Gemini API

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

# ─── محاكاة Gemini SDK (google.generativeai) ───────────────
GEMINI_MODELS = {
    "gemini-1.5-pro": {
        "ctx": 1_000_000, "in_price": 0.00350, "out_price": 0.01050,
        "capabilities": ["text", "image", "video", "audio", "pdf"],
    },
    "gemini-1.5-flash": {
        "ctx": 1_000_000, "in_price": 0.000075, "out_price": 0.000300,
        "capabilities": ["text", "image", "video"],
    },
    "gemini-2.0-flash": {
        "ctx": 1_000_000, "in_price": 0.0001, "out_price": 0.0004,
        "capabilities": ["text", "image", "video", "audio", "realtime"],
    },
}

@dataclass
class FunctionDeclaration:
    name:        str
    description: str
    parameters:  Dict[str, Any]

@dataclass
class GeminiResponse:
    text:           str
    input_tokens:   int
    output_tokens:  int
    finish_reason:  str = "STOP"
    function_call:  Optional[Dict] = None

class GeminiModel:
    def __init__(self, model_name: str = "gemini-1.5-flash"):
        if model_name not in GEMINI_MODELS:
            raise ValueError(f"النموذج '{model_name}' غير متاح")
        self.model_name  = model_name
        self._info       = GEMINI_MODELS[model_name]
        self._functions: List[FunctionDeclaration] = []
        self._total_cost = 0.0
        self._calls      = 0

    def add_function(self, fn: FunctionDeclaration):
        self._functions.append(fn)

    def generate_content(self, prompt: str, history: List[Dict] = None) -> GeminiResponse:
        self._calls += 1
        in_tok  = int(len(prompt.split()) * 1.3)
        out_tok = 70

        # محاكاة Function Calling
        fn_call = None
        if self._functions and any(kw in prompt.lower() for kw in ["ابحث", "search", "جو", "weather"]):
            fn = random.choice(self._functions)
            fn_call = {"name": fn.name, "args": {"query": prompt[:50]}}
            response_text = f"[Function Call → {fn.name}]"
        else:
            response_text = f"[Gemini/{self.model_name}] ردّي على: {prompt[:55]}..."

        cost = (in_tok/1000)*self._info["in_price"] + (out_tok/1000)*self._info["out_price"]
        self._total_cost += cost

        return GeminiResponse(response_text, in_tok, out_tok, function_call=fn_call)

    def count_tokens(self, text: str) -> int:
        return int(len(text.split()) * 1.3)

    def cost_summary(self):
        cost_s = "$" + f"{self._total_cost:.7f}"
        print(f"\n📊 ملخص Gemini:")
        print(f"   النموذج   : {self.model_name}")
        print(f"   المكالمات : {self._calls}")
        print(f"   التكلفة   : {cost_s}")

# ─── عرض النماذج ───────────────────────────────────────────
print("🌐 Gemini API — نماذج Google AI:")
print("=" * 55)

print("\n📦 النماذج المتاحة:")
print(f"  {'Model':<22} {'Context':>10} {'In/1K':>10} {'Out/1K':>10}")
print("  " + "-"*55)
for name, info in GEMINI_MODELS.items():
    ctx_s = f"{info['ctx']:,}"
    in_s  = "$" + f"{info['in_price']:.5f}"
    out_s = "$" + f"{info['out_price']:.5f}"
    caps  = ", ".join(info["capabilities"][:3])
    print(f"  {name:<22} {ctx_s:>10} {in_s:>10} {out_s:>10}")
    print(f"  {'':22} {caps}")
    print()

# ─── اختبار Gemini 1.5 Flash ──────────────────────────────
print("\n💬 اختبار Gemini 1.5 Flash:")
print("-" * 45)

model = GeminiModel("gemini-1.5-flash")

questions = [
    "ما هو الفرق بين Gemini وClaude؟",
    "اشرح Vertex AI في جملة واحدة",
    "كيف أختار بين Flash وPro؟",
]

for q in questions:
    resp = model.generate_content(q)
    tok_s = f"{resp.input_tokens}in+{resp.output_tokens}out"
    print(f"\n❓ {q}")
    print(f"💡 {resp.text}")
    print(f"   tokens: {tok_s}")

# ─── Function Calling ──────────────────────────────────────
print(f"\n\n⚡ Function Calling (استدعاء دوال خارجية):")
model_pro = GeminiModel("gemini-1.5-pro")

search_fn = FunctionDeclaration(
    name="web_search",
    description="ابحث في الويب عن معلومات محددة",
    parameters={"query": {"type": "string", "description": "نص البحث"}},
)
model_pro.add_function(search_fn)

fn_prompts = [
    "ابحث عن أحدث إصدارات Gemini",
    "ما هو الطقس في دبي اليوم؟",
    "اشرح لي Vertex AI Pipelines",
]

for prompt in fn_prompts:
    resp = model_pro.generate_content(prompt)
    if resp.function_call:
        fn_name = resp.function_call["name"]
        fn_args = json.dumps(resp.function_call["args"], ensure_ascii=False)
        print(f"  🔧 '{prompt[:40]}' → {fn_name}({fn_args})")
    else:
        print(f"  💬 '{prompt[:40]}' → {resp.text[:50]}")

model.cost_summary()
model_pro.cost_summary()
print(f"\n✅ Gemini API جاهز للاستخدام!")