Update main.py

main.py

@@ -1,26 +1,28 @@
-import os, io
+import os
 from fastapi import FastAPI, UploadFile, File, Form
 from fastapi.middleware.cors import CORSMiddleware
-from fastapi.responses import JSONResponse, HTMLResponse, FileResponse
+from fastapi.responses import JSONResponse, HTMLResponse
 from fastapi.staticfiles import StaticFiles
 from huggingface_hub import InferenceClient
 from PyPDF2 import PdfReader
 from docx import Document
 from PIL import Image
+import io
 from io import BytesIO
+import requests
+from routers import ai
 
-# -----------------------------------------------------------------------------
-# CONFIGURATION
-# -----------------------------------------------------------------------------
-HUGGINGFACE_TOKEN = os.getenv("HF_TOKEN")                # injected as a secret
-PORT               = int(os.getenv("PORT", 7860))        # HF Spaces provides it
+# Get environment variables
+HUGGINGFACE_TOKEN = os.getenv("HF_TOKEN")
+PORT = int(os.getenv("PORT", 7860))
 
 app = FastAPI(
-    title="AI‑Powered Web‑App API",
-    description="Backend endpoints for summarisation, captioning and QA",
-    version="1.1.0",
+    title="AI Web App API",
+    description="Backend API for AI-powered web application",
+    version="1.0.0"
 )
 
+# Configure CORS
 app.add_middleware(
     CORSMiddleware,
     allow_origins=["*"],
@@ -29,111 +31,156 @@ app.add_middleware(
     allow_headers=["*"],
 )
 
-# Optional: serve static assets from /static (images, css, js)
-app.mount("/static", StaticFiles(directory="static"), name="static")
+# Serve static files
+app.mount("/", StaticFiles(directory=".", html=True), name="static")
 
-# -----------------------------------------------------------------------------
-# MODEL CLIENTS (remote HuggingFace Inference API)
-# -----------------------------------------------------------------------------
-summary_client       = InferenceClient("facebook/bart-large-cnn",            token=HUGGINGFACE_TOKEN)
-qa_client            = InferenceClient("deepset/roberta-base-squad2",        token=HUGGINGFACE_TOKEN)
-image_caption_client = InferenceClient("nlpconnect/vit-gpt2-image-captioning", token=HUGGINGFACE_TOKEN)
+# Include routers
+app.include_router(ai.router)
 
-# -----------------------------------------------------------------------------
-# UTILITY FUNCTIONS
-# -----------------------------------------------------------------------------
+# Initialisation des clients Hugging Face avec authentification
+summary_client = InferenceClient(model="facebook/bart-large-cnn", token=HUGGINGFACE_TOKEN)
+qa_client = InferenceClient(model="deepset/roberta-base-squad2", token=HUGGINGFACE_TOKEN)
+image_caption_client = InferenceClient(model="nlpconnect/vit-gpt2-image-captioning", token=HUGGINGFACE_TOKEN)
 
+# Extraction du texte des fichiers
 def extract_text_from_pdf(content: bytes) -> str:
+    text = ""
     reader = PdfReader(io.BytesIO(content))
-    return "\n".join(page.extract_text() or "" for page in reader.pages).strip()
+    for page in reader.pages:
+        if page.extract_text():
+            text += page.extract_text() + "\n"
+    return text.strip()
 
 def extract_text_from_docx(content: bytes) -> str:
+    text = ""
     doc = Document(io.BytesIO(content))
-    return "\n".join(p.text for p in doc.paragraphs).strip()
+    for para in doc.paragraphs:
+        text += para.text + "\n"
+    return text.strip()
 
 def process_uploaded_file(file: UploadFile) -> str:
-    content   = file.file.read()
-    extension = file.filename.split(".")[-1].lower()
+    content = file.file.read()
+    extension = file.filename.split('.')[-1].lower()
+
     if extension == "pdf":
         return extract_text_from_pdf(content)
-    if extension == "docx":
+    elif extension == "docx":
         return extract_text_from_docx(content)
-    if extension == "txt":
+    elif extension == "txt":
         return content.decode("utf-8").strip()
-    raise ValueError("Unsupported file type")
-
-# -----------------------------------------------------------------------------
-# ROUTES
-# -----------------------------------------------------------------------------
+    else:
+        raise ValueError("Type de fichier non supporté")
 
+# Point d'entrée HTML
 @app.get("/", response_class=HTMLResponse)
-async def serve_index():
-    """Send the frontend HTML."""
-    return FileResponse("index.html")
-
-# ---------- Summarisation -----------------------------------------------------
+async def serve_homepage():
+    with open("index.html", "r", encoding="utf-8") as f:
+        return HTMLResponse(content=f.read(), status_code=200)
 
-@app.post("/api/summarize")
-async def summarize_document(file: UploadFile = File(...)):
+# Résumé
+@app.post("/analyze")
+async def analyze_file(file: UploadFile = File(...)):
     try:
         text = process_uploaded_file(file)
-        if len(text) < 20:
-            return {"result": "Document too short to summarise."}
-        summary_text = summary_client.summarization(text[:3000])
-        return {"result": str(summary_text)}
-    except Exception as exc:
-        return JSONResponse(status_code=500, content={"error": f"Analyse failure: {exc}"})
 
-# ---------- Image Caption -----------------------------------------------------
+        if len(text) < 20:
+            return {"summary": "Document trop court pour être résumé."}
 
