experimaestro 1.7.1__tar.gz → 1.8.0rc0__tar.gz

{experimaestro-1.7.1 → experimaestro-1.8.0rc0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: experimaestro
-Version: 1.7.1
+Version: 1.8.0rc0
 Summary: "Experimaestro is a computer science experiment manager"
 License: GPL-3
 Keywords: experiment manager

{experimaestro-1.7.1 → experimaestro-1.8.0rc0}/pyproject.toml

@@ -20,7 +20,7 @@ include = [
     "src/experimaestro/mkdocs/style.css",
     { path="src/experimaestro/server/data/*", format=['sdist', 'wheel']}
 ]
-version = "1.7.1"
+version = "1.8.0-rc0"
 repository = "https://github.com/experimaestro/experimaestro-python"
 documentation = "https://experimaestro-python.readthedocs.io/"
 
@@ -122,7 +122,7 @@ warn_unused_ignores = true
 
 [tool.commitizen]
 name = "cz_conventional_commits"
-version = "1.7.1"
+version = "1.8.0rc0"
 changelog_start_rev = "v1.0.0"
 tag_format = "v$major.$minor.$patch$prerelease"
 # update_changelog_on_bump = true

{experimaestro-1.7.1 → experimaestro-1.8.0rc0}/src/experimaestro/__init__.py

@@ -55,8 +55,9 @@ from .core.context import SerializationContext
 from .core.serializers import SerializationLWTask, PathSerializationLWTask
 from .core.types import Any, SubmitHook
 from .launchers import Launcher
-from .scheduler.workspace import Workspace, RunMode
 from .scheduler import Scheduler, experiment, FailedExperiment
+from .scheduler.workspace import Workspace, RunMode
+from .scheduler.state import get_experiment
 from .notifications import progress, tqdm
 from .checkers import Choices
 from .xpmutils import DirectoryContext

{experimaestro-1.7.1 → experimaestro-1.8.0rc0}/src/experimaestro/core/context.py

@@ -1,22 +1,16 @@
 from contextlib import contextmanager
-from pathlib import Path
+from pathlib import Path, UnsupportedOperation
 import shutil
 from typing import List, Optional, Protocol, Set, Union
 import os
-import sys
-
-has_hardlink_to = sys.version_info.major == 3 and sys.version_info.minor >= 10
 
 
 def shallow_copy(src_path: Path, dest_path: Path):
     """Copy a directory or file, trying to use hard links if possible"""
     if src_path.is_file():
         try:
-            if has_hardlink_to:
-                dest_path.hardlink_to(src_path)
-            else:
-                dest_path.link_to(src_path)
-        except OSError:
+            dest_path.hardlink_to(src_path)
+        except (NotImplementedError, UnsupportedOperation, OSError):
             shutil.copy(src_path, dest_path)
     else:
         if dest_path.exists():

{experimaestro-1.7.1 → experimaestro-1.8.0rc0}/src/experimaestro/core/objects.py

@@ -1079,7 +1079,7 @@ class ConfigInformation:
         if value is None:
             return None
 
-        elif isinstance(value, list):
+        elif isinstance(value, (list, tuple)):
             return [ConfigInformation._outputjsonvalue(el, context) for el in value]
 
         elif isinstance(value, dict):
@@ -1196,7 +1196,7 @@ class ConfigInformation:
         configurations if necessary"""
         if isinstance(value, Config):
             value.__xpm__.__get_objects__(objects, context)
-        elif isinstance(value, list):
+        elif isinstance(value, (list, tuple)):
             for el in value:
                 ConfigInformation.__collect_objects__(el, objects, context)
         elif isinstance(value, dict):

{experimaestro-1.7.1 → experimaestro-1.8.0rc0}/src/experimaestro/core/serialization.py

@@ -1,6 +1,6 @@
 import json
 from pathlib import Path
-from typing import Any, Dict, List, Tuple, Union, TYPE_CHECKING
+from typing import Any, Dict, List, Optional, Tuple, Union, TYPE_CHECKING
 from experimaestro.core.context import (
     SerializationContext,
     SerializedPath,
@@ -41,17 +41,20 @@ def state_dict(context: SerializationContext, obj: Any):
     return {"objects": objects, "data": data}
 
 
-def save(obj: Any, save_directory: Path):
+def save_definition(obj: Any, context: SerializationContext, path: Path):
+    data = state_dict(context, obj)
+    with path.open("wt") as out:
+        json.dump(data, out)
+
+
+def save(obj: Any, save_directory: Optional[Path]):
     """Saves an object into a disk file
 
