Update app.py

app.py

@@ -11,12 +11,19 @@ from marker.converters.pdf import PdfConverter
 from marker.models import create_model_dict
 from marker.output import text_from_rendered
 
+import tempfile
+import io
+import re
+import os
+import language_tool_python
+import html
+tool = language_tool_python.LanguageTool('de-DE')
+
 # Docling
 accelerator_options = AcceleratorOptions(
     num_threads=8, device=AcceleratorDevice.CPU
 )
 
-
 pipeline_options = PdfPipelineOptions()
 pipeline_options.accelerator_options = accelerator_options
 pipeline_options.do_ocr = True
@@ -36,6 +43,41 @@ marker_converter = PdfConverter(
     artifact_dict=create_model_dict(),
 )
 
+def check_spelling(text_input: str) -> str:
+    """Prüft den Text mit LanguageTool und markiert Fundstellen in HTML."""
+    if tool is None:
+        return "❌ **Fehler:** LanguageTool konnte nicht geladen werden."
+    text = text_input or ""
+    if not text.strip():
+        return "*Bitte lade eine Datei hoch oder füge Text ein...*"
+    if text.startswith("❌") or text.startswith("Fehler"):
+        return text
+    matches = tool.check(text)
+    if not matches:
+        return "<p style='color: green;'>✅ Keine Fehler gefunden!</p>"
+    # HTML-Aufbau für die Anzeige der Fehler
+    parts = []
+    text_len = len(text)
+    last_idx = text_len
+    for match in reversed(matches):
+        start = match.offset
+        end = match.offset + match.error_length
+        parts.append(html.escape(text[end:last_idx]))
+        word = text[start:end]
+        suggestions = ", ".join(match.replacements[:3]) if match.replacements else "keine Vorschläge"
+        error_html = (
+            f'<span style="color: red; text-decoration: underline;" '
+            f'title="{html.escape(match.message)} (Vorschläge: {html.escape(suggestions)})">'
+            f'{html.escape(word)}</span>'
+        )
+        parts.append(error_html)
+        last_idx = start
+    parts.append(html.escape(text[:last_idx]))
+    html_texte = "".join(reversed(parts)).replace("\n\n", "<br><br>")
+    html_text = html_texte.replace("\n", "<br>")
+    html_result = f"<div><strong>⚠️ {len(matches)} Fehler gefunden</strong><br><br>{html_text}</div>"
+    return html_result
+
 def convert_document(file, method):
     if method == "Docling":
         result = docling_converter.convert(file.name)