Update app.py

app.py

@@ -1,111 +1,135 @@
 
-"""Mr. Smith-Davinci v3.0 — v1 GPU Space entry point.
-
-v1 (GPU Space):  DEFAULT_BACKEND=hf_local_gguf → runs DavidAU GGUF on GPU
-"""
+"""codeMax — DavidAU coder/chat with file drop-in."""
 
 import os
 import sys
-import traceback
 import subprocess
 
-# ── Runtime install: llama-cpp-python (avoids build timeout on v1) ──
+# ── Runtime install: llama-cpp-python (CPU) ──
 try:
     import llama_cpp
 except ImportError:
-    print("[RUNTIME] Installing llama-cpp-python from pre-built wheel...", flush=True)
+    print("[RUNTIME] Installing llama-cpp-python (CPU)...", flush=True)
     subprocess.check_call([
         sys.executable, "-m", "pip", "install", "--no-cache-dir",
         "--extra-index-url", "https://abetlen.github.io/llama-cpp-python/whl/cpu",
         "llama-cpp-python"
     ])
     import llama_cpp
-    print("[RUNTIME] llama-cpp-python installed successfully", flush=True)
-
-# ── Monkey-patch: fix gradio_client 1.4.0 boolean-schema crash ──
-try:
-    from gradio_client import utils as gc_utils
-    _original_get_type = gc_utils.get_type
-
-    def _patched_get_type(schema):
-        if isinstance(schema, bool):
-            return "boolean" if schema else "null"
-        return _original_get_type(schema)
-
-    gc_utils.get_type = _patched_get_type
-except Exception:
-    pass
-
-# ── Monkey-patch: restore HfFolder for gradio 5.0.0 + hfh 1.x compat ──
-try:
-    import huggingface_hub
-    if not hasattr(huggingface_hub, "HfFolder"):
-        class _HfFolder:
-            path = None
-            @staticmethod
-            def get_token(): return None
-            @staticmethod
-            def save_token(t): pass
-            @staticmethod
-            def delete_token(): pass
-        huggingface_hub.HfFolder = _HfFolder
-except Exception:
-    pass
-
-
-def _is_on_hf_spaces() -> bool:
-    return bool(
-        os.environ.get("SPACE_ID")
-        or os.environ.get("HUGGINGFACE_SPACE_ID")
-        or os.environ.get("HF_SPACE")
-        or os.path.exists("/.hf_spaces")
+    print("[RUNTIME] llama-cpp-python installed", flush=True)
+
+import gradio as gr
+from huggingface_hub import hf_hub_download
+from llama_cpp import Llama
+
+# ── Load DavidAU ──
+REPO_ID = "DavidAU/Qwen3.6-27B-Heretic-Uncensored-FINETUNE-NEO-CODE-Di-IMatrix-MAX-GGUF"
+FILENAME = "Qwen3.6-27B-NEO-CODE-HERE-2T-OT-Q4_K_M.gguf"
+
+print(f"[MODEL] Downloading {FILENAME}...", flush=True)
+model_path = hf_hub_download(repo_id=REPO_ID, filename=FILENAME)
+print(f"[MODEL] Loading...", flush=True)
+
+llm = Llama(
+    model_path=model_path,
+    n_ctx=8192,
+    n_threads=8,
+    verbose=False,
+)
+print("[MODEL] Ready.", flush=True)
+
+
+def build_prompt(message, history, system_prompt, file_content=None):
+    """Build ChatML prompt with optional file content."""
+    parts = []
+    if system_prompt:
+        parts.append(f"<|system|>\n{system_prompt}</s>")
+    if history:
+        for msg in history:
+            role = msg.get("role", "user")
+            content = msg.get("content", "")
+            parts.append(f"<|{role}|>\n{content}</s>")
+    user_msg = message
+    if file_content:
+        user_msg = f"File contents:\n```\n{file_content}\n```\n\nUser: {message}"
+    parts.append(f"<|user|>\n{user_msg}</s>")
+    parts.append("<|assistant|>\n")
+    return "\n".join(parts)
+
+
+def respond(message, history, system_prompt, file_upload):
+    """Generate response from DavidAU."""
+    if not message.strip():
+        return ""
+
+    file_content = None
+    if file_upload is not None:
+        try:
+            if isinstance(file_upload, dict):
+                path = file_upload.get("name", file_upload.get("path", ""))
+            else:
+                path = str(file_upload)
