readme/vesseract: Added is_language_code_supported

README.md

@@ -31,16 +31,19 @@ version := vesseract.get_tesseract_version() or { panic(err) }
 // 4.1.0 installed: "Tesseract 4 - 1 - 0 detected!"
 println("Tesseract $version.major - $version.minor - $version.patch detected!")
 
+
 // Get languages supported by Tesseract
 langs := vesseract.get_languages() or { panic(err) }
 // Example: "['afr', 'amh', 'ara', 'asm', ... 'uzb_cyrl', 'vie', 'yid', 'yor']"
 println("$langs")
 
+
 // Get alto xml - Require Tesseract >4.1.0
 alto := vesseract.image_to_alto_xml_path('sample/demo.png') or { panic(err) }
 // "XML: <?xml version="1.0" encoding="UTF-8"?> ... "
 println("XML: $alto")
 
+
 // Get bounding boxes for letters
 boxes := image_to_boxes(image: 'sample/demo.png', lang: 'eng', args: '') or { panic(err) }
 println("$boxes")
@@ -53,6 +56,14 @@ println("$boxes")
     page: 0
 }, ... "
 
+
+// Check if a language model exists in Tesseract
+if (vesseract.is_language_code_supported("fra")) {
+    // Do stuff...
+} else {
+    // Use another model instead...
+}
+
 ```
 # License
 

v.mod

@@ -1,7 +1,7 @@
 Module {
 	name: 'vesseract'
 	description: 'A Tesseract-OCR wrapper for V!'
-	version: '0.0.2'
+	version: '0.0.3'
 	license: 'MIT'
 	dependencies: []
 }

vesseract.v

@@ -56,15 +56,14 @@ pub fn image_to_string(t Tesseract) ?string {
 	return str[..str.len - 2]
 }
 
-// Get installed languages from Tesseract-OCR
-// return a list of languages code
-pub fn get_languages() ?[]string {
-	// Language list
-	mut langs_supported := []string{}
-
+// Generate a map containing all of the languages supported by tesseract
+fn get_language_map() ?map[string]bool {
 	// Get tesseract langs
 	t_result := run_tesseract(['--list-langs']) or { return err }
 
+	// Language list
+	mut langs_supported := map[string]bool{}
+
 	// Split
 	content := t_result.split('\n')
 
@@ -74,13 +73,42 @@ pub fn get_languages() ?[]string {
 
 		// Filter empty lines
 		if line.len > 0 {
-			langs_supported << content[i]
+			langs_supported[content[i]] = true
 		}
 	}
 
 	return langs_supported
 }
 
+// Get installed languages from Tesseract-OCR
+// return a list of languages code
+pub fn get_languages() ?[]string {
+	// Get tesseract langs
+	t_result := run_tesseract(['--list-langs']) or { return err }
+
+	// Get language map
+	lang_map := get_language_map()  or { return err }
+
+	// Language list
+	mut langs_supported := []string{}
+
+	// Skip first line
+	for code, _ in lang_map {
+		langs_supported << code
+	}
+
+	return langs_supported
+}
+
+// Check if a language code is supported
+// No optional as this make the code easier to write
+// Return false on tesseract error (or not available), return true if supported
+pub fn is_language_code_supported(code string) bool {
+	// Get a map of langages
+	map_lang := get_language_map() or { return false }
+	return code in map_lang
+}
+
 // Get tesseract-OCR version
 pub fn get_tesseract_version() ?Tesseract_version {
 	// Get tesseract version

vesseract_test.v

@@ -10,6 +10,11 @@ fn tr_find(list []string, item string) bool {
 	return false
 }
 
+fn test_is_language_code_supported() {
+	assert is_language_code_supported("eng")
+	assert is_language_code_supported("UNKNOWN") == false
+}
+
 fn test_get_languages() {
 	langs := get_languages() or { panic(err) }