rust-whisper/rust-whisper-baked/src/main.rs

169 lines
4.3 KiB
Rust

use rust_whisper_lib;
use rust_whisper_baked_lib;
use clap::Parser;
use listen_lib;
use std::thread;
fn main() {
let flags = rust_whisper_lib::Flags::parse();
match flags.wav.clone() {
Some(_) => wav_channel(flags),
None => channel(flags),
};
}
fn wav_channel(flags: rust_whisper_lib::Flags) {
let mut w = new_destutterer();
rust_whisper_baked_lib::wav_channel(
flags.clone(),
move |result: Result<rust_whisper_lib::Transcribed, String>| {
match result {
Ok(transcribed) => {
let s = w.step(transcribed.to_string());
println!("{}", s);
},
Err(msg) => { eprintln!("error: {}", msg); },
};
},
);
}
fn wav(flags: rust_whisper_lib::Flags, _path: String) {
let mut w = new_destutterer();
rust_whisper_baked_lib::wav(flags,
move |result: Result<rust_whisper_lib::Transcribed, String>| {
match result {
Ok(transcribed) => {
let s = w.step(transcribed.to_string());
println!("{}", s);
},
Err(msg) => { eprintln!("error: {}", msg); },
};
},
);
}
fn channel(flags: rust_whisper_lib::Flags) {
let (send, recv) = std::sync::mpsc::sync_channel(100);
eprintln!("rust whisper baked lib channel...");
thread::spawn(move || {
rust_whisper_baked_lib::channel(
flags.clone(),
|result: Result<rust_whisper_lib::Transcribed, String>| {
match result {
Ok(transcribed) => { println!("{}", transcribed.to_string()); },
Err(msg) => { eprintln!("error: {}", msg); },
};
},
recv,
);
});
eprintln!("listen lib main...");
let flags = rust_whisper_lib::Flags::parse();
match flags.stream_device {
Some(device_name) => {
if device_name == "" {
for device in listen_lib::devices() {
eprintln!("{}", device);
}
} else {
listen_lib::main_with(|data| {
send.send(data).unwrap();
}, device_name);
}
},
None => {
listen_lib::main(|data| {
send.send(data).unwrap();
});
}
}
eprintln!("/listen lib main...");
}
struct Destutterer {
prevs: Vec<Word>,
}
fn new_destutterer() -> Destutterer {
Destutterer{prevs: vec![]}
}
impl Destutterer {
fn step(&mut self, next: String) -> String {
if next.len() == 0 {
return next;
}
let nexts = Word::from_string(next.clone());
let mut n = self.prevs.len().clamp(0, nexts.len());
while n > 0 {
let prev_s = Word::to_comparable_string(self.prevs[self.prevs.len() - n..].to_vec());
let next_s = Word::to_comparable_string(nexts[..n].to_vec());
if prev_s == next_s {
break;
}
n -= 1;
}
self.prevs = nexts.clone();
Word::to_string(nexts[n..].to_vec())
}
}
#[derive(Clone)]
struct Word {
raw: String,
}
impl Word {
fn from_string(s: String) -> Vec<Word> {
let mut result = vec![];
for word in s.split(" ") {
let word = word.trim();
if word.len() > 0 {
result.push(Word{raw: word.to_string()});
}
}
result
}
fn to_comparable_string(v: Vec<Word>) -> String {
v.iter()
.map(|x| x.raw.chars().filter(|c| c.is_ascii_alphanumeric()).collect())
.collect::<Vec<String>>()
.join(" ")
}
fn to_string(v: Vec<Word>) -> String {
v.iter()
.map(|x| x.raw.clone())
.collect::<Vec<String>>()
.join(" ")
}
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn test_destutterer_punctuation() {
let mut w = new_destutterer();
assert_eq!("a, b. c? d!".to_string(), w.step("a, b. c? d!".to_string()));
assert_eq!("e! f g".to_string(), w.step("d, e! f g".to_string()));
assert_eq!("hij".to_string(), w.step("f g hij".to_string()));
}
#[test]
fn test_destutterer_letters() {
let mut w = new_destutterer();
assert_eq!("a b c d e".to_string(), w.step("a b c d e".to_string()));
assert_eq!("f g".to_string(), w.step(" c d e f g".to_string()));
assert_eq!("h i j".to_string(), w.step("f g h i j ".to_string()));
assert_eq!("a g h i j".to_string(), w.step("a g h i j".to_string()));
}
}