alpiecode 0.7.0__tar.gz → 0.7.2__tar.gz

{alpiecode-0.7.0 → alpiecode-0.7.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: alpiecode
-Version: 0.7.0
+Version: 0.7.2
 Summary: AlpieCode — Autonomous AI Coding Agent CLI powered by local 169Pi GGUF VLM
 Requires-Python: >=3.9
 Requires-Dist: huggingface_hub>=0.20.0

{alpiecode-0.7.0 → alpiecode-0.7.2}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "alpiecode"
-version = "0.7.0"
+version = "0.7.2"
 description = "AlpieCode — Autonomous AI Coding Agent CLI powered by local 169Pi GGUF VLM"
 requires-python = ">=3.9"
 dependencies = [

{alpiecode-0.7.0 → alpiecode-0.7.2}/src/alpiecode.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: alpiecode
-Version: 0.7.0
+Version: 0.7.2
 Summary: AlpieCode — Autonomous AI Coding Agent CLI powered by local 169Pi GGUF VLM
 Requires-Python: >=3.9
 Requires-Dist: huggingface_hub>=0.20.0

{alpiecode-0.7.0 → alpiecode-0.7.2}/src/codeagent/config.py

@@ -75,7 +75,7 @@ def get_shared_http_client():
         import httpx
         _SHARED_HTTP_CLIENT = httpx.Client(
             http2=True,
-            timeout=httpx.Timeout(30.0, connect=3.0),
+            timeout=httpx.Timeout(120.0, connect=5.0),
             limits=httpx.Limits(max_keepalive_connections=20, max_connections=50),
         )
     return _SHARED_HTTP_CLIENT

{alpiecode-0.7.0 → alpiecode-0.7.2}/src/codeagent/local_model.py

@@ -7,6 +7,7 @@ Provides an OpenAI-compatible `create_chat_completion` interface.
 """
 
 import os
+import re
 import sys
 from pathlib import Path
 from typing import Any, Dict, List, Optional
@@ -242,30 +243,37 @@ def _ensure_llama_cpp():
     else:
         wheel_url = "https://github.com/abetlen/llama-cpp-python/releases/download/v0.3.34-vulkan/llama_cpp_python-0.3.34-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl"
 
+    install_targets = [wheel_url, "llama-cpp-python"]
     installed = False
     last_error = ""
-    for cmd_base in [["uv", "pip", "install", wheel_url], [sys.executable, "-m", "pip", "install", wheel_url]]:
-        try:
-            res = subprocess.run(cmd_base, capture_output=True, text=True, timeout=120)
-            if res.returncode == 0:
-                installed = True
-                break
-            else:
-                last_error = res.stderr.strip() or res.stdout.strip()
-        except Exception as e:
-            last_error = str(e)
-            continue
+    for target in install_targets:
+        for cmd_base in [
+            ["uv", "pip", "install", target],
+            [sys.executable, "-m", "pip", "install", target],
+        ]:
+            try:
+                res = subprocess.run(cmd_base, capture_output=True, text=True, timeout=180)
+                if res.returncode == 0:
+                    installed = True
+                    break
+                else:
+                    last_error = res.stderr.strip() or res.stdout.strip()
+            except Exception as e:
+                last_error = str(e)
+                continue
+        if installed:
+            break
 
     try:
         from llama_cpp import Llama
         print("✅ Pre-compiled local GGUF engine installed successfully!")
         return Llama
     except ImportError:
-        err_detail = f"\n   Last error: {last_error[:200]}" if last_error else ""
+        err_detail = f"\n   Last error: {last_error[:300]}" if last_error else ""
         raise RuntimeError(
             "\n╭────────────────────────────────────────────────────────────╮\n"
             "│  Failed to auto-install local GGUF engine.               │\n"
-            "│  Internet connection required for first-time setup.      │\n"
+            "│  Please check your internet connection for first setup.  │\n"
             "╰────────────────────────────────────────────────────────────╯"
             f"{err_detail}"
         )
@@ -388,8 +396,6 @@ class LocalModel:
         
         enable_thinking = kwargs.get("enable_thinking", True)
         
-        # When thinking is disabled, inject assistant prefill to skip reasoning tokens
-        # This eliminates ~6s of thinking token generation per turn
         if not enable_thinking:
             msgs = list(messages) + [{"role": "assistant", "content": "<think>\n\n</think>\n\n"}]
         else:
@@ -405,6 +411,15 @@ class LocalModel:
             params["tool_choice"] = kwargs.get("tool_choice", "auto")
 
         result_dict = llm.create_chat_completion(**params)
+
+        # Fix prefill response content so chat history remains valid across multi-turn sessions
+        if not enable_thinking and result_dict.get("choices"):
+            msg_dict = result_dict["choices"][0].get("message", {})
+            raw_content = msg_dict.get("content") or ""
+            # Strip redundant opening/closing think tags output by model after prefill
+            cleaned = re.sub(r"^\s*(?:</think>|<think>)*\s*", "", raw_content)
+            msg_dict["content"] = f"<think>\n\n</think>\n\n{cleaned}"
+
         return _DictWrapper(result_dict)