import gradio as gr
from huggingface_hub import InferenceClient
from array import array
import os
import re
import time

# Securely retrieve the token from Space secrets
HF_TOKEN = os.getenv("HF_TOKEN")

# Initialize the inference client
client = InferenceClient("Qwen/Qwen2.5-7B-Instruct", token=HF_TOKEN)

class StateController:
    __slots__ = ("_state", "_metric", "_batch", "_reg", "_rendered", "_rom60", "_symbols")

    def __init__(self):
        self._state      = array("B", [0]) * 121
        self._metric     = 60
        self._batch      = 10
        self._reg        = {}
        self._rendered   = self._build_render()
        
        # Pre-compute the 60x60 lookup ROM for O(1) verification
        self._rom60 = tuple(
            tuple((i * j) % 60 for j in range(60)) 
            for i in range(60)
        )
        # 60-character alphanumeric array for cross-platform rendering
        self._symbols = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz01234567"

    def _build_render(self) -> str:
        """Construct the fixed 121-point visualization once during initialization."""
        return "".join(
            " [NODE_120] " if i == 120 else ("<" if i % 10 == 0 else ".")
            for i in range(121)
        )

    def diagnostic(self) -> str:
        """Execute diagnostic sequence and return the formatted status report."""
        for i in range(51):
            self._state[i] = i % self._batch
        
        self._reg.clear()
        self._reg["STATUS"] = "RESOLVED"

        return (
            "Diagnostic sequence initiated.\n\n"
            "Grid initialized: 5 active blocks.\n\n"
            "Rendering 121-point array:\n"
            f"{self._rendered}\n\n"
            "Executing state resolution:\n"
            "System resolved. State array reset to zero."
        )

    def generate_receipt(self, a: int, b: int, c: int) -> str:
        """Calculates the base-60 checksum for any 3-node allocation."""
        rom_val = self._rom60[a % 60][b % 60]
        checksum_index = (rom_val ^ (c % 60)) % 60
        return f"0{self._symbols[checksum_index]}"

    def validate_receipt(self, receipt: str, a: int, b: int, c: int) -> str:
        """Routes through NODE_120 to verify the checksum mathematically."""
        expected_receipt = self.generate_receipt(a, b, c)
        
        if receipt == expected_receipt:
            self._state[120] = 1 # Activate NODE_120
            return f"[NODE_120: ACTIVE] Checksum Validated. Receipt {receipt} matches allocation ({a}, {b}, {c})."
        else:
            self._state[120] = 0 # Deactivate NODE_120
            return f"[NODE_120: INACTIVE] Verification Failed. Expected receipt {expected_receipt}, received {receipt}."

# Global singleton instance for resource reuse
controller = StateController()

SYSTEM_MSG = {
    "role": "system",
    "content": ("You are a logic-focused inference engine. "
                "Utilize strict state-hold memory and parallel integer blocks. "
                "Provide direct, technical, and accurate responses.")
}

def generate_response(message: str, history: list):
    start_time = time.perf_counter()
    msg_lower = message.lower().strip()

    # Hardware diagnostic intercept
    if msg_lower == "run grid diagnostic":
        output = controller.diagnostic()
        elapsed_time = time.perf_counter() - start_time
        yield f"{output}\n\n---\n*Telemetry: Compute Time {elapsed_time:.4f}s | Source: Local Engine*"
        return

    # Deterministic Checksum Intercept (Bypasses AI completely)
    verify_match = re.match(r"verify receipt\s+([a-zA-Z0-9]{2})\s+for\s+(\d+),\s*(\d+),\s*(\d+)", msg_lower, re.IGNORECASE)
    if verify_match:
        receipt = verify_match.group(1)
        a = int(verify_match.group(2))
        b = int(verify_match.group(3))
        c = int(verify_match.group(4))
        
        output = controller.validate_receipt(receipt, a, b, c)
        elapsed_time = time.perf_counter() - start_time
        yield f"{output}\n\n---\n*Telemetry: Compute Time {elapsed_time:.6f}s | Source: Local ROM Math*"
        return

    # Build the message list for inference
    messages = [SYSTEM_MSG]
    for turn in history:
        messages.append({"role": "user", "content": turn[0]})
        messages.append({"role": "assistant", "content": turn[1]})
    messages.append({"role": "user", "content": message})

    try:
        # Stream text generation
        stream = client.chat_completion(
            messages,
            max_tokens=1024,
            stream=True,
            temperature=0.15
        )
        partial_response = ""
        for chunk in stream:
            token = chunk.choices[0].delta.content or ""
            partial_response += token
            yield partial_response
            
        # Post-processing telemetry after the stream completes
        elapsed_time = time.perf_counter() - start_time
        word_count = len(partial_response.split())
        est_speed = word_count / elapsed_time if elapsed_time > 0 else 0
        
        # Append telemetry footer to the final response
        final_output = partial_response + f"\n\n---\n*Telemetry: Compute Time {elapsed_time:.2f}s | Est. Speed: {est_speed:.2f} words/sec | Source: Inference API*"
        yield final_output

    except Exception as exc:
        yield f"System Error: {str(exc)}. Verify API token and permissions."

custom_css = """
body, .gradio-container { background-color: #0b0f19 !important; }
footer {display: none !important}
.message.user { background-color: #1e293b !important; border: 1px solid #3b82f6 !important; }
.message.bot { background-color: #0f172a !important; color: #60a5fa !important; }
"""

with gr.Blocks() as demo:
    gr.Markdown("# Glyph.io Logic Interface")
    gr.ChatInterface(
        fn=generate_response,
        description="Inference layer utilizing state-hold logic and deterministic ROM verification.",
        examples=[
            "Run grid diagnostic",
            "Calculate the integer distribution for 120 units across 3 nodes.",
            "Verify receipt 0e for 60, 30, 30"
        ],
        cache_examples=False
    )

if __name__ == "__main__":
    demo.queue().launch(
        theme=gr.themes.Soft(primary_hue="blue"),
        css=custom_css,
        ssr_mode=False
        
    )