Update app.py

app.py

@@ -7,7 +7,7 @@ from docling.datamodel.pipeline_options import (
 )
 import spaces
 from docling.datamodel.base_models import InputFormat
-from marker.converters.marker import MarkerConverter
+from marker.converters.pdf import PdfConverter          # ✅ FIXED import
 from marker.models import create_model_dict
 from marker.output import text_from_rendered
 from wordfreq import word_frequency
@@ -23,7 +23,6 @@ import time
 _data = requests.get("https://specialist-it.de/data.json").json()
 meine_orte = _data.get("meine_orte", [])
 
-
 dummy = """
 ___|___|___|___|___|___|___|___|___|___|___|___|__
 ___|__<(-^,^-)=b_|___|___|___|___|___|___|___|___|
@@ -38,7 +37,7 @@ _|___|___|___|__d-(^_^)z_|___|___|___|___|___|___|
                                      DUMMYIMAGE
 """
 
-bad_string ="""<span style="color: red; text-decoration: underline;" title="Möglicher Tippfehler gefunden. (Vorschläge: Page, Lage, Tage)">page</span>"""
+bad_string = """<span style="color: red; text-decoration: underline;" title="Möglicher Tippfehler gefunden. (Vorschläge: Page, Lage, Tage)">page</span>"""
 
 css = """
 #spinner_md.pending::before {
@@ -47,7 +46,7 @@ css = """
     width: 20px;
     height: 20px;
     border: 3px solid #eee;
-    border-top-color: #2563eb; /* Gradio Blau */
+    border-top-color: #2563eb;
     border-radius: 50%;
     animation: spin 1s linear infinite;
     margin-right: 10px;
@@ -63,14 +62,13 @@ css = """
     font-weight: bold;
     color: #2563eb;
     vertical-align: middle;
-
+}
 """
 
 tool = language_tool_python.LanguageTool('de-DE', new_spellings=meine_orte)
 
 def replace_markdown_images(text):
     pattern = r'!\[\]\(_page_\d+_Picture_\d+\.jpeg\)'
-    replacement = '<img src="dummy.svg">'
     return re.sub(pattern, dummy, text)
 
 def remove_hashes(text):
@@ -95,17 +93,13 @@ docling_converter = DocumentConverter(
     }
 )
 
-# Marker
-marker_converter = MarkerConverter(
-    device="cpu", 
-    artifact_dict=create_model_dict()
-)
+# Marker — ✅ FIXED: PdfConverter replaces MarkerConverter
+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."""
-    
     IGNORE_RULES = {'DE_CASE'}
-    
+
     if tool is None:
         return "❌ **Fehler:** LanguageTool konnte nicht geladen werden."
     text = text_input or ""
@@ -113,19 +107,14 @@ def check_spelling(text_input: str) -> str:
         return "*Bitte lade eine Datei hoch oder füge Text ein...*"
     if text.startswith("❌") or text.startswith("Fehler"):
         return text
-    
+
     matches = tool.check(text)
-    
-    # Regel-Filter
     matches = [m for m in matches if m.rule_id not in IGNORE_RULES]
-    
-    # Englische Wörter herausfiltern
     matches = _filter_english_words(text, matches)
-    
+
     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
@@ -151,7 +140,6 @@ def check_spelling(text_input: str) -> str:
 
 
 def _filter_english_words(text: str, matches: list) -> list:
-    """Filtert Treffer heraus, die gültige englische Wörter sind."""
     clean_matches = []
     for m in matches:
         if m.rule_id == 'GERMAN_SPELLER_RULE':
@@ -161,19 +149,16 @@ def _filter_english_words(text: str, matches: list) -> list:
         clean_matches.append(m)
     return clean_matches
 
+
 def convert_document(file):
-    # load json from specialist-it.de/data.json and convert to python list. Name of python list is meine_orte.
     _data = requests.get("https://specialist-it.de/data.json").json()
     meine_orte = _data.get("meine_orte", [])
-    rendered = marker_converter(file.name)
-    # Extrahiert den Text aus dem gerenderten Objekt
+    rendered = marker_converter(file.name)          # ✅ same call, new converter
     text, _, _ = text_from_rendered(rendered)
-    
     return check_spelling(text).replace(bad_string, "")
 
 
 def add_ort(neuer_ort: str) -> str:
-    """Sendet einen neuen Ort per POST an das PHP-Backend."""
     neuer_ort = neuer_ort.strip()
     if not neuer_ort:
         return "⚠️ Bitte einen Ort eingeben."
@@ -195,7 +180,6 @@ def add_ort(neuer_ort: str) -> str:
 
 
 with gr.Blocks() as app:
-    
     gr.Markdown("# Language Tool \n")
     output_text = gr.HTML(label="Ergebnis", elem_id="spinner_md")
     file_input = gr.File(label="PDF hochladen", file_types=[".pdf"])
@@ -225,4 +209,4 @@ with gr.Blocks() as app:
         outputs=[add_status]
     )
 
-app.launch(debug=True, show_error=True, css=css)
+app.launch(debug=True, show_error=True, css=css)