Update app.py

app.py

@@ -1,26 +1,41 @@
 import gradio as gr
-import logging
-logging.basicConfig(level=logging.DEBUG)
+from mineru_vl_utils.mineru_client import MinerUClient
+from PIL import Image
+import fitz  # PyMuPDF pour lire les PDFs
+import os
 
-try:
-    from mineru_vl_utils import MinerUClient
-    print("✅ mineru-vl-utils importé")
-except Exception as e:
-    print("❌ Erreur import mineru-vl-utils:", e)
+# Init client
+model_path = "opendatalab/MinerU2.5-2509-1.2B"
+client = MinerUClient(backend="transformers", model_path=model_path)
 
-try:
-    client = MinerUClient(
-        backend="transformers",
-        model_path="opendatalab/MinerU2.5-2509-1.2B"
-    )
-    print("✅ Client MinerU initialisé")
-except Exception as e:
-    print("❌ Erreur init MinerUClient:", e)
+def extract_from_file(file):
+    # Vérifier si PDF ou image
+    ext = os.path.splitext(file.name)[-1].lower()
+    images = []
 
-def test_fn():
-    return "Space lancé avec succès ✅"
+    if ext == ".pdf":
+        doc = fitz.open(file.name)
+        for page in doc:
+            pix = page.get_pixmap()
+            img = Image.frombytes("RGB", [pix.width, pix.height], pix.samples)
+            images.append(img)
+    else:
+        images.append(Image.open(file.name))
 
-demo = gr.Interface(fn=test_fn, inputs=[], outputs="text")
+    results = []
+    for img in images:
+        blocks = client.two_step_extract(img)
+        text_blocks = [b.text for b in blocks if hasattr(b, "text")]
+        results.append("\n".join(text_blocks))
 
-if __name__ == "__main__":
-    demo.launch()
+    return "\n\n--- PAGE ---\n\n".join(results)
+
+demo = gr.Interface(
+    fn=extract_from_file,
+    inputs=gr.File(type="file", label="Upload PDF or Image"),
+    outputs=gr.Textbox(label="Extracted Text", lines=20),
+    title="MinerU2.5 Document Extractor",
+    description="Upload a PDF or Image to extract structured text using MinerU2.5."
+)
+
+demo.launch()