[MRG] Fix max hash deprecation warnings (#1301)

* swizzle property max_hash to _max_hash

* replace use of max_hash with scaled

* remove max_hash terminology

* cleanup and refactor some of the stale max_hash stuff in minahsh

* update screed req to v1.0.5

Co-authored-by: Luiz Irber <luizirber@users.noreply.github.com>

setup.cfg

@@ -33,7 +33,7 @@ packages = find:
 platforms = any
 include_package_data = True
 install_requires =
-    screed>=1.0
+    screed>=1.0.5
     cffi>=1.14.0
     numpy
     matplotlib

src/sourmash/commands.py

@@ -81,8 +81,8 @@ def compare(args):
     notify('loaded {} signatures total.'.format(len(siglist)))
 
     # check to make sure they're potentially compatible - either using
-    # max_hash/scaled, or not.
-    scaled_sigs = [s.minhash.max_hash for s in siglist]
+    # scaled, or not.
+    scaled_sigs = [s.minhash.scaled for s in siglist]
     is_scaled = all(scaled_sigs)
     is_scaled_2 = any(scaled_sigs)
 
@@ -427,7 +427,7 @@ def search(args):
 
     # downsample if requested
     if args.scaled:
-        if query.minhash.max_hash == 0:
+        if not query.minhash.scaled:
             error('cannot downsample a signature not created with --scaled')
             sys.exit(-1)
 
@@ -581,7 +581,7 @@ def gather(args):
                                              sourmash_args.get_moltype(query))
 
     # verify signature was computed right.
-    if query.minhash.scaled == 0:
+    if not query.minhash.scaled:
         error('query signature needs to be created with --scaled')
         sys.exit(-1)
 
@@ -609,7 +609,7 @@ def gather(args):
 
     found = []
     weighted_missed = 1
-    new_max_hash = query.minhash.max_hash
+    new_max_hash = query.minhash._max_hash
     next_query = query
 
     for result, weighted_missed, new_max_hash, next_query in gather_databases(query, databases, args.threshold_bp, args.ignore_abundance):
@@ -726,7 +726,7 @@ def multigather(args):
                    query.minhash.ksize, sourmash_args.get_moltype(query))
 
             # verify signature was computed right.
-            if query.minhash.max_hash == 0:
+            if not query.minhash.scaled:
                 error('query signature needs to be created with --scaled; skipping')
                 continue
 

src/sourmash/compare.py

@@ -21,7 +21,7 @@ def compare_serial(siglist, ignore_abundance, downsample=False):
 
         If the sketches are abundance weighted, calculate the angular
         similarity.
-    :param boolean downsample by max_hash if True
+    :param boolean downsample by scaled if True
     :return: np.array similarity matrix
     """
     import numpy as np
@@ -45,7 +45,7 @@ def compare_serial_containment(siglist, downsample=False):
     process. Best to use when there is few signatures.
 
     :param list siglist: list of signatures to compare
-    :param boolean downsample by max_hash if True
+    :param boolean downsample by scaled if True
     :return: np.array similarity matrix
     """
     import numpy as np
@@ -83,7 +83,7 @@ def get_similarities_at_index(index, ignore_abundance, downsample, siglist):
 
         If the sketches are abundance weighted, calculate the angular
         similarity.
-    :param boolean downsample by max_hash if True
+    :param boolean downsample by scaled if True
     :param siglist list of signatures
     :return: list of similarities for the combinations of signature at index
         with rest of the signatures from index+1
@@ -114,7 +114,7 @@ def compare_parallel(siglist, ignore_abundance, downsample, n_jobs):
 
         If the sketches are abundance weighted, calculate the angular
         similarity.
-    :param boolean downsample by max_hash if True
+    :param boolean downsample by scaled if True
     :param int n_jobs number of processes to run the similarity calculations on
     :return: np.array similarity matrix
     """
@@ -194,7 +194,7 @@ def compare_all_pairs(siglist, ignore_abundance, downsample=False, n_jobs=None):
 
         If the sketches are abundance weighted, calculate the angular
         similarity.
-    :param boolean downsample by max_hash if True
+    :param boolean downsample by scaled if True
     :param int n_jobs number of processes to run the similarity calculations on,
     if number of jobs is None or 1, compare serially, otherwise parallely.
     :return: np.array similarity matrix

src/sourmash/minhash.py

@@ -217,7 +217,7 @@ def __copy__(self):
             hp=self.hp,
             track_abundance=self.track_abundance,
             seed=self.seed,
-            max_hash=self.max_hash,
+            max_hash=self._max_hash,
         )
         a.merge(self)
         return a
@@ -233,7 +233,7 @@ def __getstate__(self):
             self.hashes,
             None,
             self.track_abundance,
-            self.max_hash,
+            self._max_hash,
             self.seed,
         )
 
