learned images

config.json

@@ -43,5 +43,6 @@
             "height": 43
         }
     ],
-    "ocr_server_endpoint": "http://localhost:3000/"
+    
+    "ocr_server_endpoint": "https://tesserver.spruett.dev/"
 }

learned.json

@@ -1,4 +1,5 @@
 {
-    "learned_images": {},
-    "learned_tracks": {}
+  "learned_images": {
+  },
+  "learned_tracks": {}
 }

src/capture.rs

@@ -21,9 +21,8 @@ fn get_raw_frame(capturer: &mut Capturer) -> Result<Vec<u8>> {
     }
 }
 
-pub fn get_frame() -> Result<RgbImage> {
-    let mut capturer = Capturer::new(Display::primary()?)?;
-    let frame = get_raw_frame(&mut capturer)?;
+pub fn get_frame(capturer: &mut Capturer) -> Result<RgbImage> {
+    let frame = get_raw_frame(capturer)?;
     let mut image = RgbImage::new(capturer.width() as u32, capturer.height() as u32);
 
     let stride = frame.len() / capturer.height();

src/config.rs

@@ -41,7 +41,7 @@ fn load_or_make_default<T: DeserializeOwned>(path: &str, default: &str) -> Resul
     if !file_path.exists() {
         std::fs::write(&path, default)?;
     }
-    load_json_config(&path)
+    load_json_config(path)
 }
 
 fn load_json_config<T: DeserializeOwned>(path: &str) -> Result<T> {

src/configs/config.default.json

@@ -43,5 +43,5 @@
             "height": 43
         }
     ],
-    "ocr_server_endpoint": "http://localhost:3000/"
+    "ocr_server_endpoint": "https://tesserver.spruett.dev/"
 }

src/control_loop.rs

@@ -1,6 +1,6 @@
 use std::{
     collections::HashMap,
-    time::{Duration, Instant},
+    time::{Duration, Instant}, thread,
 };
 
 use anyhow::Result;
@@ -8,9 +8,10 @@ use egui_extras::RetainedImage;
 use image::RgbImage;
 use scrap::{Capturer, Display};
 
+
 use crate::{
     capture,
-    image_processing::{self, Region},
+    image_processing::{self, hash_image},
     ocr,
     state::{AppState, DebugOcrFrame, ParsedFrame, RaceState, SharedAppState},
 };
@@ -103,12 +104,15 @@ fn handle_new_frame(state: &mut AppState, frame: ParsedFrame, image: &RgbImage)
     }
 }
 
-async fn run_loop_once(state: &SharedAppState) -> Result<()> {
+fn run_loop_once(state: &SharedAppState) -> Result<()> {
+    let mut capturer = Capturer::new(Display::primary()?)?;
     let config = state.lock().unwrap().config.clone();
-    let frame = capture::get_frame()?;
-    let ocr_results = ocr::ocr_all_regions(&frame, &config.ocr_regions).await;
+    let learned_config = state.lock().unwrap().learned.clone();
+    let frame = capture::get_frame(&mut capturer)?;
+    let ocr_results = ocr::ocr_all_regions(&frame, config.clone(), learned_config.clone());
 
     let mut saved_frames = HashMap::new();
+
     if state.lock().unwrap().debug_frames {
         let hasher = img_hash::HasherConfig::new().to_hasher();
         for region in &config.ocr_regions {
@@ -119,13 +123,7 @@ async fn run_loop_once(state: &SharedAppState) -> Result<()> {
                 &image_processing::to_png_bytes(&extracted),
             )
             .unwrap();
-            let have_to_use_other_image_library_version = img_hash::image::RgbImage::from_raw(
-                extracted.width(),
-                extracted.height(),
-                extracted.as_raw().to_vec(),
-            )
-            .unwrap();
-            let hash = hasher.hash_image(&have_to_use_other_image_library_version);
+            let hash = hash_image(&extracted);
             saved_frames.insert(
                 region.name.clone(),
                 DebugOcrFrame {
@@ -146,11 +144,11 @@ async fn run_loop_once(state: &SharedAppState) -> Result<()> {
     Ok(())
 }
 
