Xi~[ 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;y))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 s t d i }| d| j d D ] }|j 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 ^/home/obispo/Crisostomo_bridge/mision_env/lib/python3.12/site-packages/edge_tts/communicate.pyget_headers_and_datar- ) sw dE ",--G^m$**73 ZZa(
U D *,- -- stringc J t | t r| j d } t | t s t d t | }t
| D ]> \ }}t | }d|cxk rdk s n d|cxk rdk sn d|cxk rdk s7n :d ||< @ d
j | S ) aS
The service does not support a couple character ranges.
Most important being the vertical tab character which is
commonly present in OCR-ed PDFs. Not doing this will
result in an error from the service.
Args:
string (str or bytes): The string to be cleaned.
Returns:
str: The cleaned string.
utf-8zstring must be str or bytesr ) r$ r% decodestrr& list enumerateordjoin)r/ charsidxcharcodes r, remove_incompatible_charactersrC A s &% w' fc "566F|E u% TI
NNd 0b 0bD6FB6F E#J
775>r. c > t j j S )zZ
Returns a UUID without dashes.
Returns:
str: A UUID without dashes.
)uuiduuid4hex r. r,
connect_idrI ] s ::< r. textlimitc \ | j dd| }|dk r| j 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 s4 " zz%E*H!|::dAu- Or. text_segmentc t | }|dkD r | d| j d |S |S # t $ r |dz }Y nw xY w|dkD r3 )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 3O
O " MH Q,s * ;;rP c |dkD rBd| d| v r;| j dd| }| j d|| dk7 r |S |}|dkD rd| d| v r;|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^ sm " Q,44 ?2 ++dAx8 99T?H 5 ;
O # Q,44 ?2 Or. byte_lengthc # K t | t r| j d } t | t s t d |dk r t d t
| |kD rlt | | }|dk r t | }t | | }|dk r t d | d| j }|r| | |dkD r|ndd } t
| |kD rl| j }|r| yyw)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 "344a=>>
d)k
!;D+N a<248H 6dHE a< C
Yh%%' K 1 H!565 d)k
!: jjlO s CC! C!tc escaped_textc t |t r|j 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 sd ,&#**73 z " 88*HRWWIZ {" . r. c R 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 ==Ddkkm 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 ' s) zl # k " & r. c B 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 Zy) 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 s t d t t
t | d | _ |t |t s t d || _ t | t s t d t |
t s t d t j d d | |
| _
|%t |t j s t d || _ dd d d
d | _ y ) 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__> s $E4I $$ ./ / / 1$7 8
Zs%; /0 0$)
/3/ 9: :/3/ 9: : & 4 4(%
Iw?T?T)U EF F:C #$$%!& (
r. r r c t j | d D ]X }|d }|dv r;|d d | j d z }|d d }|||t |d d d d
c S |d v rLt 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 IH (I @@ V$X. jA |j( d d d } | t; |j( kD r t= d t- |j( | \ }} |j1 d d k7 r t= d |j1 d d }
|
d vr t= d |
t; | d!k( rt= d" t; | d!k( r t= d# d}d$| d% &|j" t j$ jB k( sOtE |j( r|j( d& |s tG d' d d d d { d d d d { y 7 67 7 7 7 6 ?7 &# 1 d { 7 sw Y 6xY w7 -# 1 d { 7 sw Y y xY ww))Nc K j j dk( } | rdnd}| sdnd}j dt d| d| d d{ y7 w) 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 sn OO 4 4 FM(gB ,'B$$ ~ / 0 1. /1T1J2$ O
s A AAAc K j t t t t j
j d d{ y7 w)z&Sends the SSML request to the service.r N)r rx rI rs rn r r )r r s r, send_ssml_requestz/Communicate.__stream..send_ssml_request sM $$& L"$
>2
s AA AA 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*GL';$l&7&7 &D($J &>>'48D 00*.*?*?*E -- ,H 5
8S S
#9: ,<@JJ 2=
#89
#89YF9 %AA -.E FF ]]g&7&7&>&>>8==)A- 0 `
%(NN8==!3De$LM$s8=='99 0 W
(<
}($J
"~~g.(: 0 Q $.>>/4#HL#+@@ 0 [
$+ t9> $ 1 Z
4yA~ 0 X
*.&#*D99 ]]g&7&7&=&==()1
IIgx0 NN4(
) ) ) ) ) ) )sb C
CB1B&B$
B&.B14B1<(B1$B&&B1( C1B: 6C= CC Cc # K dt ddf fd }t }t j j 5 }|j || |j }|n| ddd y# 1 sw Y yxY ww)z-Synchronous interface for async stream methodqueuer Nc d fd }t j }t j | |j | |j y )Nc K j 2 3 d { } j | 7 6 j d y wN)r put)itemr r s r, get_itemszECommunicate.stream_sync..fetch_async_items..get_items\ s: "&++- $ $$IIdO$- $s A1/1A1Ar )asyncionew_event_loopset_event_looprun_until_completeclose)r r loopr s ` r, fetch_async_itemsz2Communicate.stream_sync..fetch_async_items[ s?
))+D " "4 ( # #IK 0 JJLr. )r
concurrentfuturesThreadPoolExecutorsubmitr )r r r executorr s ` r, stream_synczCommunicate.stream_syncX s U t w
2
2
4 OO -u 5 yy{<
s 7A<,A0' A<0A95A<c t j j 5 }|j t j
| j
|| }|j ddd y# 1 sw Y yxY w)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 sZ
2
2
4 __ TYY{N CF
MMO s AA))A2r )__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
G 1 127
}7
"# 7
"# 7
rCU Cx C&Ux~ > Un " $ ' "D 7;)3:&) !sEz!23)
)0Yxt'; < 8 7;
3:&
!sEz!23
r. rz )r s I
" . % N N # /.
. #.
4u u $%.05e+< 8 C 5 0e 8U c c