def test_AMCuervo(display):
try:
assert (os.path.exists("../testTextPyData/Cuervo/audio"))
except AssertionError:
os.mkdir("../testTextPyData/Cuervo/audio")
audioFilename = "../testData/Cargos.ogg"
elanXmlFilename = "newcuervo.eaf"
targetDirectory = "../testTextPyData/Cuervo/audio"
soundFile = os.path.join(targetDirectory, "Cargos.ogg")
projectDirectory = "../testTextPyData/Cuervo"
tierGuideFile = "../testTextPyData/Cuervo/tierGuide.yaml"
elanFile = os.path.join(projectDirectory, elanXmlFilename)
ae = AudioExtractor(audioFilename, elanFile, targetDirectory)
ae.determineStartAndEndTimes()
ae.extract()
times = ae.startStopTable
text = Text(elanFile,
audioFilename,
grammaticalTermsFile=None,
tierGuideFile=tierGuideFile,
projectDirectory=projectDirectory)
print("--- test_toHTML")
text.getLineAsTable(0)
htmlText = text.toHTML()
# filename = "../testTextPyData/Cuervo/CuervoNew.html"
# f = open(filename, "w")
# f.write(indent(htmlText))
# f.close()
if (display):
os.system("open %s" % filename)
def test_praying_6_Line_from_webapp(display):
print("--- test_praying_6_Line_from_webappn")
audioFilename = "../testData/Cargos.ogg"
elanXmlFilename = "../testData/s/praying.eaf"
targetDirectory = "../testData/s/audio"
soundFile = os.path.join(targetDirectory, "Cargos.ogg")
projectDirectory = "../testData/s"
tierGuideFile = "../testData/s/tierGuide.yaml"
grammaticalTermsFile = "../testData/s/grammaticalTerms.txt"
ae = AudioExtractor(audioFilename, elanXmlFilename, targetDirectory)
ae.determineStartAndEndTimes()
times = ae.startStopTable
text = Text(elanXmlFilename,
audioFilename,
grammaticalTermsFile=grammaticalTermsFile,
tierGuideFile=tierGuideFile,
projectDirectory=projectDirectory)
# text.getTable(1)
display = False
htmlText = text.toHTML()
if (display):
filename = "6test.html"
f = open(filename, "w")
f.write(indent(htmlText))
f.close()
os.system("open %s" % filename)
def test_GhostWagon(display,extract):
'''tests .eaf file with empty and missing line or translation annotations'''
print("--- test_GhostWagon")
audioFilename = "../testTextPyData/GhostInWagon/GhostInWagon.ogg"
elanXmlFilename = "../testTextPyData/GhostInWagon/GhostInWagon_original.eaf"
targetDirectory = "../testTextPyData/GhostInWagon/audio"
soundFile = os.path.join(targetDirectory, "GhostInWagon.ogg")
projectDirectory = "../testTextPyData/GhostInWagon"
tierGuideFile = "../testTextPyData/Cuervo/tierGuide.yaml"
ae = AudioExtractor(audioFilename, elanXmlFilename, targetDirectory)
if extract:
ae.extract()
times = ae.startStopTable
text = Text(elanXmlFilename,
audioFilename,
grammaticalTermsFile=None,
tierGuideFile=tierGuideFile,
projectDirectory=projectDirectory)
htmlText = text.toHTML()
if (display):
filename = "../testTextPyData/GhostInWagon/GhostInWagon.html"
f = open(filename, "w")
f.write(indent(htmlText))
f.close()
os.system("open %s" % filename)
def test_AYAFW(display,audioFilename,extract):
print("--- test_AYAFW")
# audioFilename = "../testData/Cargos.ogg"
elanXmlFilename = "../testTextPyData/AYAFW/AYAFW.eaf"
targetDirectory = "../testTextPyData/AYAFW/audio"
soundFile = os.path.join(targetDirectory, "Cargos.ogg")
projectDirectory = "../testTextPyData/AYAFW"
tierGuideFile = "../testTextPyData/AYAFW/tierGuide.yaml"
grammaticalTermsFile = "../testTextPyData/AYAFW/abbreviations.txt"
ae = AudioExtractor(audioFilename, elanXmlFilename, targetDirectory)
