Migrated bin_ids_forming_sequence from sequences.seqops to new sequen…

…ces.choices, updating dependencies

polymerist/genutils/sequences/choices.py

@@ -0,0 +1,45 @@
+'''Enumeration algorithms for choosing a sequence of symbols out of a sequence of sets of those symbols'''
+
+from typing import Generator, Iterable, Sequence, TypeVar
+T = TypeVar('T') # generic type for sequence element
+
+from copy import deepcopy
+from containers import defaultdict, Counter
+from itertools import product as cartesian_product
+
+from .seqops import is_unique
+
+
+def bin_ids_forming_sequence(sequence : Sequence[T], choice_bins : Sequence[Iterable[T]], draw_without_repeats : bool=True, unique_bins : bool=False) -> Generator[tuple[int, ...], None, None]:
+    '''
+    Takes an ordered sequence of N objects of a given type and an ordered of any number of bins, each containing an arbitary amount of unordered objects of the same type
+    Generates all possible N-tuples of bin indices which could produce the target sequence when drawing from those bins in the 
+    
+    if draw_without_repeats=True, will respect the multiplicity of elements in each bin when drawing
+    (i.e. will never have a bin position appear for a given object more times that that object appears in the corresponding bin)
+
+    if unique_bins=True, will only allow each bin to be sampled from once, EVEN if that bin contains elements which may occur later in the sequence
+    '''
+    symbol_inventory = defaultdict(Counter) # keys are objects of type T ("symbols"), values give multiplicities of symbols keyed by bin position
+    for i, choice_bin in enumerate(choice_bins):
+        for sym in choice_bin:
+            symbol_inventory[sym][i] += 1 # NOTE : implementation here requires that T be a hashable type
+
+    for idxs in cartesian_product(*(symbol_inventory[item].keys() for item in sequence)): # generate every valid sequence of bin positions WITHOUT regard to repetition or uniqueness
+        if unique_bins and not is_unique(idxs):
+            continue # skip non-unique bin choices if the option is set
+
+        if draw_without_repeats:
+            choice_inventory = deepcopy(symbol_inventory) # make a new copy for each path check
+            for i, sym in zip(idxs, sequence, strict=True):
+                if choice_inventory[sym][i] == 0:
+                    overdrawn = True
+                    break
+                choice_inventory[sym][i] -= 1
+            else:
+                overdrawn = False
+
+            if overdrawn:
+                continue
+
+        yield idxs # only yield if all specified conditions are met

polymerist/genutils/sequences/seqops.py

@@ -1,12 +1,9 @@
 '''Generic operations for indexing, generating, and iterating over sequences'''
 
-from typing import Generator, Iterable, Sequence, TypeVar, Union
+from typing import Generator, Sequence, TypeVar, Union
 T = TypeVar('T') # generic type for sequence element
 S = TypeVar('S') # generic type for a distinct sequence element
-
-from copy import deepcopy
-from collections import defaultdict, Counter
-from itertools import count, product as cartesian_product
+from itertools import count
 
 
 def is_unique(seq : Sequence) -> bool:
@@ -39,42 +36,8 @@ def cycle_items(seq : Sequence[T], places : int=1) -> list[T]:
         cycle_items([1,2,3,4],  2) -> [3,4,1,2]
         cycle_items([1,2,3,4], -1) -> [4,1,2,3]
     '''
-    n_items = len(seq) # this length call is what requires the input to be a Seuquence and not just an Iterable
+    n_items = len(seq) # this length call is what requires the input to be a Sequence and not just an Iterable
     return [
         seq[i % n_items]
             for i in range(places, places + n_items)
-    ]
-
-def bin_ids_forming_sequence(sequence : Sequence[T], choice_bins : Sequence[Iterable[T]], draw_without_repeats : bool=True, unique_bins : bool=False) -> Generator[tuple[int, ...], None, None]:
-    '''
-    Takes an ordered sequence of N objects of a given type and an ordered of any number of bins, each containing an arbitary amount of unordered objects of the same type
-    Generates all possible N-tuples of bin indices which could produce the target sequence when drawing from those bins in the 
-    
-    if draw_without_repeats=True, will respect the multiplicity of elements in each bin when drawing
-    (i.e. will never have a bin position appear for a given object more times that that object appears in the corresponding bin)
-
-    if unique_bins=True, will only allow each bin to be sampled from once, EVEN if that bin contains elements which may occur later in the sequence
-    '''
-    symbol_inventory = defaultdict(Counter) # keys are objects of type T ("symbols"), values give multiplicities of symbols keyed by bin position
-    for i, choice_bin in enumerate(choice_bins):
-        for sym in choice_bin:
-            symbol_inventory[sym][i] += 1 # NOTE : implementation here requires that T be a hashable type
-
-    for idxs in cartesian_product(*(symbol_inventory[item].keys() for item in sequence)): # generate every valid sequence of bin positions WITHOUT regard to repetition or uniqueness
-        if unique_bins and not is_unique(idxs):
-            continue # skip non-unique bin choices if the option is set
-
-        if draw_without_repeats:
-            choice_inventory = deepcopy(symbol_inventory) # make a new copy for each path check
-            for i, sym in zip(idxs, sequence, strict=True):
-                if choice_inventory[sym][i] == 0:
-                    overdrawn = True
-                    break
-                choice_inventory[sym][i] -= 1
-            else:
-                overdrawn = False
-
-            if overdrawn:
-                continue
-
-        yield idxs # only yield if all specified conditions are met
+    ]

polymerist/polymers/monographs.py

@@ -6,7 +6,7 @@
 from itertools import product as cartesian_product
 
 from ..genutils.iteration import asiterable
-from ..genutils.sequences.seqops import bin_ids_forming_sequence
+from ..genutils.sequences.choices import bin_ids_forming_sequence
 from ..genutils.textual.delimiters import validate_braces
 from ..genutils.fileutils.jsonio.serialize import TypeSerializer
 

polymerist/rdutils/reactions/reactions.py

@@ -16,7 +16,7 @@
 from ..labeling.bondwise import get_bonded_pairs_by_map_nums
 
 from ...genutils.decorators.functional import allow_string_paths, allow_pathlib_paths
-from ...genutils.sequences.seqops import bin_ids_forming_sequence
+from ...genutils.sequences.choices import bin_ids_forming_sequence
 from ...smileslib.substructures import matching_labels_from_substruct_dict