CLI using flags

README.md

@@ -304,4 +304,22 @@ python -m carps.utils.check_missing rundir
 
 # Gather logs from files
 python -m carps.analysis.gather_data rundir
+```
+
+# Commands using Flags
+```bash
+# Fill database
+carps --create_cluster_configs --optimizer DUMMY/config --problem DUMMY/config
+
+# Run from database (with local env)
+carps --run_from_db
+
+# Run local with filelogging
+carps --run --optimizer DUMMY/config --problem DUMMY/config --seed 0
+
+# Check missing runs (creates runcommands_missing.sh)
+carps --check_missing
+
+# Gather logs from files
+carps --gather_data
 ```

carps/cli.py

@@ -0,0 +1,113 @@
+"""
+# CLI.
+
+This module defines command-line options using flags.
+
+This includes the entry point for the programs execution.
+"""
+
+import subprocess
+from typing import Any
+
+from absl import app, flags
+
+FLAGS = flags.FLAGS
+flags.DEFINE_boolean(
+    "create_cluster_configs",
+    short_name="c",
+    default=False,
+    help="Create cluster configs in the database. Passing paths to the "
+    "optimizer and problem config to be used is required.",
+)
+flags.DEFINE_boolean(
+    "run_from_db",
+    short_name="db",
+    default=False,
+    help="Start runs for all configs in the database.",
+)
+flags.DEFINE_boolean(
+    "run",
+    short_name="r",
+    default=False,
+    help="Start run for provided optimizer, problem, and seed. Passing paths to "
+    "the optimizer and problem config to be used is required, as well as "
+    "providing a seed.",
+)
+flags.DEFINE_boolean(
+    "reset_experiments",
+    short_name="re",
+    default=False,
+    help="Reset status of experiments in the database.",
+)
+flags.DEFINE_boolean(
+    "check_missing", short_name="cm", default=False, help="Check for missing runs."
+)
+flags.DEFINE_boolean(
+    "gather_data",
+    short_name="gd",
+    default=False,
+    help="Gather logs from the run files.",
+)
+flags.DEFINE_string(
+    "optimizer", short_name="o", default=None, help="Path to optimizer config."
+)
+flags.DEFINE_string(
+    "problem", short_name="p", default=None, help="Path to problem config."
+)
+flags.DEFINE_string(
+    "seed",
+    short_name="s",
+    default=None,
+    help="Seed to be used when running the experiment.",
+)
+flags.DEFINE_string(
+    "rundir",
+    short_name="rd",
+    default="runs",
+    help="Path to the run directory where results and logs are stored.",
+)
+
+
+def main(argv: Any) -> None:
+    """Call the function to execute."""
+    if FLAGS.create_cluster_configs:
+        if FLAGS.optimizer is None or FLAGS.problem is None:
+            print("Please provide optimizer and problem.")
+            return
+        subprocess.call(
+            [
+                "python",
+                "-m",
+                "carps.container.create_cluster_configs",
+                f"+optimizer={FLAGS.optimizer}",
+                f"+problem={FLAGS.problem}",
+            ]
+        )
+    if FLAGS.run_from_db:
+        subprocess.call(["python", "-m", "carps.run_from_db"])
+    if FLAGS.run:
+        if FLAGS.optimizer is None or FLAGS.problem is None or FLAGS.seed is None:
+            print("Please provide optimizer, problem and seed.")
+            return
+        subprocess.call(
+            [
+                "python",
+                "-m",
+                "carps.run",
+                f"+optimizer={FLAGS.optimizer}",
+                f"+problem={FLAGS.problem}",
+                f"seed={FLAGS.seed}",
+            ]
+        )
+    if FLAGS.reset_experiments:
+        subprocess.call(["python", "-m", "carps.utils.database.reset_experiments"])
+    if FLAGS.check_missing:
+        subprocess.call(["python", "-m", "carps.utils.check_missing", FLAGS.rundir])
+    if FLAGS.gather_data:
+        subprocess.call(["python", "-m", "carps.analysis.gather_data", FLAGS.rundir])
+
+
+if __name__ == "__main__":
+    pass
+
+app.run(main)

carps/container/create_cluster_configs.py

@@ -25,22 +25,33 @@ def main(cfg: DictConfig) -> None:
     """
     cfg_dict = OmegaConf.to_container(cfg=cfg, resolve=True)
 
-    experiment_configuration_file_path = cfg.pyexperimenter_configuration_file_path or Path(__file__).parent / "py_experimenter.yaml"
-
-    database_credential_file_path = cfg.database_credential_file_path or Path(__file__).parent / "credentials.yaml"
-    if database_credential_file_path is not None and not database_credential_file_path.exists():
+    experiment_configuration_file_path = (
+        cfg.pyexperimenter_configuration_file_path
+        or Path(__file__).parent / "py_experimenter.yaml"
+    )
+
