mirror of
https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs.git
synced 2025-01-02 23:38:45 +00:00
net/aws: use aws-sdk-transcribestreaming
Switch from manual webservice client impl to `aws-sdk-transcribestreaming`. Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs/-/merge_requests/1104>
This commit is contained in:
parent
57f365979c
commit
00153754bb
5 changed files with 473 additions and 807 deletions
|
@ -619,7 +619,7 @@
|
|||
"rank": "none"
|
||||
},
|
||||
"awstranscriber": {
|
||||
"author": "Jordan Petridis <jordan@centricular.com>, Mathieu Duponchelle <mathieu@centricular.com>",
|
||||
"author": "Jordan Petridis <jordan@centricular.com>, Mathieu Duponchelle <mathieu@centricular.com>, François Laignel <francois@centricular.com>",
|
||||
"description": "Speech to Text filter, using AWS transcribe",
|
||||
"hierarchy": [
|
||||
"GstAwsTranscriber",
|
||||
|
|
|
@ -11,36 +11,30 @@ edition = "2021"
|
|||
rust-version = "1.66"
|
||||
|
||||
[dependencies]
|
||||
bytes = "1.0"
|
||||
futures = "0.3"
|
||||
gst = { package = "gstreamer", git = "https://gitlab.freedesktop.org/gstreamer/gstreamer-rs" }
|
||||
gst-base = { package = "gstreamer-base", git = "https://gitlab.freedesktop.org/gstreamer/gstreamer-rs" }
|
||||
gst-audio = { package = "gstreamer-audio", git = "https://gitlab.freedesktop.org/gstreamer/gstreamer-rs", features = ["v1_16"] }
|
||||
async-stream = "0.3.4"
|
||||
base32 = "0.4"
|
||||
aws-config = "0.54.0"
|
||||
aws-sdk-s3 = "0.24.0"
|
||||
aws-sdk-transcribe = "0.24.0"
|
||||
aws-sdk-transcribestreaming = "0.24.0"
|
||||
aws-types = "0.54.0"
|
||||
aws-credential-types = "0.54.0"
|
||||
aws-sig-auth = "0.54.0"
|
||||
aws-smithy-http = { version = "0.54.0", features = [ "rt-tokio" ] }
|
||||
aws-smithy-types = "0.54.0"
|
||||
bytes = "1.0"
|
||||
futures = "0.3"
|
||||
gio = { git = "https://github.com/gtk-rs/gtk-rs-core.git", package = "gio" }
|
||||
gst = { package = "gstreamer", git = "https://gitlab.freedesktop.org/gstreamer/gstreamer-rs" }
|
||||
gst-base = { package = "gstreamer-base", git = "https://gitlab.freedesktop.org/gstreamer/gstreamer-rs" }
|
||||
gst-audio = { package = "gstreamer-audio", git = "https://gitlab.freedesktop.org/gstreamer/gstreamer-rs", features = ["v1_16"] }
|
||||
http = "0.2.7"
|
||||
chrono = "0.4"
|
||||
url = "2"
|
||||
once_cell = "1.0"
|
||||
percent-encoding = "2"
|
||||
tokio = { version = "1.0", features = [ "full" ] }
|
||||
async-tungstenite = { version = "0.20", features = ["tokio", "tokio-runtime", "tokio-native-tls"] }
|
||||
nom = "7"
|
||||
crc = "3"
|
||||
byteorder = "1.3.4"
|
||||
once_cell = "1.0"
|
||||
serde = "1"
|
||||
serde_derive = "1"
|
||||
serde_json = "1"
|
||||
atomic_refcell = "0.1"
|
||||
base32 = "0.4"
|
||||
backoff = { version = "0.4", features = [ "futures", "tokio" ] }
|
||||
gio = { git = "https://github.com/gtk-rs/gtk-rs-core.git", package = "gio" }
|
||||
url = "2"
|
||||
|
||||
[dev-dependencies]
|
||||
chrono = { version = "0.4", features = [ "alloc" ] }
|
||||
|
|
File diff suppressed because it is too large
Load diff
|
@ -10,7 +10,8 @@ use gst::glib;
|
|||
use gst::prelude::*;
|
||||
|
||||
mod imp;
|
||||
mod packet;
|
||||
|
||||
use aws_sdk_transcribestreaming::model::{PartialResultsStability, VocabularyFilterMethod};
|
||||
|
||||
#[derive(Debug, Eq, PartialEq, Ord, PartialOrd, Hash, Clone, Copy, glib::Enum)]
|
||||
#[repr(u32)]