@@ -274,7 +274,7 @@ def copy_and_clear(self):
             self.hp,
             self.track_abundance,
             self.seed,
-            self.max_hash,
+            self._max_hash,
         )
         return a
 
@@ -401,6 +401,11 @@ def ksize(self):
     def max_hash(self):
         return self._methodcall(lib.kmerminhash_max_hash)
 
+    # a non-deprecated `max_hash` property for internal testing purposes only
+    @property
+    def _max_hash(self):
+        return self._methodcall(lib.kmerminhash_max_hash)
+
     @property
     def track_abundance(self):
         return self._methodcall(lib.kmerminhash_track_abundance)
@@ -458,11 +463,7 @@ def downsample(self, num=None, scaled=None):
         elif scaled is not None:
             if self.num:
                 raise ValueError("num != 0 - cannot downsample a standard MinHash")
-            max_hash = self.max_hash
-            if max_hash is None:
-                raise ValueError("no max_hash available - cannot downsample")
-
-            old_scaled = _get_scaled_for_max_hash(self.max_hash)
+            old_scaled = self.scaled
             if old_scaled > scaled:
                 raise ValueError(
                     "new scaled {} is lower than current sample scaled {}".format(
@@ -492,7 +493,7 @@ def flatten(self):
         # create new object:
         a = MinHash(
             self.num, self.ksize, self.is_protein, self.dayhoff, self.hp,
-            False, self.seed, self.max_hash
+            False, self.seed, self._max_hash
         )
         a.add_many(self)
 

tests/test__minhash.py

@@ -241,7 +241,7 @@ def test_scaled(track_abundance):
     scaled = _get_scaled_for_max_hash(35)
     print('XX', scaled, _get_max_hash_for_scaled(scaled))
     mh = MinHash(0, 4, track_abundance=track_abundance, scaled=scaled)
-    assert mh.max_hash == 35
+    assert mh._max_hash == 35
 
     mh.add_hash(10)
     mh.add_hash(20)
@@ -1177,7 +1177,7 @@ def test_mh_copy_and_clear(track_abundance):
     b = a.copy_and_clear()
     assert a.ksize == b.ksize
     assert b.num == a.num
-    assert b.max_hash == 0
+    assert b._max_hash == 0
     assert not b.is_protein
     assert b.track_abundance == track_abundance
     assert b.seed == a.seed
@@ -1196,7 +1196,7 @@ def test_mh_copy_and_clear_with_max_hash(track_abundance):
     b = a.copy_and_clear()
     assert a.ksize == b.ksize
     assert b.num == a.num
-    assert b.max_hash == 20
+    assert b._max_hash == 20
     assert not b.is_protein
     assert b.track_abundance == track_abundance
     assert b.seed == a.seed
@@ -1220,8 +1220,8 @@ def test_pickle_max_hash(track_abundance):
     b = pickle.loads(pickle.dumps(a))
     assert a.ksize == b.ksize
     assert b.num == a.num
-    assert b.max_hash == a.max_hash
-    assert b.max_hash == 20
+    assert b._max_hash == a._max_hash
+    assert b._max_hash == 20
     assert not b.is_protein
     assert b.track_abundance == track_abundance
     assert b.seed == a.seed
@@ -1239,8 +1239,8 @@ def test_pickle_scaled(track_abundance):
     b = pickle.loads(pickle.dumps(a))
     assert a.ksize == b.ksize
     assert b.num == a.num
-    assert b.max_hash == a.max_hash
-    assert b.max_hash == 20
+    assert b._max_hash == a._max_hash
+    assert b._max_hash == 20
     assert not b.is_protein
     assert b.track_abundance == track_abundance
     assert b.seed == a.seed
@@ -1401,7 +1401,7 @@ def test_flatten():
     # test behavior with scaled
     scaled = _get_scaled_for_max_hash(35)
     mh = MinHash(0, 4, track_abundance=True, scaled=scaled)
-    assert mh.max_hash == 35
+    assert mh._max_hash == 35
 
     mh.add_hash(10)
     mh.add_hash(10)
@@ -1512,7 +1512,7 @@ def test_downsample_scaled(track_abundance):
     assert list(sorted(mh.hashes)) == list(mins)
 
     mh2 = mh.downsample(scaled=2)
-    print(mh.max_hash, mh2.max_hash)
+    print(mh._max_hash, mh2._max_hash)
 
     assert len(mh2) == 3
     assert list(sorted(mh2.hashes)) == list(mins[:3])

tests/test_cmd_signature.py

@@ -1180,7 +1180,7 @@ def test_sig_downsample_2_num_to_scaled(c):
     actual_mins = actual_downsample_sig.minhash.hashes.keys()
 
