ml-dash 0.4.0__py3-none-any.whl → 0.5.1__py3-none-any.whl

ml_dash/run.py

@@ -1,331 +0,0 @@
-"""Experiment class - main API for ML-Logger."""
-
-import json
-import os
-import socket
-import time
-import uuid
-from contextlib import contextmanager
-from datetime import datetime
-from functools import wraps
-from pathlib import Path
-from typing import Any, Callable, Dict, Optional
-
-from .backends.base import StorageBackend
-from .backends.local_backend import LocalBackend
-from .backends.dash_backend import DashBackend
-from .components.parameters import ParameterManager
-from .components.metrics import MetricsLogger
-from .components.files import FileManager
-from .components.logs import LogManager
-
-
-class Experiment:
-    """Main experiment tracking class.
-
-    Represents a single training execution with parameters, metrics, files, and logs.
-
-    Args:
-        namespace: User/team namespace (required)
-        workspace: Project workspace (required)
-        prefix: Experiment path (required)
-        remote: Remote server URL (optional)
-        local_root: Local storage directory (default: ".ml-logger")
-        directory: Directory path for organizing experiments (optional)
-        readme: Searchable description (optional)
-        experiment_id: Server-side experiment ID (optional)
-    """
-
-    def __init__(
-        self,
-        namespace: str,
-        workspace: str,
-        prefix: str,
-        remote: Optional[str] = None,
-        local_root: str = ".ml-logger",
-        directory: Optional[str] = None,
-        readme: Optional[str] = None,
-        experiment_id: Optional[str] = None,
-        tags: Optional[list] = None,
-    ):
-        """Initialize experiment.
-
-        Args:
-            namespace: User/team namespace
-            workspace: Project workspace
-            prefix: Experiment path (used as experiment name)
-            remote: Remote server URL (optional)
-            local_root: Local storage directory
-            directory: Directory path for organizing experiments (e.g., "dir1/dir2")
-            readme: Searchable description
-            experiment_id: Server-side experiment ID
-            tags: Experiment tags
-        """
-        self.namespace = namespace
-        self.workspace = workspace
-        self.prefix = prefix
-        self.remote = remote
-        self.local_root = local_root
-        self.directory = directory
-        self.readme = readme or ""
-        self.experiment_id = experiment_id
-        self.run_id: Optional[str] = None
-        self.charts: Dict[str, Any] = {}
-        self.tags = tags or []
-
-        # Full path: {local_root}/{namespace}/{workspace}/{directory}/{prefix}
-        # If directory is provided, insert it before prefix
-        if directory:
-            self.local_path = f"{namespace}/{workspace}/{directory}/{prefix}"
-        else:
-            self.local_path = f"{namespace}/{workspace}/{prefix}"
-
-        # Initialize backend
-        if remote:
-            # Use remote DashBackend
-            self.backend: StorageBackend = DashBackend(
-                server_url=remote,
-                namespace=namespace,
-                workspace=workspace,
-                experiment_name=prefix,
-                experiment_id=experiment_id,
-                directory=directory,
-            )
-            # Initialize experiment on server
-            try:
-                exp_data = self.backend.initialize_experiment(description=readme, tags=tags)
-                self.experiment_id = exp_data.get("id")
-                print(f"✓ Initialized experiment on remote server: {self.experiment_id}")
-            except Exception as e:
-                print(f"Warning: Failed to initialize experiment on remote server: {e}")
-                # Fall back to local backend
-                self.backend = LocalBackend(local_root)
-        else:
-            # Use local backend
-            self.backend = LocalBackend(local_root)
-
-        # Initialize components
-        self.params = ParameterManager(self.backend, self.local_path)
-        self.metrics = MetricsLogger(self.backend, self.local_path)
-        self.files = FileManager(self.backend, self.local_path)
-        self.logs = LogManager(self.backend, self.local_path)
-
-        # Metadata
-        self._meta_file = f"{self.local_path}/.ml-logger.meta.json"
-        self._status = "created"
-        self._started_at: Optional[float] = None
-        self._completed_at: Optional[float] = None
-        self._hostname = socket.gethostname()
-
-        # Load or create metadata (only for local backend)
-        if not remote:
-            self._load_metadata()
-
-    def _load_metadata(self) -> None:
-        """Load experiment metadata from file."""
-        if self.backend.exists(self._meta_file):
-            try:
-                content = self.backend.read_text(self._meta_file)
-                meta = json.loads(content)
-                self._status = meta.get("status", "created")
-                self._started_at = meta.get("started_at")
-                self._completed_at = meta.get("completed_at")
-                self.readme = meta.get("readme", self.readme)
-                self.charts = meta.get("charts", {})
-            except Exception:
-                pass
-
-    def _save_metadata(self) -> None:
-        """Save experiment metadata to file."""
-        meta = {
-            "namespace": self.namespace,
-            "workspace": self.workspace,
-            "prefix": self.prefix,
-            "remote": self.remote,
-            "experiment_id": self.experiment_id,
-            "readme": self.readme,
-            "charts": self.charts,
-            "status": self._status,
-            "started_at": self._started_at,
-            "completed_at": self._completed_at,
-            "hostname": self._hostname,
-            "updated_at": time.time(),
-        }
-        content = json.dumps(meta, indent=2)
-        self.backend.write_text(self._meta_file, content)
-
-    def run(self, func: Optional[Callable] = None):
-        """Mark experiment as started (supports 3 patterns).
