vec-inf 0.6.0__py3-none-any.whl → 0.7.0__py3-none-any.whl

vec_inf/client/_utils.py

@@ -14,10 +14,11 @@ from typing import Any, Optional, Union, cast
 import requests
 import yaml
 
-from vec_inf.client._client_vars import MODEL_READY_SIGNATURE
+from vec_inf.client._client_vars import MODEL_READY_SIGNATURE, REQUIRED_ARGS
+from vec_inf.client._exceptions import MissingRequiredFieldsError
+from vec_inf.client._slurm_vars import CACHED_CONFIG_DIR
 from vec_inf.client.config import ModelConfig
 from vec_inf.client.models import ModelStatus
-from vec_inf.client.slurm_vars import CACHED_CONFIG
 
 
 def run_bash_command(command: str) -> tuple[str, str]:
@@ -41,9 +42,9 @@ def run_bash_command(command: str) -> tuple[str, str]:
 
 def read_slurm_log(
     slurm_job_name: str,
-    slurm_job_id: int,
+    slurm_job_id: str,
     slurm_log_type: str,
-    log_dir: Optional[Union[str, Path]],
+    log_dir: str,
 ) -> Union[list[str], str, dict[str, str]]:
     """Read the slurm log file.
 
@@ -51,12 +52,12 @@ def read_slurm_log(
     ----------
     slurm_job_name : str
         Name of the SLURM job
-    slurm_job_id : int
+    slurm_job_id : str
         ID of the SLURM job
     slurm_log_type : str
         Type of log file to read ('out', 'err', or 'json')
-    log_dir : Optional[Union[str, Path]]
-        Directory containing log files, if None uses default location
+    log_dir : str
+        Directory containing log files
 
     Returns
     -------
@@ -66,31 +67,11 @@ def read_slurm_log(
         - dict[str, str] for 'json' logs
         - str for error messages if file not found
     """
-    if not log_dir:
-        # Default log directory
-        models_dir = Path.home() / ".vec-inf-logs"
-        # Iterate over all dirs in models_dir, sorted by dir name length in desc order
-        for directory in sorted(
-            [d for d in models_dir.iterdir() if d.is_dir()],
-            key=lambda d: len(d.name),
-            reverse=True,
-        ):
-            if directory.name in slurm_job_name:
-                log_dir = directory
-                break
-    else:
-        log_dir = Path(log_dir)
-
-    # If log_dir is still not set, then didn't find the log dir at default location
-    if not log_dir:
-        return "LOG DIR NOT FOUND"
-
     try:
-        file_path = (
-            log_dir
-            / Path(f"{slurm_job_name}.{slurm_job_id}")
-            / f"{slurm_job_name}.{slurm_job_id}.{slurm_log_type}"
-        )
+        if "+" in slurm_job_id:
+            main_job_id, het_job_id = slurm_job_id.split("+")
+            slurm_job_id = str(int(main_job_id) + int(het_job_id))
+        file_path = Path(log_dir, f"{slurm_job_name}.{slurm_job_id}.{slurm_log_type}")
         if slurm_log_type == "json":
             with file_path.open("r") as file:
                 json_content: dict[str, str] = json.load(file)
@@ -103,7 +84,7 @@ def read_slurm_log(
 
 
 def is_server_running(
-    slurm_job_name: str, slurm_job_id: int, log_dir: Optional[str]
+    slurm_job_name: str, slurm_job_id: str, log_dir: str
 ) -> Union[str, ModelStatus, tuple[ModelStatus, str]]:
     """Check if a model is ready to serve requests.
 
@@ -111,9 +92,9 @@ def is_server_running(
     ----------
     slurm_job_name : str
         Name of the SLURM job
-    slurm_job_id : int
+    slurm_job_id : str
         ID of the SLURM job
-    log_dir : Optional[str]
+    log_dir : str
         Directory containing log files
 
     Returns
@@ -138,16 +119,16 @@ def is_server_running(
     return status
 
 
-def get_base_url(slurm_job_name: str, slurm_job_id: int, log_dir: Optional[str]) -> str:
+def get_base_url(slurm_job_name: str, slurm_job_id: str, log_dir: str) -> str:
     """Get the base URL of a model.
 