-@app.post("/api/caption")
-async def caption_image(file: UploadFile = File(...)):
-    try:
-        image_bytes = await file.read()
-        image_pil   = Image.open(io.BytesIO(image_bytes)).convert("RGB")
-        image_pil.thumbnail((1024, 1024))
-        buf = BytesIO(); image_pil.save(buf, format="JPEG"); img = buf.getvalue()
-        result = image_caption_client.image_to_text(img)
-        if isinstance(result, dict):
-            caption = result.get("generated_text") or result.get("caption") or "No caption found."
-        elif isinstance(result, list):
-            caption = result[0].get("generated_text", "No caption found.")
-        else:
-            caption = str(result)
-        return {"result": str(caption)}
-    except Exception as exc:
-        return JSONResponse(status_code=500, content={"error": f"Caption failure: {exc}"})
+        summary = summary_client.summarization(text[:3000])
+        return {"summary": summary}
 
-# ---------- Question Answering ----------------------------------------------
+    except Exception as e:
+        return JSONResponse(status_code=500, content={"error": f"Erreur lors de l'analyse: {str(e)}"})
 
-@app.post("/api/qa")
-async def question_answering(file: UploadFile = File(...), question: str = Form(...)):
+# Question-Réponse
+@app.post("/ask")
+async def ask_question(file: UploadFile = File(...), question: str = Form(...)):
     try:
-        if file.content_type.startswith("image/"):
+        # Determine if the file is an image
+        content_type = file.content_type
+        if content_type.startswith("image/"):
             image_bytes = await file.read()
-            pil        = Image.open(io.BytesIO(image_bytes)).convert("RGB"); pil.thumbnail((1024, 1024))
-            buf = BytesIO(); pil.save(buf, format="JPEG"); img = buf.getvalue()
-            res = image_caption_client.image_to_text(img)
-            context = res.get("generated_text") if isinstance(res, dict) else str(res)
+            image_pil = Image.open(io.BytesIO(image_bytes)).convert("RGB")
+            image_pil.thumbnail((1024, 1024))
+
+            img_byte_arr = BytesIO()
+            image_pil.save(img_byte_arr, format='JPEG')
+            img_byte_arr = img_byte_arr.getvalue()
+
+            # Generate image description
+            result = image_caption_client.image_to_text(img_byte_arr)
+            if isinstance(result, dict):
+                context = result.get("generated_text") or result.get("caption") or ""
+            elif isinstance(result, list) and len(result) > 0:
+                context = result[0].get("generated_text", "")
+            elif isinstance(result, str):
+                context = result
+            else:
+                context = ""
+
         else:
-            context = process_uploaded_file(file)[:3000]
+            # Not an image, process as document
+            text = process_uploaded_file(file)
+            if len(text) < 20:
+                return {"answer": "Document trop court pour répondre à la question."}
+            context = text[:3000]
+
         if not context:
-            return {"result": "No context – cannot answer."}
-        answer = qa_client.question_answering(question=question, context=context)
-        return {"result": str(answer.get("answer", "No answer found."))}
-    except Exception as exc:
-        return JSONResponse(status_code=500, content={"error": f"QA failure: {exc}"})
+            return {"answer": "Aucune information disponible pour répondre à la question."}
 
-# ---------- Health check ------------------------------------------------------
+        result = qa_client.question_answering(question=question, context=context)
+        return {"answer": result.get("answer", "Aucune réponse trouvée.")}
 
-@app.get("/api/health")
-async def health():
-    return {"status": "healthy", "hf_token_set": bool(HUGGINGFACE_TOKEN)}
+    except Exception as e:
+        return JSONResponse(status_code=500, content={"error": f"Erreur lors de la recherche de réponse: {str(e)}"})
+        
+# Interprétation d'Image
+@app.post("/interpret_image")
+async def interpret_image(image: UploadFile = File(...)):
+    try:
+        # Lire l'image
+        image_bytes = await image.read()
+
+        # Ouvrir l'image avec PIL
+        image_pil = Image.open(io.BytesIO(image_bytes))
+        image_pil = image_pil.convert("RGB")
+        image_pil.thumbnail((1024, 1024))
 
-# -----------------------------------------------------------------------------
-# ENTRYPOINT
-# -----------------------------------------------------------------------------
+        # Convertir en bytes (JPEG)
+        img_byte_arr = BytesIO()
+        image_pil.save(img_byte_arr, format='JPEG')
+        img_byte_arr = img_byte_arr.getvalue()
 
+        # Appeler le modèle
+        result = image_caption_client.image_to_text(img_byte_arr)
+
+        # 🔍 Affichage du résultat brut pour débogage
+        print("Résultat brut du modèle image-to-text:", result)
+
+        # Extraire la description si disponible
+        if isinstance(result, dict):
+            description = result.get("generated_text") or result.get("caption") or "Description non trouvée."
+        elif isinstance(result, list) and len(result) > 0:
+            description = result[0].get("generated_text", "Description non trouvée.")
+        elif isinstance(result, str):
+            description = result
+        else:
+            description = "Description non trouvée."
+
+        return {"description": description}
+
+    except Exception as e:
+        return JSONResponse(status_code=500, content={"error": f"Erreur lors de l'interprétation de l'image: {str(e)}"})
+
+@app.get("/api/health")
+async def health_check():
+    return {
+        "status": "healthy",
+        "version": "1.0.0",
+        "hf_token_set": bool(HUGGINGFACE_TOKEN)
+    }
+
+# Démarrage local
 if __name__ == "__main__":
     import uvicorn
-    uvicorn.run(app, host="0.0.0.0", port=PORT)
+    uvicorn.run(app, host="0.0.0.0", port=PORT)