General improvements (#4)

* Add assertions to the tests

This way, we can verify they work

* Require mutable references for tesseract

This way, we prevent changing the tesseract object unless you have a
mutable reference to it

* Remove the libc dependency

It isn't required

* Add myself as an author

* Follow Cargo clippy suggestions

* Investigate memory safety

We should free the image after we pass it into Tesseract.
It makes its own copy.
https://tesseract-ocr.github.io/4.0.0/a01625.html#ga0c4c7f05fd58b3665b123232a05545ad

I'm not sure how to free the memory from get_text. Advice or pull
requests around this will be appreciated.

* Link to tesseract's github

* Bump to 0.2.0

We're allowed to make breaking changes in the before 1.0.0 releases.
This is a breaking change because many of the methods now require a
mutable pointer.

Author: ccouzens

Cargo.toml

@@ -1,16 +1,15 @@
 [package]
 name          = "tesseract"
-version       = "0.1.0"
-authors       = ["Kevin Kwok <[email protected]>"]
+version       = "0.2.0"
+authors       = ["Kevin Kwok <[email protected]>", "Chris Couzens <[email protected]>"]
 documentation = "https://docs.rs/tesseract"
 repository    = "https://github.com/antimatter15/tesseract-rs"
 description   = "Higher-level bindings for Tesseract OCR"
 license       = "MIT"
 keywords      = ["tesseract", "OCR", "bindings"]
-
+categories    = ["api-bindings", "multimedia::images"]
 
 [dependencies]
-libc = "0.1"
 leptonica-sys = "0.1.0"
 
 [dependencies.tesseract-sys]

README.md

@@ -1,2 +1,2 @@
 # tesseract-rs
-Rust bindings for Tesseract
+Rust bindings for [Tesseract](https://github.com/tesseract-ocr/tesseract)

src/lib.rs

@@ -1,8 +1,7 @@
 extern crate leptonica_sys;
-extern crate libc;
 extern crate tesseract_sys;
 
-use leptonica_sys::pixRead;
+use leptonica_sys::{pixRead, pixFreeData};
 use std::ffi::CStr;
 use std::ffi::CString;
 use std::ptr;
@@ -18,11 +17,16 @@ pub struct Tesseract {
 
 impl Drop for Tesseract {
     fn drop(&mut self) {
-        println!("Ave Imperator! Nos morituri te salutamus.");
         unsafe { TessBaseAPIDelete(self.raw) }
     }
 }
 
+impl Default for Tesseract {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
 fn cs(string: &str) -> CString {
     // do not call as_ptr yet, since the data will be freed before we return
     CString::new(string).unwrap()
@@ -34,50 +38,62 @@ impl Tesseract {
             raw: unsafe { TessBaseAPICreate() },
         }
     }
-    pub fn set_lang(&self, language: &str) -> i32 {
+    pub fn set_lang(&mut self, language: &str) -> i32 {
         let cs_language = cs(language);
         unsafe { TessBaseAPIInit3(self.raw, ptr::null(), cs_language.as_ptr()) }
     }
-    pub fn set_image(&self, filename: &str) {
+    pub fn set_image(&mut self, filename: &str) {
         let cs_filename = cs(filename);
         unsafe {
             let img = pixRead(cs_filename.as_ptr());
             TessBaseAPISetImage2(self.raw, img);
+            pixFreeData(img);
         }
     }
-    pub fn set_variable(&self, name: &str, value: &str) -> i32 {
+    pub fn set_variable(&mut self, name: &str, value: &str) -> i32 {
         let cs_name = cs(name);
         let cs_value = cs(value);
         unsafe { TessBaseAPISetVariable(self.raw, cs_name.as_ptr(), cs_value.as_ptr()) }
     }
-    pub fn recognize(&self) -> i32 {
+    pub fn recognize(&mut self) -> i32 {
         unsafe { TessBaseAPIRecognize(self.raw, ptr::null_mut()) }
     }
     pub fn get_text(&self) -> &str {
+        // I think this leaks memory
+        // The c++ documentation
+        // https://tesseract-ocr.github.io/4.0.0/a01625.html#ga115ef656f83352ba608b4f0bf9cfa2c4
+        // says memory must be freed "with the delete [] operator".
+        // The c documentation doesn't say much
+        // https://tesseract-ocr.github.io/4.0.0/a00014.html#a624731fab8a0107a6949195f62d63710
+        // But the code directly calls the c++ function.
+        // Rust can't call `delete []`, so I'm not sure how rust is meant to free this string.
         unsafe {
             str::from_utf8(CStr::from_ptr(TessBaseAPIGetUTF8Text(self.raw)).to_bytes()).unwrap()
         }
     }
 }
 
 pub fn ocr(filename: &str, language: &str) -> String {
-    let cube = Tesseract::new();
+    let mut cube = Tesseract::new();
     cube.set_lang(language);
     cube.set_image(filename);
     cube.recognize();
-    return cube.get_text().to_string();
+    cube.get_text().to_string()
 }
 
 #[test]
-fn blah() {
-    ocr("img.png", "eng");
+fn ocr_test() {
+    assert_eq!(
+        ocr("img.png", "eng"),
+        include_str!("../img.txt").to_string()
+    );
 }
 
 #[test]
-fn it_works() {
-    let cube = Tesseract::new();
+fn expanded_test() {
+    let mut cube = Tesseract::new();
     cube.set_lang("eng");
     cube.set_image("img.png");
     cube.recognize();
-    println!("{:?}", cube.get_text());
+    assert_eq!(cube.get_text(), include_str!("../img.txt").to_string())
 }

wrinkle/src/main.rs

@@ -3,7 +3,7 @@ use tesseract::*;
 
 fn main() {
     println!("Hello, world!");
-    let cube = Tesseract::new();
+    let mut cube = Tesseract::new();
     let filename = "img.png";
     let language = "eng";
 	cube.set_lang(language);