctuation(removePunctuation)
api.setVerbalizePunctuation(verbalizePunctuation)
api.setSegmentWithNLTK(not rawSeg)
api.setExpandNumberInWords(expandNumberInWords)
if language == 0:
api.trainClassifier()
# Main processing
MyFile.checkDirExists(outputDir)
io = Ioread()
inputList = io.readFileContentList(inputList)
for i, f in enumerate(inputList):
api.setInputFile(f)
api.prepareDocument(language)
strUnformatted = api.getCleanedText()
outputFile = "%s/%s.lab" % (outputDir,
os.path.splitext(os.path.basename(f))[0])
io.writeFileContent(outputFile, strUnformatted + "\n")
setupLogging(logging.INFO, outputDir + "/data_preparation_log.txt")
#Api setup
api = DataPreparationAPI(None, outputDir)
api.setRegexFile(regexFile)
api.setFilterSentences(filterSentences)
api.setFilterSentences2ndStage(filterSentences2ndStage)
api.setLMModeling(lmModeling)
api.setRemovePunctuation(removePunctuation)
api.setVerbalizePunctuation(verbalizePunctuation)
api.setSegmentWithNLTK(not rawSeg)
api.setKeepNewWords(keepNewWords)
if language == 0:
api.trainClassifier()
#Main processing
MyFile.checkDirExists(outputDir)
io = Ioread()
inputList = io.readFileContentList(inputList)
for i, f in enumerate(inputList):
api.setInputFile(f)
api.prepareDocument(language)
strUnformatted = api.getCleanedText()
outputFile = "%s/%s.lab" % (outputDir, os.path.splitext(os.path.basename(f))[0])
io.writeFileContent(outputFile, strUnformatted + u"\n")
commonTestSuite = CommonTestSuite.getCommonTestSuite(unitTestList)
frenchTestSuite = FrenchTestSuite.getFrenchTestSuite(unitTestList)
germanTestSuite = GermanTestSuite.getGermanTestSuite(unitTestList)
formulaTestSuite = FormulaTestSuite.getFormulaTestSuite(unitTestList)
allTestSuite = []
if commonTestSuite is not None:
allTestSuite.extend(commonTestSuite)
if formulaTestSuite is not None:
allTestSuite.extend(formulaTestSuite)
if frenchTestSuite is not None:
allTestSuite.extend(frenchTestSuite)
if germanTestSuite is not None:
allTestSuite.extend(germanTestSuite)
allTests = unittest.TestSuite(allTestSuite)
return allTests
if __name__ == "__main__":
if len(sys.argv) < 2:
print getUsage()
print " usage: %s 'unit test name 1 or all' 'unit test name 2' " % sys.argv[0]
print ""
sys.exit(0)
MyFile.checkDirExists(TEMPDIRUNITTEST)
runner = unittest.TextTestRunner(verbosity = 2)
runner.run(asrtTestSuite(sys.argv[1:]))
allTestSuite = []
if commonTestSuite is not None:
allTestSuite.extend(commonTestSuite)
if formulaTestSuite is not None:
allTestSuite.extend(formulaTestSuite)
if frenchTestSuite is not None:
allTestSuite.extend(frenchTestSuite)
if germanTestSuite is not None:
allTestSuite.extend(germanTestSuite)
if englishTestSuite is not None:
allTestSuite.extend(englishTestSuite)
allTests = unittest.TestSuite(allTestSuite)
return allTests
if __name__ == "__main__":
if len(sys.argv) < 2:
print((getUsage()))
print((" usage: %s 'unit test name 1 or all' 'unit test name 2' " %
sys.argv[0]))
print("")
sys.exit(0)
MyFile.checkDirExists(TEMPDIRUNITTEST)
runner = unittest.TextTestRunner(verbosity=2)
runner.run(asrtTestSuite(sys.argv[1:]))
