def __append_input(self, base_fn, base_ext):
for ae in base_ext:
if sppasFileUtils(base_fn + ae).exists() \
and base_fn not in self._inputs:
self._inputs.append(base_fn)
return True
return False
def extract_aligned(self, trstier, nametier):
""" Extract from a time-aligned transcription file.
:returns: a tuple with tracks and silences lists
"""
trstracks = []
silences = []
self._units = list()
self._names = list()
i = 0
last = trstier.GetSize()
while i < last:
# Set the current annotation values
ann = trstier[i]
# Save information
if ann.GetLabel().IsSilence():
start = ann.GetLocation().GetBegin().GetMidpoint()
end = ann.GetLocation().GetEnd().GetMidpoint()
# Verify next annotations (concatenate all silences between 2 tracks)
if (i + 1) < last:
nextann = trstier[i + 1]
while (i + 1) < last and nextann.GetLabel().IsSilence():
end = nextann.GetLocation().GetEnd().GetMidpoint()
i += 1
if (i + 1) < last:
nextann = trstier[i + 1]
silences.append([start, end])
else:
start = ann.GetLocation().GetBegin().GetMidpoint()
end = ann.GetLocation().GetEnd().GetMidpoint()
trstracks.append([start, end])
self._units.append(ann.GetLabel().GetValue())
if nametier is not None:
aname = nametier.Find(ann.GetLocation().GetBegin().GetMidpoint(),
ann.GetLocation().GetEnd().GetMidpoint(), True)
if len(aname) == 0:
trstracks.pop()
self._units.pop()
else:
sf = sppasFileUtils(aname[0].GetLabel().GetValue())
# We have to take care in case of duplicated names
filename = sf.clear_whitespace()
if len(filename) == 0:
filename = "unnamed_track"
new_name = filename
idx = 2
while new_name in self._names:
new_name = u"%s_%.06d" % (filename, idx)
idx += 1
self._names.append(new_name)
# Continue
i += 1
return trstracks, silences
def fix_workingdir(inputaudio=""):
"""Fix the working directory to store temporarily the data.
:param inputaudio: (str) Audio file name
"""
sf = sppasFileUtils()
workdir = sf.set_random()
while os.path.exists(workdir) is True:
workdir = sf.set_random()
audio_file = os.path.basename(inputaudio)
sf = sppasFileUtils(audio_file)
formatted_audio_file = sf.format()
os.mkdir(workdir)
shutil.copy(inputaudio, os.path.join(workdir, formatted_audio_file))
return workdir
def extract_aligned(self, trstier, nametier):
""" Extract from a time-aligned transcription file.
:returns: a tuple with tracks and silences lists
"""
trstracks = []
silences = []
self._units = []
self._names = []
i = 0
last = trstier.GetSize()
while i < last:
# Set the current annotation values
ann = trstier[i]
# Save information
if ann.GetLabel().IsSilence():
start = ann.GetLocation().GetBegin().GetMidpoint()
end = ann.GetLocation().GetEnd().GetMidpoint()
# Verify next annotations (concatenate all silences between 2 tracks)
if (i + 1) < last:
nextann = trstier[i + 1]
while (i + 1) < last and nextann.GetLabel().IsSilence():
end = nextann.GetLocation().GetEnd().GetMidpoint()
i += 1
if (i + 1) < last:
nextann = trstier[i + 1]
silences.append([start, end])
else:
start = ann.GetLocation().GetBegin().GetMidpoint()
end = ann.GetLocation().GetEnd().GetMidpoint()
trstracks.append([start, end])
self._units.append(ann.GetLabel().GetValue())
if nametier is not None:
aname = nametier.Find(ann.GetLocation().GetBegin().GetMidpoint(),
ann.GetLocation().GetEnd().GetMidpoint(), True)
if len(aname) == 0:
trstracks.pop()
self._units.pop()
else:
sf = sppasFileUtils(aname[0].GetLabel().GetValue())
self._names.append(sf.clear_whitespace())
# Continue
i += 1
return trstracks, silences
def fix_audioinput(inputaudioname):
""" Fix the audio file name that will be used.
An only-ascii-based file name without whitespace is set if the
current audio file name does not fit in these requirements.
