current main is now successfully merged into branch

This commit is contained in:
Paul Koudelka committed 2026-05-27 20:02:43 +02:00
1 parent 00b6c98f3b
commit ea87d79ca9
32 files changed
+607 -478

No files matched your search

+2 -1
View File
@@ -1,6 +1,6 @@
[package]
name = "mindwork-ai-studio"
version = "26.4.1"
version = "26.5.5"
edition = "2024"
description = "MindWork AI Studio"
authors = ["Thorsten Sommer"]
@@ -48,6 +48,7 @@ tempfile = "3.27.0"
strum_macros = "0.28.0"
sysinfo = "0.39.1"
bytes = "1.11.1"
tokenizers = "0.23.1"
[target.'cfg(target_os = "windows")'.dependencies]
windows-registry = "0.6.1"
+2 -2
View File
@@ -28,7 +28,7 @@ use crate::pdfium::PDFIUM_LIB_PATH;
use crate::qdrant::{start_qdrant_server, stop_qdrant_server};
#[cfg(debug_assertions)]
use crate::dotnet::create_startup_env_file;
use crate::tokenizer::set_path_resolver;
use crate::tokenizer::set_default_tokenizer_path;
/// The Tauri main window.
pub static MAIN_WINDOW: Lazy<Mutex<Option<WebviewWindow>>> = Lazy::new(|| Mutex::new(None));
@@ -151,7 +151,7 @@ pub fn start_tauri() {
start_qdrant_server(app.handle().clone());
set_path_resolver(app.path_resolver());
set_default_tokenizer_path(app.handle().clone());
info!(Source = "Bootloader Tauri"; "Reconfigure the file logger to use the app data directory {data_path:?}");
switch_to_file_logging(data_path).map_err(|e| error!("Failed to switch logging to file: {e}")).unwrap();
+6
View File
@@ -33,6 +33,7 @@ pub fn start_runtime_api() {
.route("/system/dotnet/port", get(crate::dotnet::dotnet_port))
.route("/system/dotnet/ready", get(crate::dotnet::dotnet_ready))
.route("/system/qdrant/info", get(crate::qdrant::qdrant_port))
.route("/system/tokenizer/info", get(crate::tokenizer::tokenizer_info))
.route("/clipboard/set", post(crate::clipboard::set_clipboard))
.route("/events", get(crate::app_window::get_event_stream))
.route("/updates/check", get(crate::app_window::check_for_update))
@@ -56,6 +57,11 @@ pub fn start_runtime_api() {
.route("/retrieval/fs/extract", get(crate::file_data::extract_data))
.route("/log/paths", get(crate::log::get_log_paths))
.route("/log/event", post(crate::log::log_event))
.route("/tokenizer/count", post(crate::tokenizer::token_count))
.route("/tokenizer/validate", post(crate::tokenizer::validate_tokenizer))
.route("/tokenizer/store", post(crate::tokenizer::store_tokenizer))
.route("/tokenizer/delete", post(crate::tokenizer::delete_tokenizer))
.route("/tokenizer/set", post(crate::tokenizer::set_tokenizer))
.route("/shortcuts/register", post(crate::app_window::register_shortcut))
.route("/shortcuts/validate", post(crate::app_window::validate_shortcut))
.route("/shortcuts/suspend", post(crate::app_window::suspend_shortcuts))
+260 -144
View File
@@ -1,24 +1,43 @@
use std::fs;
use std::path::PathBuf;
use std::sync::{OnceLock, RwLock};
use log::warn;
use rocket::post;
use rocket::serde::json::Json;
use rocket::serde::Serialize;
use serde::Deserialize;
use tauri::PathResolver;
use tokenizers::Error;
use tokenizers::tokenizer::{Tokenizer, Error as TokenizerError};
use std::sync::{Mutex, OnceLock, RwLock};
use axum::Json;
use log::{error, warn};
use once_cell::sync::Lazy;
use serde::{Deserialize, Serialize};
use tauri::path::BaseDirectory;
use tauri::Manager;
use tokenizers::tokenizer::Tokenizer;
use crate::api_token::APIToken;
use crate::environment::DATA_DIRECTORY;
static TOKENIZER: OnceLock<RwLock<Option<Tokenizer>>> = OnceLock::new();
const DEFAULT_TOKENIZER_RESOURCE_PATH: &str = "resources/tokenizers/tokenizer.json";
const NO_TOKENIZER_LOADED_MESSAGE: &str = "Tokenizer must be set before counting tokens.";
static TOKENIZER_PATH_RESOLVER: OnceLock<PathResolver> = OnceLock::new();
static TOKENIZER: OnceLock<RwLock<Option<Tokenizer>>> = OnceLock::new();
