Fix check for swmr compatibility in h5 files (#368)

* Check compatability via super block version

Previous check of SWMR compatibility was trying to open the file with
old HDF5 library versions, to see if an error was raised. With HDF5>2.0
this no longer works because a fix was implemented to only raise an
error when trying to create or write new objects that are incompatible
rather than when opening the file.

* Add test for swmr compatability function

Author: pblowey

src/dlstbx/services/validation.py

@@ -86,7 +86,7 @@ def validate(self, rw, header, message):
 
         fail = functools.partial(self.fail_validation, rw, header, output)
 
-        # Verify HDF5 file version is _not_ compatible with HDF5 1.8 format
+        # Verify HDF5 file version is SWMR compatible (doesn't require SWMR mode to be enabled)
         if filename.endswith((".h5", ".nxs")):
             if not hdf5_util.is_readable(filename):
                 return fail(f"{filename} is an invalid HDF5 file")
@@ -106,17 +106,16 @@ def validate(self, rw, header, message):
                     return fail(
                         f"HDF5 file {filename} links to invalid file(s) {', '.join(errors)}"
                     )
-                hdf_18 = [
+                hdf_swmr_incompatible = [
                     link
                     for link, image_count in hdf5_util.find_all_references(
                         filename
                     ).items()
-                    if image_count and hdf5_util.is_HDF_1_8_compatible(link)
+                    if image_count and not hdf5_util.is_SWMR_compatible(link)
                 ]
-                if hdf_18:
+                if hdf_swmr_incompatible:
                     return fail(
-                        f"HDF5 file {filename} links to HDF5 1.8 format data in %s"
-                        % ", ".join(hdf_18)
+                        f"HDF5 file {filename} links to non-SWMR compatible file(s) {', '.join(hdf_swmr_incompatible)}"
                     )
                 try:
                     hdf5_util.validate_pixel_mask(filename)

src/dlstbx/util/hdf5.py

@@ -104,14 +104,15 @@ def is_readable(filename: str) -> bool:
         return False
 
 
-def is_HDF_1_8_compatible(filename: str) -> bool:
-    """Check if a file can be read with HDF 1.8. This excludes all SWMR formatted files."""
-
-    try:
-        with h5py.File(filename, "r", libver=("earliest", "v108")):
-            return True
-    except OSError:
-        return False
+def is_SWMR_compatible(filename: str) -> bool:
+    """Check if a file format is SWMR compatible by checking superblock version."""
+
+    with h5py.File(filename, "r") as f:
+        fcpl = f.id.get_create_plist()
+        # returns a tuple of (superblock_version, freelist_version, symbol_table_version, shared_header_version)
+        versions = fcpl.get_version()
+        superblock_version = versions[0]
+        return superblock_version >= 3
 
 
 def validate_pixel_mask(filename: str) -> bool:

tests/util/test_hdf5.py

@@ -2,6 +2,7 @@
 
 import os
 
+import h5py
 import pytest
 
 import dlstbx.util.hdf5
@@ -61,3 +62,15 @@ def test_validate_pixel_mask_shape_int64():
         dlstbx.util.hdf5.validate_pixel_mask(
             "/dls/mx/data/cm31104/cm31104-1/020222/RT/ToNV_WT/x_27_master.h5"
         )
+
+
+def test_is_SWMR_compatible(tmp_path):
+    # Test that the function correctly identifies SWMR compatible and incompatible files
+    swmr_compatible_file = tmp_path / "swmr_compatible.h5"
+    with h5py.File(swmr_compatible_file.as_posix(), "w", libver=("v110", "latest")):
+        pass
+    assert dlstbx.util.hdf5.is_SWMR_compatible(swmr_compatible_file.as_posix())
+    swmr_incompatible_file = tmp_path / "swmr_incompatible.h5"
+    with h5py.File(swmr_incompatible_file.as_posix(), "w", libver=("earliest", "v108")):
+        pass
+    assert not dlstbx.util.hdf5.is_SWMR_compatible(swmr_incompatible_file.as_posix())