app.py

import gradio as gr
from huggingface_hub import InferenceClient
import os
import json
import time
from pathlib import Path
from typing import List, Tuple

# ==================== CONFIGURATION ====================
HF_TOKEN = os.getenv("HF_TOKEN")
if not HF_TOKEN:
    raise ValueError("HF_TOKEN environment variable is required! Add it in Space Settings → Repository secrets.")

# Kimi-K2.6 served via the Novita inference provider on HF
# Note: provider="novita" is set in InferenceClient, so NO :novita suffix here
MODEL_NAME = "moonshotai/Kimi-K2.6"

HISTORY_FILE = "chat_history.json"
MAX_HISTORY_LENGTH = 20  # Keep last 20 messages

# Token limits for response lengths
RESPONSE_LENGTHS = {
    "Short": 512,
    "Medium": 1024,
    "Long": 4096,
}

MAX_FILE_SIZE_MB = 5

# ==================== INFERENCE CLIENT ====================
def create_client():
    """Create and validate the HF Inference Client with graceful fallback."""
    try:
        c = InferenceClient(
            provider="novita",
            api_key=HF_TOKEN,
        )
        print(f"✅ Connected to {MODEL_NAME} via Novita")
        return c
    except Exception as e:
        print(f"⚠️  Client initialization warning: {e}")
        return None

client = create_client()

# ==================== MEMORY MANAGEMENT ====================

def load_history() -> List[dict]:
    """Load conversation history from JSON file."""
    if not os.path.exists(HISTORY_FILE):
        return []
    try:
        with open(HISTORY_FILE, "r", encoding="utf-8") as f:
            history = json.load(f)
        if len(history) > MAX_HISTORY_LENGTH:
            history = history[-MAX_HISTORY_LENGTH:]
            save_history(history)
        return history
    except Exception as e:
        print(f"Error loading history: {e}")
        return []


def save_history(history: List[dict]):
    """Save conversation history to JSON file."""
    try:
        if len(history) > MAX_HISTORY_LENGTH:
            history = history[-MAX_HISTORY_LENGTH:]
        with open(HISTORY_FILE, "w", encoding="utf-8") as f:
            json.dump(history, f, indent=2, ensure_ascii=False)
    except Exception as e:
        print(f"Error saving history: {e}")


def clear_memory() -> str:
    """Clear conversation history file."""
    if os.path.exists(HISTORY_FILE):
        os.remove(HISTORY_FILE)
    return "✅ Memory cleared successfully!"


# ==================== FILE PROCESSING ====================

def extract_text_from_pdf(file_path: str) -> str:
    """Extract text from PDF using PyPDF2."""
    try:
        import PyPDF2
        text = ""
        with open(file_path, "rb") as f:
            reader = PyPDF2.PdfReader(f)
            for page in reader.pages:
                text += page.extract_text() + "\n"
        return text.strip()
    except ImportError:
        return "❌ PyPDF2 not installed. PDF reading is unavailable."
    except Exception as e:
        return f"❌ Error reading PDF: {str(e)}"


def process_uploaded_file(file) -> Tuple[str, str]:
    """Process an uploaded file and return (content_string, status_message)."""
    if file is None:
        return "", ""
    try:
        file_path = file.name
        file_size_mb = os.path.getsize(file_path) / (1024 * 1024)

        if file_size_mb > MAX_FILE_SIZE_MB:
            return "", f"❌ File too large ({file_size_mb:.1f} MB). Max: {MAX_FILE_SIZE_MB} MB"

        file_ext = Path(file_path).suffix.lower()
        file_name = Path(file_path).name

        if file_ext == ".pdf":
            content = extract_text_from_pdf(file_path)
        elif file_ext in [".txt", ".py", ".js", ".ts", ".html", ".css",
                          ".json", ".md", ".java", ".cpp", ".c", ".rs", ".go"]:
            with open(file_path, "r", encoding="utf-8", errors="ignore") as f:
                content = f.read()
        else:
            return "", f"❌ Unsupported file type: {file_ext}"

        # Truncate very long files
        if len(content) > 50_000:
            content = content[:50_000] + "\n\n[... Content truncated due to length ...]"

