feat: cache index of configs automatically instead of generating

Author: jmtoepperwien

carps/analysis/calc_hypervolume.py

@@ -337,7 +337,7 @@ def apply_calc_hv_low_mem(logs: pd.DataFrame, on_key: str = "trial_value__cost")
 
     def _save_group_to_filesystem(group: pd.DataFrame, input_directory: Path) -> pd.DataFrame:
         group_id = group[run_id].iloc[0].to_list()
-        gdf_fn = input_directory / f"{'_'.join([str(g) for g in group_id]).replace('/','_')}.parquet"
+        gdf_fn = input_directory / f"{'_'.join([str(g) for g in group_id]).replace('/', '_')}.parquet"
         if gdf_fn.is_file():
             return gdf_fn
         gdf_fn.parent.mkdir(parents=True, exist_ok=True)

carps/analysis/gather_data.py

@@ -25,6 +25,7 @@
 from carps.utils.task import Task
 from carps.utils.trials import TrialInfo
 from carps.utils.types import RunStatus
+from carps.utils.index_configs import get_index_config
 
 if TYPE_CHECKING:
     from carps.objective_functions.objective_function import ObjectiveFunction
@@ -404,9 +405,7 @@ def maybe_postadd_task(logs: pd.DataFrame, overwrite: bool = False) -> pd.DataFr
         logger.debug("No task_id in logs. Can't add task info.")
         return logs
     index_fn = Path(__file__).parent.parent / "configs/task/index.csv"
-    if not index_fn.is_file():
-        raise ValueError("ObjectiveFunction ids have not been indexed. Run `python -m carps.utils.index_configs`.")
-    task_index = pd.read_csv(index_fn)
+    task_index = get_index_config(index_fn)
 
     new_logs = []
     for gid, gdf in logs.groupby(by=["task_id", "seed"]):

carps/analysis/generate_report.py

@@ -631,9 +631,7 @@ def plot_performance_per_task(
         _annotate_heatmap(ax0, mesh, {"fontsize": 8}, ".6g", annot_data)
 
         title = get_figure_title(groupers, gid)
-        ax0.set_title(
-            f"Final Performance per Task for {title}\n" "Annotations: Raw Values, Colormap: Normalized Values"
-        )
+        ax0.set_title(f"Final Performance per Task for {title}\nAnnotations: Raw Values, Colormap: Normalized Values")
 
         ax0.set_title(f"Final Performance per Task for {title}")
         ax0.text(

carps/analysis/process_data.py

@@ -10,6 +10,7 @@
 from omegaconf import DictConfig, ListConfig, OmegaConf
 
 from carps.utils.loggingutils import get_logger, setup_logging
+from carps.utils.index_configs import get_index_config
 
 setup_logging()
 logger = get_logger(__file__)
@@ -69,9 +70,7 @@ def maybe_postadd_task(logs: pd.DataFrame) -> pd.DataFrame:
         Logs with task columns.
     """
     index_fn = Path(__file__).parent.parent / "configs/task/index.csv"
-    if not index_fn.is_file():
-        raise ValueError("Task ids have not been indexed. Run `python -m carps.utils.index_configs`.")
-    task_index = pd.read_csv(index_fn)
+    task_index = get_index_config(index_fn)
 
     def load_task_cfg(task_id: str) -> DictConfig:
         config_fn = task_index["config_fn"][task_index["task_id"] == task_id].iloc[0]