Adding operator testing

Author: rocky

mathics_scanner/data/named-characters.yml

@@ -75,6 +75,19 @@ AGrave:
   unicode-equivalent-name: LATIN SMALL LETTER A WITH GRAVE
   wl-unicode: "\xE0"
   wl-unicode-name: LATIN SMALL LETTER A WITH GRAVE
+Apply:
+  ascii: "@@"
+  has-unicode-inverse: false
+  is-letter-like: false
+  operator-name: Apply
+# @@@ replaces heads at level 1 of expr by f.
+# We add "1" at the end so keys are unique, but
+# the WL operator is still "Apply".
+Apply1:
+  ascii: "@@@"
+  has-unicode-inverse: false
+  is-letter-like: false
+  operator-name: Apply
 AHat:
   esc-alias: a^
   has-unicode-inverse: false
@@ -107,6 +120,7 @@ Aleph:
   unicode-equivalent-name: ALEF SYMBOL
   wl-unicode: "\u2135"
   wl-unicode-name: ALEF SYMBOL
+# Note: the unicode ofr AliasDelimiter doesn't look like "-"
 AliasDelimiter:
   has-unicode-inverse: false
   is-letter-like: false
@@ -1430,6 +1444,7 @@ Distributed:
   is-letter-like: false
   wl-unicode: "\uF3D2"
 Divide:
+  ascii: "/"
   esc-alias: div
   has-unicode-inverse: true
   is-letter-like: false
@@ -4065,6 +4080,12 @@ Infinity:
   unicode-equivalent-name: INFINITY
   wl-unicode: "\u221E"
   wl-unicode-name: INFINITY
+# Infix isn't really an operator
+# Infix:
+#  ascii: "~"
+#  has-unicode-inverse: false
+#  is-letter-like: false
+#  operator-name: Infix
 Integral:
   esc-alias: int
   has-unicode-inverse: false
@@ -4405,12 +4426,13 @@ LeoSign:
   wl-unicode: "\u264C"
   wl-unicode-name: LEO
 Less:
+  ascii: "<"
   has-unicode-inverse: true
   is-letter-like: false
   operator-name: Less
-  unicode-equivalent: <
+  unicode-equivalent: "<"
   unicode-equivalent-name: LESS-THAN SIGN
-  wl-unicode: <
+  wl-unicode: "<"
   wl-unicode-name: LESS-THAN SIGN
 LessEqual:
   ascii: "<="
@@ -5313,6 +5335,11 @@ Pluto:
   unicode-equivalent-name: PLUTO
   wl-unicode: "\u2647"
   wl-unicode-name: PLUTO
+Postfix:
+  ascii: "//"
+  has-unicode-inverse: false
+  is-letter-like: false
+  operator-name: Postfix
 Power:
   ascii: "^"
   has-unicode-inverse: true
@@ -5350,6 +5377,12 @@ PrecedesTilde:
   unicode-equivalent-name: PRECEDES OR EQUIVALENT TO
   wl-unicode: "\u227E"
   wl-unicode-name: PRECEDES OR EQUIVALENT TO
+# Prefix isn't really an operator
+# Prefix:
+#  ascii: "@"
+#  has-unicode-inverse: false
+#  is-letter-like: false
+#  operator-name: Prefix
 Prime:
   esc-alias: ''''
   has-unicode-inverse: false
@@ -5556,10 +5589,11 @@ RawSemicolon:
   unicode-equivalent-name: SEMICOLON
   wl-unicode: ;
   wl-unicode-name: SEMICOLON
+# RawSlash is not an operator. Divide which
+# looks the same, however is.
 RawSlash:
   has-unicode-inverse: true
   is-letter-like: false
-  operator-name: Divide
   unicode-equivalent: /
   unicode-equivalent-name: SOLIDUS
   wl-unicode: /
@@ -5898,15 +5932,17 @@ RoundSpaceIndicator:
   is-letter-like: false
   wl-unicode: "\uF3B2"
 Rule:
-  esc-alias: ->
+  ascii: "->"
+  esc-alias: "->"
   has-unicode-inverse: false
   is-letter-like: false
   operator-name: Rule
   unicode-equivalent: "\u2192"
   unicode-equivalent-name: RIGHTWARDS ARROW
   wl-unicode: "\uF522"
 RuleDelayed:
-  esc-alias: :>
+  ascii: ":>"
+  esc-alias: ":>"
   has-unicode-inverse: true
   operator-name: RuleDelayed
   is-letter-like: false
@@ -6722,6 +6758,11 @@ TabKey:
   has-unicode-inverse: false
   is-letter-like: false
   wl-unicode: "\uF7BE"
+TagSet:
+  ascii: "/:"
+  has-unicode-inverse: false
+  is-letter-like: false
+  operator-name: TagSet
 Tau:
   esc-alias: t
   has-unicode-inverse: false
@@ -6930,11 +6971,6 @@ UnSameQ:
   has-unicode-inverse: false
   is-letter-like: false
   operator-name: UnSameQ
-UnSet:
-  ascii: "=."
-  has-unicode-inverse: false
-  is-letter-like: false
-  operator-name: UnSet
 Union:
   esc-alias: un
   has-unicode-inverse: false
@@ -6950,6 +6986,11 @@ UnionPlus:
   unicode-equivalent-name: MULTISET UNION
   wl-unicode: "\u228E"
   wl-unicode-name: MULTISET UNION
+Unset:
+  ascii: "=."
+  has-unicode-inverse: false
+  is-letter-like: false
+  operator-name: Unset
 UpArrow:
   has-unicode-inverse: false
   is-letter-like: false
@@ -7031,6 +7072,16 @@ Upsilon:
   unicode-equivalent-name: GREEK SMALL LETTER UPSILON
   wl-unicode: "\u03C5"
   wl-unicode-name: GREEK SMALL LETTER UPSILON
+UpSet:
+  ascii: "^="
+  has-unicode-inverse: false
+  is-letter-like: false
+  operator-name: UpSet
+UpSetDelayed:
+  ascii: "^:="
+  has-unicode-inverse: false
+  is-letter-like: false
+  operator-name: UpSetDelayed
 