        formatted = f"📎 **File: {file_name}**\n```\n{content}\n```\n\n"
        return formatted, f"✅ File loaded: {file_name} ({file_size_mb:.2f} MB)"
    except Exception as e:
        return "", f"❌ Error processing file: {str(e)}"


# ==================== CHAT ENGINE ====================

def build_messages(history: List[dict], full_message: str) -> List[dict]:
    """
    Convert Gradio history format into the OpenAI-compatible messages list
    that the Novita/Kimi API expects.
    """
    system_prompt = (
        "You are Kimi, a highly intelligent and helpful AI assistant created by Moonshot AI. "
        "You excel at reasoning, coding, mathematics, and analysis. "
        "Provide clear, structured, and accurate responses."
    )
    messages = [{"role": "system", "content": system_prompt}]

    # Add existing conversation history
    for turn in history:
        if turn.get("role") in ("user", "assistant") and turn.get("content"):
            # Strip out any <thought> blocks from assistant history
            # (Kimi docs: do NOT include reasoning in multi-turn history)
            content = turn["content"]
            messages.append({"role": turn["role"], "content": content})

    # Add current user message
    messages.append({
        "role": "user",
        "content": [{"type": "text", "text": full_message}]
    })
    return messages


def parse_kimi_response(completion) -> Tuple[str, str]:
    """
    Parse the API response.
    Returns (reasoning_text, final_answer_text).
    reasoning_text may be empty if the model didn't return it.
    """
    try:
        msg = completion.choices[0].message
        reasoning = getattr(msg, "reasoning", "") or ""
        content = msg.content or ""
        return reasoning.strip(), content.strip()
    except Exception as e:
        return "", f"⚠️ Error parsing response: {str(e)}"


def chat_engine(
    message: str,
    history: List[dict],
    response_length: str,
    file_content: str = "",
):
    """
    Main chat function. Returns updated Gradio history.
    history is a list of {role, content} dicts (Gradio messages format).
    """
    if not message.strip():
        return history

    if client is None:
        err = (
            "⚠️ Model client not initialized.\n\n"
            "Please check that:\n"
            "1. `HF_TOKEN` is set in your Space Secrets.\n"
            "2. The token has **read** access and Inference API is enabled."
        )
        history.append({"role": "user",    "content": message})
        history.append({"role": "assistant", "content": err})
        return history

    try:
        full_message = (file_content + message) if file_content else message
        messages = build_messages(history, full_message)
        max_tokens = RESPONSE_LENGTHS.get(response_length, 1024)

        # ── API call ────────────────────────────────────────────────────────
        completion = client.chat.completions.create(
            model=MODEL_NAME,
            messages=messages,
            max_tokens=max_tokens,
            temperature=1.0,
            top_p=0.95,
            stream=False,
        )
        # ───────────────────────────────────────────────────────────────────

        reasoning, answer = parse_kimi_response(completion)

        # Format the bot reply: show reasoning in a collapsible block if present
        if reasoning:
            bot_reply = (
                f"<details>\n"
                f"<summary>🧠 <b>Kimi's Reasoning (click to expand)</b></summary>\n\n"
                f"{reasoning}\n\n"
                f"</details>\n\n"
                f"---\n\n{answer}"
            )
        else:
            bot_reply = answer

        # Save to persistent file (clean format, no reasoning blocks)
        persistent = load_history()
        persistent.append({"user": message, "bot": answer})
        save_history(persistent)

        # Update Gradio history
        history.append({"role": "user",    "content": message})
        history.append({"role": "assistant", "content": bot_reply})
        return history

    except Exception as e:
        error_str = str(e)