static DEFAULT_TOKENIZER_PATH: OnceLock<PathBuf> = OnceLock::new();
static TOKENIZER_STATUS: Lazy<Mutex<TokenizerStatusInfo>> = Lazy::new(|| Mutex::new(TokenizerStatusInfo::default()));
static TOKENIZER_OPERATION_LOCK: Lazy<Mutex<()>> = Lazy::new(|| Mutex::new(()));
#[derive(Clone, Copy, Default, Serialize, PartialEq, Eq)]
pub enum TokenizerStatus {
#[default]
Unavailable,
Running,
Available,
}
#[derive(Default)]
struct TokenizerStatusInfo {
status: TokenizerStatus,
unavailable_reason: Option<String>,
}
#[derive(Deserialize)]
pub struct SetTokenText {
pub text: String,
text: String,
}
#[derive(Clone, Deserialize)]
@@ -42,94 +61,205 @@ pub struct TokenizerResponse {
success: bool,
token_count: usize,
message: String,
status: TokenizerStatus,
stored_path: String,
}
impl From<Result<usize, TokenizerError>> for TokenizerResponse {
fn from(result: Result<usize, TokenizerError>) -> Self {
match result {
Ok(count) => TokenizerResponse {
success: true,
token_count: count,
message: "Success".to_string(),
},
Err(e) => TokenizerResponse {
success: false,
token_count: 0,
message: e.to_string(),
},
impl TokenizerResponse {
fn available(token_count: usize) -> Self {
TokenizerResponse {
success: true,
token_count,
message: String::new(),
status: TokenizerStatus::Available,
stored_path: String::new(),
}
}
fn stored(stored_path: String) -> Self {
TokenizerResponse {
success: true,
token_count: 0,
message: String::new(),
status: TokenizerStatus::Available,
stored_path,
}
}
fn unavailable(reason: String) -> Self {
TokenizerResponse {
success: false,
token_count: 0,
message: reason,
status: TokenizerStatus::Unavailable,
stored_path: String::new(),
}
}
}
pub fn set_path_resolver(path_resolver: PathResolver) {
match TOKENIZER_PATH_RESOLVER.set(path_resolver) {
pub fn set_default_tokenizer_path(app_handle: tauri::AppHandle) {
let tokenizer_path = match app_handle
.path()
.resolve(DEFAULT_TOKENIZER_RESOURCE_PATH, BaseDirectory::Resource)
{
Ok(path) => path,
Err(e) => {
let reason = format!("The default tokenizer file '{DEFAULT_TOKENIZER_RESOURCE_PATH}' could not be resolved: {e}");
error!(Source = "Tokenizer"; "{reason}");
set_tokenizer_unavailable(reason);
return;
}
};
if !tokenizer_path.is_file() {
let reason = format!("The default tokenizer file was not found: {}", tokenizer_path.display());
error!(Source = "Tokenizer"; "{reason}");
set_tokenizer_unavailable(reason);
return;
}
match DEFAULT_TOKENIZER_PATH.set(tokenizer_path) {
Ok(_) => (),
Err(e) => warn!(Source = "Tokenizer"; "Could not set the path resolver: {:?}", e),
Err(e) => warn!(Source = "Tokenizer"; "Could not set the default tokenizer path: {:?}", e),
}
}
fn tokenizer_state() -> &'static RwLock<Option<Tokenizer>> {
TOKENIZER.get_or_init(|| RwLock::new(None))
pub async fn tokenizer_info(_token: APIToken) -> Json<TokenizerResponse> {
let status = TOKENIZER_STATUS.lock().unwrap();
match status.status {
TokenizerStatus::Available => Json(TokenizerResponse::available(0)),
TokenizerStatus::Running => Json(TokenizerResponse {
success: false,
token_count: 0,
message: String::new(),
status: TokenizerStatus::Running,
stored_path: String::new(),
}),
TokenizerStatus::Unavailable => Json(TokenizerResponse::unavailable(status.unavailable_reason.clone().unwrap_or_default())),
}
}
pub fn handle_tokenizer_set(path: &str) -> Result<(), Error> {
let tokenizer_path = if path.trim().is_empty() {
let relative_source_path = String::from("resources/tokenizers/tokenizer.json");
let path_resolver = TOKENIZER_PATH_RESOLVER
.get()
.ok_or_else(|| Error::from("Tokenizer path resolver is not initialized"))?;
path_resolver
.resolve_resource(relative_source_path)
.ok_or_else(|| Error::from("Failed to resolve default tokenizer resource path"))?