if extract:
ae.extract()
text = Text(elanXmlFilename,
audioFilename,
grammaticalTermsFile=grammaticalTermsFile,
tierGuideFile=tierGuideFile,
projectDirectory=projectDirectory) # ,
# startStopTable=times)
# IjalLine.getTable(1)
htmlText = text.toHTML()
if (display):
filename = "../testTextPyData/AYAFW/test_AYAFW.html"
f = open(filename, "w")
f.write(indent(htmlText))
f.close()
os.system("open %s" % filename)
def __init__(self,
xmlFilename,
soundFileName,
grammaticalTermsFile,
tierGuideFile,
projectDirectory,
quiet=True):
self.xmlFilename = xmlFilename
self.soundFileName = soundFileName
self.audioPath = "audio"
self.audioFileType = self.soundFileName[-3:]
self.grammaticalTermsFile = grammaticalTermsFile
self.tierGuideFile = tierGuideFile
self.projectDirectory = projectDirectory
self.validInputs()
self.quiet = quiet
self.xmlDoc = etree.parse(self.xmlFilename)
with open(tierGuideFile, 'r') as f:
self.tierGuide = yaml.safe_load(f)
self.speechTier = self.tierGuide['speech']
self.speechTierList = identifyLines.getList(self.xmlDoc,
self.tierGuide)
self.lineCount = len(self.speechTierList)
if os.path.isfile(os.path.join(projectDirectory, "ERRORS.log")):
os.remove(os.path.join(projectDirectory, "ERRORS.log"))
logging.basicConfig(filename=os.path.join(projectDirectory,
"ERRORS.log"),
format="%(levelname)s %(message)s")
logging.getLogger().setLevel(logging.WARNING)
targetDirectory = os.path.join(projectDirectory, "audio")
self.audio = AudioExtractor(soundFileName, xmlFilename,
targetDirectory)
def test_Inferno(display,audioFilename,extract):
print("--- test_Inferno")
try:
assert(os.path.exists("../testTextPyData/Inferno/audio"))
except AssertionError:
os.mkdir("../testTextPyData/Inferno/audio")
# audioFilename = "../testTextPyData/Inferno/inferno-threeLines.wav"
elanXmlFilename = "../testTextPyData/Inferno/inferno-threeLines.eaf"
targetDirectory = "../testTextPyData/Inferno/audio"
soundFile = os.path.join(targetDirectory, audioFilename)
projectDirectory = "../testTextPyData/Inferno"
tierGuideFile = "../testTextPyData/Inferno/tierGuide.yaml"
grammaticalTermsFile = "../testTextPyData/Inferno/abbreviations.txt"
ae = AudioExtractor(audioFilename, elanXmlFilename, targetDirectory)
if extract:
ae.extract()
text = Text(elanXmlFilename,
audioFilename,
grammaticalTermsFile=grammaticalTermsFile,
tierGuideFile=tierGuideFile,
projectDirectory=projectDirectory)
htmlText = text.toHTML()
if (display):
filename = "../testTextPyData/Inferno/inferno-threeLines.html"
f = open(filename, "w")
f.write(indent(htmlText))
f.close()
os.system("open %s" % filename)
def test_Cuervo(display,audioFilename,extract):
print("--- test_Cuervo")
# audioFilename = "../testData/Cargos.ogg"
elanXmlFilename = "../testTextPyData/Cuervo/newcuervo.eaf"
targetDirectory = "../testTextPyData/Cuervo/audio"
soundFile = os.path.join(targetDirectory, "Cargos.ogg")
projectDirectory = "../testTextPyData/Cuervo"
tierGuideFile = "../testTextPyData/Cuervo/tierGuide.yaml"
ae = AudioExtractor(audioFilename, elanXmlFilename, targetDirectory)
if extract:
ae.extract()
text = Text(elanXmlFilename,
audioFilename,
grammaticalTermsFile=None,
tierGuideFile=tierGuideFile,
projectDirectory=projectDirectory)
htmlText = text.toHTML()
if (display):
filename = "../testTextPyData/Cuervo/newcuervo.html"
f = open(filename, "w")
f.write(indent(htmlText))
f.close()
