Create HIPS-style properties file

benchmarks/benchmarks.py

@@ -9,9 +9,8 @@
 
 import hats.pixel_math as hist
 import hats.pixel_math.healpix_shim as hp
-from hats.catalog import Catalog, PartitionInfo
+from hats.catalog import Catalog, PartitionInfo, TableProperties
 from hats.catalog.association_catalog.partition_join_info import PartitionJoinInfo
-from hats.catalog.catalog_info import CatalogInfo
 from hats.io.paths import pixel_catalog_files
 from hats.pixel_math import HealpixPixel
 from hats.pixel_tree import PixelAlignment, align_trees
@@ -29,7 +28,7 @@ def time_test_alignment_even_sky():
 
 def time_test_cone_filter_multiple_order():
     """Create a catalog cone filter where we have multiple orders in the catalog"""
-    catalog_info = CatalogInfo(
+    catalog_info = TableProperties(
         **{
             "catalog_name": "test_name",
             "catalog_type": "object",

docs/guide/directory_scheme.rst

@@ -22,8 +22,8 @@ structure:
     __ /path/to/catalogs/<catalog_name>/
        |__ _common_metadata
        |__ _metadata
-       |__ catalog_info.json
        |__ partition_info.csv
+       |__ properties
        |__ Norder=1/
        |   |__ Dir=0/
        |       |__ Npix=0.parquet

pyproject.toml

@@ -22,12 +22,14 @@ dependencies = [
     "astropy",
     "fsspec>=2023.10.0", # Used for abstract filesystems
     "healpy",
+    "jproperties",
     "matplotlib>=3.3,<3.9",
     "mocpy",
     "numba>=0.58",
     "numpy<3", 
     "pandas",
     "pyarrow>=14.0.1",
+    "pydantic",
     "typing-extensions>=4.3.0",
     "universal-pathlib",
 ]

src/hats/catalog/__init__.py

@@ -1,8 +1,9 @@
 """Catalog data wrappers"""
 
-from .association_catalog.association_catalog import AssociationCatalog
+from .association_catalog import AssociationCatalog
 from .catalog import Catalog
 from .catalog_type import CatalogType
 from .dataset.dataset import Dataset
+from .dataset.table_properties import TableProperties
 from .margin_cache.margin_catalog import MarginCatalog
 from .partition_info import PartitionInfo

src/hats/catalog/association_catalog/__init__.py

@@ -1,3 +1,2 @@
 from .association_catalog import AssociationCatalog
-from .association_catalog_info import AssociationCatalogInfo
 from .partition_join_info import PartitionJoinInfo

src/hats/catalog/association_catalog/association_catalog.py

@@ -6,12 +6,10 @@
 import pandas as pd
 import pyarrow as pa
 from mocpy import MOC
-from typing_extensions import TypeAlias
 from upath import UPath
 
-from hats.catalog.association_catalog.association_catalog_info import AssociationCatalogInfo
 from hats.catalog.association_catalog.partition_join_info import PartitionJoinInfo
-from hats.catalog.catalog_type import CatalogType
+from hats.catalog.dataset.table_properties import TableProperties
 from hats.catalog.healpix_dataset.healpix_dataset import HealpixDataset, PixelInputTypes
 from hats.io import file_io, paths
 
@@ -24,24 +22,17 @@ class AssociationCatalog(HealpixDataset):
     Catalog, corresponding to each pair of partitions in each catalog that contain rows to join.
     """
 
-    # Update CatalogInfoClass, used to check if the catalog_info is the correct type, and
-    # set the catalog info to the correct type
-    CatalogInfoClass: TypeAlias = AssociationCatalogInfo
-    catalog_info: CatalogInfoClass
-
     JoinPixelInputTypes = Union[list, pd.DataFrame, PartitionJoinInfo]
 
     def __init__(
         self,
-        catalog_info: CatalogInfoClass,
+        catalog_info: TableProperties,
         pixels: PixelInputTypes,
         join_pixels: JoinPixelInputTypes,
         catalog_path=None,
         moc: MOC | None = None,
         schema: pa.Schema | None = None,
     ) -> None:
-        if not catalog_info.catalog_type == CatalogType.ASSOCIATION:
-            raise ValueError("Catalog info `catalog_type` must be 'association'")
         super().__init__(catalog_info, pixels, catalog_path, moc=moc, schema=schema)
         self.join_info = self._get_partition_join_info_from_pixels(join_pixels)
 