} else {
PathBuf::from(path)
};
pub async fn token_count(_token: APIToken, req: Json<SetTokenText>) -> Json<TokenizerResponse> {
match get_token_count(&req.text) {
Ok(count) => Json(TokenizerResponse::available(count)),
Err(e) => Json(TokenizerResponse::unavailable(e)),
}
}
let tokenizer = Tokenizer::from_file(tokenizer_path)?;
let mut tokenizer_guard = tokenizer_state()
.write()
.map_err(|_| Error::from("Tokenizer state lock is poisoned"))?;
*tokenizer_guard = Some(tokenizer);
pub async fn validate_tokenizer(_token: APIToken, payload: Json<TokenizerPath>) -> Json<TokenizerResponse> {
match handle_tokenizer_validate(&PathBuf::from(payload.file_path.clone())) {
Ok(count) => Json(TokenizerResponse::available(count)),
Err(e) => Json(TokenizerResponse::unavailable(e)),
}
}
pub async fn store_tokenizer(_token: APIToken, payload: Json<TokenizerStorage>) -> Json<TokenizerResponse> {
match handle_tokenizer_store(&payload) {
Ok(dest_path) => Json(TokenizerResponse::stored(dest_path)),
Err(e) => Json(TokenizerResponse::unavailable(e.to_string())),
}
}
pub async fn delete_tokenizer(_token: APIToken, payload: Json<TokenizerDelete>) -> Json<TokenizerResponse> {
match handle_tokenizer_delete(&payload) {
Ok(_) => Json(TokenizerResponse::stored(String::new())),
Err(e) => Json(TokenizerResponse::unavailable(e.to_string())),
}
}
pub async fn set_tokenizer(_token: APIToken, payload: Json<TokenizerPath>) -> Json<TokenizerResponse> {
match handle_tokenizer_set(&payload.file_path) {
Ok(_) => Json(TokenizerResponse::available(0)),
Err(e) => Json(TokenizerResponse::unavailable(e)),
}
}
pub fn handle_tokenizer_set(path: &str) -> Result<(), String> {
let _operation_guard = begin_tokenizer_operation()?;
set_tokenizer_running();
let tokenizer_path = resolve_tokenizer_path(path).map_err(|e| {
error!(Source = "Tokenizer"; "{e} Starting the app without a tokenizer.");
unavailable_with_status_update(&e)
})?;
let tokenizer = load_tokenizer_from_file(&tokenizer_path).map_err(|e| {
error!(Source = "Tokenizer"; "{e}");
unavailable_with_status_update(&e)
})?;
match tokenizer_state().write() {
Ok(mut tokenizer_guard) => *tokenizer_guard = Some(tokenizer),
Err(_) => return Err(unavailable_with_status_update("Tokenizer state lock is poisoned.")),
}
set_tokenizer_available();
Ok(())
}
fn handle_tokenizer_validate(path: &PathBuf) -> Result<usize, TokenizerError> {
if !path.is_file() {
return Err(TokenizerError::from(format!(
"Tokenizer file was not found: {}",
path.display()
)));
fn handle_tokenizer_validate(path: &PathBuf) -> Result<usize, String> {
let _operation_guard = begin_tokenizer_operation()?;
set_tokenizer_running();
let result = validate_tokenizer_file(path);
match tokenizer_state().read() {
Ok(tokenizer_guard) if tokenizer_guard.is_some() => set_tokenizer_available(),
Ok(_) => set_tokenizer_unavailable(NO_TOKENIZER_LOADED_MESSAGE.to_string()),
Err(_) => set_tokenizer_unavailable("Tokenizer state lock is poisoned.".to_string()),
}
let tokenizer = Tokenizer::from_file(path).map_err(|e| {
TokenizerError::from(format!(
"Failed to load tokenizer from '{}': {}",
path.display(),
e
))
})?;
result
}
pub fn get_token_count(text: &str) -> Result<usize, String> {
if text.trim().is_empty() {
return Ok(0);
}
let _operation_guard = begin_tokenizer_operation()?;
{
let status = TOKENIZER_STATUS.lock().unwrap();
if status.status != TokenizerStatus::Available {
return Err(status.unavailable_reason.clone().unwrap_or_else(|| NO_TOKENIZER_LOADED_MESSAGE.to_string()));
