Merge pull request #33 from ASaiM/dmnddb

Add sniffer for dmnd files
jmchilton · Jan 9, 2017 · f50272b · f50272b
2 parents 5f0a413 + 50f9f74
commit f50272b
Show file tree

Hide file tree

Showing 4 changed files with 38 additions and 1 deletion.
diff --git a/config/datatypes_conf.xml.sample b/config/datatypes_conf.xml.sample
@@ -38,7 +38,7 @@
     <datatype extension="len" type="galaxy.datatypes.chrominfo:ChromInfo" display_in_upload="true">
       <converter file="len_to_linecount.xml" target_datatype="linecount" />
     </datatype>
-    <datatype extension="dmnddb" type="galaxy.datatypes.binary:Binary" subclass="True" display_in_upload="false"/>
+    <datatype extension="dmnd" type="galaxy.datatypes.binary:Binary:DMND" subclass="True" display_in_upload="false"/>
     <datatype extension="bigbed" type="galaxy.datatypes.binary:BigBed" mimetype="application/octet-stream" display_in_upload="true">
       <display file="ucsc/bigbed.xml" />
       <display file="igb/bb.xml" />
@@ -593,6 +593,7 @@
     <sniffer type="galaxy.datatypes.binary:Sff"/>
     <sniffer type="galaxy.datatypes.binary:Sra"/>
     <sniffer type="galaxy.datatypes.binary:NetCDF"/>
+    <sniffer type="galaxy.datatypes.binary:DMND" />
     <sniffer type="galaxy.datatypes.triples:Rdf"/>
     <sniffer type="galaxy.datatypes.blast:BlastXml"/>
     <sniffer type="galaxy.datatypes.xml:Phyloxml"/>

diff --git a/lib/galaxy/datatypes/binary.py b/lib/galaxy/datatypes/binary.py
@@ -1436,3 +1436,36 @@ def sniff( self, filename ):
 
 
 Binary.register_sniffable_binary_format("netcdf", "netcdf", NetCDF)
+
+
+class DMND( Binary ):
+    """
+    Class describing an DMND file
+    >>> from galaxy.datatypes.sniff import get_test_fname
+    >>> fname = get_test_fname( 'diamond_db.dmnd' )
+    >>> DMND().sniff( fname )
+    True
+    >>> fname = get_test_fname( 'interval.interval' )
+    >>> DMND().sniff( fname )
+    False
+    """
+    file_ext = "dmnd"
+    edam_format = ""
+
+    def __init__( self, **kwd ):
+        Binary.__init__( self, **kwd )
+        self._magic = binascii.unhexlify("6d18ee15a4f84a02")
+
+    def sniff( self, filename ):
+        # The first 8 bytes of any dmnd file are 0x24af8a415ee186d
+
+        try:
+            header = open( filename, 'rb' ).read(8)
+            if header == self._magic:
+                return True
+            return False
+        except:
+            return False
+
+
+Binary.register_sniffable_binary_format("dmnd", "dmnd", DMND)
diff --git a/lib/galaxy/datatypes/sniff.py b/lib/galaxy/datatypes/sniff.py
@@ -356,6 +356,9 @@ def guess_ext( fname, sniff_order, is_multi_byte=False ):
     >>> fname = get_test_fname('1.gg')
     >>> guess_ext(fname, sniff_order)
     'gg'
+    >>> fname = get_test_fname('diamond_db.dmnd')
+    >>> guess_ext(fname, sniff_order)
+    'dmnd'
     """
     file_ext = None
     for datatype in sniff_order:

diff --git a/lib/galaxy/datatypes/test/diamond_db.dmnd b/lib/galaxy/datatypes/test/diamond_db.dmnd