feat: Added RFC-9 Zipped OME-Zarr

Signed-off-by: Sricharan Reddy Varra <sricharan.varra@biohub.org>

Author: srivarra

src/iohub/cli/cli.py

@@ -5,12 +5,15 @@
 from iohub import __version__, open_ome_zarr
 from iohub.cli.parsing import input_position_dirpaths
 from iohub.convert import TIFFConverter
+from iohub.core.ozx import pack_ozx, summarize_ozx
 from iohub.reader import print_info
 from iohub.rename_wells import rename_wells
 
 VERSION = __version__
 
 _DATASET_PATH = click.Path(exists=True, file_okay=False, resolve_path=True, path_type=pathlib.Path)
+# Like _DATASET_PATH but also accepts files (e.g. ``.ozx`` archives).
+_OME_ZARR_PATH = click.Path(exists=True, resolve_path=True, path_type=pathlib.Path)
 
 
 @click.group()
@@ -26,7 +29,7 @@ def cli():
     "files",
     nargs=-1,
     required=True,
-    type=_DATASET_PATH,
+    type=_OME_ZARR_PATH,
 )
 @click.option(
     "--verbose",
@@ -35,11 +38,11 @@ def cli():
     help="Show usage guide to open dataset in Python and full tree for HCS Plates in OME-Zarr",
 )
 def info(files, verbose):
-    """View basic metadata of a list of FILES.
+    """View metadata for one or more FILES.
 
-    Supported formats are Micro-Manager-acquired TIFF datasets
-    (single-page TIFF, multi-page OME-TIFF, NDTIFF)
-    and OME-Zarr (v0.1 linear HCS layout and all v0.4 layouts).
+    Supports Micro-Manager TIFF datasets (multi-page OME-TIFF, NDTIFF),
+    OME-Zarr directory stores (v0.4 and v0.5), and RFC-9 zipped
+    OME-Zarr archives (``.ozx``).
     """
     for file in files:
         click.echo(f"Reading file:\t {file}")
@@ -195,3 +198,91 @@ def rename_wells_command(zarrfile, csvfile):
     ```
     """
     rename_wells(zarrfile, csvfile)
+
+
+def _echo_ozx_summary(verb: str, path: pathlib.Path) -> None:
+    """Print the one-liner ``ozx pack`` shows on success."""
+    s = summarize_ozx(path)
+    click.echo(f"{verb}: {path}  (ome version={s.version}, jsonFirst={s.json_first})")
+
+
+@cli.group(name="ozx")
+@click.help_option("-h", "--help")
+def ozx_group():
+    """RFC-9 zipped OME-Zarr (``.ozx``) operations.
+
+    A ``.ozx`` is a single-file OME-Zarr archive (ZIP_STORED + ZIP64)
+    designed for distribution - S3, AWS Open Data, HPC↔compute-cluster
+    transfers - not as a live mutable store. One file to upload,
+    download, and serve.
+
+    Use ``ozx pack`` to bundle a directory store for export and
+    ``ozx info`` to inspect an archive's RFC-9 properties. ``iohub
+    info`` works on ``.ozx`` paths directly for the usual FOV summary.
+    """
+
+
+@ozx_group.command(name="pack")
+@click.help_option("-h", "--help")
+@click.option(
+    "-i",
+    "--input",
+    "src",
+    required=True,
+    type=click.Path(exists=True, file_okay=False, dir_okay=True, resolve_path=True, path_type=pathlib.Path),
+    help="Input OME-Zarr directory store.",
+)
+@click.option(
+    "-o",
+    "--output",
+    "dst",
+    required=True,
+    type=click.Path(exists=False, resolve_path=True, path_type=pathlib.Path),
+    help="Output .ozx path (must not exist).",
+)
+@click.option(
+    "--version",
+    "ome_version",
+    default=None,
+    help="OME-NGFF version to record. Sniffed from source if omitted.",
+)
+def ozx_pack(src, dst, ome_version):
+    """Pack an OME-Zarr directory into an RFC-9 ``.ozx`` archive.
+
+    Writes a ZIP_STORED + ZIP64 archive in one pass with entries in BFS
+    order - root ``zarr.json`` first, other ``zarr.json`` next, chunks
+    last - and ``jsonFirst:true`` in the archive comment. HTTP-range
+    readers can then fetch all metadata in one contiguous Range request
+    on cold opens, the win that matters for S3 and AWS Open Data.
+    """
+    out = pack_ozx(src, dst, version=ome_version)
+    _echo_ozx_summary("packed", out)
+
+
+@ozx_group.command(name="info")
+@click.help_option("-h", "--help")
+@click.argument(
+    "path",
+    type=click.Path(exists=True, dir_okay=False, resolve_path=True, path_type=pathlib.Path),
+)
+def ozx_info(path):
+    """Print RFC-9 archive metadata: version, ``jsonFirst``, entry count, size.
+
+    Use it to sanity-check an archive before publishing (``jsonFirst``
+    should be ``True`` after ``iohub ozx pack``) or to inspect archives
+    received from collaborators or downloaded from public datasets.
+    """
+    import zipfile
+
+    # One zip open for the namelist, one for the comment.
+    with zipfile.ZipFile(path) as zf:
+        names = zf.namelist()
+    summary = summarize_ozx(path)
+    n_entries = len(names)
+    n_meta = sum(1 for n in names if n.rsplit("/", 1)[-1] == "zarr.json")
+    n_dupes = n_entries - len(set(names))
+    click.echo(f"path:        {path}")
+    click.echo(f"size:        {path.stat().st_size:,} bytes")
+    click.echo(f"entries:     {n_entries:,}  ({n_meta} zarr.json, {n_dupes} duplicate names)")
+    click.echo(f"ome version: {summary.version}")
+    click.echo(f"jsonFirst:   {summary.json_first}")

src/iohub/core/__init__.py

@@ -15,6 +15,18 @@
     PathNormalizationError,
     StoreOpenError,
 )
+from iohub.core.ozx import (
+    OZX_EXTENSION,
+    OzxStore,
+    OzxSummary,
+    is_ozx_path,
+    pack_ozx,
+    read_ozx_comment,
+    read_ozx_json_first,
+    read_ozx_version,
+    summarize_ozx,
+    write_ozx_comment,
+)
 from iohub.core.protocol import (
     ArrayBackend,
     ArrayIO,
@@ -37,6 +49,7 @@
 from iohub.core.utils import normalize_path, pad_shape
 
 __all__ = [
+    "OZX_EXTENSION",
     # Types
     "AccessMode",
     "ArrayBackend",
@@ -55,6 +68,9 @@
     # Arrays
     "NGFFArray",
     "NGFFVersion",
+    # OZX (RFC-9 zipped OME-Zarr)
+    "OzxStore",
+    "OzxSummary",
     "PathNormalizationError",
     "StoreOpenError",
     "StorePath",
@@ -65,12 +81,19 @@
     # Registry
     "available_implementations",
     "get_implementation",
+    "is_ozx_path",
     "ngff_version_for_format",
     # Utils
     "normalize_path",
+    "pack_ozx",
     "pad_shape",
+    "read_ozx_comment",
+    "read_ozx_json_first",
+    "read_ozx_version",
     "register_implementation",
     "set_default_implementation",
+    "summarize_ozx",
+    "write_ozx_comment",
     # Compat
     "zarr_format_for_version",
 ]