feat: use rusty-tesseract as another backend

Also use OnceLock & LazyLock

swordfish-common/src/lib.rs

@@ -1,3 +1,4 @@
+#![feature(lazy_cell)]
 pub use log;
 pub use tracing::{debug, error, info, trace, warn};
 use tracing_subscriber::{self, fmt, EnvFilter};
@@ -10,6 +11,7 @@ pub fn setup_logger(level: &str) -> Result<(), fern::InitError> {
         .with_level(true)
         .with_target(true)
         .with_thread_ids(false)
+        .with_line_number(true)
         .with_thread_names(false);
     let filter = EnvFilter::builder()
         .from_env()

swordfish-common/src/tesseract.rs

@@ -1,18 +0,0 @@
-pub use leptess::{LepTess, Variable};
-
-pub fn init_tesseract(numeric_only: bool) -> Result<LepTess, String> {
-    let mut lep_tess = match LepTess::new(None, "eng") {
-        Ok(lep_tess) => lep_tess,
-        Err(why) => return Err(format!("Failed to initialize Tesseract: {:?}", why)),
-    };
-    lep_tess.set_variable(Variable::TesseditPagesegMode, "6").unwrap();
-    // Use LSTM only.
-    lep_tess.set_variable(Variable::TesseditOcrEngineMode, "1").unwrap();
-    if numeric_only {
-        match lep_tess.set_variable(Variable::TesseditCharWhitelist, "0123456789") {
-            Ok(_) => (),
-            Err(why) => return Err(format!("Failed to set whitelist: {:?}", why)),
-        };
-    }
-    Ok(lep_tess)
-}

swordfish-common/src/tesseract/libtesseract.rs

@@ -0,0 +1,60 @@
+pub use leptess::{LepTess, Variable};
+use std::{sync::{
+    Arc, Mutex, LazyLock
+}, thread};
+
+static TESSERACT: LazyLock<Arc<Mutex<LepTess>>> = LazyLock::new(|| {
+    let mut lep_tess = match LepTess::new(None, "eng") {
+        Ok(lep_tess) => lep_tess,
+        Err(why) => panic!("{}", format!("Failed to initialize Tesseract: {:?}", why)),
+    };
+    // lep_tess.set_variable(Variable::TesseditPagesegMode, "6").unwrap();
+    // Use LSTM only.
+    lep_tess.set_variable(Variable::TesseditOcrEngineMode, "2").unwrap();
+    Arc::new(Mutex::new(lep_tess))
+});
+
+static mut TESSERACT_VEC: Vec<Arc<Mutex<LepTess>>> = Vec::new();
+
+pub fn get_tesseract(numeric_only: bool) -> Arc<Mutex<LepTess>> {
+    TESSERACT.clone()
+}
+
+pub unsafe fn get_tesseract_from_vec(numeric_only: bool) -> Arc<Mutex<LepTess>> {
+    let lep_tess: Arc<Mutex<LepTess>>;
+    if TESSERACT_VEC.len() == 0 {
+        for _ in 0..3 {
+            let num_only = numeric_only.clone();
+            thread::spawn(move || {
+                let ocr = init_tesseract(num_only).unwrap();
+                TESSERACT_VEC.push(Arc::new(Mutex::new(ocr)));
+            });
+        }
+        lep_tess = Arc::new(Mutex::new(init_tesseract(numeric_only).unwrap()));
+    } 
+    else {
+        lep_tess = TESSERACT_VEC.pop().unwrap();
+        thread::spawn(move || unsafe {
+            let ocr = init_tesseract(numeric_only).unwrap();
+            TESSERACT_VEC.push(Arc::new(Mutex::new(ocr)));
+        });
+    }
+    lep_tess
+}
+
+pub fn init_tesseract(numeric_only: bool) -> Result<LepTess, String> {
+    let mut lep_tess = match LepTess::new(None, "eng") {
+        Ok(lep_tess) => lep_tess,
+        Err(why) => return Err(format!("Failed to initialize Tesseract: {:?}", why)),
+    };
+    lep_tess.set_variable(Variable::TesseditPagesegMode, "6").unwrap();
+    // Use LSTM only.
+    lep_tess.set_variable(Variable::TesseditOcrEngineMode, "1").unwrap();
+    if numeric_only {
+        match lep_tess.set_variable(Variable::TesseditCharWhitelist, "0123456789") {
+            Ok(_) => (),
+            Err(why) => return Err(format!("Failed to set whitelist: {:?}", why)),
+        };
+    }
+    Ok(lep_tess)
+}

swordfish-common/src/tesseract/mod.rs

@@ -0,0 +1,2 @@
+pub mod subprocess;
+pub mod libtesseract;

swordfish-common/src/tesseract/subprocess.rs

@@ -0,0 +1,36 @@
+pub use rusty_tesseract;
+pub use rusty_tesseract::{Args, Image};
+use std::{collections::HashMap, sync::LazyLock};
+
+static TESSERACT_ARGS: LazyLock<Args> = LazyLock::new(|| Args {
+    lang: "eng".to_string(),
+    config_variables: HashMap::new(),
+    psm: Some(6),
+    dpi: None,
+    oem: Some(1),
+});
+
+static TESSERACT_NUMERIC_ARGS: LazyLock<Args> = LazyLock::new(|| Args {
+    lang: "eng".to_string(),
+    config_variables: HashMap::from([(
+        "tessedit_char_whitelist".into(),
+        "0123456789".into(),
+    )]),
+    psm: Some(6),
+    dpi: None,
+    oem: Some(1),
+});
+
+pub fn image_to_string(image: &Image) -> Result<String, String> {
+    match rusty_tesseract::image_to_string(image, &TESSERACT_ARGS) {
+        Ok(text) => Ok(text),
+        Err(why) => Err(format!("Failed to OCR image: {:?}", why)),
+    }
+}
+
+pub fn image_to_numeric_string(image: &Image) -> Result<String, String> {
+    match rusty_tesseract::image_to_string(image, &TESSERACT_NUMERIC_ARGS) {
+        Ok(text) => Ok(text),
+        Err(why) => Err(format!("Failed to OCR image: {:?}", why)),
+    }
+}