     # select those mins that are beneath the new max hash...
-    max_hash = actual_downsample_sig.minhash.max_hash
+    max_hash = actual_downsample_sig.minhash._max_hash
     test_mins_down = { k for k in test_mins if k < max_hash }
     assert test_mins_down == set(actual_mins)
 

tests/test_jaccard.py

@@ -124,7 +124,8 @@ def test_pickle(track_abundance):
     assert e1.num == e2.num
     assert e1.ksize == e2.ksize
     assert e1.is_protein == e2.is_protein
-    assert e1.max_hash == e2.max_hash
+    assert e1.scaled == e2.scaled
+    assert e1.scaled == 0
     assert e1.seed == e2.seed
 
 

tests/test_signature.py

@@ -129,7 +129,7 @@ def test_roundtrip_empty(track_abundance):
     assert sig2.similarity(sig) == 0
 
 
-def test_roundtrip_max_hash(track_abundance):
+def test_roundtrip_scaled(track_abundance):
     e = sourmash.MinHash(n=0, ksize=20, track_abundance=track_abundance,
                          max_hash=10)
     e.add_hash(5)
@@ -139,7 +139,7 @@ def test_roundtrip_max_hash(track_abundance):
     sig2 = siglist[0]
     e2 = sig2.minhash
 
-    assert e.max_hash == e2.max_hash
+    assert e.scaled == e2.scaled
 
     assert sig.similarity(sig2) == 1.0
     assert sig2.similarity(sig) == 1.0

tests/test_sourmash_compute.py

@@ -580,9 +580,9 @@ def test_do_sourmash_compute_with_scaled_1():
         siglist = list(signature.load_signatures(outfile))
         assert len(siglist) == 2
 
-        max_hashes = [ x.minhash.max_hash for x in siglist ]
-        assert len(max_hashes) == 2
-        assert set(max_hashes) == { sourmash.MAX_HASH }
+        scaled_vals = [ x.minhash.scaled for x in siglist ]
+        assert len(scaled_vals) == 2
+        assert set(scaled_vals) == { 1 }
 
 
 def test_do_sourmash_compute_with_scaled_2():
@@ -599,7 +599,7 @@ def test_do_sourmash_compute_with_scaled_2():
         siglist = list(signature.load_signatures(outfile))
         assert len(siglist) == 2
 
-        max_hashes = [ x.minhash.max_hash for x in siglist ]
+        max_hashes = [ x.minhash._max_hash for x in siglist ]
         assert len(max_hashes) == 2
         assert set(max_hashes) == set([ int(2**64 /2.) ])
 
@@ -618,7 +618,7 @@ def test_do_sourmash_compute_with_scaled():
         siglist = list(signature.load_signatures(outfile))
         assert len(siglist) == 2
 
-        max_hashes = [ x.minhash.max_hash for x in siglist ]
+        max_hashes = [ x.minhash._max_hash for x in siglist ]
         assert len(max_hashes) == 2
         assert set(max_hashes) == set([ int(2**64 /100.) ])
 

tests/test_sourmash_sketch.py

@@ -638,9 +638,9 @@ def test_do_sourmash_sketchdna_with_scaled_1():
         siglist = list(signature.load_signatures(outfile))
         assert len(siglist) == 2
 
-        max_hashes = [ x.minhash.max_hash for x in siglist ]
-        assert len(max_hashes) == 2
-        assert set(max_hashes) == { sourmash.MAX_HASH }
+        scaled_vals = [ x.minhash.scaled for x in siglist ]
+        assert len(scaled_vals) == 2
+        assert set(scaled_vals) == { 1 }
 
 
 def test_do_sourmash_sketchdna_with_scaled_2():
@@ -657,7 +657,7 @@ def test_do_sourmash_sketchdna_with_scaled_2():
         siglist = list(signature.load_signatures(outfile))
         assert len(siglist) == 2
 
-        max_hashes = [ x.minhash.max_hash for x in siglist ]
+        max_hashes = [ x.minhash._max_hash for x in siglist ]
         assert len(max_hashes) == 2
         assert set(max_hashes) == set([ int(2**64 /2.) ])
 
@@ -676,7 +676,7 @@ def test_do_sourmash_sketchdna_with_scaled():
         siglist = list(signature.load_signatures(outfile))
         assert len(siglist) == 2
 
-        max_hashes = [ x.minhash.max_hash for x in siglist ]
+        max_hashes = [ x.minhash._max_hash for x in siglist ]
         assert len(max_hashes) == 2
         assert set(max_hashes) == set([ int(2**64 /100.) ])