o
4ѹgL @ s d Z ddlZddlZddlZddlZddlZddlZddlm Z ddl
m Z ddl m
Z
ddlmZmZmZmZmZmZmZmZ ddlmZ ddlZddlZdd lm Z m Z m Z m Z dd
l m!Z! dd l"m#Z# dd l$m%Z%m&Z&m'Z'm(Z( dd
lm)Z)m*Z* de+de,deee+e+f e+f fddZ-dee.e+f de.fddZ/de.fddZ0dee.e+f de,dee+ddf fddZ1d e!d ee.e+f de.fd d Z2de.fd d!Z3d"e.d#e.d$e.de.fd%d&Z4d'e!de,fd(d)Z5G d*d+ d+Z6dS ),zCommunicate with the service. Only the Communicate class should be used by
end-users. The other classes and functions are for internal use only. N) nullcontext)
TextIOWrapper)Queue)AsyncGeneratorContextManagerDict GeneratorListOptionalTupleUnion)escape )
DEFAULT_VOICESEC_MS_GEC_VERSION WSS_HEADERSWSS_URL) TTSConfig)DRM)NoAudioReceivedUnexpectedResponseUnknownResponseWebSocketError)CommunicateStateTTSChunkdata
header_lengthreturnc C sZ t | ts tdi }| d| dD ]}|dd\}}|||< q|| |d d fS )z
Returns the headers and data from the given data.
Args:
data (bytes): The data to be parsed.
header_length (int): The length of the header.
Returns:
tuple: The headers and data to be used in the request.
zdata must be bytesNs
:r )
isinstancebytes TypeErrorsplit)r r headerslinekeyvalue r( X/Users/admin/.pyenv/versions/3.10.0/lib/python3.10/site-packages/edge_tts/communicate.pyget_headers_and_data( s
r* stringc C s t | tr
| d} t | tstdt| }t|D ].\}}t|}d| kr-dksEn d| kr8dksEn d| krCdkrIn qd ||< qd
|S ) aS
The service does not support a couple character ranges.
Most important being the vertical tab character which is
commonly present in OCR-ed PDFs. Not doing this will
result in an error from the service.
Args:
string (str or bytes): The string to be cleaned.
Returns:
str: The cleaned string.
utf-8zstring must be str or bytesr N) r r! decodestrr" list enumerateordjoin)r+ charsidxcharcoder( r( r) remove_incompatible_characters@ s
D
r> c C s t t ddS )zZ
Returns a UUID without dashes.
Returns:
str: A UUID without dashes.
-r3 N)r5 uuiduuid4replacer( r( r( r)
connect_id\ s rC text byte_lengthc c s t | tr | d} t | tstd|dkr tdt| |kr| dd|}|dkr/|n|}d| d| v rc| dd|}| d ||dkrJqc|d
}|dk rVtd |dkr[qcd| d| v s9| d|
}|rp|V |dkrvd
}| |d } t| |ks"|
}|r|V dS dS ) a
Splits a string into a list of strings of a given byte length
while attempting to keep words together. This function assumes
text will be inside of an XML tag.
Args:
text (str or bytes): The string to be split. If bytes, it must be UTF-8 encoded.
byte_length (int): The maximum byte length of each string in the list.
Yield:
bytes: The next string in the list.
r, ztext must be str or bytesr z"byte_length must be greater than 0 &N ;r z0Maximum byte length is too small or invalid text) r r5 encoder! r"
ValueErrorlenrfindrindexfindstrip)rD rE Zsplit_atZampersand_indexZnew_textr( r( r) split_text_by_byte_lengthf s>
rQ tc escaped_textc C s@ t |tr
|d}d| j d| j d| j d| j d| d S ) z
Creates a SSML string from the given parameters.
Args:
tc (TTSConfig): The TTS configuration.
escaped_text (str or bytes): The escaped text. If bytes, it must be UTF-8 encoded.
Returns:
str: The SSML string.
r, z_zN)r r! r4 voicepitchratevolume)rR rS r( r( r) mkssml s
rX c C s t dt S )zg
Return Javascript-style date string.
Returns:
str: Javascript-style date string.
z:%a %b %d %Y %H:%M:%S GMT+0000 (Coordinated Universal Time)N)timestrftimegmtimer( r( r( r) date_to_string s r\
request_id timestampssmlc C s d| d| d| S )z
Returns the headers and data to be used in the request.
Returns:
str: The headers and data to be used in the request.
z X-RequestId:z1
Content-Type:application/ssml+xml
X-Timestamp:zZ
Path:ssml
Nr( )r] r^ r_ r( r( r) ssml_headers_plus_data s r`
tts_configc C s* d}t tt t t| dd }|| S )zCalculates the maximum message size for the given voice, rate, and volume.
