def run(self, filepath):
self.log(logging.INFO, "Starting %s" % (filepath))
out_file = self.derive_new_file_path(filepath, ".csv")
if file_utils.should_run(filepath, out_file):
t = transcript.PlaintextTranscript(
filepath=filepath,
label=None,
pos_tagger_path=self.pos_tagger_path)
transcript_utterances_fillers = None
if self.filler_dir:
file_id = os.path.basename(file_utils.strip_ext(filepath))
if file_id in self.filler_files:
filler_file = os.path.join(self.filler_dir,
self.filler_files[file_id])
filler_transcript = transcript.PlaintextTranscript(
filepath=filler_file,
label=None,
pos_tagger_path=self.pos_tagger_path)
transcript_utterances_fillers = filler_transcript.tokens
self.feature_extractor.extract(
t,
out_csv=out_file,
transcript_utterances_fillers=transcript_utterances_fillers)
self.log(logging.INFO, "Done %s -> %s" % (filepath, out_file))
self.emit(out_file)
def run(self, in_file):
self.log(logging.INFO, "Starting %s" % (in_file))
out_file = self.derive_new_file_path(in_file, 'txt')
if file_utils.should_run(in_file, out_file):
cmd = self.cmd.replace("**in_placeholder**",
os.path.abspath(in_file))
process = subprocess.Popen(cmd,
shell=True,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
out, err = process.communicate()
res = process.returncode
if res == 0:
hyp = self.regex.search(str(err).replace("\\n", "\n")).group(1)
with open(out_file, "w") as f:
f.write(hyp)
self.log(logging.INFO, "Done %s -> %s" % (in_file, out_file))
else:
error_lines = self.error_regex.findall(
str(err).replace("\\n", "\n"))
self.log(
logging.ERROR,
"Failed %s -> %s with error code %i. cmd: %s. Error message: %s"
% (in_file, out_file, res, cmd, "\n\t".join(error_lines)))
return
self.emit([out_file])
def run(self):
if file_utils.should_run(self.filepath, self.out_file):
self.features['FileID'] = self.filepath
with open(self.filepath) as f:
self.tokens = f.read()
self.compute_basic_word_stats()
self.compute_word_frequency_norms()
self._run_chinese_corenlp(self.filepath)
self._parse_corenlp_output()
self.write_features(self.out_file, debug=False)
def run(self, in_file):
self.log(logging.INFO, "Starting %s" % (in_file))
out_file = self.derive_new_file_path(in_file, self.out_ext)
if file_utils.should_run(in_file, out_file):
try:
self.matlab_function(in_file, out_file)
except Exception as e:
self.log(logging.ERROR, e)
self.log(logging.INFO, "Done %s -> %s" % (in_file, out_file))
self.emit(out_file)
def run(self, in_file):
self.log(logging.INFO, "Starting %s" % (in_file))
out_file = self.derive_new_file_path(in_file, self.ext)
if file_utils.should_run(in_file, out_file):
cmd = self.command.format(in_file=in_file, out_file=out_file)
res = shell_run(cmd.split(" "))
if res != 0:
self.log(
logging.ERROR,
"Failed %s -> %s with error code %i. cmd: %s" %
(in_file, out_file, res, cmd))
return
self.log(logging.INFO, "Done %s -> %s" % (in_file, out_file))
self.emit(out_file)
def run(self, mp3_file):
self.log(logging.INFO, "Starting %s" % (mp3_file))
if not mp3_file.endswith(".mp3"):
self.log(logging.ERROR, "Failed %s. Not mp3 file" % (mp3_file))
return
wav_file = self.derive_new_file_path(mp3_file, "wav")
if file_utils.should_run(mp3_file, wav_file):
res = shell_run(["lame", "--decode", mp3_file, wav_file])
if res != 0:
self.log(
logging.ERROR, "Failed %s -> %s with lame error code %i" %
(mp3_file, wav_file, res))
return
self.log(logging.INFO, "Done %s -> %s" % (mp3_file, wav_file))
self.emit(wav_file)
def run(self, in_file):
self.log(logging.INFO, "Starting %s" % (in_file))
out_file = self.derive_new_file_path(in_file, self.out_ext)
if file_utils.should_run(in_file, out_file):
cmd = [
self.opensmile_exec, "-C", self.conf_file, "-I", in_file,
self.out_flag, out_file
] + self.extra_flags
res = shell_run(cmd)
if res != 0:
self.log(
logging.ERROR,
"Failed %s -> %s with SmileExtract error code %i. cmd: %s"
% (in_file, out_file, res, " ".join(cmd)))
return
self.log(logging.INFO, "Done %s -> %s" % (in_file, out_file))
self.emit([out_file])
def run(self, in_file):
self.log(logging.INFO, "Starting %s" % (in_file))
out_file = self.derive_new_file_path(in_file, 'csv')
if file_utils.should_run(in_file, out_file):
cmd = [
'praat', '--run', 'scripts/syllable_nuclei_v2.praat',
os.path.abspath(in_file)
]
res = shell_run(cmd, stdout=out_file)
if res != 0:
self.log(
logging.ERROR,
"Failed %s -> %s with error code %i. cmd: %s" %
(in_file, out_file, res, " ".join(cmd)))
return
self.log(logging.INFO, "Done %s -> %s" % (in_file, out_file))
self.emit([out_file])
def run(self, wav_file):
self.log(logging.INFO, "Starting %s" % (wav_file))
if not wav_file.endswith(".wav"):
self.log(logging.ERROR, "Failed %s. Not wav file" % (wav_file))
return
new_wav_file = self.derive_new_file_path(wav_file, "wav")
if file_utils.should_run(wav_file, new_wav_file):
res = shell_run(
["sox", wav_file, "--rate",
str(self.new_sr), new_wav_file])
if res != 0:
self.log(
logging.ERROR, "Failed %s -> %s with lame error code %i" %
(wav_file, new_wav_file, res))
return
self.log(logging.INFO, "Done %s -> %s" % (wav_file, new_wav_file))
self.emit(new_wav_file)
