alpiecode 10.6.9__tar.gz → 10.7.0__tar.gz

{alpiecode-10.6.9 → alpiecode-10.7.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: alpiecode
-Version: 10.6.9
+Version: 10.7.0
 Summary: AlpieCode
 Requires-Python: >=3.9
 Requires-Dist: huggingface_hub>=0.20.0

{alpiecode-10.6.9 → alpiecode-10.7.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "alpiecode"
-version = "10.6.9"
+version = "10.7.0"
 description = "AlpieCode"
 requires-python = ">=3.9"
 dependencies = [

{alpiecode-10.6.9 → alpiecode-10.7.0}/src/alpiecode.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: alpiecode
-Version: 10.6.9
+Version: 10.7.0
 Summary: AlpieCode
 Requires-Python: >=3.9
 Requires-Dist: huggingface_hub>=0.20.0

{alpiecode-10.6.9 → alpiecode-10.7.0}/src/codeagent/backends/local_backend.py

@@ -1007,27 +1007,60 @@ class AndroidBackend(BasePlatformBackend):
     def _get_bridge_url(self) -> str:
         return os.environ.get("ANDROID_LLAMA_BRIDGE", "http://127.0.0.1:8080").rstrip("/")
 
+    def _find_llama_server_bin(self) -> Optional[Path]:
+        custom = os.environ.get("LLAMA_SERVER_PATH") or os.environ.get("LLAMA_BIN")
+        if custom and Path(custom).is_file():
+            return Path(custom)
+        which_bin = shutil.which("llama-server")
+        if which_bin:
+            return Path(which_bin)
+        candidates = [
+            Path("/data/data/com.termux/files/usr/bin/llama-server"),
+            Path.home() / "llama.cpp" / "build" / "bin" / "llama-server",
+            Path.home() / "llama.cpp" / "build" / "bin" / "Release" / "llama-server",
+            Path.home() / ".alpiecode" / "bin" / "llama-server",
+            Path.cwd() / "llama.cpp" / "build" / "bin" / "llama-server",
+        ]
+        for c in candidates:
+            if c.is_file() and os.access(c, os.X_OK):
+                return c
+        return None
+
+    def _find_gguf_model(self) -> Optional[Path]:
+        from ..local_model import get_model_cache_dir
+        candidate_dirs = [
+            get_model_cache_dir(),
+            Path.home() / ".alpiecode" / "models",
+            Path.home() / "models",
+            Path.cwd() / "models",
+            Path.cwd() / ".alpiecode" / "models",
+        ]
+        for d in candidate_dirs:
+            if d.exists():
+                ggufs = list(d.glob("*.gguf"))
+                mains = [g for g in ggufs if "mmproj" not in g.name.lower()]
+                if mains:
+                    mains.sort(key=lambda p: p.stat().st_size)
+                    return mains[0]
+        return None
+
     def _ensure_local_server(self) -> bool:
         """Auto-detect or auto-start local llama-server in background if needed."""
         if self._is_bridge_online():
             return True
-        server_bin = shutil.which("llama-server")
+        server_bin = self._find_llama_server_bin()
         if not server_bin:
+            self._last_server_error = "llama-server binary was not found in PATH or ~/llama.cpp/build/bin/llama-server."
             return False
-        from ..local_model import get_model_cache_dir
-        cache_dir = get_model_cache_dir()
-        if not cache_dir.exists():
+        model_path = self._find_gguf_model()
+        if not model_path:
+            self._last_server_error = "No GGUF model found in ~/.alpiecode/models or ~/models."
             return False
-        ggufs = list(cache_dir.glob("*.gguf"))
-        main_models = [g for g in ggufs if "mmproj" not in g.name.lower()]
-        if not main_models:
-            return False
-        model_path = main_models[0]
-        mmprojs = [g for g in ggufs if "mmproj" in g.name.lower()]
+        mmprojs = list(model_path.parent.glob("*mmproj*.gguf"))
         cmd = [
-            server_bin,
+            str(server_bin),
             "-m", str(model_path),
-            "-c", "4096",
+            "-c", "2048",
             "-t", "4",
             "-np", "1",
             "-b", "512",
@@ -1039,6 +1072,7 @@ class AndroidBackend(BasePlatformBackend):
         if mmprojs:
             cmd += ["--mmproj", str(mmprojs[0])]
         try:
+            print(f"⏳ Starting local llama-server in background ({model_path.name})...")
             subprocess.Popen(
                 cmd,
                 stdout=subprocess.DEVNULL,
@@ -1046,12 +1080,13 @@ class AndroidBackend(BasePlatformBackend):
                 start_new_session=True,
             )
             import time
-            for _ in range(25):
+            for _ in range(35):
                 if self._is_bridge_online():
                     return True
-                time.sleep(0.4)
-        except Exception:
-            pass
+                time.sleep(0.5)
+        except Exception as e:
+            self._last_server_error = f"Failed to launch llama-server: {e}"
+            return False
         return self._is_bridge_online()
 
     @property
@@ -1060,7 +1095,21 @@ class AndroidBackend(BasePlatformBackend):
             import llama_cpp
             return True
         except ImportError:
-            return self._is_bridge_online() or bool(shutil.which("llama-server") or shutil.which("llama-cli"))
+            return self._is_bridge_online() or bool(self._find_llama_server_bin() or shutil.which("llama-cli"))
+
+    def _raise_missing_engine_error(self) -> None:
+        err_detail = getattr(self, "_last_server_error", None)
+        detail_msg = f"\n│  Reason: {err_detail}" if err_detail else ""
+        raise RuntimeError(
+            "\n╭────────────────────────────────────────────────────────────╮\n"
+            "│  Android / Termux Local Inference:                         │\n"
+            "│  Local inference server (http://127.0.0.1:8080) is offline.│"
+            f"{detail_msg}\n"
+            "│                                                            │\n"
+            "│  To start it manually in Termux:                           │\n"
+            "│    llama-server -m ~/.alpiecode/models/*.gguf -c 2048 -t 4 │\n"
+            "╰────────────────────────────────────────────────────────────╯"
+        )
 
     def chat_completion(
         self,
@@ -1079,15 +1128,7 @@ class AndroidBackend(BasePlatformBackend):
         if self._ensure_local_server():
             return self._bridge_chat_completion(messages, tools, temperature, max_tokens, enable_thinking)
 
-        raise RuntimeError(
-            "\n╭────────────────────────────────────────────────────────────╮\n"
-            "│  Android / Termux Native Inference:                        │\n"
-            "│  No local engine is available.                             │\n"
-            "│  To enable local inference on Termux, run:                 │\n"
-            "│    pkg install llama-cpp                                   │\n"
-            "│    alpiecode init                                          │\n"
-            "╰────────────────────────────────────────────────────────────╯"
-        )
+        self._raise_missing_engine_error()
 
     def chat_completion_stream(
         self,
@@ -1108,15 +1149,7 @@ class AndroidBackend(BasePlatformBackend):
             yield from self._bridge_chat_completion_stream(messages, tools, temperature, max_tokens, enable_thinking)
             return
 
-        raise RuntimeError(
-            "\n╭────────────────────────────────────────────────────────────╮\n"
-            "│  Android / Termux Native Inference:                        │\n"
-            "│  No local engine is available.                             │\n"
-            "│  To enable local inference on Termux, run:                 │\n"
-            "│    pkg install llama-cpp                                   │\n"
-            "│    alpiecode init                                          │\n"
-            "╰────────────────────────────────────────────────────────────╯"
-        )
+        self._raise_missing_engine_error()
 
 
 class MacOSBackend(BasePlatformBackend):