Sܶi~[ : d Z ddlZddlZddlZddlZddlZddlZddlm Z ddl
m Z ddl m
Z
ddlmZmZmZmZmZmZmZmZ ddlmZmZ ddlZddlZddl m Z d d
l m Z m Z m!Z!m"Z" d d l#m$Z$ d d l%m&Z& d d
l'm(Z(m)Z)m*Z*m+Z+ d dlm,Z,m-Z- de.de/deee.e.f e.f fdZ0dee1e.f de1fdZ2de1fdZ3de.de/de/fdZ4de.de/fdZ5de.de/de/fd Z6dee1e.f d e/dee.ddf fd Z7d e$d ee1e.f de1fd!Z8de1fd"Z9d#e1d$e1d%e1de1fd&Z: G d' d( Z;dS ))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)escapeunescape)Literal )
DEFAULT_VOICESEC_MS_GEC_VERSION WSS_HEADERSWSS_URL) TTSConfig)DRM)NoAudioReceivedUnexpectedResponseUnknownResponseWebSocketError)CommunicateStateTTSChunkdata
header_lengthreturnc t | t st d i }| d| d D ] }| dd \ }}|||< !|| |dz 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 headerslinekeyvalues W/root/projects/openclaw-proxy/venv/lib/python3.11/site-packages/edge_tts/communicate.pyget_headers_and_datar- ) s dE " " .,---G^m^$**733 ZZa((
U D *,,- -- stringc t | t r| d } t | t st d t | }t
| D ]G\ }}t | }d|cxk rdk s!n d|cxk rdk sn d|cxk rdk rn Bd ||< Hd
| S ) aS
The service does not support a couple character ranges.
Most important being the vertical tab character which is
commonly present in OCR-ed PDFs. Not doing this will
result in an error from the service.
Args:
string (str or bytes): The string to be cleaned.
Returns:
str: The cleaned string.
utf-8zstring must be str or bytesr ) r$ r% decodestrr& list enumerateordjoin)r/ charsidxcharcodes r, remove_incompatible_charactersrC A s &% ( w'' fc " " 75666F||E u%% TII
NNNNNNNNd 0 0 0 0b 0 0 0 0bD6F6F6F6FB6F6F6F6F6F E#J
775>>r. c 2 t j j S )zZ
Returns a UUID without dashes.
Returns:
str: A UUID without dashes.
)uuiduuid4hex r. r,
connect_idrI ] s :<< r. textlimitc n | dd| }|dk r| dd| }|S )a
Finds the index of the rightmost preferred split character (newline or space)
within the initial `limit` bytes of the text.
This helps find a natural word or sentence boundary for splitting, prioritizing
newlines over spaces.
Args:
text (bytes): The byte string to search within.
limit (int): The maximum index (exclusive) to search up to.
Returns:
int: The index of the last found newline or space within the limit,
or -1 if neither is found in that range.
r )rfind)rJ rK split_ats r, (_find_last_newline_or_space_within_limitrQ g s< " zz%E**H!||::dAu-- Or. text_segmentc t | }|dk r; | d| d |S # t $ r |dz }Y nw xY w|dk ;|S )a
Finds the rightmost possible byte index such that the
segment `text_segment[:index]` is a valid UTF-8 sequence.
This prevents splitting in the middle of a multi-byte UTF-8 character.
Args:
text_segment (bytes): The byte segment being considered for splitting.
Returns:
int: The index of the safe split point. Returns 0 if no valid split
point is found (e.g., if the first byte is part of a multi-byte
sequence longer than the limit allows).
r Nr1 r )lenr9 UnicodeDecodeError)rR rP s r, _find_safe_utf8_split_pointrV s{ < H
Q,, ( # * *7 3 3 3O! MHHH Q,, Os 6 AArP c |dk rSd| d| v rG| dd| }| d|| dk rn|}|dk r d| d| v G|S )a
Adjusts a proposed split point backward to prevent splitting inside an XML entity.
For example, if `text` is `b"this & that"` and `split_at` falls between
`&` and `;`, this function moves `split_at` to the index before `&`.
Args:
text (bytes): The text segment being considered.
split_at (int): The proposed split point index, determined by whitespace
or UTF-8 safety.
Returns:
int: The adjusted split point index. It will be moved to the '&'
if an unterminated entity is detected right before the original `split_at`.
Otherwise, the original `split_at` is returned.
r &N ;)rindexfind)rJ rP ampersand_indexs r, "_adjust_split_point_for_xml_entityr^ s} " Q,,44 ?22 ++dAx88 99T?H 5 5 ; ; # Q,,44 ?22 Or. byte_lengthc # P K t | t r| d } t | t st d |dk rt d t
| |k rt | | }|dk rt | }t | | }|dk rt d | d|
}|r|V | |dk r|ndd } t
| |k |
}|r|V dS dS )a
Splits text into chunks, each not exceeding a maximum byte length.
