def _tokenize(self, text):
# Pre-clean
text = text.strip()
# Apply pre-processors
for pp in self.pre_processor_funcs:
log.debug("pre-processing: %s", pp)
text = pp(text)
if _len(text) <= self.GOOGLE_TTS_MAX_CHARS:
return _clean_tokens([text])
# Tokenize
log.debug("tokenizing: %s", self.tokenizer_func)
tokens = self.tokenizer_func(text)
# Clean
tokens = _clean_tokens(tokens)
# Minimize
min_tokens = []
for t in tokens:
min_tokens += _minimize(t, ' ', self.GOOGLE_TTS_MAX_CHARS)
# Filter empty tokens, post-minimize
tokens = [t for t in min_tokens if t]
return min_tokens
def url(self):
"""
generate url to call google tts
"""
try:
# Calculate token
part_tk = self.token.calculate_token(self.text)
except requests.exceptions.RequestException as e: # pragma: no cover
log.debug(str(e), exc_info=True)
raise gTTSError("Connection error during token calculation: %s" %
str(e))
payload = {
'ie': 'UTF-8',
'q': self.text,
'tl': self.lang,
'ttsspeed': self.speed,
'total': 1,
'idx': 0,
'client': 'tw-ob',
'textlen': _len(self.text),
'tk': part_tk
}
return self.GOOGLE_TTS_URL + "?" + urllib.parse.urlencode(payload)
def _prepare_requests(self):
"""Created the TTS API the request(s) without sending them.
Returns:
list: ``requests.PreparedRequests_``. <https://2.python-requests.org/en/master/api/#requests.PreparedRequest>`_``.
"""
# TTS API URL
translate_url = _translate_url(tld=self.tld, path="translate_tts")
text_parts = self._tokenize(self.text)
log.debug("text_parts: %i", len(text_parts))
assert text_parts, 'No text to send to TTS API'
prepared_requests = []
for idx, part in enumerate(text_parts):
try:
# Calculate token
part_tk = self.token.calculate_token(part)
except requests.exceptions.RequestException as e: # pragma: no cover
log.debug(str(e), exc_info=True)
raise gTTSError(
"Connection error during token calculation: %s" % str(e))
payload = {
'ie': 'UTF-8',
'q': part,
'tl': self.lang,
'ttsspeed': self.speed,
'total': len(text_parts),
'idx': idx,
'client': 'tw-ob',
'textlen': _len(part),
'tk': part_tk
}
log.debug("payload-%i: %s", idx, payload)
# Request
r = requests.Request(method='GET',
url=translate_url,
params=payload,
headers=self.GOOGLE_TTS_HEADERS)
# Prepare request
prepared_requests.append(r.prepare())
return prepared_requests
def stream_to_fps(self, fps):
if not isinstance(fps, list):
fps = [fps]
urllib3.disable_warnings(urllib3.exceptions.InsecureRequestWarning)
text_parts = self._tokenize(self.text)
assert text_parts, 'No text to send to TTS API'
for idx, part in enumerate(text_parts):
try:
# Calculate token
part_tk = self.token.calculate_token(part)
except requests.exceptions.RequestException as e: # pragma: no cover
raise gtts.gTTSError(
"Connection error during token calculation: %s" %
str(e))
payload = {'ie': 'UTF-8',
'q': part,
'tl': self.lang,
'ttsspeed': self.speed,
'total': len(text_parts),
'idx': idx,
'client': 'tw-ob',
'textlen': _len(part),
'tk': part_tk}
r = None
try:
# Request
r = requests.get(self.GOOGLE_TTS_URL,
params=payload,
headers=self.GOOGLE_TTS_HEADERS,
proxies=urllib.request.getproxies(),
verify=False,
stream=True)
r.raise_for_status()
except requests.exceptions.HTTPError:
# Request successful, bad response
raise gtts.gTTSError(tts=self, response=r)
except requests.exceptions.RequestException as e: # pragma: no cover
# Request failed
raise gtts.gTTSError(str(e))
for chunk in r.iter_content(chunk_size=1024):
for f in fps:
f.write(chunk)
def write_to_fp(self, fp):
"""Do the TTS API request and write bytes to a file-like object.
Args:
fp (file object): Any file-like object to write the ``mp3`` to.
Raises:
:class:`gTTSError`: When there's an error with the API request.
TypeError: When ``fp`` is a file-like object that takes bytes.
"""
# When disabling ssl verify in requests (for proxies and firewalls),
# urllib3 prints an insecure warning on stdout. We disable that.
urllib3.disable_warnings(urllib3.exceptions.InsecureRequestWarning)
text_parts = self._tokenize(self.text)
log.debug("text_parts: %i", len(text_parts))
assert text_parts, 'No text to send to TTS API'
for idx, part in enumerate(text_parts):
try:
# Calculate token
part_tk = self.token.calculate_token(part)
except requests.exceptions.RequestException as e: # pragma: no cover
log.debug(str(e), exc_info=True)
raise gTTSError(
"Connection error during token calculation: %s" % str(e))
payload = {
'ie': 'UTF-8',
'q': part,
'tl': self.lang,
'ttsspeed': self.speed,
'total': len(text_parts),
'idx': idx,
'client': 'tw-ob',
'textlen': _len(part),
'tk': part_tk
}
log.debug("payload-%i: %s", idx, payload)
try:
# Request
r = requests.get(self.GOOGLE_TTS_URL,
params=payload,
headers=self.GOOGLE_TTS_HEADERS,
proxies=urllib.request.getproxies(),
verify=False)
log.debug("headers-%i: %s", idx, r.request.headers)
log.debug("url-%i: %s", idx, r.request.url)
log.debug("status-%i: %s", idx, r.status_code)
r.raise_for_status()
except requests.exceptions.HTTPError as e:
# Request successful, bad response
raise gTTSError(tts=self, response=r)
except requests.exceptions.RequestException as e: # pragma: no cover
# Request failed
raise gTTSError(str(e))
try:
# Write
for chunk in r.iter_content(chunk_size=1024):
fp.write(chunk)
log.debug("part-%i written to %s", idx, fp)
except (AttributeError, TypeError) as e:
raise TypeError(
"'fp' is not a file-like object or it does not take bytes: %s"
% str(e))
def test_len_unicode():
text = u"但在一个重要的任务上"
assert _len(text) == 10
def test_len_ascii():
text = "Bacon ipsum dolor sit amet flank corned beef."
assert _len(text) == 45
def test_unicode(self):
text = u"但在一个重要的任务上"
self.assertEqual(_len(text), 10)
def test_ascii(self):
text = "Bacon ipsum dolor sit amet flank corned beef."
self.assertEqual(_len(text), 45)