        # ── Friendly error hints ────────────────────────────────────────────
        if "401" in error_str or "Unauthorized" in error_str:
            hint = (
                "🔑 **401 Unauthorized** — Your HF token is invalid or missing.\n\n"
                "Fix: Go to Space Settings → Secrets → add `HF_TOKEN` with a valid token."
            )
        elif "403" in error_str or "Forbidden" in error_str:
            hint = (
                "🚫 **403 Forbidden** — Your token doesn't have access to this model or provider.\n\n"
                "Fix: Make sure your HF token has `read` permission and you have accepted "
                "the model's license on Hugging Face."
            )
        elif "429" in error_str or "rate limit" in error_str.lower():
            hint = (
                "⏳ **429 Rate Limited** — Too many requests. Please wait 30 seconds and try again.\n\n"
                "This is a Novita provider limit, not a code error."
            )
        elif "503" in error_str or "loading" in error_str.lower():
            hint = (
                "⏳ **503 Model Loading** — The model is warming up on the server.\n\n"
                "Wait 30–60 seconds and resend your message."
            )
        elif "model" in error_str.lower() and "not found" in error_str.lower():
            hint = (
                "❓ **Model Not Found** — The model ID or provider tag may have changed.\n\n"
                f"Current model: `{MODEL_NAME}`\n"
                "Check the HF model page for the latest provider tag."
            )
        else:
            hint = f"⚠️ **Unexpected Error:**\n```\n{error_str}\n```\n\nPlease try again in a moment."

        history.append({"role": "user",    "content": message})
        history.append({"role": "assistant", "content": hint})
        return history


# ==================== GRADIO INTERFACE ====================

def create_interface():
    """Build and return the Gradio Blocks interface."""

    css = """
    /* ── Google Font ── */
    @import url('https://fonts.googleapis.com/css2?family=Inter:wght@400;500;600;700&display=swap');

    /* ── Root / Body ── */
    body, .gradio-container {
        font-family: 'Inter', sans-serif !important;
        background: #0d0f14 !important;
        color: #e2e8f0 !important;
    }

    /* ── Header block ── */
    .kimi-header {
        background: linear-gradient(135deg, #1a1f2e 0%, #0f1623 50%, #1a1035 100%);
        border: 1px solid rgba(139, 92, 246, 0.25);
        border-radius: 16px;
        padding: 28px 36px;
        margin-bottom: 20px;
        text-align: center;
        box-shadow: 0 4px 32px rgba(139, 92, 246, 0.12);
    }
    .kimi-header h1 {
        font-size: 2rem !important;
        font-weight: 700 !important;
        background: linear-gradient(90deg, #a78bfa, #7c3aed, #c4b5fd) !important;
        -webkit-background-clip: text !important;
        -webkit-text-fill-color: transparent !important;
        margin: 0 0 6px 0 !important;
    }
    .kimi-header p {
        color: #94a3b8 !important;
        font-size: 0.9rem !important;
        margin: 0 !important;
    }
    .kimi-header .badge {
        display: inline-block;
        background: rgba(124, 58, 237, 0.18);
        border: 1px solid rgba(139, 92, 246, 0.35);
        color: #c4b5fd;
        border-radius: 999px;
        padding: 3px 12px;
        font-size: 0.78rem;
        margin-top: 10px;
    }

    /* ── Chat bubble overrides ── */
    .message-wrap {
        padding: 6px 0 !important;
    }
    .user .message-bubble-border, .user .message {
        background: linear-gradient(135deg, #3b1d8a, #5b21b6) !important;
        border: none !important;
        color: #f5f3ff !important;
        border-radius: 18px 18px 4px 18px !important;
    }
    .bot .message-bubble-border, .bot .message {
        background: #1e2333 !important;
        border: 1px solid rgba(139, 92, 246, 0.2) !important;
        color: #e2e8f0 !important;
        border-radius: 18px 18px 18px 4px !important;
    }

    /* ── Input textbox ── */
    .input-wrap textarea {
        background: #141824 !important;
        border: 1px solid rgba(139, 92, 246, 0.3) !important;
        border-radius: 12px !important;
        color: #e2e8f0 !important;
        font-size: 0.95rem !important;
        padding: 12px 16px !important;
        resize: none !important;
        transition: border-color 0.2s;
    }
    .input-wrap textarea:focus {
        border-color: rgba(139, 92, 246, 0.7) !important;
        outline: none !important;
        box-shadow: 0 0 0 3px rgba(139, 92, 246, 0.12) !important;
    }