+    database_credential_file_path = (
+        cfg.database_credential_file_path or Path(__file__).parent / "credentials.yaml"
+    )
+    if (
+        database_credential_file_path is not None
+        and not database_credential_file_path.exists()
+    ):
         database_credential_file_path = None
 
-    experimenter = PyExperimenter(experiment_configuration_file_path=experiment_configuration_file_path,
-                                  name="carps",
-                                  database_credential_file_path=database_credential_file_path,
-                                  log_level=logging.INFO,
-                                  use_ssh_tunnel=OmegaConf.load(experiment_configuration_file_path).PY_EXPERIMENTER.Database.use_ssh_tunnel
-                                  )
+    experimenter = PyExperimenter(
+        experiment_configuration_file_path=experiment_configuration_file_path,
+        name="carps",
+        database_credential_file_path=database_credential_file_path,
+        log_level=logging.INFO,
+        use_ssh_tunnel=OmegaConf.load(
+            experiment_configuration_file_path
+        ).PY_EXPERIMENTER.Database.use_ssh_tunnel,
+    )
 
     cfg_json = OmegaConf.to_container(cfg, resolve=True)
 
-    # This value will always be unique so it 
+    # This value will always be unique so it
     # disables duplicate checking when adding entries to the database.
     # Py_experimenter will add a creation date so the information
     # is not lost.
@@ -54,19 +65,23 @@ def main(cfg: DictConfig) -> None:
     cfg_str = json.dumps(cfg_json)
     cfg_hash = hashlib.sha256(cfg_str.encode()).hexdigest()
 
-    rows = [{
-        "config": cfg_str,
-        "config_hash": cfg_hash,
-        "benchmark_id": cfg_dict["benchmark_id"],
-        "problem_id": cfg_dict["problem_id"],
-        "optimizer_id": cfg_dict["optimizer_id"],
-        "optimizer_container_id": cfg_dict["optimizer_container_id"],
-        "seed": cfg_dict["seed"],
-        "n_trials": cfg_dict["task"]["n_trials"],
-        "time_budget": cfg_dict["task"]["time_budget"],
-    }]
-
-    column_names = list(experimenter.db_connector.database_configuration.keyfields.keys())
+    rows = [
+        {
+            "config": cfg_str,
+            "config_hash": cfg_hash,
+            "benchmark_id": cfg_dict["benchmark_id"],
+            "problem_id": cfg_dict["problem_id"],
+            "optimizer_id": cfg_dict["optimizer_id"],
+            "optimizer_container_id": cfg_dict["optimizer_container_id"],
+            "seed": cfg_dict["seed"],
+            "n_trials": cfg_dict["task"]["n_trials"],
+            "time_budget": cfg_dict["task"]["time_budget"],
+        }
+    ]
+
+    column_names = list(
+        experimenter.db_connector.database_configuration.keyfields.keys()
+    )
 
     exists = False
 
@@ -78,16 +93,22 @@ def main(cfg: DictConfig) -> None:
         for e in existing_rows:
             if e["config_hash"] == cfg_hash:
                 exists = True
-                logger.info("Experiment not added to the database because config hash already exists!")
+                logger.info(
+                    "Experiment not added to the database because config hash already exists!"
+                )
     except DatabaseConnectionError as e:
         if "1146" in e.args[0] or "no such table" in e.args[0]:
-            logger.info("Database empty, will fill.:)")
+            logger.info("Database does not exist and will be created.")
         else:
             raise e
 
     if not exists:
         experimenter.fill_table_with_rows(rows)
-
+        logger.info("Experiment was added to the database.")
+        logger.info(
+            "Next, add more experiments or run all experiments in the database "
+            "by calling 'carps --run_from_db'."
+        )
     return None
 
 

pyproject.toml

@@ -3,8 +3,8 @@ name = "CARP-S"
 version = "0.1.0"
 dependencies = [
   "typing_extensions", # Better typing
-  "tomli", 
-  "pre-commit", 
+  "tomli",
+  "pre-commit",
   "pytest==6.2.4",
   "coverage==4.5.4",
   "ruff",
@@ -45,6 +45,9 @@ classifiers = [
 ]
 license = { file = "LICENSE" }
 
+[project.scripts]
+carps = "carps.cli:main"
+
 [project.optional-dependencies]
 dev = ["carps[doc, tooling, test, examples]"]
 tooling = ["commitizen", "pre-commit", "ruff"]

setup.py

@@ -58,6 +58,9 @@ def read_file(filepath: str) -> str:
         "numpy"
     ],
     extras_require=extras_require,
+    entry_points={
+        "console_scripts": ["carps = carps.cli:main"],
+    },
     test_suite="pytest",
     platforms=["Linux"],
     classifiers=[