This function prioritizes splitting at natural boundaries (newlines, spaces)
while ensuring that:
1. No chunk exceeds `byte_length` bytes.
2. Chunks do not end with an incomplete UTF-8 multi-byte character.
3. Chunks do not split XML entities (like `&`) in the middle.
Args:
text (str or bytes): The input text. If str, it's encoded to UTF-8.
byte_length (int): The maximum allowed byte length for any yielded chunk.
Must be positive.
Yields:
bytes: Text chunks (UTF-8 encoded, stripped of leading/trailing whitespace)
that conform to the byte length and integrity constraints.
Raises:
TypeError: If `text` is not str or bytes.
ValueError: If `byte_length` is not positive, or if a split point
cannot be determined (e.g., due to extremely small byte_length
relative to character/entity sizes).
r1 ztext must be str or bytesr z"byte_length must be greater than 0zTMaximum byte length is too small or invalid text structure near '&' or invalid UTF-8Nr ) r$ r: encoder% r&
ValueErrorrT rQ rV r^ strip)rJ r_ rP chunkremaining_chunks r, split_text_by_byte_lengthrf s` 6 $ ${{7## dE " " 53444a=>>>
d))k
!
!;D+NN a<<2488H 6dHEE a<< C
YhY%%'' KKK 1 HH!5565 d))k
!
!: jjllO r. tc escaped_textc t |t r| d }d| j d| j d| j d| j d| d S )z
Creates a SSML string from the given parameters.
Args:
tc (TTSConfig): The TTS configuration.
escaped_text (str or bytes): The escaped text. If bytes, it must be UTF-8 encoded.
Returns:
str: The SSML string.
r1 z_z)r$ r% r9 voicepitchratevolume)rg rh s r, mkssmlrn s ,&& 4#**733 8 -/W @B r. c N t j dt j 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))timestrftimegmtimerH r. r, date_to_stringrs s# =Ddkmm r.
request_id timestampssmlc 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
rH )rt ru rv s r, ssml_headers_plus_datarx ' s4 z r. c e Zd ZdZefdddddddddd ed
ed ed ed
eded deej dee dee
dee
fdZ de de
fdZdee
df fdZdee
df fdZ d deee f deeee f ddfd Zdee
ddf fd Z d deee f deeee f ddfd ZdS ) Communicatez'
Communicate with the service.
z+0%z+0HzSentenceBoundaryN
< )rl rm rk boundary connectorproxyconnect_timeoutreceive_timeoutrJ rj rl rm rk r~ WordBoundaryr{ r r r r c | t ||||| | _ t |t st d t t
t | d | _ |$t |t st d || _ t | t st d t |
t st d t j d d | |
| _
|)t |t j st d || _ dd d d
d | _ d S ) Nztext must be stri zproxy must be strzconnect_timeout must be intzreceive_timeout must be int)totalconnect sock_connect sock_readz'connector must be aiohttp.BaseConnectorr. r F) partial_textoffset_compensationlast_duration_offsetstream_was_called)r
tts_configr$ r: r& rf r rC textsr intaiohttp
ClientTimeoutsession_timeout
BaseConnectorr state) selfrJ rj rl rm rk r~ r r r r s r, __init__zCommunicate.__init__> sO $E4II $$$ 0 .// / / 1$77 8 8
Zs%;%; /00 0$)
/3// ; 9:: :/3// ; 9:: : & 4(%
Iw?T)U)U EFF F:C #$$%!& (
(
r. r r c B t j | d D ]v}|d }|dv rR|d d | j d z }|d d }|||t |d d d d
c S |d v ret d | t d
)NMetadataTyper DataOffsetr DurationrJ Text)typeoffsetdurationrJ )
SessionEndzUnknown metadata type: z No WordBoundary metadata found)jsonloadsr r r r )r r meta_obj meta_typecurrent_offsetcurrent_durations r, __parse_metadataz Communicate.__parse_metadataw s
4((4 I IH (I @@@ V$X. |j d d d } | t; |j k rt= d t- |j | \ }} | d d k rt= d | d d }
|
d vrt= d! |
$t; | d"k r t= d# t; | d"k rt= d$ d}d%| d&W V F|j t j j! k r tE |j r|j nd' 6 |stG d( d d d d {V n# 1 d {V swxY w Y d d d d {V d S # 1 d {V swxY w Y d S )*Nr c K j j dk } | rdnd}| sdnd} dt d| d| d d{V dS ) z)Sends the command request to the service.r truefalsez X-Timestamp:z
Content-Type:application/json; charset=utf-8