}
}
let tokenizer_guard = tokenizer_state()
.read()
.map_err(|_| unavailable_with_status_update("Tokenizer state lock is poisoned."))?;
let tokenizer = match tokenizer_guard.as_ref() {
Some(tokenizer) => tokenizer,
None => {
drop(tokenizer_guard);
return Err(unavailable_with_status_update("Tokenizer not initialized."));
}
};
let token_count = match tokenizer.encode(text, true) {
Ok(enc) => enc.len(),
Err(e) => {
let reason = format!("Failed to tokenize text: {e}");
drop(tokenizer_guard);
return Err(unavailable_with_status_update(&reason));
}
};
Ok(token_count)
}
fn validate_tokenizer_file(path: &PathBuf) -> Result<usize, String> {
let tokenizer = load_tokenizer_from_file(path)?;
let test_string = "Hello, world! This is a test string for tokenizer validation.";
let encoding = tokenizer.encode(test_string, true).map_err(|e| {
TokenizerError::from(format!(
"Tokenizer failed to encode validation string: {}",
e
))
})?;
let encoding = tokenizer
.encode(test_string, true)
.map_err(|e| format!("Tokenizer failed to encode validation string: {e}"))?;
let token_count = encoding.len();
if token_count == 0 {
return Err(TokenizerError::from(
"Tokenizer produced 0 tokens for test string. The tokenizer is likely invalid or misconfigured.",
));
return Err("Tokenizer produced 0 tokens for test string. The tokenizer is likely invalid or misconfigured.".to_string());
}
if encoding.get_tokens().iter().any(|t| t.is_empty()) {
return Err(TokenizerError::from(
"Tokenizer produced empty tokens. The tokenizer is invalid.",
));
return Err("Tokenizer produced empty tokens. The tokenizer is invalid.".to_string());
}
Ok(token_count)
@@ -147,26 +277,25 @@ fn handle_tokenizer_store(payload: &TokenizerStorage) -> Result<String, std::io:
.file_name()
.and_then(|n| n.to_str())
.ok_or_else(|| std::io::Error::new(std::io::ErrorKind::InvalidInput, "Invalid tokenizer file path"))?;
let model_path = &base_path.join(&payload.model_id);
let destination_path = &model_path.join(source_name);
let model_path = base_path.join(&payload.model_id);
let destination_path = model_path.join(source_name);
if source_path.eq(destination_path) {
return Ok(destination_path.to_str().unwrap().to_string());
if source_path.eq(&destination_path) {
return Ok(destination_path.to_string_lossy().to_string());
}
match model_path.try_exists()? {
true => fs::remove_dir_all(model_path.clone())?,
false => (),
if model_path.try_exists()? {
fs::remove_dir_all(&model_path)?;
}
if payload.file_path.trim().is_empty() {
return Ok(String::from(""));
return Ok(String::new());
}
fs::create_dir_all(model_path)?;
fs::create_dir_all(&model_path)?;
fs::copy(&source_path, &destination_path)?;
Ok(destination_path.to_str().unwrap().to_string())
Ok(destination_path.to_string_lossy().to_string())
}
fn handle_tokenizer_delete(payload: &TokenizerDelete) -> Result<(), std::io::Error> {
@@ -189,74 +318,61 @@ fn handle_tokenizer_delete(payload: &TokenizerDelete) -> Result<(), std::io::Err
Ok(())
}
pub fn get_token_count(text: &str) -> Result<usize, TokenizerError> {
if text.trim().is_empty() {
return Err(TokenizerError::from("Input text is empty"));
fn tokenizer_state() -> &'static RwLock<Option<Tokenizer>> {
TOKENIZER.get_or_init(|| RwLock::new(None))
}
fn begin_tokenizer_operation() -> Result<std::sync::MutexGuard<'static, ()>, String> {
TOKENIZER_OPERATION_LOCK
.lock()
.map_err(|_| unavailable_with_status_update("Tokenizer operation lock is poisoned."))
