İçeriğini ocr yapamıyor.

script.js

@@ -114,7 +114,20 @@ document.addEventListener('DOMContentLoaded', function() {
             processBtn.innerHTML = '<i data-feather="cpu" class="mr-2"></i> Process Files';
             feather.replace();
         }
-    });
+        
+        // Load additional Turkish language data
+        function loadTurkishLanguageData() {
+            if (!window.tesseractTurDataLoaded) {
+                Tesseract.addLanguageData('tur', {
+                    data: '/static/tesseract/tur.traineddata.gz'
+                });
+                window.tesseractTurDataLoaded = true;
+            }
+        }
+        
+        loadTurkishLanguageData();
+}
+);
     
     async function processFile(file) {
         const format = outputFormat.value;
@@ -249,9 +262,13 @@ async function extractTextFromImage(file) {
                     logger: m => console.log(m),
                     preserve_interword_spaces: true,
                     tessedit_pageseg_mode: 6, // Assume a single uniform block of text 
-                    tessedit_char_whitelist: 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789.,!?-(){}[]/\\\'" ',
-                    tessedit_create_hocr: 1 // Include formatting info
-                }
+                    tessedit_char_whitelist: 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789.,!?-(){}[]/\\\'" ğüşıöçĞÜŞİÖÇ', // Added Turkish chars
+                    tessedit_create_hocr: 1, // Include formatting info
+                    load_system_dawg: 1,
+                    load_freq_dawg: 1,
+                    user_words_suffix: 'tur',
+                    user_patterns_suffix: 'tur'
+}
             ).then(({ data: { text, hocr } }) => {
                 if (outputFormat.value === 'formatted') {
                     // Process formatted output similar to Adobe/Abbyy
@@ -262,7 +279,6 @@ async function extractTextFromImage(file) {
                 }
             }).catch(reject);
         });
-
         function processFormattedOCR(hocr) {
             // Parse hOCR output to preserve formatting and layout
             const parser = new DOMParser();
@@ -282,24 +298,27 @@ async function extractTextFromImage(file) {
                         const wordConfidence = parseFloat(word.getAttribute('title')
                             .match(/x_wconf (\d+)/)[1]);
                         
-                        // Apply formatting based on confidence and context
-                        if (wordConfidence < 60) {
+                        // Better handling of Turkish characters and confidence
+                        if (wordConfidence < 50) {
                             lineText += `[${wordText}] `;
-                        } else if (wordConfidence < 80 && index > 0) {
+                        } else if (wordConfidence < 70) {
+                            lineText += `<span confidence-medium>${wordText}</span> `;
+                        } else if (wordConfidence < 85 && /[ğüşıöçĞÜŞİÖÇ]/.test(wordText)) {
                             lineText += `${wordText}`;
                         } else {
                             lineText += `${wordText} `;
                         }
                     });
                     
-                    formattedText += lineText.trim() + '\n';
+                    // Better line spacing for Turkish text
+                    formattedText += lineText.trim() + '\n\n';
                 });
                 
                 formattedText += '\n';
             });
             
             return formattedText;
-        }
+}
 }
     
     function displayResult(result) {

style.css

@@ -39,9 +39,21 @@ pre [confidence-low] {
     color: #c62828;
     padding: 0 2px;
 }
-
 pre [confidence-medium] {
     background-color: #fff8e1;
     color: #f57f17;
     padding: 0 2px;
 }
+
+/* Turkish specific OCR styling */
+.turkish-text {
+    font-family: 'Noto Sans', sans-serif;
+    line-height: 1.8;
+    letter-spacing: 0.5px;
+}
+
+.ocr-result {
+    white-space: pre-wrap;
+    font-size: 1.1rem;
+    font-family: 'Segoe UI', Tahoma, sans-serif;
+}