add query opt $exists

h5rdmtoolbox/database/file.py

@@ -69,8 +69,15 @@ def _regex(value, pattern) -> bool:
         return False
     return True
 
+def _exists(value, tf: bool) -> bool:
+    if tf:
+        return value is not None
+    else:
+        return value is None
+
 
-operator = {'$regex': _regex, '$eq': _eq, '$gt': _gt, '$gte': _gte, '$lt': _lt, '$lte': _lte}
+operator = {'$regex': _regex, '$eq': _eq, '$gt': _gt, '$gte': _gte, '$lt': _lt, '$lte': _lte,
+            '$exists': _exists}
 value_operator = {'$eq': _arreq, '$gt': _gt, '$gte': _gte, '$lt': _lt, '$lte': _lte}
 
 

setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = h5rdmtoolbox
-version = 0.12.1
+version = 0.12.2
 author = Matthias Probst
 author_email = matthias.probst@kit.edu
 description = Supporting a FAIR Research Data lifecycle using Python and HDF5.

tests/database/test_filequery.py

@@ -1,6 +1,5 @@
-import unittest
-
 import numpy as np
+import unittest
 
 import h5rdmtoolbox as h5tbx
 from h5rdmtoolbox.wrapper.core import File
@@ -109,6 +108,17 @@ def test_Folder(self):
         res = fd.find_one_per_file({'$basename': {'$regex': 'ds[0-9]'}})
         self.assertEqual(2, len(res))
 
+    def test_exists(self):
+        with h5tbx.File() as h5:
+            h5.create_dataset('ds1', shape=(1, 2, 3), attrs=dict(units='', long_name='long name 1'))
+            h5.create_dataset('ds2', shape=(1, 2, 3), attrs=dict(units='', long_name='long name 1'))
+
+            self.assertTrue('long_name' in h5.find_one({'long_name': {'$exists': True}}).attrs)
+            self.assertIsInstance(h5.find_one({'long_name': {'$exists': True}}), h5tbx.wrapper.core.Dataset)
+
+            mres = h5.find({'long_name': {'$exists': True}})
+            self.assertEqual(2, len(mres))
+
     def test_chained_find(self):
         with h5tbx.File() as h5:
             g = h5.create_group('grp1')