os.system("open %s" % filename)
def createText():
audioFilename = "../testData/Cargos.ogg"
print("--- testing normal file")
elanXmlFilename = "../infernoDemo/inferno-threeLines.eaf"
validFile = validate_EAF(elanXmlFilename)
print("--- testing file with broken HTML tags")
elanXmlFilename = "../testData/praying/Prayer_broken-HTML_tags.eaf"
validFile = validate_EAF(elanXmlFilename)
if not validFile:
return False
print("--- testing file with HTML tags")
elanXmlFilename = "../testData/praying/Prayer_HTML_tags.eaf"
validFile = validate_EAF(elanXmlFilename)
if not validFile:
print("=== failed to validate file with tags")
targetDirectory = "../testData/praying/audio"
soundFile = os.path.join(targetDirectory, "Cargos.ogg")
projectDirectory = "../testData/praying"
ae = AudioExtractor(audioFilename, elanXmlFilename, targetDirectory)
ae.determineStartAndEndTimes()
text = Text(
elanXmlFilename,
audioFilename,
grammaticalTermsFile="../testData/praying/grammaticalTerms.txt",
tierGuideFile="../testData/praying/tierGuide.yaml",
projectDirectory=projectDirectory,
quiet=True)
return (text)
def test_ogg(display):
print("--- test_ogg")
audioFilename = "inferno-threeLines.ogg"
elanXmlFilename = "../testData/inferno-threeLines/inferno-threeLines.eaf"
targetDirectory = "../testData/inferno-threeLines/audio"
projectDirectory = "../testData/inferno-threeLines"
soundFile = os.path.join(projectDirectory, audioFilename)
print(soundFile)
tierGuideFile = "../testData/inferno-threeLines/tierGuide.yaml"
grammaticalTermsFile = "../testData/inferno-threeLines/grammaticalTerms.txt"
ae = AudioExtractor(soundFile, elanXmlFilename, targetDirectory)
ae.validInputs()
ae.extract(False)
times = ae.startStopTable
text = Text(elanXmlFilename,
soundFile,
grammaticalTermsFile=grammaticalTermsFile,
tierGuideFile=tierGuideFile,
projectDirectory=projectDirectory)
htmlText = text.toHTML()
if (display):
filename = "../testData/inferno-threeLines/inferno-threeLinesogg.html"
f = open(filename, "w")
f.write(indent(htmlText))
f.close()
os.system("open %s" % filename)
def test_Lazy_extraction():
elanXmlFilename = "../testTextPyData/Lazybones/Lazybones.eaf"
targetDirectory = "../testTextPyData/Lazybones/audio"
projectDirectory = "../testTextPyData/Lazybones"
tierGuideFile = "../testTextPyData/Lazybones/tierGuide.yaml"
grammaticalTermsFile = "../testTextPyData/Lazybones/abbreviations.txt"
audioFilename = "/Users/David/OpenSource/github/OldSlexilTestData/TEX_Lazy/4_TEX_TheLazyWoman.wav"
ae = AudioExtractor(audioFilename, elanXmlFilename, targetDirectory)
ae.extract()
text = Text(elanXmlFilename,
audioFilename,
grammaticalTermsFile=grammaticalTermsFile,
tierGuideFile=tierGuideFile,
projectDirectory=projectDirectory)
def createText(): audioFilename = "../testData/Cargos.ogg" elanXmlFilename="../testData/featherSnake/featherSnake.eaf" targetDirectory = "../testData/featherSnake/audio" soundFile = os.path.join(targetDirectory,"Cargos.ogg") ae = AudioExtractor(audioFilename, elanXmlFilename, targetDirectory) ae.determineStartAndEndTimes() times = ae.startStopTable text = Text(elanXmlFilename, audioFilename, grammaticalTermsFile="../testData/featherSnake/grammaticalTerms.txt", tierGuideFile="../testData/featherSnake/tierGuide.yaml", projectDirectory="../testData/featherSnake", #startStopTable=times, quiet=False) return(text)