     Parameters
     ----------
     slurm_job_name : str
         Name of the SLURM job
-    slurm_job_id : int
+    slurm_job_id : str
         ID of the SLURM job
-    log_dir : Optional[str]
+    log_dir : str
         Directory containing log files
 
     Returns
@@ -164,7 +145,7 @@ def get_base_url(slurm_job_name: str, slurm_job_id: int, log_dir: Optional[str])
 
 
 def model_health_check(
-    slurm_job_name: str, slurm_job_id: int, log_dir: Optional[str]
+    slurm_job_name: str, slurm_job_id: str, log_dir: str
 ) -> tuple[ModelStatus, Union[str, int]]:
     """Check the health of a running model on the cluster.
 
@@ -172,9 +153,9 @@ def model_health_check(
     ----------
     slurm_job_name : str
         Name of the SLURM job
-    slurm_job_id : int
+    slurm_job_id : str
         ID of the SLURM job
-    log_dir : Optional[str]
+    log_dir : str
         Directory containing log files
 
     Returns
@@ -199,12 +180,17 @@ def model_health_check(
         return (ModelStatus.FAILED, str(e))
 
 
-def load_config() -> list[ModelConfig]:
+def load_config(config_path: Optional[str] = None) -> list[ModelConfig]:
     """Load the model configuration.
 
     Loads configuration from default and user-specified paths, merging them
     if both exist. User configuration takes precedence over default values.
 
+    Parameters
+    ----------
+    config_path : Optional[str]
+        Path to the configuration file
+
     Returns
     -------
     list[ModelConfig]
@@ -213,44 +199,80 @@ def load_config() -> list[ModelConfig]:
     Notes
     -----
     Configuration is loaded from:
-    1. Default path: package's config/models.yaml
-    2. User path: specified by VEC_INF_CONFIG environment variable
+    1. User path: specified by config_path
+    2. Default path: package's config/models.yaml or CACHED_CONFIG if it exists
+    3. Environment variable: specified by VEC_INF_CONFIG environment variable
+        and merged with default config
 
     If user configuration exists, it will be merged with default configuration,
     with user values taking precedence for overlapping fields.
     """
+
+    def load_yaml_config(path: Path) -> dict[str, Any]:
+        """Load YAML config with error handling."""
+        try:
+            with path.open() as f:
+                return yaml.safe_load(f) or {}
+        except FileNotFoundError as err:
+            raise FileNotFoundError(f"Could not find config: {path}") from err
+        except yaml.YAMLError as err:
+            raise ValueError(f"Error parsing YAML config at {path}: {err}") from err
+
+    def process_config(config: dict[str, Any]) -> list[ModelConfig]:
+        """Process the config based on the config type."""
+        return [
+            ModelConfig(model_name=name, **model_data)
+            for name, model_data in config.get("models", {}).items()
+        ]
+
+    def resolve_config_path_from_env_var() -> Path | None:
+        """Resolve the config path from the environment variable."""
+        config_dir = os.getenv("VEC_INF_CONFIG_DIR")
+        config_path = os.getenv("VEC_INF_MODEL_CONFIG")
+        if config_path:
+            return Path(config_path)
+        if config_dir:
+            return Path(config_dir, "models.yaml")
+        return None
+
+    def update_config(
+        config: dict[str, Any], user_config: dict[str, Any]
+    ) -> dict[str, Any]:
+        """Update the config with the user config."""
+        for name, data in user_config.get("models", {}).items():
+            if name in config.get("models", {}):
+                config["models"][name].update(data)
+            else:
+                config.setdefault("models", {})[name] = data
+
+        return config
+
+    # 1. If config_path is given, use only that
+    if config_path:
+        config = load_yaml_config(Path(config_path))
+        return process_config(config)
+
+    # 2. Otherwise, load default config
     default_path = (
-        CACHED_CONFIG
-        if CACHED_CONFIG.exists()
+        CACHED_CONFIG_DIR / "models.yaml"
+        if CACHED_CONFIG_DIR.exists()
         else Path(__file__).resolve().parent.parent / "config" / "models.yaml"
     )
+    config = load_yaml_config(default_path)
+
+    # 3. If user config exists, merge it
+    user_path = resolve_config_path_from_env_var()
+    if user_path and user_path.exists():
+        user_config = load_yaml_config(user_path)
+        config = update_config(config, user_config)
+    elif user_path:
+        warnings.warn(
+            f"WARNING: Could not find user config: {str(user_path)}, revert to default config located at {default_path}",
+            UserWarning,
+            stacklevel=2,
+        )
 