|
||||
|
@ -31,6 +32,17 @@ pub enum AwsTranscriberResultStability {
|
|||
Low = 2,
|
||||
}
|
||||
|
||||
impl From<AwsTranscriberResultStability> for PartialResultsStability {
|
||||
fn from(val: AwsTranscriberResultStability) -> Self {
|
||||
use AwsTranscriberResultStability::*;
|
||||
match val {
|
||||
High => PartialResultsStability::High,
|
||||
Medium => PartialResultsStability::Medium,
|
||||
Low => PartialResultsStability::Low,
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Debug, Eq, PartialEq, Ord, PartialOrd, Hash, Clone, Copy, glib::Enum)]
|
||||
#[repr(u32)]
|
||||
#[enum_type(name = "GstAwsTranscriberVocabularyFilterMethod")]
|
||||
|
@ -44,6 +56,17 @@ pub enum AwsTranscriberVocabularyFilterMethod {
|
|||
Tag = 2,
|
||||
}
|
||||
|
||||
impl From<AwsTranscriberVocabularyFilterMethod> for VocabularyFilterMethod {
|
||||
fn from(val: AwsTranscriberVocabularyFilterMethod) -> Self {
|
||||
use AwsTranscriberVocabularyFilterMethod::*;
|
||||
match val {
|
||||
Mask => VocabularyFilterMethod::Mask,
|
||||
Remove => VocabularyFilterMethod::Remove,
|
||||
Tag => VocabularyFilterMethod::Tag,
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
glib::wrapper! {
|
||||
pub struct Transcriber(ObjectSubclass<imp::Transcriber>) @extends gst::Element, gst::Object;
|
||||
}
|
||||
|
|
|
@ -1,174 +0,0 @@
|
|||
// Copyright (C) 2020 Jordan Petridis <jordan@centricular.com>
|
||||
//
|
||||
// This Source Code Form is subject to the terms of the Mozilla Public License, v2.0.
|
||||
// If a copy of the MPL was not distributed with this file, You can obtain one at
|
||||
// <https://mozilla.org/MPL/2.0/>.
|
||||
//
|
||||
// SPDX-License-Identifier: MPL-2.0
|
||||
|
||||
use byteorder::{BigEndian, WriteBytesExt};
|
||||
use nom::{
|
||||
self, bytes::complete::take, combinator::map_res, multi::many0, number::complete::be_u16,
|
||||
number::complete::be_u32, number::complete::be_u8, sequence::tuple, IResult,
|
||||
};
|
||||
use std::borrow::Cow;
|
||||
use std::io::{self, Write};
|
||||
|
||||
const CRC: crc::Crc<u32> = crc::Crc::<u32>::new(&crc::CRC_32_ISO_HDLC);
|
||||
|
||||
#[derive(Debug)]
|
||||
struct Prelude {
|
||||
total_bytes: u32,
|
||||
header_bytes: u32,
|
||||
#[allow(dead_code)]
|
||||
prelude_crc: u32,
|
||||
}
|
||||
|
||||
#[derive(Debug)]
|
||||
pub struct Header {
|
||||
pub name: Cow<'static, str>,
|
||||
pub value_type: u8,
|
||||
pub value: Cow<'static, str>,
|
||||
}
|
||||
|
||||
#[derive(Debug)]
|
||||
pub struct Packet<'a> {
|
||||
#[allow(dead_code)]
|
||||
prelude: Prelude,
|
||||
headers: Vec<Header>,
|
||||
pub payload: &'a [u8],
|
||||
#[allow(dead_code)]
|
||||
msg_crc: u32,
|
||||
}
|
||||
|
||||
fn write_header<W: Write>(w: &mut W, header: &Header) -> Result<(), io::Error> {
|
||||
w.write_u8(header.name.len() as u8)?;
|
||||
w.write_all(header.name.as_bytes())?;
|
||||
w.write_u8(header.value_type)?;
|
||||
w.write_u16::<BigEndian>(header.value.len() as u16)?;
|
||||
w.write_all(header.value.as_bytes())?;
|
||||
Ok(())
|
||||
}
|
||||
|
||||
fn write_headers<W: Write>(w: &mut W, headers: &[Header]) -> Result<(), io::Error> {
|
||||
for header in headers {
|
||||
write_header(w, header)?;
|
||||
}
|
||||
Ok(())
|
||||
}
|
||||
|
||||
pub fn encode_packet(payload: &[u8], headers: &[Header]) -> Result<Vec<u8>, io::Error> {
|
||||
let mut res = Vec::with_capacity(1024);
|
||||
|
||||
// Total length