}
fn set_tokenizer_available() {
let mut status = TOKENIZER_STATUS.lock().unwrap();
status.status = TokenizerStatus::Available;
status.unavailable_reason = None;
}
fn set_tokenizer_running() {
let mut status = TOKENIZER_STATUS.lock().unwrap();
status.status = TokenizerStatus::Running;
status.unavailable_reason = None;
}
fn set_tokenizer_unavailable(reason: String) {
let mut status = TOKENIZER_STATUS.lock().unwrap();
status.status = TokenizerStatus::Unavailable;
status.unavailable_reason = Some(reason);
}
fn unavailable_with_status_update(reason: &str) -> String {
let reason = reason.to_string();
match tokenizer_state().write() {
Ok(mut tokenizer_guard) => *tokenizer_guard = None,
Err(_) => set_tokenizer_unavailable("Tokenizer state lock is poisoned.".to_string()),
}
let tokenizer = tokenizer_state()
.read()
.map_err(|_| TokenizerError::from("Tokenizer state lock is poisoned"))?
.clone()
.ok_or_else(|| TokenizerError::from("Tokenizer not initialized"))?;
let enc = tokenizer.encode(text, true)?;
Ok(enc.len())
set_tokenizer_unavailable(reason.clone());
reason
}
#[post("/tokenizer/count", data = "<req>")]
pub fn token_count(_token: APIToken, req: Json<SetTokenText>) -> Json<TokenizerResponse> {
Json(get_token_count(&req.text).into())
}
#[post("/tokenizer/validate", data = "<payload>")]
pub fn validate_tokenizer(_token: APIToken, payload: Json<TokenizerPath>) -> Json<TokenizerResponse> {
Json(handle_tokenizer_validate(&PathBuf::from(payload.file_path.clone())).into())
}
#[post("/tokenizer/store", data = "<payload>")]
pub fn store_tokenizer(_token: APIToken, payload: Json<TokenizerStorage>) -> Json<TokenizerResponse> {
match handle_tokenizer_store(&payload) {
Ok(dest_path) => Json(TokenizerResponse {
success: true,
token_count: 0,
message: dest_path,
}),
Err(e) => Json(TokenizerResponse {
success: false,
token_count: 0,
message: e.to_string(),
}),
fn resolve_tokenizer_path(path: &str) -> Result<PathBuf, String> {
if !path.trim().is_empty() {
return Ok(PathBuf::from(path));
}
DEFAULT_TOKENIZER_PATH
.get()
.cloned()
.ok_or_else(|| "Default tokenizer path is not initialized.".to_string())
}
#[post("/tokenizer/delete", data = "<payload>")]
pub fn delete_tokenizer(_token: APIToken, payload: Json<TokenizerDelete>) -> Json<TokenizerResponse> {
match handle_tokenizer_delete(&payload) {
Ok(_) => Json(TokenizerResponse {
success: true,
token_count: 0,
message: "Success".to_string(),
}),
Err(e) => Json(TokenizerResponse {
success: false,
token_count: 0,
message: e.to_string(),
}),
fn load_tokenizer_from_file(path: &PathBuf) -> Result<Tokenizer, String> {
if !path.is_file() {
return Err(format!("Tokenizer file was not found: {}", path.display()));
}
}
#[post("/tokenizer/set", data = "<payload>")]
pub fn set_tokenizer(_token: APIToken, payload: Json<TokenizerPath>) -> Json<TokenizerResponse> {
match handle_tokenizer_set(&payload.file_path) {
Ok(_) => Json(TokenizerResponse {
success: true,
token_count: 0,
message: "Success".to_string(),
}),
Err(e) => Json(TokenizerResponse {
success: false,
token_count: 0,
message: e.to_string(),
}),
}
Tokenizer::from_file(path)
.map_err(|e| format!("Failed to load tokenizer from '{}': {e}", path.display()))
}