def createText(): audioFilename = "../testData/Cargos.ogg" elanXmlFilename="../testData/loco/loco.eaf" targetDirectory = "../testData/loco/audio" soundFile = os.path.join(targetDirectory,"Cargos.ogg") projectDirectory="../testData/loco" tierGuideFile="../testData/loco/tierGuide.yaml" grammaticalTermsFile="../testData/loco/grammaticalTerms.txt" ae = AudioExtractor(audioFilename, elanXmlFilename, targetDirectory) ae.determineStartAndEndTimes() times = ae.startStopTable text = Text(elanXmlFilename, audioFilename, grammaticalTermsFile=grammaticalTermsFile, tierGuideFile=tierGuideFile, projectDirectory=projectDirectory)# , # startStopTable=times) return(text)
def createText():
audioFilename = "../testData/aktzini/18-06-03Aktzini-GA.wav"
elanXmlFilename = "../testData/aktzini/18-06-03Aktzini-GA.eaf"
targetDirectory = "../testData/aktzini/audio"
soundFile = os.path.join(targetDirectory, audioFilename)
projectDirectory = "../testData/aktzini"
tierGuideFile = "../testData/aktzini/tierGuide.yaml"
grammaticalTermsFile = "../testData/aktzini/grammaticalTerms.txt"
ae = AudioExtractor(audioFilename, elanXmlFilename, targetDirectory)
ae.extract()
ae.determineStartAndEndTimes()
times = ae.startStopTable
text = Text(elanXmlFilename,
audioFilename,
grammaticalTermsFile=grammaticalTermsFile,
tierGuideFile=tierGuideFile,
projectDirectory=projectDirectory,
quiet=False)
return (text)
def createText():
audioFilename = "inferno-threeLines.wav"
elanXmlFilename = "../testData/inferno-threeLines/inferno-threeLines.eaf"
targetDirectory = "../testData/inferno-threeLines/audio"
soundFile = os.path.join(targetDirectory, audioFilename)
projectDirectory = "../testData/inferno-threeLines"
tierGuideFile = "../testData/inferno-threeLines/tierGuide.yaml"
grammaticalTermsFile = "../testData/inferno-threeLines/grammaticalTerms.txt"
ae = AudioExtractor(audioFilename, elanXmlFilename, targetDirectory)
ae.determineStartAndEndTimes()
text = Text(elanXmlFilename,
audioFilename,
grammaticalTermsFile=grammaticalTermsFile,
tierGuideFile=tierGuideFile,
projectDirectory=projectDirectory,
quiet=True)
return (text)
def test_inferno_plus():
print("--- test_inferno_plus")
elanXmlFilename = "../testData/test_alignable/inferno-threeLines_plus_one.eaf"
audioFilename = "../testData/test_alignable/inferno-threeLines.wav"
targetDirectory = "../testData/test_alignable/audio"
fileList = os.listdir(targetDirectory)
for f in fileList:
target = os.path.join(targetDirectory, f)
os.remove(target)
ae = AudioExtractor(audioFilename, elanXmlFilename, targetDirectory)
ae.determineStartAndEndTimes()
ae.extract()
fileList = [
f for f in os.listdir(targetDirectory) if not f.startswith('.')
]
try:
assert (len(fileList) == 3)
print("There is the correct number of audiophrases.")
except AssertionError as e:
print("Error: There are %d audiophrases rather than 3" % len(fileList))
def test_aktzini():
print("--- test_aktzini")
elanXmlFilename = "../testData/aktzini/18-06-03Aktzini-GA.eaf"
audioFilename = "../testData/aktzini/18-06-03Aktzini-GA.wav"
targetDirectory = "../testData/aktzini/audio"
fileList = os.listdir(targetDirectory)
for f in fileList:
target = os.path.join(targetDirectory, f)
os.remove(target)
ae = AudioExtractor(audioFilename, elanXmlFilename, targetDirectory)
ae.determineStartAndEndTimes()
ae.extract()
fileList = [
f for f in os.listdir(targetDirectory) if not f.startswith('.')
