From e61f63e9bb006f25925c7e6040102c15bf1c0c35 Mon Sep 17 00:00:00 2001
From: Jayaram Kancherla <jayaram.kancherla@gmail.com>
Date: Fri, 1 Nov 2024 11:47:18 -0700
Subject: [PATCH] Parse H5 backed delayed objects

---
 src/rds2py/generics.py            |  2 ++
 src/rds2py/read_delayed_matrix.py | 31 +++++++++++++++++++++++++++++++
 2 files changed, 33 insertions(+)
 create mode 100644 src/rds2py/read_delayed_matrix.py

diff --git a/src/rds2py/generics.py b/src/rds2py/generics.py
index 565d1cd..cdd1e92 100644
--- a/src/rds2py/generics.py
+++ b/src/rds2py/generics.py
@@ -58,6 +58,8 @@
     # multi assay experiment
     "MultiAssayExperiment": "rds2py.read_mae.read_multi_assay_experiment",
     "ExperimentList": "rds2py.read_dict.read_dict",
+    # delayed matrices
+    "H5SparseMatrix": "rds2py.read_delayed_matrix.read_hdf5_sparse",
 }
 
 
diff --git a/src/rds2py/read_delayed_matrix.py b/src/rds2py/read_delayed_matrix.py
new file mode 100644
index 0000000..c7a9ce4
--- /dev/null
+++ b/src/rds2py/read_delayed_matrix.py
@@ -0,0 +1,31 @@
+"""Functions and classes for parsing R delayed matrix objects from HDF5Array."""
+
+from typing import Literal
+
+from hdf5array import Hdf5CompressedSparseMatrix
+
+from .generics import _dispatcher
+from .rdsutils import get_class
+
+__author__ = "jkanche"
+__copyright__ = "jkanche"
+__license__ = "MIT"
+
+
+def read_hdf5_sparse(robject: dict, **kwargs) -> Hdf5CompressedSparseMatrix:
+    _cls = get_class(robject)
+
+    if _cls not in ["H5SparseMatrix"]:
+        raise RuntimeError(f"`robject` does not contain not a 'H5SparseMatrix' object, contains `{_cls}`.")
+
+    by_column = False
+    # get seed package name
+    _seed_cls = get_class(robject["attributes"]["seed"])
+    if _seed_cls in ["CSC_H5SparseMatrixSeed"]:
+        by_column = True
+
+    shape = _dispatcher(robject["attributes"]["seed"]["dim"], **kwargs)
+    fpath = list(_dispatcher(robject["attributes"]["seed"]["filepath"], **kwargs))[0]
+    group_name = list(_dispatcher(robject["attributes"]["seed"]["group"], **kwargs))[0]
+
+    return Hdf5CompressedSparseMatrix(path=fpath, group_name=group_name, shape=shape, by_column=by_column)