    /* ── Send button ── */
    #send-btn {
        background: linear-gradient(135deg, #7c3aed, #5b21b6) !important;
        border: none !important;
        border-radius: 12px !important;
        color: #fff !important;
        font-weight: 600 !important;
        font-size: 0.95rem !important;
        letter-spacing: 0.02em !important;
        transition: all 0.2s ease !important;
        box-shadow: 0 4px 14px rgba(124, 58, 237, 0.4) !important;
    }
    #send-btn:hover {
        background: linear-gradient(135deg, #6d28d9, #4c1d95) !important;
        box-shadow: 0 6px 20px rgba(124, 58, 237, 0.55) !important;
        transform: translateY(-1px) !important;
    }

    /* ── Settings accordion ── */
    .gr-accordion {
        background: #141824 !important;
        border: 1px solid rgba(139, 92, 246, 0.2) !important;
        border-radius: 12px !important;
        margin-top: 12px !important;
    }
    .gr-accordion .label-wrap {
        color: #a78bfa !important;
        font-weight: 600 !important;
    }

    /* ── Radio buttons ── */
    .gr-radio label {
        background: #1e2333 !important;
        border: 1px solid rgba(139, 92, 246, 0.2) !important;
        border-radius: 8px !important;
        color: #cbd5e1 !important;
        padding: 6px 14px !important;
        transition: all 0.15s;
    }
    .gr-radio label:hover {
        border-color: rgba(139, 92, 246, 0.5) !important;
        color: #e2e8f0 !important;
    }
    .gr-radio input:checked + label {
        background: rgba(124, 58, 237, 0.25) !important;
        border-color: #7c3aed !important;
        color: #c4b5fd !important;
    }

    /* ── Action buttons ── */
    #clear-file-btn {
        background: #1e2333 !important;
        border: 1px solid rgba(139, 92, 246, 0.25) !important;
        color: #94a3b8 !important;
        border-radius: 8px !important;
        font-size: 0.82rem !important;
        transition: all 0.15s;
    }
    #clear-file-btn:hover {
        border-color: rgba(139, 92, 246, 0.5) !important;
        color: #c4b5fd !important;
    }
    #clear-chat-btn {
        background: rgba(239, 68, 68, 0.1) !important;
        border: 1px solid rgba(239, 68, 68, 0.3) !important;
        color: #f87171 !important;
        border-radius: 8px !important;
        font-size: 0.82rem !important;
        transition: all 0.15s;
    }
    #clear-chat-btn:hover {
        background: rgba(239, 68, 68, 0.2) !important;
        border-color: rgba(239, 68, 68, 0.5) !important;
    }

    /* ── Status messages ── */
    .file-status p { color: #4ade80 !important; font-size: 0.82rem !important; }

    /* ── Footer tips ── */
    .tips-block {
        background: #141824;
        border: 1px solid rgba(139, 92, 246, 0.15);
        border-radius: 12px;
        padding: 14px 20px;
        margin-top: 12px;
    }
    .tips-block p, .tips-block li { color: #64748b !important; font-size: 0.82rem !important; }

    /* ── Chatbot container ── */
    .chatbot-wrap {
        border: 1px solid rgba(139, 92, 246, 0.2) !important;
        border-radius: 14px !important;
        background: #0f1219 !important;
        overflow: hidden !important;
    }
    """

    with gr.Blocks(
        title="Kimi K2.6 · AI Reasoning Chatbot",
        css=css,
        theme=gr.themes.Base(
            primary_hue="violet",
            neutral_hue="slate",
            font=gr.themes.GoogleFont("Inter"),
        ).set(
            body_background_fill="#0d0f14",
            body_text_color="#e2e8f0",
            block_background_fill="#141824",
            block_border_color="rgba(139,92,246,0.2)",
            input_background_fill="#141824",
            button_primary_background_fill="linear-gradient(135deg,#7c3aed,#5b21b6)",
            button_primary_text_color="#ffffff",
        ),
    ) as demo:

        # ── Header ─────────────────────────────────────────────────────────
        gr.HTML("""
        <div class="kimi-header">
            <h1>🌙 Kimi K2.6</h1>
            <p>Moonshot AI's 1-Trillion-Parameter Reasoning Model</p>
            <span class="badge">⚡ Powered by Novita Inference API · 256K Context Window</span>
        </div>
        """)