-pub async fn run_control_loop(state: SharedAppState) -> Result<()> {
+pub fn run_control_loop(state: SharedAppState) {
     loop {
-        if let Err(e) = run_loop_once(&state).await {
+        if let Err(e) = run_loop_once(&state) {
             eprintln!("Error in control loop: {:?}", e)
         }
-        tokio::time::sleep(Duration::from_millis(500)).await;
+        thread::sleep(Duration::from_millis(500));
     }
 }

src/image_processing.rs

@@ -1,4 +1,3 @@
-use anyhow::Result;
 use image::{codecs::png::PngEncoder, ColorType, ImageEncoder, Rgb, RgbImage};
 use serde::{Deserialize, Serialize};
 
@@ -64,3 +63,12 @@ pub fn to_png_bytes(image: &RgbImage) -> Vec<u8> {
         .expect("failed encoding image to PNG");
     buffer
 }
+
+pub fn hash_image(image: &RgbImage) -> String {
+    let hasher = img_hash::HasherConfig::new().to_hasher();
+    let have_to_use_other_image_library_version =
+        img_hash::image::RgbImage::from_raw(image.width(), image.height(), image.as_raw().to_vec())
+            .unwrap();
+    let hash = hasher.hash_image(&have_to_use_other_image_library_version);
+    hash.to_base64()
+}

src/main.rs

@@ -9,7 +9,7 @@ mod config;
 
 use std::{
     sync::{Arc, Mutex},
-    time::Duration,
+    time::Duration, thread,
 };
 
 use config::{Config, LearnedConfig};
@@ -19,18 +19,15 @@ use eframe::{
 };
 use state::{AppState, RaceState, SharedAppState};
 
-#[tokio::main(flavor = "multi_thread", worker_threads = 8)]
-async fn main() -> anyhow::Result<()> {
+fn main() -> anyhow::Result<()> {
     let mut app_state = AppState::default();
     app_state.config = Arc::new(Config::load().unwrap());
     app_state.learned = Arc::new(LearnedConfig::load().unwrap());
     let state = Arc::new(Mutex::new(app_state));
     {
         let state = state.clone();
-        let _ = tokio::spawn(async move {
-            control_loop::run_control_loop(state)
-                .await
-                .expect("control loop failed");
+        let _ = thread::spawn(move || {
+            control_loop::run_control_loop(state);
         });
     }
 
@@ -91,11 +88,16 @@ fn label_time_delta(ui: &mut Ui, time: Duration, old: Option<Duration>) {
 
 struct MyApp {
     state: SharedAppState,
+
+    config_load_err: Option<String>,
+    
+    hash_to_learn: String,
+    value_to_learn: String,
 }
 
 impl MyApp {
     pub fn new(state: SharedAppState) -> Self {
-        Self { state }
+        Self { state, config_load_err: None, hash_to_learn: "".to_owned(), value_to_learn: "".to_owned() }
     }
 }
 
@@ -110,7 +112,7 @@ fn show_race_state(ui: &mut Ui, race: &RaceState) {
         ui.label("Tyres");
         ui.end_row();
         for (i, lap) in race.laps.iter().enumerate() {
-            if let Some(lap_time) = *&lap.lap_time {
+            if let Some(lap_time) = lap.lap_time {
                 let prev_lap = race.laps.get(i - 1);
                 ui.label(format!("#{}", lap.lap.unwrap_or(i + 1)));
 
@@ -211,9 +213,41 @@ impl eframe::App for MyApp {
                 screenshots_sorted.sort_by_key(|(name, _)| name.clone());
                 for (name, image) in screenshots_sorted {
                     ui.label(name);
-                    ui.label(image.img_hash.to_base64());
+                    if ui.button(&image.img_hash).on_hover_text("Copy").clicked() {
+                        ui.output().copied_text = image.img_hash.clone();
+                    }
                     image.image.show_max_size(ui, ui.available_size());
                 }
+
+                if ui.button("Reload config").clicked() {
+                    match Config::load() {
+                        Ok(c) => {
+                            state.config = Arc::new(c);
+                            self.config_load_err = None;
+                        }
+                        Err(e) => {
+                            self.config_load_err = Some(format!("failed to load config: {:?}", e));
+                        }
+                    }
+                }
+                if let Some(e) = &self.config_load_err {
+                    ui.colored_label(Color32::RED, e);
+                }
+
+                ui.separator();
+                ui.label("Hash");
+                ui.text_edit_singleline(&mut self.hash_to_learn);
+                ui.label("Value");
+                ui.text_edit_singleline(&mut self.value_to_learn);
+                if ui.button("Learn").clicked() {
+                    let mut learned_config = (*state.learned).clone();
+                    learned_config.learned_images.insert(self.hash_to_learn.clone(), self.value_to_learn.clone());
+                    learned_config.save().unwrap();
+                    state.learned = Arc::new(learned_config);
+
+                    self.hash_to_learn = "".to_owned();
+                    self.value_to_learn = "".to_owned();
+                }
             });
         }
 