-    config: dict[str, Any] = {}
-    with open(default_path) as f:
-        config = yaml.safe_load(f) or {}
-
-    user_path = os.getenv("VEC_INF_CONFIG")
-    if user_path:
-        user_path_obj = Path(user_path)
-        if user_path_obj.exists():
-            with open(user_path_obj) as f:
-                user_config = yaml.safe_load(f) or {}
-                for name, data in user_config.get("models", {}).items():
-                    if name in config.get("models", {}):
-                        config["models"][name].update(data)
-                    else:
-                        config.setdefault("models", {})[name] = data
-        else:
-            warnings.warn(
-                f"WARNING: Could not find user config: {user_path}, revert to default config located at {default_path}",
-                UserWarning,
-                stacklevel=2,
-            )
-
-    return [
-        ModelConfig(model_name=name, **model_data)
-        for name, model_data in config.get("models", {}).items()
-    ]
+    return process_config(config)
 
 
 def parse_launch_output(output: str) -> tuple[str, dict[str, str]]:
@@ -285,3 +307,100 @@ def parse_launch_output(output: str) -> tuple[str, dict[str, str]]:
             config_dict[key.lower().replace(" ", "_")] = value
 
     return slurm_job_id, config_dict
+
+
+def is_power_of_two(n: int) -> bool:
+    """Check if a number is a power of two.
+
+    Parameters
+    ----------
+    n : int
+        The number to check
+    """
+    return n > 0 and (n & (n - 1)) == 0
+
+
+def find_matching_dirs(
+    log_dir: Path,
+    model_family: Optional[str] = None,
+    model_name: Optional[str] = None,
+    job_id: Optional[int] = None,
+    before_job_id: Optional[int] = None,
+) -> list[Path]:
+    """
+    Find log directories based on filtering criteria.
+
+    Parameters
+    ----------
+    log_dir : Path
+        The base directory containing model family directories.
+    model_family : str, optional
+        Filter to only search inside this family.
+    model_name : str, optional
+        Filter to only match model names.
+    job_id : int, optional
+        Filter to only match this exact SLURM job ID.
+    before_job_id : int, optional
+        Filter to only include job IDs less than this value.
+
+    Returns
+    -------
+    list[Path]
+        List of directories that match the criteria and can be deleted.
+    """
+    matched = []
+
+    if not log_dir.exists() or not log_dir.is_dir():
+        raise FileNotFoundError(f"Log directory does not exist: {log_dir}")
+
+    if not model_family and not model_name and not job_id and not before_job_id:
+        return [log_dir]
+
+    for family_dir in log_dir.iterdir():
+        if not family_dir.is_dir():
+            continue
+        if model_family and family_dir.name != model_family:
+            continue
+
+        if model_family and not model_name and not job_id and not before_job_id:
+            return [family_dir]
+
+        for job_dir in family_dir.iterdir():
+            if not job_dir.is_dir():
+                continue
+
+            try:
+                name_part, id_part = job_dir.name.rsplit(".", 1)
+                parsed_id = int(id_part)
+            except ValueError:
+                continue
+
+            if model_name and name_part != model_name:
+                continue
+            if job_id is not None and parsed_id != job_id:
+                continue
+            if before_job_id is not None and parsed_id >= before_job_id:
+                continue
+
+            matched.append(job_dir)
+
+    return matched
+
+
+def check_required_fields(params: dict[str, Any]) -> None:
+    """Check for required fields without default vals and their corresponding env vars.
+
+    Parameters
+    ----------
+    params : dict[str, Any]
+        Dictionary of parameters to check.
+    """
+    for arg in REQUIRED_ARGS:
+        if not params.get(arg):
+            default_value = os.getenv(REQUIRED_ARGS[arg])
+            if default_value:
+                params[arg] = default_value
+            else:
+                raise MissingRequiredFieldsError(
+                    f"{arg} is required, please set it in the command arguments or environment variables"
+                )