     :param save_directory: The directory in which the object and its data will
         be saved (by default, the object is saved in "definition.json")
     """
     context = SerializationContext(save_directory=save_directory)
-
-    data = state_dict(context, obj)
-    with (save_directory / "definition.json").open("wt") as out:
-        json.dump(data, out)
+    save_definition(obj, context, save_directory / "definition.json")
 
 
 def get_data_loader(path: Union[str, Path, SerializedPathLoader]):

{experimaestro-1.7.1 → experimaestro-1.8.0rc0}/src/experimaestro/experiments/cli.py

@@ -59,6 +59,7 @@ class ConfigurationLoader:
     def __init__(self):
         self.yamls = []
         self.python_path = set()
+        self.yaml_module_file: None | Path = None
 
     def load(self, yaml_file: Path):
         """Loads a YAML file, and parents one if they exist"""
@@ -73,6 +74,10 @@ class ConfigurationLoader:
             if not path.is_absolute():
                 _data["file"] = str((yaml_file.parent / path).resolve())
 
+        if "module" in _data:
+            # Keeps track of the YAML file where the module was defined
+            self.yaml_module_file = yaml_file
+
         if parent := _data.get("parent", None):
             self.load(yaml_file.parent / parent)
 
@@ -121,7 +126,10 @@ class ConfigurationLoader:
     help="Port for monitoring (can be defined in the settings.yaml file)",
 )
 @click.option(
-    "--file", "xp_file", help="The file containing the main experimental code"
+    "--file",
+    "xp_file",
+    type=Path,
+    help="The file containing the main experimental code",
 )
 @click.option(
     "--module-name", "module_name", help="Module containing the experimental code"
@@ -202,8 +210,6 @@ def experiments_cli(  # noqa: C901
             logging.info(
                 "Using python path: %s", ", ".join(str(s) for s in python_path)
             )
-    else:
-        xp_file = Path(xp_file)
 
     assert (
         module_name or xp_file
@@ -221,6 +227,19 @@ def experiments_cli(  # noqa: C901
     for path in python_path:
         sys.path.append(str(path))
 
+    # --- Adds automatically the experiment module if not found
+    if module_name and conf_loader.yaml_module_file:
+        try:
+            importlib.import_module(module_name)
+        except ModuleNotFoundError:
+            # Try to setup a path
+            path = conf_loader.yaml_module_file.resolve()
+            for _ in range(len(module_name.split("."))):
+                path = path.parent
+
+            logging.info("Appending %s to python path", path)
+            sys.path.append(str(path))
+
     if xp_file:
         if not xp_file.exists() and xp_file.suffix != ".py":
             xp_file = xp_file.with_suffix(".py")
@@ -243,17 +262,20 @@ def experiments_cli(  # noqa: C901
 
     # --- ... and runs it
     if helper is None:
-        raise ValueError(f"Could not find run function in {xp_file}")
+        raise click.ClickException(
+            f"Could not find run function in {xp_file if xp_file else module_name}"
+        )
 
     if not isinstance(helper, ExperimentHelper):
         helper = ExperimentHelper(helper)
 
     parameters = inspect.signature(helper.callable).parameters
     list_parameters = list(parameters.values())
-    assert len(list_parameters) == 2, (
-        "Callable function should only "
-        f"have two arguments (got {len(list_parameters)})"
-    )
+    if len(list_parameters) != 2:
+        raise click.ClickException(
+            f"run in {xp_file if xp_file else module_name} function should only "
+            f"have two arguments (got {len(list_parameters)}), "
+        )
 
     schema = list_parameters[1].annotation
     omegaconf_schema = OmegaConf.structured(schema())
@@ -263,7 +285,7 @@ def experiments_cli(  # noqa: C901
             configuration = OmegaConf.merge(omegaconf_schema, configuration)
         except omegaconf.errors.ConfigKeyError as e:
             cprint(f"Error in configuration:\n\n{e}", "red", file=sys.stderr)
-            sys.exit(1)
+            raise click.ClickException("Error in configuration")
 
     if show:
         print(json.dumps(OmegaConf.to_container(configuration)))  # noqa: T201

{experimaestro-1.7.1 → experimaestro-1.8.0rc0}/src/experimaestro/scheduler/base.py

@@ -1057,6 +1057,15 @@ class experiment:
                 logger.info("Stopping web server")
                 self.server.stop()
 
