def printVocab(self, lexiconPath):
xml = openXml(lexiconPath, 'utf-8')
xml.open('lexicon')
xml.empty('phoneme-inventory')
xml.comment('special lemmas')
for special, orthEval in self.special.iteritems():
xml.open('lemma', special=special)
for orth in orthEval[0]:
xml.element('orth', orth)
xml.empty('phon')
if orthEval[1] is None:
xml.empty('eval')
else:
xml.open('eval')
xml.element('tok', orthEval[1])
xml.close()
xml.close()
xml.comment('regular lemmas')
for orth in self.vocab:
xml.open('lemma')
xml.element('orth', orth)
xml.empty('phon')
xml.close()
xml.close()
closeXml(xml)
if self.isLog:
print >> sys.stderr, '-->', lexiconPath
def parse(self, pathIn, pathOut, encoding): self.category_ = [] self.processLine_ = self.processHeaderLine_ fd = uopen(pathIn, encoding, 'r') for line in fd: self.processLine_(line) uclose(fd) xml = openXml(pathOut, encoding) self.writeBliss_(xml) closeXml(xml) if self.isLog: print >> sys.stderr, pathIn, '-->', pathOut
def convert(self):
""" reads the old cart tree and converts it to the new xml-format
"""
self._writer = openXml(self._pathToNewTree, self._newEncoding)
self._setNumberOfCartClasses()
self._reader = uopen(self._pathToOldTree)
self._writer.open("decision-tree")
self._convertPhonemeList()
self._convertQuestionList()
self._convertBinaryTree()
self._writer.close("decision-tree")
closeXml(self._writer)
uclose(self._reader)
## print >> stderr, 'Warning: exit status of \"' + aligner + '\" is', str(exitCode)
cmd = aligner + ' ' + ' '.join(argv)
print >> stderr, 'Run', '"' + cmd + '"'
os.system(cmd)
assert os.path.exists(alignmentPath)
print >> stderr, '\"' + alignmentPath + '\" created'
else:
print >> stderr, 'Alignment file \"' + alignmentPath + '\" is up to date'
print >> stderr
# calculate overall wer
resultPath = getNormalizedPath(
getValue(options.resultPath,
getDefaultResultPath(os.path.curdir, corpusFilename)))
if options.force or not valid(resultPath, alignmentPath):
print >> stderr, 'Calculate result from \"' + alignmentPath + '\" ...'
accu, stats = calculateErrorStatistics(alignmentPath, True, True,
corpusPath)
xml = openXml(resultPath, accu.encoding)
stats.writeXml(xml)
closeXml(xml)
assert os.path.exists(resultPath)
print >> stderr, '\"' + resultPath + '\" created'
del accu
del stats
else:
print >> stderr, 'Result file \"' + resultPath + '\" is up to date'
print >> stderr
uclose(stderr)
optparser.add_option("-s",
"--speaker",
dest="bySpeaker",
action="store_true",
default=False,
help="dump statistics by speaker")
optparser.add_option("-c",
"--condition",
dest="byCondition",
action="store_true",
default=False,
help="dump statistics by condition")
optparser.add_option(
"-b",
"--corpus",
dest="corpusPath",
default=None,
help="bliss corpus; used to get speaker and condition names",
metavar="FILE")
if len(sys.argv) == 1:
optparser.print_help()
sys.exit(0)
options, args = optparser.parse_args()
accu, stats = calculateErrorStatistics(args[0], options.bySpeaker,
options.byCondition,
options.corpusPath)
xml = openXml(options.output, accu.encoding)
stats.writeXml(xml)
closeXml(xml)
def openCorpus(self):
self.xml = openXml(self.corpusFile)
self.xml.open('corpus', name=self.name)
# speaker
# condition
self.segFd = uwrite(self.segFile)
def startFile(self, path, encoding): self.xml = openXml(self.outPath, encoding)