vec_inf/client/api.py

@@ -10,8 +10,10 @@ vec_inf.client._helper : Helper classes for model inference server management
 vec_inf.client.models : Data models for API responses
 """
 
+import shutil
 import time
 import warnings
+from pathlib import Path
 from typing import Any, Optional, Union
 
 from vec_inf.client._exceptions import (
@@ -19,14 +21,16 @@ from vec_inf.client._exceptions import (
     SlurmJobError,
 )
 from vec_inf.client._helper import (
+    BatchModelLauncher,
     ModelLauncher,
     ModelRegistry,
     ModelStatusMonitor,
     PerformanceMetricsCollector,
 )
-from vec_inf.client._utils import run_bash_command
+from vec_inf.client._utils import find_matching_dirs, run_bash_command
 from vec_inf.client.config import ModelConfig
 from vec_inf.client.models import (
+    BatchLaunchResponse,
     LaunchOptions,
     LaunchResponse,
     MetricsResponse,
@@ -60,6 +64,9 @@ class VecInfClient:
     wait_until_ready(slurm_job_id, timeout_seconds, poll_interval_seconds, log_dir)
         Wait for a model to become ready
 
+    cleanup_logs(log_dir, model_name, model_family, job_id, dry_run)
+        Remove logs from the log directory.
+
     Examples
     --------
     >>> from vec_inf.api import VecInfClient
@@ -145,17 +152,42 @@ class VecInfClient:
         model_launcher = ModelLauncher(model_name, options_dict)
         return model_launcher.launch()
 
-    def get_status(
-        self, slurm_job_id: int, log_dir: Optional[str] = None
-    ) -> StatusResponse:
+    def batch_launch_models(
+        self,
+        model_names: list[str],
+        batch_config: Optional[str] = None,
+        account: Optional[str] = None,
+        work_dir: Optional[str] = None,
+    ) -> BatchLaunchResponse:
+        """Launch multiple models on the cluster.
+
+        Parameters
+        ----------
+        model_names : list[str]
+            List of model names to launch
+
+        Returns
+        -------
+        BatchLaunchResponse
+            Response containing launch details for each model
+
+        Raises
+        ------
+        ModelConfigurationError
+            If the model configuration is invalid
+        """
+        model_launcher = BatchModelLauncher(
+            model_names, batch_config, account, work_dir
+        )
+        return model_launcher.launch()
+
+    def get_status(self, slurm_job_id: str) -> StatusResponse:
         """Get the status of a running model.
 
         Parameters
         ----------
-        slurm_job_id : int
+        slurm_job_id : str
             The SLURM job ID to check
-        log_dir : str, optional
-            Path to the SLURM log directory. If None, uses default location
 
         Returns
         -------
@@ -167,20 +199,16 @@ class VecInfClient:
             - Base URL (if ready)
             - Error information (if failed)
         """
-        model_status_monitor = ModelStatusMonitor(slurm_job_id, log_dir)
+        model_status_monitor = ModelStatusMonitor(slurm_job_id)
         return model_status_monitor.process_model_status()
 
-    def get_metrics(
-        self, slurm_job_id: int, log_dir: Optional[str] = None
-    ) -> MetricsResponse:
+    def get_metrics(self, slurm_job_id: str) -> MetricsResponse:
         """Get the performance metrics of a running model.
 