|
||||
res.write_u32::<BigEndian>(0)?;
|
||||
// Header length
|
||||
res.write_u32::<BigEndian>(0)?;
|
||||
// Prelude CRC32 placeholder
|
||||
res.write_u32::<BigEndian>(0)?;
|
||||
|
||||
// Write all headers
|
||||
write_headers(&mut res, headers)?;
|
||||
|
||||
// Rewrite header length
|
||||
let header_length = res.len() - 12;
|
||||
(&mut res[4..8]).write_u32::<BigEndian>(header_length as u32)?;
|
||||
|
||||
// Write payload
|
||||
res.write_all(payload)?;
|
||||
|
||||
// Rewrite total length
|
||||
let total_length = res.len() + 4;
|
||||
(&mut res[0..4]).write_u32::<BigEndian>(total_length as u32)?;
|
||||
|
||||
// Rewrite the prelude crc since we replaced the lengths
|
||||
let prelude_crc = CRC.checksum(&res[0..8]);
|
||||
(&mut res[8..12]).write_u32::<BigEndian>(prelude_crc)?;
|
||||
|
||||
// Message CRC
|
||||
let message_crc = CRC.checksum(&res);
|
||||
res.write_u32::<BigEndian>(message_crc)?;
|
||||
|
||||
Ok(res)
|
||||
}
|
||||
|
||||
fn parse_prelude(input: &[u8]) -> IResult<&[u8], Prelude> {
|
||||
map_res(
|
||||
tuple((be_u32, be_u32, be_u32)),
|
||||
|(total_bytes, header_bytes, prelude_crc)| {
|
||||
let sum = CRC.checksum(&input[0..8]);
|
||||
if prelude_crc != sum {
|
||||
return Err(nom::Err::Error((
|
||||
"Prelude CRC doesn't match",
|
||||
nom::error::ErrorKind::MapRes,
|
||||
)));
|
||||
}
|
||||
|
||||
Ok(Prelude {
|
||||
total_bytes,
|
||||
header_bytes,
|
||||
prelude_crc,
|
||||
})
|
||||
},
|
||||
)(input)
|
||||
}
|
||||
|
||||
fn parse_header(input: &[u8]) -> IResult<&[u8], Header> {
|
||||
let (input, header_length) = be_u8(input)?;
|
||||
let (input, name) = map_res(take(header_length), std::str::from_utf8)(input)?;
|
||||
let (input, value_type) = be_u8(input)?;
|
||||
let (input, value_length) = be_u16(input)?;
|
||||
let (input, value) = map_res(take(value_length), std::str::from_utf8)(input)?;
|
||||
|
||||
let header = Header {
|
||||
name: name.to_string().into(),
|
||||
value_type,
|
||||
value: value.to_string().into(),
|
||||
};
|
||||
|
||||
Ok((input, header))
|
||||
}
|
||||
|
||||
pub fn packet_is_exception(packet: &Packet) -> bool {
|
||||
for header in &packet.headers {
|
||||
if header.name == ":message-type" && header.value == "exception" {
|
||||
return true;
|
||||
}
|
||||
}
|
||||
|
||||
false
|
||||
}
|
||||
|
||||
pub fn parse_packet(input: &[u8]) -> IResult<&[u8], Packet> {
|
||||
let (remainder, prelude) = parse_prelude(input)?;
|
||||
|
||||
// Check the crc of the whole input
|
||||
let sum = CRC.checksum(&input[..input.len() - 4]);
|
||||
let (_, msg_crc) = be_u32(&input[input.len() - 4..])?;
|
||||
|
||||
if msg_crc != sum {
|
||||
return Err(nom::Err::Error(nom::error::Error::new(
|
||||
b"Prelude CRC doesn't match",
|
||||
nom::error::ErrorKind::MapRes,
|
||||
)));
|
||||
}
|
||||
|
||||
let (remainder, header_input) = take(prelude.header_bytes)(remainder)?;
|
||||
let (_, headers) = many0(parse_header)(header_input)?;
|
||||
|
||||
let payload_length = prelude.total_bytes - prelude.header_bytes - 4 - 12;
|
||||
let (remainder, payload) = take(payload_length)(remainder)?;
|
||||
|
||||
// only the message_crc we check before should be remaining now
|
||||
assert_eq!(remainder.len(), 4);
|
||||
|
||||
Ok((
|
||||
input,
|
||||
Packet {
|
||||
prelude,
|
||||
headers,
|
||||
payload,
|
||||
msg_crc,
|
||||
},
|
||||
))
|
||||
}
|
Loading…
Reference in a new issue