:param inputaudioname: (str) Audio file name
"""
sf = sppasFileUtils(inputaudioname)
inputaudio = sf.format()
if inputaudio != inputaudioname:
shutil.copy(inputaudioname, inputaudio)
return inputaudio
def fix_audioinput(inputaudioname):
""" Fix the audio file name that will be used.
An only-ascii-based file name without whitespace is set if the
current audio file name does not fit in these requirements.
:param inputaudioname: (str) Audio file name
"""
sf = sppasFileUtils(inputaudioname)
inputaudio = sf.format()
if inputaudio != inputaudioname:
shutil.copy(inputaudioname, inputaudio)
return inputaudio
def read(self, filename):
""" Read an ANT file and fill the Transcription.
:param filename: (str)
"""
zf = zipfile.ZipFile(filename, 'r')
unzip_dir = sppasFileUtils().set_random()
zf.extractall(unzip_dir)
zf.close()
antx_filename = os.path.join(unzip_dir, "annotation.xml")
antx = sppasANTX()
antx.read(antx_filename)
self.set(antx)
def _get_filename(self, filename, extensions):
""" Return a filename corresponding to one of extensions.
:param filename: input file name
:param extensions: the list of expected extension
:returns: a file name of the first existing file with an expected extension or None
"""
for ext in extensions:
ext_filename = os.path.splitext(filename)[0] + ext
new_filename = sppasFileUtils(ext_filename).exists()
if new_filename is not None and os.path.isfile(new_filename):
return new_filename
return None
def _get_filename(filename, extensions):
"""Return a filename corresponding to one of the extensions.
:param filename: input file name
:param extensions: the list of expected extension
:returns: a file name of the first existing file with an expected
extension or None
"""
base_name = os.path.splitext(filename)[0]
for ext in extensions:
ext_filename = base_name + ext
new_filename = sppasFileUtils(ext_filename).exists()
if new_filename is not None and os.path.isfile(new_filename):
return new_filename
return None
def fix_workingdir(inputaudio=""):
""" Fix the working directory to store temporarily the data.
:param inputaudio: (str) Audio file name
"""
if len(inputaudio) == 0:
# Notice that the following generates a directory that the
# aligners won't be able to access under Windows.
# No problem with MacOS or Linux.
sf = sppasFileUtils()
workdir = sf.set_random()
while os.path.exists(workdir) is True:
workdir = sf.set_random()
else:
workdir = os.path.splitext(inputaudio)[0]+"-temp"
os.mkdir(workdir)
return workdir
def fix_workingdir(inputaudio=""):
""" Fix the working directory to store temporarily the data.
:param inputaudio: (str) Audio file name
"""
if len(inputaudio) == 0:
# Notice that the following generates a directory that the
# aligners won't be able to access under Windows.
# No problem with MacOS or Linux.
sf = sppasFileUtils()
workdir = sf.set_random()
while os.path.exists(workdir) is True:
workdir = sf.set_random()
else:
workdir = os.path.splitext(inputaudio)[0] + "-temp"
i = 1
while os.path.exists(workdir) is True:
workdir = os.path.splitext(inputaudio)[0] + "-temp" + str(i)
i = i + 1
os.mkdir(workdir)
return workdir
def run_ipusegmentation(self, stepidx):
"""
Execute the SPPAS-IPUSegmentation program.
@return number of files processed successfully
"""
# Initializations
step = self.parameters.get_step(stepidx)
stepname = self.parameters.get_step_name(stepidx)
files_processed_success = 0
self._progress.set_header(stepname)
self._progress.update(0,"")
# Get the list of input file names, with the ".wav" extension
filelist = self.set_filelist(".wav")
if len(filelist) == 0:
return 0
total = len(filelist)
# Create annotation instance, and fix options
try:
seg = sppasIPUseg(self._logfile)
except Exception as e:
if self._logfile is not None:
self._logfile.print_message("%s\n"%str(e), indent=1,status=4)
return 0
# Execute the annotation for each file in the list
for i,f in enumerate(filelist):
# fix the default values
seg.reset()
seg.fix_options(step.get_options())
# Indicate the file to be processed
if self._logfile is not None:
self._logfile.print_message(stepname+" of file "+f, indent=1)
self._progress.set_text(os.path.basename(f)+" ("+str(i+1)+"/"+str(total)+")")
# Fix input/output file name
outname = os.path.splitext(f)[0] + self.parameters.get_output_format()