]
try:
assert (len(fileList) == 16)
print("There is the correct number of audiophrases.")
except AssertionError as e:
print("Error: There are %d audiophrases rather than 3" % len(fileList))
def test_Lazybones():
print("--- test_Lazybones")
elanXmlFilename = "../testTextPyData/Lazybones/Lazybones.eaf"
audioFilename = "../testData/Cargos.ogg"
targetDirectory = "../testTextPyData/Lazybones/audio"
fileList = os.listdir(targetDirectory)
for f in fileList:
target = os.path.join(targetDirectory, f)
os.remove(target)
ae = AudioExtractor(audioFilename, elanXmlFilename, targetDirectory)
ae.determineStartAndEndTimes()
ae.extract()
fileList = [
f for f in os.listdir(targetDirectory) if not f.startswith('.')
]
try:
assert (len(fileList) == 396)
print("There is the correct number of audiophrases.")
except AssertionError as e:
print("Error: There are %d audiophrases rather than 396" %
len(fileList))
def test_Prayer(display,audioFilename,extract):
print("--- test_Prayer")
# audioFilename = "../testData/Cargos.ogg"
elanXmlFilename = "../testTextPyData/Prayer_superscript/praying.eaf"
targetDirectory = "../testTextPyData/Prayer_superscript/audio"
soundFile = os.path.join(targetDirectory, "Cargos.ogg")
projectDirectory = "../testTextPyData/Prayer_superscript"
tierGuideFile = "../testTextPyData/Prayer_superscript/tierGuide.yaml"
grammaticalTermsFile = "../testTextPyData/Prayer_superscript/grammaticalTerms.txt"
ae = AudioExtractor(audioFilename, elanXmlFilename, targetDirectory)
if extract:
ae.extract()
text = Text(elanXmlFilename,
audioFilename,
grammaticalTermsFile=grammaticalTermsFile,
tierGuideFile=tierGuideFile,
projectDirectory=projectDirectory) # ,
# startStopTable=times)
# IjalLine.getTable(1)
try:
htmlText = text.toHTML()
if (display):
filename = "../testTextPyData/Prayer_superscript/test_Prayer.html"
f = open(filename, "w")
f.write(indent(htmlText))
f.close()
os.system("open %s" % filename)
except TooManyMorphsError as e:
print(
"EAF error: There are more morphs (%d) than glosses (%d) in line %s." % (e.morphs, e.glosses, e.lineNumber))
except TooManyGlossesError as e:
print(
"EAF error: There are more glosses (%d) than morphs (%d) in line %s." % (e.glosses, e.morphs, e.lineNumber))
except EmptyTiersError as e:
print("EAF error: There are empty tiers or incomplete glosses after line %s" % e.lineNumber)
def test_Lazybones(display,audioFilename,extract):
print("--- test_Lazybones")
# audioFilename = "../testData/Cargos.ogg"
elanXmlFilename = "../testTextPyData/Lazybones/Lazybones.eaf"
targetDirectory = "../testTextPyData/Lazybones/audio"
soundFile = os.path.join(targetDirectory, "Cargos.ogg")
projectDirectory = "../testTextPyData/Lazybones"
tierGuideFile = "../testTextPyData/Lazybones/tierGuide.yaml"
grammaticalTermsFile = "../testTextPyData/Lazybones/abbreviations.txt"
# audioFilename ="/Users/David/OpenSource/github/OldSlexilTestData/TEX_Lazy/4_TEX_TheLazyWoman.wav"
ae = AudioExtractor(audioFilename, elanXmlFilename, targetDirectory)
if extract:
ae.extract()
text = Text(elanXmlFilename,
audioFilename,
grammaticalTermsFile=grammaticalTermsFile,
tierGuideFile=tierGuideFile,
projectDirectory=projectDirectory)
try:
htmlText = text.toHTML()
if (display):
filename = "../testTextPyData/Lazybones/test_Lazybones.html"
f = open(filename, "w")
f.write(indent(htmlText))
f.close()
os.system("open %s" % filename)
except TooManyMorphsError as e:
print(
"EAF error: There are more morphs (%d) than glosses (%d) in line %s." % (e.morphs, e.glosses, e.lineNumber))
except TooManyGlossesError as e:
print(
"EAF error: There are more glosses (%d) than morphs (%d) in line %s." % (e.glosses, e.morphs, e.lineNumber))
except EmptyTiersError as e:
print("EAF error: There are empty tiers or incomplete glosses after line %s" % e.lineNumber)