Path:speech.config
{"context":{"synthesis":{"audio":{"metadataoptions":{"sentenceBoundaryEnabled":"z","wordBoundaryEnabled":"z9"},"outputFormat":"audio-24khz-48kbitrate-mono-mp3"}}}}
N)r r~ send_strrs )
word_boundarywdsqr websockets r, send_command_requestz2Communicate.__stream..send_command_request s O 4 FM(5gB ,9'B$$~ / / /1 LN
r. c K t t t t j j d d{V dS )z&Sends the SSML request to the service.r N)r rx rI rs rn r r )r r s r, send_ssml_requestz/Communicate.__stream..send_ssml_request s $$& LL"$$
>2
r. F)cafileT)r trust_envtimeoutz&ConnectionId=z &Sec-MS-GEC=z&Sec-MS-GEC-Version= )compressr r( sslr1 s
s Paths audio.metadatar r r s turn.endr 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*>w*G*GL';$l&7&7 &D&D( ($J &>>'488D 000*.*?*?*E*E ----- ,H 5
8S S
#9:: ,,<@J 2=
#89
#8999YF999 %AAA -.E F FF B ]g&7&>>>8=))A-- 0 `
%(NN8=!3De$L$LM$s8='9'999 0 W
(<
}( ($J
"~~g..(:: 0 Q $.>>/4#H#HL#+@@@ 0 [
$+ t99>> $ 1 Z
4yyA~~ 0 X
*.&#*D9999999 ]g&7&===()1 K
O >o #,x &
%\ ]p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p sD #A/N>#N6M5
N$ $N>'N$ (N>>
O Oc | K | j d rt d d| j d< | j D ]| j d< | 2 3 d{V }|W V
6 .# t j $ rL}|j dk r t j | | 2 3 d{V }|W V
6 Y d}~d}~ww xY wdS )au
Streams audio and metadata from the service.
Raises:
NoAudioReceived: If no audio is received from the service.
UnexpectedResponse: If the response from the service is unexpected.
UnknownResponse: If the response from the service is unknown.
WebSocketError: If there is an error with the websocket.
r z stream can only be called once.Tr Ni ) r RuntimeErrorr _Communicate__streamr ClientResponseErrorstatusr handle_client_response_error)r messagees r, streamzCommunicate.stream! s/ :) * B @AA A*.
&' +/*
"
"DJ~ &
"%)]]__ " " " " " " "'!MMMMM &5_.
"
"
"8s??0333%)]]__ " " " " " " "'!MMMMM &5_____
"
"
"s/ A A A B9-4B4!B.'B44B9 audio_fnamemetadata_fnamec K |t |dd n
t }|5 t |d 5 }| 2 3 d{V }|d dk r | |d /t |t
r4|d d v r*t
j || | d
y6 ddd n # 1 swxY w Y ddd dS # 1 swxY w Y dS ) zE
Save the audio and metadata to the specified files.
Nwr1 )encodingwbr r r r
)openr r writer$ r r dump)r r r metadatar r s r, savezCommunicate.save@ s )
w 7 7 7 7
) )tK.. )%!%
)
)
)
)
)
)
)g6?g--KK0000 -88 )WV_ Q > > Igx000 NN4((( "/ ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) )sA C,C
CA4C C,C C,C C,,C03C0c # K dt ddf fd }t }t j 5 }| || | }|n|V ddd dS # 1 swxY w Y dS )z-Synchronous interface for async stream methodqueuer Nc d fd }t j }t j | | | | d S )Nr c K 2 3 d {V } | 6 d d S N)r put)itemr r s r, get_itemszECommunicate.stream_sync..fetch_async_items..get_items\ s` "&++-- $ $ $ $ $ $ $$IIdOOOO #0 $s 5r )asyncionew_event_loopset_event_looprun_until_completeclose)r r loopr s ` r, fetch_async_itemsz2Communicate.stream_sync..fetch_async_items[ ss
)++D "4 ( ( ( # #IIKK 0 0 0 JJLLLLLr. )r
concurrentfuturesThreadPoolExecutorsubmitr )r r r executorr s ` r, stream_synczCommunicate.stream_syncX s U t ww
2
2
4
4 OO -u 5 5 5
yy{{<
s 4BBBc t j 5 }| t j | || }| ddd dS # 1 swxY w Y dS )z,Synchronous interface for async save method.N)r r r r r runr result)r r r r futures r, save_synczCommunicate.save_syncq s
2
2
4
4 __ TYY{N C C F
MMOOO s A A55A9<A9r )__name__
__module__ __qualname____doc__r r: r r r r r r r% r r r r r r
r r r r rH r. r, rz rz 8 s #7
@R59 #)+)+7
7
7
7
7
7
7
7
<=7
G 127
}7
"# 7
"# 7
7
7
7
rCU Cx C C C C&Ux~ > U U U Un " $ ' " " " "D 7;) )3:&) !sEz!23)
) ) ) )0Yxt'; < 8 7;
3:&
!sEz!23
r. rz )r s I I
" " " " " " . - - - - - - - % % % % % % N N N N N N N N N N N N # # # # # # / . . . . . . ..
. #.
4u u $%. . . .05e+< 8 C 5 0e 8U c c