src/ocr.rs

@@ -1,10 +1,16 @@
-use std::{collections::HashMap, sync::{Arc, Mutex}};
+use std::{
+    collections::HashMap,
+    sync::{Arc, Mutex},
+};
 
 use anyhow::Result;
 use image::RgbImage;
 use serde::{Deserialize, Serialize};
 
-use crate::image_processing::{Region, extract_region, filter_to_white};
+use crate::{
+    config::{Config, LearnedConfig},
+    image_processing::{extract_region, filter_to_white, hash_image, Region},
+};
 
 #[derive(Serialize, Deserialize, Debug)]
 pub struct OcrRegion {
@@ -18,46 +24,56 @@ struct OcrResult {
     error: Option<String>,
 }
 
-async fn run_ocr(image: &RgbImage) -> Result<Vec<OcrRegion>> {
+async fn run_ocr(image: &RgbImage, url: &str) -> Result<Option<String>> {
     let client = reqwest::Client::new();
     let response = client
-        .post("http://localhost:3000/")
-        .body(crate::image_processing::to_png_bytes(&image))
+        .post(url)
+        .body(crate::image_processing::to_png_bytes(image))
         .send()
         .await?;
 
     if !response.status().is_success() {
+        eprintln!("failed to run OCR query");
         anyhow::bail!("failed to run OCR query")
     }
     let result: OcrResult = response.json().await?;
-    Ok(result.regions)
+    let result = if result.regions.is_empty() {
+        None
+    } else {
+        let mut buffer = String::new();
+        for r in &result.regions {
+            buffer += &r.value;
+        }
+        Some(buffer)
+    };
+    Ok(result)
 }
 
-pub async fn ocr_all_regions(image: &RgbImage, regions: &[Region]) -> HashMap<String, Option<String>> {
+#[tokio::main(flavor = "current_thread")]
+pub async fn ocr_all_regions(
+    image: &RgbImage,
+    config: Arc<Config>,
+    learned: Arc<LearnedConfig>,
+) -> HashMap<String, Option<String>> {
     let results = Arc::new(Mutex::new(HashMap::new()));
 
     let mut handles = Vec::new();
-    for region  in regions {
+    for region in &config.ocr_regions {
         let filtered_image = extract_region(image, region);
         let region = region.clone();
         let results = results.clone();
+        let config = config.clone();
+        let learned = learned.clone();
         handles.push(tokio::spawn(async move {
             let mut image = filtered_image;
             filter_to_white(&mut image);
-            let ocr_results = run_ocr(&image).await;
-            let value = match ocr_results {
-                Ok(ocr_regions) => {
-                    if ocr_regions.is_empty() {
-                        None
-                    } else {
-                        let mut out = String::new();
-                        for r in ocr_regions {
-                            out += &r.value;
-                        }
-                        Some(out)
-                    }
-                }
-                Err(_) => None
+            let hash = hash_image(&image);
+            let value = if let Some(learned_value) = learned.learned_images.get(&hash) {
+                Some(learned_value.clone())
+            } else {
+                run_ocr(&image, &config.ocr_server_endpoint)
+                    .await
+                    .unwrap_or(None)
             };
             results.lock().unwrap().insert(region.name, value);
         }));

src/state.rs

@@ -73,7 +73,7 @@ pub struct RaceState {
 pub struct DebugOcrFrame {
     pub image: RetainedImage,
     pub rgb_image: RgbImage,
-    pub img_hash: img_hash::ImageHash,
+    pub img_hash: String,
 }
 
 #[derive(Default)]