Azure OpenAI Service

import json
from dataclasses import dataclass, field
from typing import List, Dict, Optional

# ─── محاكاة Azure OpenAI SDK ───────────────────────────────
@dataclass
class AzureOpenAIDeployment:
    name:       str
    model:      str
    capacity:   int    # 1 capacity unit = 1000 TPM
    mode:       str    = "Standard"  # Standard, Provisioned

MODELS_PRICING = {
    "gpt-4o":              {"in": 0.005,  "out": 0.015,  "ctx": "128K"},
    "gpt-4-turbo":         {"in": 0.01,   "out": 0.03,   "ctx": "128K"},
    "gpt-35-turbo":        {"in": 0.0005, "out": 0.0015, "ctx": "16K"},
    "text-embedding-3-large": {"in": 0.00013, "out": 0, "ctx": "8K"},
}

class AzureOpenAIClient:
    """محاكاة openai.AzureOpenAI SDK"""

    def __init__(self, endpoint: str, api_version: str = "2024-02-01"):
        self.endpoint    = endpoint
        self.api_version = api_version
        self.deployments: Dict[str, AzureOpenAIDeployment] = {}
        self._total_cost = 0.0
        self._calls      = 0

    def create_deployment(self, name: str, model: str, capacity: int = 1):
        d = AzureOpenAIDeployment(name, model, capacity)
        self.deployments[name] = d
        ctx = MODELS_PRICING.get(model, {}).get("ctx", "?")
        print(f"  ✅ Deployment: {name} ({model}, ctx={ctx})")
        return d

    def chat(self, deployment: str, messages: List[Dict],
             system: str = "", temperature: float = 0.7) -> Dict:
        if deployment not in self.deployments:
            raise ValueError(f"Deployment '{deployment}' غير موجود")
        d   = self.deployments[deployment]
        mdl = MODELS_PRICING.get(d.model, {"in": 0.01, "out": 0.03})

        # محاكاة الرد
        last_msg = messages[-1].get("content", "") if messages else ""
        response = f"[{d.model}] ردّي على: {last_msg[:55]}..."

        in_tok  = int(sum(len(m.get("content","").split()) * 1.3 for m in messages))
        out_tok = 60
        cost    = (in_tok / 1000) * mdl["in"] + (out_tok / 1000) * mdl["out"]
        self._total_cost += cost
        self._calls      += 1

        return {
            "content":     response,
            "usage":       {"prompt_tokens": in_tok, "completion_tokens": out_tok},
            "cost_usd":    round(cost, 7),
            "model":       d.model,
            "deployment":  deployment,
        }

    def embed(self, deployment: str, text: str) -> List[float]:
        """توليد Embedding"""
        import math
        # محاكاة embedding 8-dimensional
        embed = [math.sin(i * hash(text) % 100 * 0.1) for i in range(8)]
        mag   = math.sqrt(sum(x**2 for x in embed))
        return [x / mag for x in embed]

    def cost_summary(self):
        total_str = "$" + f"{self._total_cost:.6f}"
        print(f"\n📊 ملخص الاستخدام:")
        print(f"   المكالمات  : {self._calls}")
        print(f"   التكلفة    : {total_str}")

# ─── إعداد Azure OpenAI ────────────────────────────────────
print("🔵 Azure OpenAI Service:")
print("=" * 52)

client = AzureOpenAIClient(
    endpoint="https://my-openai.openai.azure.com/",
    api_version="2024-02-01",
)

print("\n1️⃣  إنشاء Deployments:")
client.create_deployment("gpt4o-prod",   "gpt-4o",       capacity=10)
client.create_deployment("gpt35-fast",   "gpt-35-turbo", capacity=50)
client.create_deployment("embed-large",  "text-embedding-3-large", capacity=5)

# ─── استخدام Chat Completions ──────────────────────────────
print(f"\n2️⃣  اختبار Chat Completions:")
convos = [
    ("gpt4o-prod",  "ما هو Azure OpenAI؟ أجب في جملتين"),
    ("gpt35-fast",  "اعطني مثالاً على use case لـ Azure OpenAI"),
    ("gpt4o-prod",  "كيف أختار بين Standard و Provisioned deployments؟"),
]
for dep, msg in convos:
    resp = client.chat(dep, [{"role": "user", "content": msg}],
                       system="أجب بالعربية باختصار")
    cost_s = "$" + f"{resp['cost_usd']:.7f}"
    print(f"\n  [{dep}] {msg[:50]}")
    print(f"  💬 {resp['content']}")
    print(f"  🔢 {resp['usage']['prompt_tokens']}in+{resp['usage']['completion_tokens']}out | {cost_s}")

# ─── Embeddings ────────────────────────────────────────────
print(f"\n\n3️⃣  Embeddings للبحث الدلالي:")
texts = ["Azure OpenAI للمؤسسات", "AWS Bedrock لـ Amazon", "GCP Vertex AI لـ Google"]
embeddings = {t: client.embed("embed-large", t) for t in texts}

import math
def cosine_sim(a, b):
    dot  = sum(x*y for x,y in zip(a,b))
    norm = math.sqrt(sum(x**2 for x in a)) * math.sqrt(sum(x**2 for x in b))
    return dot / norm if norm > 0 else 0.0

query = "خدمة AI سحابية"
q_emb = client.embed("embed-large", query)
print(f"  الاستعلام: '{query}'")
for text, emb in embeddings.items():
    sim = cosine_sim(q_emb, emb)
    bar = "█" * int(sim * 10)
    print(f"  {sim:.3f} {bar} {text}")

client.cost_summary()
print(f"\n✅ Azure OpenAI Service جاهز!")