# Is there already an existing IPU-seg (in any format)!
ext = []
for e in sppas.src.annotationdata.aio.extensions_in:
if not e in ['.txt','.hz', '.PitchTier']:
ext.append(e)
existoutname = self._get_filename(f, ext)
# it's existing... but not in the expected format: convert!
if existoutname is not None and existoutname != outname:
# just copy the file!
if self._logfile is not None:
self._logfile.print_message('Export '+existoutname, indent=2)
self._logfile.print_message('into '+outname, indent=2)
try:
t = sppas.src.annotationdata.aio.read(existoutname)
sppas.src.annotationdata.aio.write(outname,t)
# OK, now outname is as expected! (or not...)
except Exception:
pass
# Execute annotation
tgfname = sppasFileUtils(outname).exists()
if tgfname is None:
# No already existing IPU seg., but perhaps a txt.
txtfile = self._get_filename(f, [".txt"])
if self._logfile is not None:
if txtfile:
self._logfile.print_message("A transcription was found, perform Silence/Speech segmentation time-aligned with a transcription %s"%txtfile, indent=2,status=3)
else:
self._logfile.print_message("No transcription was found, perform Silence/Speech segmentation only.", indent=2,status=3)
try:
seg.run(f, trsinputfile=txtfile, ntracks=None, diroutput=None, tracksext=None, trsoutput=outname)
files_processed_success += 1
if self._logfile is not None:
self._logfile.print_message(outname, indent=2,status=0)
except Exception as e:
if self._logfile is not None:
self._logfile.print_message("%s for file %s\n" % (str(e),outname), indent=2,status=-1)
else:
if seg.get_option('dirtracks') is True:
self._logfile.print_message("A time-aligned transcription was found, split into multiple files", indent=2,status=3)
try:
seg.run(f, trsinputfile=tgfname, ntracks=None, diroutput=None, tracksext=None, trsoutput=None)
files_processed_success += 1
if self._logfile is not None:
self._logfile.print_message(tgfname, indent=2,status=0)
except Exception as e:
if self._logfile is not None:
self._logfile.print_message("%s for file %s\n"%(str(e),tgfname), indent=2,status=-1)
else:
if self._logfile is not None:
self._logfile.print_message("because a previous segmentation is existing.", indent=2,status=2)
# Indicate progress
self._progress.set_fraction(float((i+1))/float(total))
if self._logfile is not None:
self._logfile.print_newline()
# Indicate completed!
self._progress.update(1,"Completed (%d files successfully over %d files).\n"%(files_processed_success,total))
self._progress.set_header("")
return files_processed_success
# -*- coding: utf8 -*-
import unittest
import os
import shutil
from ..label.label import Label
from ..ptime.point import TimePoint
from ..ptime.interval import TimeInterval
from ..annotation import Annotation
from ..aio.praat import TextGrid
from sppas.src.utils.fileutils import sppasFileUtils
# ---------------------------------------------------------------------------
TEMP = sppasFileUtils().set_random()
DATA = os.path.join(os.path.dirname(os.path.abspath(__file__)), "data")
# ---------------------------------------------------------------------------
class TestTextGrid(unittest.TestCase):
"""
Test reader/writers of TextGrid files from Praat.
"""
def setUp(self):
if os.path.exists(TEMP) is False:
os.mkdir(TEMP)
def tearDown(self):
src.resources.tests.test_vocab.py
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"""
import unittest
import os.path
import shutil
from sppas import RESOURCES_PATH
from sppas.src.utils.fileutils import sppasFileUtils
from sppas.src.utils.makeunicode import u
from ..vocab import sppasVocabulary
# ---------------------------------------------------------------------------
TEMP = sppasFileUtils().set_random()
VOCAB = os.path.join(os.path.dirname(os.path.abspath(__file__)), "data",
"vocab.txt")
ITA = os.path.join(RESOURCES_PATH, "vocab", "ita.vocab")
VOCAB_TEST = os.path.join(TEMP, "vocab.txt")
# ---------------------------------------------------------------------------
class TestVocabulary(unittest.TestCase):
def setUp(self):
if os.path.exists(TEMP) is False:
os.mkdir(TEMP)
def tearDown(self):
