Merge pull request #18507 from TomHarrop/datatype_pod5_backport

[24.1] Backport pod5 datatype

lib/galaxy/config/sample/datatypes_conf.xml.sample

@@ -361,6 +361,7 @@
     <datatype extension="fast5.tar.gz" type="galaxy.datatypes.binary:Fast5ArchiveGz" display_in_upload="true"/>
     <datatype extension="fast5.tar.bz2" type="galaxy.datatypes.binary:Fast5ArchiveBz2" display_in_upload="true"/>
     <datatype extension="fast5.tar.xz" type="galaxy.datatypes.binary:Fast5ArchiveXz" display_in_upload="true"/>
+    <datatype extension="pod5" type="galaxy.datatypes.binary:Pod5" display_in_upload="true"/>
     <datatype extension="peptideshaker_archive" type="galaxy.datatypes.binary:CompressedArchive" subclass="true" display_in_upload="true"/>
     <datatype extension="percin" type="galaxy.datatypes.tabular:Tabular" subclass="true"/>
     <datatype extension="percout" type="galaxy.datatypes.xml:GenericXml" subclass="true"/>
@@ -1079,6 +1080,7 @@
     <sniffer type="galaxy.datatypes.binary:Fast5ArchiveXz"/>
     <sniffer type="galaxy.datatypes.binary:Fast5ArchiveBz2"/>
     <sniffer type="galaxy.datatypes.binary:Fast5Archive"/>
+    <sniffer type="galaxy.datatypes.binary:Pod5"/>
     <sniffer type="galaxy.datatypes.binary:Meryldb" />
     <sniffer type="galaxy.datatypes.binary:Bref3" />
     <sniffer type="galaxy.datatypes.binary:PostgresqlArchive"/>

lib/galaxy/datatypes/binary.py

@@ -3759,6 +3759,31 @@ class Fast5ArchiveBz2(Fast5Archive):
         return Fast5Archive.sniff(self, filename)
 
 
+class Pod5(Binary):
+    """
+    Class describing a POD5 file. The POD5 Format Specification is at
+    https://pod5-file-format.readthedocs.io/en/latest/SPECIFICATION.html
+
+    >>> from galaxy.datatypes.sniff import get_test_fname
+    >>> fname = get_test_fname('test.pod5')
+    >>> Pod5().sniff(fname)
+    True
+    >>> fname = get_test_fname('test.fast5.tar')
+    >>> Pod5().sniff(fname)
+    False
+    """
+
+    file_ext = "pod5"
+
+    def sniff(self, filename: str) -> bool:
+        expected_signature = bytes([0x8B, 0x50, 0x4F, 0x44, 0x0D, 0x0A, 0x1A, 0x0A])
+        with open(filename, "rb") as f:
+            first_8_bytes = f.read(8)
+            f.seek(-8, 2)
+            last_8_bytes = f.read(8)
+            return first_8_bytes == expected_signature and last_8_bytes == expected_signature
+
+
 class SearchGuiArchive(CompressedArchive):
     """Class describing a SearchGUI archive"""