Alphabetic canonical order

mathics/core/atoms/strings.py

@@ -8,7 +8,7 @@
 import sympy
 
 from mathics.core.element import BoxElementMixin
-from mathics.core.keycomparable import BASIC_ATOM_STRING_ELT_ORDER
+from mathics.core.keycomparable import BASIC_ATOM_STRING_ELT_ORDER, wma_str_sort_key
 from mathics.core.symbols import Atom, Symbol, SymbolFalse, SymbolTrue, symbol_set
 from mathics.core.systemsymbols import SymbolFullForm, SymbolInputForm
 
@@ -69,7 +69,7 @@ def element_order(self) -> tuple:
         """
         return (
             BASIC_ATOM_STRING_ELT_ORDER,
-            self.value,
+            wma_str_sort_key(self.value),
             0,
             1,
         )

mathics/core/expression.py

@@ -45,6 +45,7 @@
     GENERAL_EXPRESSION_ELT_ORDER,
     GENERAL_NUMERIC_EXPRESSION_ELT_ORDER,
     Monomial,
+    wma_str_sort_key,
 )
 from mathics.core.structure import LinkedStructure
 from mathics.core.symbols import (
@@ -893,8 +894,9 @@ def element_order(self) -> tuple:
         3: tuple:        list of Elements
         4: 1:        No clue...
         """
-        exps: Dict[str, Union[float, complex]] = {}
+        exps: Dict[Tuple[str, str], Union[float, complex]] = {}
         head = self._head
+
         if head is SymbolTimes:
             for element in self.elements:
                 name = element.get_name()
@@ -904,8 +906,10 @@ def element_order(self) -> tuple:
                     assert isinstance(expr, (Expression, NumericOperators))
                     exp = expr.round_to_float()
                     if var and exp is not None:
+                        var = wma_str_sort_key(var)
                         exps[var] = exps.get(var, 0) + exp
                 elif name:
+                    name = wma_str_sort_key(name)
                     exps[name] = exps.get(name, 0) + 1
         elif self.has_form("Power", 2):
             var = self.elements[0].get_name()
@@ -917,6 +921,7 @@ def element_order(self) -> tuple:
             except AttributeError:
                 exp = None
             if var and exp is not None:
+                var = wma_str_sort_key(var)
                 exps[var] = exps.get(var, 0) + exp
         if exps:
             return (

mathics/core/keycomparable.py

@@ -2,6 +2,7 @@
 Base classes for canonical order.
 
 """
+from typing import Tuple
 
 
 class KeyComparable:
@@ -287,3 +288,38 @@ def __ne__(self, other) -> bool:
 
 BASIC_EXPRESSION_ELT_ORDER = 0x22
 GENERAL_EXPRESSION_ELT_ORDER = 0x23
+
+
+def wma_str_sort_key(s: str) -> Tuple[str, str]:
+    """
+    Return a Tuple providing the sort key
+    reproduce the order of strings and symbols
+    in WMA.
+    For example, the following is a list of sorted
+    strings in the WMA order:
+    `{Abeja, ABEJA, ave de paso, Ave de paso, Ave de Paso, AVe}`
+    The order criteria is: first sort case insensitive, then
+    for the first different character in the original string,
+    lower case comes before upper case.
+    """
+    # An alternative to this implementation would be to map the
+    # characters in a way that
+    # a -> A
+    # A -> B
+    # b -> C
+    # B -> D
+    #  ...
+    # m -> Z
+    # M -> a
+    # n -> b
+    # N -> c
+    #  ...
+    # z -> y
+    # Z -> z
+    # so the result is again a string. Another possibility would be
+    # to return a wrapper class that implement this special comparison
+    # on the fly through the method `__lt__`.
+    return (
+        s.lower(),
+        s.swapcase(),
+    )

mathics/core/symbols.py

@@ -19,6 +19,7 @@
     BASIC_EXPRESSION_ELT_ORDER,
     BASIC_NUMERIC_EXPRESSION_ELT_ORDER,
     Monomial,
+    wma_str_sort_key,
 )
 from mathics.eval.tracing import trace_evaluate
 
@@ -556,15 +557,17 @@ def element_order(self) -> tuple:
         Return a tuple value that is used in ordering elements
         of an expression. The tuple is ultimately compared lexicographically.
         """
+        name = self.name
+        name_key = wma_str_sort_key(name)
         return (
             (
                 BASIC_NUMERIC_EXPRESSION_ELT_ORDER
                 if self.is_numeric()
                 else BASIC_EXPRESSION_ELT_ORDER
             ),
-            Monomial({self.name: 1}),
+            Monomial({name_key: 1}),
             0,
-            self.name,
+            name,
             1,
         )
 

test/builtin/test_sort.py

@@ -1,10 +1,49 @@
 # -*- coding: utf-8 -*-
 
+from test.helper import check_evaluation
 
 from mathics.core.expression import Expression
 from mathics.core.symbols import Symbol, SymbolPlus, SymbolTimes
 
 
+def test_sort_wma():
+    """Test the alphabetic order in WMA for Strings and Symbols"""
+    # In Python, str are ordered as tuples of
+    # ascii codes of the characters. So,
+    #
+    # "Abeja" <"Ave"<"aVe"<"abeja"
+    #
+    # In WMA, strings and symbols are sorted in alphabetical order, with
+    # lowercaps characters coming before than the corresponding upper case.
+    # Then, the same words are sorted in WMA as
+    #
+    # "abeja"< "Abeja"<"aVe"<"Ave"
+    #
+    # Such order is equivalent to use
+    # `lambda s: (s.lower(), s.swapcaps(),)` as sort key.
+    #
+    # Finally, String atoms comes before than Symbols. The following test
+    # reinforce this order.
+    str_expr = (
+        '{"Ave", "aVe", "abeja", AVe, ave, aVe, "Abeja", "ABEJA", '
+        '"AVe", "ave del paraíso", "Ave del paraíso", '
+        '"Ave del Paraíso"} // Sort // InputForm'
+    )
+    str_expected = (
+        '{"abeja", "Abeja", "ABEJA", "aVe", "Ave", "AVe", '
+        '"ave del paraíso", "Ave del paraíso", "Ave del Paraíso", '
+        "ave, aVe, AVe}//InputForm"
+    )
+    check_evaluation(
+        str_expr,
+        str_expected,
+        # to_string_expr=True,
+        # to_string_expected=True,
+        # hold_expected=True,
+        failure_message="WMA order",
+    )
+
+
 def test_Expression_sameQ():
     """
     Test Expression.SameQ