@@ -67,7 +58,7 @@ def _get_partition_join_info_from_pixels(
     @classmethod
     def _read_args(
         cls, catalog_base_dir: str | Path | UPath
-    ) -> Tuple[CatalogInfoClass, PixelInputTypes, JoinPixelInputTypes]:  # type: ignore[override]
+    ) -> Tuple[TableProperties, PixelInputTypes, JoinPixelInputTypes]:  # type: ignore[override]
         args = super()._read_args(catalog_base_dir)
         partition_join_info = PartitionJoinInfo.read_from_dir(catalog_base_dir)
         return args + (partition_join_info,)

src/hats/catalog/association_catalog/partition_join_info.py

@@ -73,7 +73,7 @@ def primary_to_join_map(self) -> Dict[HealpixPixel, List[HealpixPixel]]:
         return primary_to_join
 
     def write_to_metadata_files(self, catalog_path: str | Path | UPath | None = None):
-        """Generate parquet metadata, using the known partitions.
+        """Generate parquet metadata, using the known joint partitions.
 
         Args:
             catalog_path (UPath): base path for the catalog
@@ -86,7 +86,7 @@ def write_to_metadata_files(self, catalog_path: str | Path | UPath | None = None
         """
         if catalog_path is None:
             if self.catalog_base_dir is None:
-                raise ValueError("catalog_path is required if info was not loaded from a directory")
+                raise ValueError("catalog_path is required if join info was not loaded from a directory")
             catalog_path = self.catalog_base_dir
 
         batches = [

src/hats/catalog/catalog.py

@@ -2,19 +2,12 @@
 
 from __future__ import annotations
 
-from pathlib import Path
 from typing import List, Tuple
 
 import numpy as np
-import pyarrow as pa
-from mocpy import MOC
-from typing_extensions import TypeAlias
-from upath import UPath
 
 import hats.pixel_math.healpix_shim as hp
-from hats.catalog.catalog_info import CatalogInfo
-from hats.catalog.catalog_type import CatalogType
-from hats.catalog.healpix_dataset.healpix_dataset import HealpixDataset, PixelInputTypes
+from hats.catalog.healpix_dataset.healpix_dataset import HealpixDataset
 from hats.pixel_math import HealpixPixel
 from hats.pixel_math.box_filter import generate_box_moc, wrap_ra_angles
 from hats.pixel_math.cone_filter import generate_cone_moc
@@ -36,45 +29,6 @@ class Catalog(HealpixDataset):
     `Norder=/Dir=/Npix=.parquet`
     """
 
-    HIPS_CATALOG_TYPES = [CatalogType.OBJECT, CatalogType.SOURCE]
-
-    # Update CatalogInfoClass, used to check if the catalog_info is the correct type, and
-    # set the catalog info to the correct type
-    CatalogInfoClass: TypeAlias = CatalogInfo
-    catalog_info: CatalogInfoClass
-
-    def __init__(
-        self,
-        catalog_info: CatalogInfoClass,
-        pixels: PixelInputTypes,
-        catalog_path: str | Path | UPath | None = None,
-        moc: MOC | None = None,
-        schema: pa.Schema | None = None,
-    ) -> None:
-        """Initializes a Catalog
-
-        Args:
-            catalog_info: CatalogInfo object with catalog metadata
-            pixels: Specifies the pixels contained in the catalog. Can be either a
-                list of HealpixPixel, `PartitionInfo object`, or a `PixelTree` object
-            catalog_path: If the catalog is stored on disk, specify the location of the catalog
-                Does not load the catalog from this path, only store as metadata
-            moc (mocpy.MOC): MOC object representing the coverage of the catalog
-            schema (pa.Schema): The pyarrow schema for the catalog
-        """
-        if catalog_info.catalog_type not in self.HIPS_CATALOG_TYPES:
-            raise ValueError(
-                f"Catalog info `catalog_type` must be one of "
-                f"{', '.join([t.value for t in self.HIPS_CATALOG_TYPES])}"
-            )
-        super().__init__(
-            catalog_info,
-            pixels,
-            catalog_path=catalog_path,
-            moc=moc,
-            schema=schema,
-        )
-
     def filter_by_cone(self, ra: float, dec: float, radius_arcsec: float) -> Catalog:
         """Filter the pixels in the catalog to only include the pixels that overlap with a cone
 

src/hats/catalog/dataset/__init__.py

@@ -1,2 +1 @@
-from .base_catalog_info import BaseCatalogInfo
 from .dataset import Dataset