+        if self.workspace.run_mode == RunMode.NORMAL:
+            # Write the state
+            logging.info("Saving the experiment state")
+            from experimaestro.scheduler.state import ExperimentState
+
+            ExperimentState.save(
+                self.workdir / "state.json", self.scheduler.jobs.values()
+            )
+
     async def update_task_output_count(self, delta: int):
         """Change in the number of task outputs to process"""
         async with self.central.exitCondition:

experimaestro-1.8.0rc0/src/experimaestro/scheduler/state.py

@@ -0,0 +1,75 @@
+from dataclasses import dataclass, field
+import json
+from pathlib import Path
+from typing import Iterable, Optional, Type
+from experimaestro import Task
+
+from experimaestro.core.context import SerializationContext
+from experimaestro.scheduler.base import Job, JobDependency
+from experimaestro.settings import find_workspace
+from experimaestro.core.serialization import from_state_dict, save_definition
+
+
+@dataclass
+class JobInformation:
+    id: str
+    path: Path
+    task: Task
+    tags: dict[str, str]
+    depends_on: list["JobInformation"] = field(default_factory=list)
+
+    def __post_init__(self):
+        self.path = Path(self.path)
+
+
+class ExperimentState:
+    def __init__(self, workdir: Path, name: str):
+        path = workdir / "xp" / name / "state.json"
+        with path.open("rt") as fh:
+            content = json.load(fh)
+
+        self.states: dict[str, JobInformation] = {
+            state_dict["id"]: JobInformation(**state_dict)
+            for state_dict in from_state_dict(content, as_instance=False)
+        }
+
+        for state in self.states.values():
+            state.depends_on = [self.states[key] for key in state.depends_on]
+
+    def get_jobs(self, task_class: Type[Task]) -> list[JobInformation]:
+        if task_class is None:
+            return list(self.data.values())
+
+        tasks = []
+        for job_state in self.states.values():
+            if isinstance(job_state.task, task_class):
+                tasks.append(job_state)
+        return tasks
+
+    @staticmethod
+    def save(path: Path, jobs: Iterable[Job]):
+        save_definition(
+            [
+                {
+                    "id": job.identifier,
+                    "path": str(job.path),
+                    "task": job.config,
+                    "tags": job.config.__xpm__.tags(),
+                    "depends_on": list(
+                        dep.origin.identifier
+                        for dep in job.dependencies
+                        if isinstance(dep, JobDependency)
+                    ),
+                }
+                for job in jobs
+            ],
+            SerializationContext(),
+            path,
+        )
+
+
+def get_experiment(
+    name: str, *, workspace: Optional[str] = None, workdir: Optional[Path] = None
+) -> ExperimentState:
+    ws = find_workspace(workspace=workspace, workdir=workdir)
+    return ExperimentState(ws.path, name)

experimaestro-1.8.0rc0/src/experimaestro/tests/test_experiment.py

@@ -0,0 +1,50 @@
+from experimaestro import Task, Param, get_experiment, tag
+from experimaestro.tests.utils import TemporaryDirectory, TemporaryExperiment
+
+
+class TaskA(Task):
+    def execute(self):
+        pass
+
+
+class TaskB(Task):
+    task_a: Param[TaskA]
+    x: Param[int]
+
+    def execute(self):
+        pass
+
+
+# xp = get_experiment(id="my-xp-1")
+
+# # Returns a list of tasks which were submitted and successful
+# tasks = xp.get_tasks(myxps.evaluation.Evaluation, status=Job.DONE)
+
+# for task in tasks:
+#     # Look at the tags
+#     print(task.tags)
+
+#     # Get some information
+#     print("Task ran in {task.workdir}")
+
+#     # Look at the parent jobs
+#    print(task.depends_on)
+
+#    # Look at the dependant
+#    print(task.dependents)
+
+
+def test_experiment_history():
+    """Test retrieving experiment history"""
+    with TemporaryDirectory() as workdir:
+        with TemporaryExperiment("experiment", workdir=workdir):
+            task_a = TaskA().submit()
+            TaskB(task_a=task_a, x=tag(1)).submit()
+
+    # Look at the experiment
+    xp = get_experiment("experiment", workdir=workdir)
+
+    (task_a_info,) = xp.get_jobs(TaskA)
+    (task_b_info,) = xp.get_jobs(TaskB)
+    assert task_b_info.tags == {"x": 1}
+    assert task_b_info.depends_on == [task_a_info]

{experimaestro-1.7.1 → experimaestro-1.8.0rc0}/src/experimaestro/typingutils.py

@@ -8,14 +8,7 @@ if sys.version_info.major == 3:
     else:
         from typing import _collect_parameters
 
-    if sys.version_info.minor < 9:
-        from typing_extensions import (
-            _AnnotatedAlias as AnnotatedAlias,
-            get_args,
-            get_origin,
-        )
-    else:
-        from typing import _AnnotatedAlias as AnnotatedAlias, get_args, get_origin
+    from typing import _AnnotatedAlias as AnnotatedAlias, get_args, get_origin
 
 GenericAlias = typing._GenericAlias