         Parameters
         ----------
-        slurm_job_id : int
+        slurm_job_id : str
             The SLURM job ID to get metrics for
-        log_dir : str, optional
-            Path to the SLURM log directory. If None, uses default location
 
         Returns
         -------
@@ -190,9 +218,7 @@ class VecInfClient:
             - Performance metrics or error message
             - Timestamp of collection
         """
-        performance_metrics_collector = PerformanceMetricsCollector(
-            slurm_job_id, log_dir
-        )
+        performance_metrics_collector = PerformanceMetricsCollector(slurm_job_id)
 
         metrics: Union[dict[str, float], str]
         if not performance_metrics_collector.metrics_url.startswith("http"):
@@ -206,12 +232,12 @@ class VecInfClient:
             timestamp=time.time(),
         )
 
-    def shutdown_model(self, slurm_job_id: int) -> bool:
+    def shutdown_model(self, slurm_job_id: str) -> bool:
         """Shutdown a running model.
 
         Parameters
         ----------
-        slurm_job_id : int
+        slurm_job_id : str
             The SLURM job ID to shut down
 
         Returns
@@ -232,23 +258,20 @@ class VecInfClient:
 
     def wait_until_ready(
         self,
-        slurm_job_id: int,
+        slurm_job_id: str,
         timeout_seconds: int = 1800,
         poll_interval_seconds: int = 10,
-        log_dir: Optional[str] = None,
     ) -> StatusResponse:
         """Wait until a model is ready or fails.
 
         Parameters
         ----------
-        slurm_job_id : int
+        slurm_job_id : str
             The SLURM job ID to wait for
         timeout_seconds : int, optional
             Maximum time to wait in seconds, by default 1800 (30 mins)
         poll_interval_seconds : int, optional
             How often to check status in seconds, by default 10
-        log_dir : str, optional
-            Path to the SLURM log directory. If None, uses default location
 
         Returns
         -------
@@ -273,7 +296,7 @@ class VecInfClient:
         start_time = time.time()
 
         while True:
-            status_info = self.get_status(slurm_job_id, log_dir)
+            status_info = self.get_status(slurm_job_id)
 
             if status_info.server_status == ModelStatus.READY:
                 return status_info
@@ -300,3 +323,51 @@ class VecInfClient:
 
             # Wait before checking again
             time.sleep(poll_interval_seconds)
+
+    def cleanup_logs(
+        self,
+        log_dir: Optional[Union[str, Path]] = None,
+        model_family: Optional[str] = None,
+        model_name: Optional[str] = None,
+        job_id: Optional[int] = None,
+        before_job_id: Optional[int] = None,
+        dry_run: bool = False,
+    ) -> list[Path]:
+        """Remove logs from the log directory.
+
+        Parameters
+        ----------
+        log_dir : str or Path, optional
+            Root directory containing log files. Defaults to ~/.vec-inf-logs.
+        model_family : str, optional
+            Only delete logs for this model family.
+        model_name : str, optional
+            Only delete logs for this model name.
+        job_id : int, optional
+            If provided, only match directories with this exact SLURM job ID.
+        before_job_id : int, optional
+            If provided, only delete logs with job ID less than this value.
+        dry_run : bool
+            If True, return matching files without deleting them.
+
+        Returns
+        -------
+        list[Path]
+            List of deleted (or matched if dry_run) log file paths.
+        """
+        log_root = Path(log_dir) if log_dir else Path.home() / ".vec-inf-logs"
+        matched = find_matching_dirs(
+            log_dir=log_root,
+            model_family=model_family,
+            model_name=model_name,
+            job_id=job_id,
+            before_job_id=before_job_id,
+        )
+
+        if dry_run:
+            return matched
+
+        for path in matched:
+            shutil.rmtree(path)
+
+        return matched