        # ── State ──────────────────────────────────────────────────────────
        file_content_state = gr.State("")

        # ── Chat window ────────────────────────────────────────────────────
        chatbot = gr.Chatbot(
            label="",
            height=520,
            show_label=False,
            type="messages",
            render_markdown=True,
            bubble_full_width=False,
            elem_classes=["chatbot-wrap"],
            avatar_images=(
                None,
                "https://huggingface.co/moonshotai/Kimi-K2.6/resolve/main/figures/kimi-logo.png",
            ),
        )

        # ── Input row ──────────────────────────────────────────────────────
        with gr.Row(equal_height=True):
            msg = gr.Textbox(
                label="",
                placeholder="✦  Ask Kimi anything — coding, math, reasoning, analysis...",
                lines=2,
                max_lines=6,
                scale=8,
                show_label=False,
                elem_classes=["input-wrap"],
                container=False,
            )
            send_btn = gr.Button(
                "Send  🚀",
                variant="primary",
                scale=1,
                min_width=100,
                elem_id="send-btn",
            )

        # ── Settings accordion ─────────────────────────────────────────────
        with gr.Accordion("⚙️  Settings & File Upload", open=False):

            with gr.Row():
                response_length = gr.Radio(
                    choices=["Short", "Medium", "Long"],
                    value="Medium",
                    label="📏  Response Length",
                    info="Short · 512 tok   Medium · 1024 tok   Long · 4096 tok",
                    scale=2,
                )

            gr.HTML("<hr style='border-color:rgba(139,92,246,0.15);margin:12px 0;'>")

            file_upload = gr.File(
                label="📎  Upload File for Analysis  (PDF · Code · Text  —  max 5 MB)",
                file_types=[".txt", ".pdf", ".py", ".js", ".ts",
                            ".html", ".css", ".json", ".md", ".java",
                            ".cpp", ".c", ".rs", ".go"],
                type="filepath",
            )
            file_status = gr.Markdown("", elem_classes=["file-status"])

            with gr.Row():
                clear_file_btn = gr.Button(
                    "🗑️  Clear File",
                    size="sm",
                    variant="secondary",
                    elem_id="clear-file-btn",
                )
                clear_btn = gr.Button(
                    "🧹  Clear Chat",
                    size="sm",
                    variant="stop",
                    elem_id="clear-chat-btn",
                )

            clear_status = gr.Markdown("")

        # ── Tips footer ────────────────────────────────────────────────────
        gr.HTML("""
        <div class="tips-block">
            <b style="color:#7c3aed;">💡 Tips</b>
            <ul style="margin:6px 0 0 18px;padding:0;">
                <li>Upload a PDF or code file, then ask Kimi to summarize, review, or debug it.</li>
                <li>Use <b>Long</b> mode for complex coding or multi-step math problems.</li>
                <li>Kimi's internal <b>Reasoning</b> block (if shown) reveals step-by-step thinking.</li>
                <li>Press <b>Enter</b> or click <b>Send</b> to submit your message.</li>
            </ul>
        </div>
        """)

        # ── Event handler functions ────────────────────────────────────────
        def handle_file_upload(file):
            content, status = process_uploaded_file(file)
            return content, status

        def clear_file():
            return None, "", ""

        def clear_conversation():
            result = clear_memory()
            return [], result

        # ── Wire up events ─────────────────────────────────────────────────
        file_upload.change(
            fn=handle_file_upload,
            inputs=[file_upload],
            outputs=[file_content_state, file_status],
        )

        clear_file_btn.click(
            fn=clear_file,
            outputs=[file_upload, file_content_state, file_status],
        )

        send_btn.click(
            fn=chat_engine,
            inputs=[msg, chatbot, response_length, file_content_state],
            outputs=[chatbot],
            api_name="chat",
        )

        msg.submit(
            fn=chat_engine,
            inputs=[msg, chatbot, response_length, file_content_state],
            outputs=[chatbot],
            api_name="chat",
        )

        clear_btn.click(
            fn=clear_conversation,
            outputs=[chatbot, clear_status],
            api_name="clear_memory",
        )

    return demo


# ==================== LAUNCH ====================

if __name__ == "__main__":
    demo = create_interface()
    demo.launch(
        server_name="0.0.0.0",
        server_port=7860,
        share=False,
    )