-
-        Pattern 1 - Direct call:
-            experiment.run()
-            # ... training code ...
-            experiment.complete()
-
-        Pattern 2 - Context manager:
-            with experiment.run():
-                # ... training code ...
-
-        Pattern 3 - Decorator:
-            @experiment.run
-            def train():
-                # ... training code ...
-
-        Args:
-            func: Function to wrap (for decorator pattern)
-
-        Returns:
-            Context manager or decorated function
-        """
-        if func is None:
-            # Pattern 1 (direct) or Pattern 2 (context manager)
-            self._status = "running"
-            self._started_at = time.time()
-            # Only save metadata for local backends
-            if not isinstance(self.backend, DashBackend):
-                self._save_metadata()
-            return self._run_context()
-        else:
-            # Pattern 3 (decorator)
-            @wraps(func)
-            def wrapper(*args, **kwargs):
-                with self.run():
-                    return func(*args, **kwargs)
-            return wrapper
-
-    @contextmanager
-    def _run_context(self):
-        """Context manager for run lifecycle."""
-        try:
-            # Create run on remote server if using DashBackend
-            if isinstance(self.backend, DashBackend) and not self.run_id:
-                try:
-                    run_data = self.backend.create_run(name=self.prefix, tags=self.tags)
-                    self.run_id = run_data.get("id")
-                    print(f"✓ Created run on remote server: {self.run_id}")
-                except Exception as e:
-                    print(f"Warning: Failed to create run on remote server: {e}")
-
-            yield self
-            self.complete()
-        except Exception as e:
-            self.fail(str(e))
-            raise
-
-    def complete(self) -> None:
-        """Mark experiment as completed."""
-        self._status = "completed"
-        self._completed_at = time.time()
-
-        # Update run status on remote server
-        if isinstance(self.backend, DashBackend) and self.run_id:
-            try:
-                self.backend.update_run(status="COMPLETED")
-                print(f"✓ Marked run as COMPLETED on remote server")
-            except Exception as e:
-                print(f"Warning: Failed to update run status: {e}")
-
-        # Save metadata locally
-        if not isinstance(self.backend, DashBackend):
-            self._save_metadata()
-
-    def fail(self, error: str) -> None:
-        """Mark experiment as failed.
-
-        Args:
-            error: Error message
-        """
-        self._status = "failed"
-        self._completed_at = time.time()
-
-        # Log error
-        self.logs.error("Experiment failed", error=error)
-
-        # Update run status on remote server
-        if isinstance(self.backend, DashBackend) and self.run_id:
-            try:
-                self.backend.update_run(status="FAILED", metadata={"error": error})
-                print(f"✓ Marked run as FAILED on remote server")
-            except Exception as e:
-                print(f"Warning: Failed to update run status: {e}")
-
-        # Save metadata locally
-        if not isinstance(self.backend, DashBackend):
-            self._save_metadata()
-
-    # Convenience methods for logging
-    def info(self, message: str, **context) -> None:
-        """Log info message (convenience method).
-
-        Args:
-            message: Log message
-            **context: Additional context
-        """
-        self.logs.info(message, **context)
-
-    def error(self, message: str, **context) -> None:
-        """Log error message (convenience method).
-
-        Args:
-            message: Log message
-            **context: Additional context
-        """
-        self.logs.error(message, **context)
-
-    def warning(self, message: str, **context) -> None:
-        """Log warning message (convenience method).
-
-        Args:
-            message: Log message
-            **context: Additional context
-        """
-        self.logs.warning(message, **context)
-
-    def debug(self, message: str, **context) -> None:
-        """Log debug message (convenience method).
-
-        Args:
-            message: Log message
-            **context: Additional context
-        """
-        self.logs.debug(message, **context)
-
-    @classmethod
-    def _auto_configure(cls) -> "Experiment":
-        """Create auto-configured experiment from environment.
-
-        Reads configuration from:
-        - ML_LOGGER_NAMESPACE (default: "default")
-        - ML_LOGGER_WORKSPACE (default: "experiments")
-        - ML_LOGGER_PREFIX (default: timestamp+uuid)
-        - ML_LOGGER_REMOTE (optional)
-
-        Returns:
-            Auto-configured Experiment instance
-        """
-        namespace = os.environ.get("ML_LOGGER_NAMESPACE", "default")
-        workspace = os.environ.get("ML_LOGGER_WORKSPACE", "experiments")
-
-        # Generate default prefix with timestamp + short UUID
-        timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
-        short_id = str(uuid.uuid4())[:8]
-        default_prefix = f"{timestamp}_{short_id}"
-
-        prefix = os.environ.get("ML_LOGGER_PREFIX", default_prefix)
-        remote = os.environ.get("ML_LOGGER_REMOTE")
-
-        return cls(
-            namespace=namespace,
-            workspace=workspace,
-            prefix=prefix,
-            remote=remote,
-        )
-
-    def __repr__(self) -> str:
-        """String representation."""
-        return (
-            f"Experiment(namespace='{self.namespace}', "
-            f"workspace='{self.workspace}', "
-            f"prefix='{self.prefix}', "
-            f"status='{self._status}')"
-        )