+            with open(path, "r", encoding="utf-8", errors="replace") as f:
+                file_content = f.read()
+        except Exception:
+            pass
+
+    # Convert history to dict format
+    history_dicts = []
+    if history:
+        for h in history:
+            if isinstance(h, (list, tuple)) and len(h) == 2:
+                history_dicts.append({"role": "user", "content": h[0]})
+                history_dicts.append({"role": "assistant", "content": h[1]})
+
+    prompt = build_prompt(message, history_dicts, system_prompt, file_content)
+
+    try:
+        stream = llm(
+            prompt,
+            max_tokens=2048,
+            stop=["</s>", "<|user|>"],
+            stream=True,
+        )
+        response = ""
+        for output in stream:
+            token = output["choices"][0].get("text", "")
+            response += token
+            yield response
+    except Exception as e:
+        yield f"[Error: {e}]"
+
+
+# ── UI ──
+with gr.Blocks(title="codeMax — DavidAU Coder", css="""
+    :root { --bg: #0d1117; --fg: #c9d1d9; --accent: #58a6ff; }
+    body, .gradio-container { background: var(--bg) !important; color: var(--fg) !important; }
+    .gr-button-primary { background: var(--accent) !important; }
+""") as demo:
+    gr.Markdown("# codeMax\n### DavidAU Qwen3.6-27B — Coder + File Drop")
+    with gr.Row():
+        with gr.Column(scale=1):
+            system_box = gr.Textbox(
+                label="System Prompt",
+                value="You are an expert software engineer. Be concise, write clean code.",
+                lines=3,
+            )
+            file_input = gr.File(label="Drop code file here", file_types=[".py", ".js", ".ts", ".tsx", ".json", ".txt", ".md", ".html", ".css"])
+        with gr.Column(scale=3):
+            chatbot = gr.Chatbot(label="Chat", height=480, type="messages")
+            msg_box = gr.Textbox(label="Message", placeholder="Ask me to code something...", lines=2)
+            send_btn = gr.Button("Send", variant="primary")
+
+    send_btn.click(
+        respond,
+        inputs=[msg_box, chatbot, system_box, file_input],
+        outputs=[chatbot],
     )
-
-
-def _get_default_backend() -> str:
-    return "hf_local_gguf"
-
-
-def _get_chairman_space_id() -> str:
-    return os.environ.get("CHAIRMAN_SPACE_ID", "kirikir13/mr-SmithDavinci_Full-Harness_v1")
-
-
-try:
-    print("Importing smith_davinci...", file=sys.stderr)
-    from smith_davinci.engine import SmithEngine
-    from ui.app import create_ui
-
-    print("Creating engine...", file=sys.stderr)
-    engine = SmithEngine()
-
-    from smith_davinci.backends.hf_local_gguf import HFLocalGGUFBackend
-    engine.registry._backends["hf_local_gguf"] = HFLocalGGUFBackend
-    print("[REGISTRY] Force-registered hf_local_gguf backend", file=sys.stderr)
-
-    engine.default_backend = _get_default_backend()
-    engine.chairman_space_id = _get_chairman_space_id()
-    engine.is_hf_spaces = _is_on_hf_spaces()
-
-    print(
-        f"[ENV] Backend: {engine.default_backend} | "
-        f"Chairman Space: {engine.chairman_space_id} | "
-        f"On HF: {engine.is_hf_spaces}",
-        file=sys.stderr,
+    msg_box.submit(
+        respond,
+        inputs=[msg_box, chatbot, system_box, file_input],
+        outputs=[chatbot],
     )
 
-    print("Creating UI...", file=sys.stderr)
-    demo = create_ui(engine)
-
-    print("DEMO CREATED — handing off to runtime", file=sys.stderr)
-except Exception as e:
-    print("FATAL STARTUP ERROR:", repr(e), file=sys.stderr)
-    traceback.print_exc(file=sys.stderr)
-    raise
-
 if __name__ == "__main__":
-    demo.launch(
-        debug=True,
-        share=False,
-        server_name="0.0.0.0",
-        server_port=7860,
-    )
+    demo.launch(server_name="0.0.0.0", server_port=7860)