Returns:
int: The maximum message size.
i r3 2 N)rL r` rC r\ rX )ra Zwebsocket_max_sizeZoverhead_per_messager( r( r) calc_max_mesg_size s
rc c @ s e Zd ZdZefddddddddded ed
ed ed ed
eej dee dee dee fddZ
de de fddZ
dee df fddZdee df fddZ d#deee f d eeee f ddfd d Zdee ddf fd d Z d#deee f d eeee f ddfd!d"ZdS )$ Communicatez'
Communicate with the service.
z+0%z+0HzN
< )rV rW rU connectorproxyconnect_timeoutreceive_timeoutrD rT rV rW rU rg rh ri rj c
C s t ||||| _t|tstdttt|t| j| _ |d ur+t|ts+td|| _
t|t s7tdt| t s@tdt j
d d || d| _|d urXt|t jsXtd|| _dddd d
| _d S ) Nztext must be strzproxy must be strzconnect_timeout must be intzreceive_timeout must be int)totalconnect sock_connectZ sock_readz'connector must be aiohttp.BaseConnector r F) partial_textoffset_compensationlast_duration_offsetstream_was_called)r ra r r5 r" rQ r
r> rc textsrh intaiohttpZ
ClientTimeoutsession_timeout
BaseConnectorrg state)
selfrD rT rV rW rU rg rh ri rj r( r( r) __init__ s8
zCommunicate.__init__r r c C s t |d D ]5}|d }|dkr1|d d | jd }|d d }||||d d d d
S |d v r6qtd | td
)NZMetadataType WordBoundaryZDataZOffsetrp ZDurationrD Text)typeoffsetdurationrD )Z
SessionEndzUnknown metadata type: z No WordBoundary metadata found)jsonloadsrx r r )ry r Zmeta_objZ meta_typecurrent_offsetZcurrent_durationr( r( r) Z__parse_metadata( s
z Communicate.__parse_metadatac
s d+fdd }d+ fdd }d}t jt d}tj jd jd 4 I d H 8}|jt d
t
d t d t
d
jt|d4 I d H | I d H | I d H 2 z3 d H W }|jtjjkr|jd}t||d\}} |dd }
|
dkr | } | V | d | d jd< qW|
dkr jd jd< jd d7 < q8|
dvrtdqW|jtjjkr"t |jdk rt d t |jd d d } | t |jkrt d t|j| \}} |dd krt d |d!d }
|
d"vrt d#|
d u rt | d$kr
qWt d%t | d$krt d&d}d'| d(V qW|jtjj kr6t!|jr3|jd)qW6 |s?t"d*W d I d H qV1 I d H sQw Y W d I d H d S 1 I d H shw Y d S ),Nr c s dt dI dH dS )z)Sends the command request to the service.z X-Timestamp:z
Content-Type:application/json; charset=utf-8
Path:speech.config
{"context":{"synthesis":{"audio":{"metadataoptions":{"sentenceBoundaryEnabled":"false","wordBoundaryEnabled":"true"},"outputFormat":"audio-24khz-48kbitrate-mono-mp3"}}}}
N)send_strr\ r( ) websocketr( r) send_command_request< s z2Communicate.__stream..send_command_requestc s0 tt t t j jd I dH dS )z&Sends the SSML request to the service.ro N)r r` rC r\ rX ra rx r( ry r r( r) send_ssml_requestH s z/Communicate.__stream..send_ssml_requestF)cafileT)rg trust_envtimeoutz &Sec-MS-GEC=z&Sec-MS-GEC-Version=z&ConnectionId= )compressrh r$ sslr, s
s Paths audio.metadatar r rq s turn.endrp i )s responses
turn.startzUnknown path receivedr zBWe received a binary message, but it is missing the header length.bigz9The header length is greater than the length of the data.s audioz3Received binary message, but the path is not audio.s Content-Type)s
audio/mpegNz=Received binary message, but with an unexpected Content-Type.r z.fetch_async_items..get_itemsr )asyncionew_event_loopset_event_looprun_until_completeclose)r r loopry )r r) fetch_async_items s
z2Communicate.stream_sync..fetch_async_itemsT)r
concurrentfuturesThreadPoolExecutorsubmitr )ry r r executorr r( r r) stream_sync s "zCommunicate.stream_syncc C sN t j }|tj| ||}| W d dS 1 s w Y dS )z,Synchronous interface for async save method.N)r r r r r runr result)ry r r r futurer( r( r) save_sync s
"zCommunicate.save_syncr )__name__
__module__ __qualname____doc__r r5 r
ru rw rt rz r! r r r r r r r r r r r( r( r( r) rd sj
8
"
rd )7r r concurrent.futuresr r r rY r@
contextlibr ior r r typingr r r r r r
r r xml.sax.saxutilsr
ru r constantsr r r r Z data_classesr Zdrmr
exceptionsr r r r r r r! rt r* r5 r> rC rQ rX r\ r` rc rd r( r( r( r) sP (
7