 # The rendering of Uranus at https://reference.wolfram.com/language/ref/character/Uranus.html
 # looks more like U+26E2 (Astronomical Symbol for Uranus) than the Standard Unicode equavalent

test/test_general_yaml_sanity.py

@@ -5,6 +5,7 @@
 
 yaml_data = load_mathics_character_yaml()
 
+
 def check_attr_is_invertible(attr: str):
     for v in yaml_data.values():
         if attr in v:
@@ -39,6 +40,43 @@ def test_yaml_field_names():
         assert diff == set(), f"Item {k} has unknown fields {diff}"
 
 
+def test_operators():
+    ascii_seen = set()
+    operator_name_seen = set()
+    dup_operators = set(["Apply"])
+    for k, v in yaml_data.items():
+        if "ascii" in v:
+            if len(v["ascii"]) > 1:
+                assert (
+                    "operator-name" in v
+                ), "In %s: ASCII with more than one characters must be an operator" % (
+                    k
+                )
+                pass
+        else:
+            assert "wl-unicode" in v, (
+                "In %s: there must be either an ascii name or have a wl-unicode"
+                % k
+            )
+        if "operator-name" not in v:
+            continue
+
+        assert not v["is-letter-like"], "Operator %s should not be letter-like" % k
+
+        assert "ascii" in v, 'Operator %s should have an "ascii" field' % k
+        ascii = v["ascii"]
+        assert ascii not in ascii_seen
+        ascii_seen.add(ascii)
+
+        operator_name = v["operator-name"]
+        if operator_name in dup_operators:
+            continue
+        assert (
+            operator_name not in operator_name_seen
+        ), "Operator name %s has operator %s already been seen" % (operator_name, k)
+        operator_name_seen.add(operator_name)
+
+
 def test_wl_unicode_name():
     for k, v in yaml_data.items():
         if "wl-unicode" not in v:
@@ -77,7 +115,9 @@ def test_unicode_name():
             try:
                 expected_name = " + ".join(unicodedata.name(c) for c in uni)
             except ValueError:
-                import pdb; pdb.set_trace()
+                import pdb
+
+                pdb.set_trace()
                 raise ValueError(
                     f"{k}'s unicode-equivalent doesn't have a unicode name (it's not valid unicode)"
                 )
@@ -97,6 +137,7 @@ def test_unicode_name():
                 "unicode-equivalent-name" not in v
             ), f"{k} has unicode-equivalent-name set to {v['unicode-equivalent-name']} but it doesn't have a unicode equivalent"
 
+
 def test_wl_unicode():
     for k, v in yaml_data.items():
         if "operator-name" in v:
@@ -106,6 +147,7 @@ def test_wl_unicode():
                 continue
         assert "wl-unicode" in v, f"{k} has no wl-unicode attribute"
 
+
 def test_general_yaml_sanity():
     # Check if required attributes are in place
     check_has_attr("is-letter-like")