allow unicode modifiers after ' (JuliaLang#37247)

Author: simeonschaub

NEWS.md

@@ -25,6 +25,9 @@ New language features
   can now be used to rename imported modules and identifiers ([#1255]).
 * Unsigned literals (starting with `0x`) which are too big to fit in an `UInt128` object
   are now interpreted as `BigInt` ([#23546]).
+* The postfix conjugate transpose operator `'` now accepts Unicode modifiers as
+  suffixes, so e.g. `a'ᵀ` is parsed as `var"'ᵀ"(a)`, which can be defined by the
+  user. `a'ᵀ` parsed as `a' * ᵀ` before, so this is a minor change ([#37247]).
 
 Language changes
 ----------------

base/show.jl

@@ -1131,6 +1131,10 @@ function isidentifier(s::AbstractString)
 end
 isidentifier(s::Symbol) = isidentifier(string(s))
 
+is_op_suffix_char(c::AbstractChar) = ccall(:jl_op_suffix_char, Cint, (UInt32,), c) != 0
+
+_isoperator(s) = ccall(:jl_is_operator, Cint, (Cstring,), s) != 0
+
 """
     isoperator(s::Symbol)
 
@@ -1142,7 +1146,7 @@ julia> Base.isoperator(:+), Base.isoperator(:f)
 (true, false)
 ```
 """
-isoperator(s::Union{Symbol,AbstractString}) = ccall(:jl_is_operator, Cint, (Cstring,), s) != 0
+isoperator(s::Union{Symbol,AbstractString}) = _isoperator(s) || ispostfixoperator(s)
 
 """
     isunaryoperator(s::Symbol)
@@ -1169,7 +1173,26 @@ julia> Base.isbinaryoperator(:-), Base.isbinaryoperator(:√), Base.isbinaryoper
 (true, false, false)
 ```
 """
-isbinaryoperator(s::Symbol) = isoperator(s) && (!isunaryoperator(s) || is_unary_and_binary_operator(s))
+function isbinaryoperator(s::Symbol)
+    return _isoperator(s) && (!isunaryoperator(s) || is_unary_and_binary_operator(s)) &&
+        s !== Symbol("'")
+end
+
+"""
+    ispostfixoperator(s::Union{Symbol,AbstractString})
+
+Return `true` if the symbol can be used as a postfix operator, `false` otherwise.
+
+# Examples
+```jldoctest
+julia> Base.ispostfixoperator(Symbol("'")), Base.ispostfixoperator(Symbol("'ᵀ")), Base.ispostfixoperator(:-)
+(true, true, false)
+```
+"""
+function ispostfixoperator(s::Union{Symbol,AbstractString})
+    s = String(s)
+    return startswith(s, '\'') && all(is_op_suffix_char, SubString(s, 2))
+end
 
 """
     operator_precedence(s::Symbol)
@@ -1594,6 +1617,20 @@ function show_unquoted(io::IO, ex::Expr, indent::Int, prec::Int, quote_level::In
         end
         head !== :row && print(io, cl)
 
+    # transpose
+    elseif (head === Symbol("'") && nargs == 1) || (
+        # ' with unicode suffix is a call expression
+        head === :call && nargs == 2 && args[1] isa Symbol &&
+        ispostfixoperator(args[1]) && args[1] !== Symbol("'")
+    )
+        op, arg1 = head === Symbol("'") ? (head, args[1]) : (args[1], args[2])
+        if isa(arg1, Expr) || (isa(arg1, Symbol) && isoperator(arg1))
+            show_enclosed_list(io, '(', [arg1], ", ", ')', indent, 0)
+        else
+            show_unquoted(io, arg1, indent, 0, quote_level)
+        end
+        print(io, op)
+
     # function call
     elseif head === :call && nargs >= 1
         func = args[1]
@@ -1622,7 +1659,7 @@ function show_unquoted(io::IO, ex::Expr, indent::Int, prec::Int, quote_level::In
             end
 
         # unary operator (i.e. "!z")
-        elseif isa(func,Symbol) && func in uni_ops && length(func_args) == 1
+        elseif isa(func,Symbol) && length(func_args) == 1 && func in uni_ops
             show_unquoted(io, func, indent, 0, quote_level)
             arg1 = func_args[1]
             if isa(arg1, Expr) || (isa(arg1, Symbol) && isoperator(arg1))
@@ -1959,17 +1996,6 @@ function show_unquoted(io::IO, ex::Expr, indent::Int, prec::Int, quote_level::In
             parens && print(io, ")")
         end
 
-    # transpose
-    elseif head === Symbol('\'') && nargs == 1
-        if isa(args[1], Symbol)
-            show_unquoted(io, args[1], 0, 0, quote_level)
-        else
-            print(io, "(")
-            show_unquoted(io, args[1], 0, 0, quote_level)
-            print(io, ")")
-        end
-        print(io, head)
-
     # `where` syntax
     elseif head === :where && nargs > 1
         parens = 1 <= prec

src/flisp/julia_extensions.c

@@ -152,7 +152,7 @@ JL_DLLEXPORT int jl_id_char(uint32_t wc)
 #include "julia_opsuffs.h"
 
 // chars that can follow an operator (e.g. +) and be parsed as part of the operator
-int jl_op_suffix_char(uint32_t wc)
+JL_DLLEXPORT int jl_op_suffix_char(uint32_t wc)
 {
     static htable_t jl_opsuffs; // XXX: requires uv_once
     if (!jl_opsuffs.size) { // initialize hash table of suffixes

src/julia-parser.scm

@@ -228,12 +228,12 @@
 
 (define (op-or-sufchar? c) (or (op-suffix-char? c) (opchar? c)))
 
-(define (read-operator port c)
-  (if (and (eqv? c #\*) (eqv? (peek-char port) #\*))
+(define (read-operator port c0 (postfix? #f))
+  (if (and (eqv? c0 #\*) (eqv? (peek-char port) #\*))
       (error "use \"x^y\" instead of \"x**y\" for exponentiation, and \"x...\" instead of \"**x\" for splatting."))
   (if (or (eof-object? (peek-char port)) (not (op-or-sufchar? (peek-char port))))
-      (symbol (string c)) ; 1-char operator
-      (let ((str (let loop ((str (string c))
+      (symbol (string c0)) ; 1-char operator
+      (let ((str (let loop ((str (string c0))
                             (c   (peek-char port))
                             (in-suffix? #f))
                    (if (eof-object? c)
@@ -257,7 +257,10 @@
                                        (and (or (eq? opsym '<-) (eq? opsym '.<-))
                                             (read-char port)
                                             (begin0 (eqv? (peek-char port) #\-)
-                                                    (io.ungetc port #\-))))
+                                                    (io.ungetc port #\-)))
+                                       ;; consume suffixes after ', only if parsing a call chain
+                                       ;; otherwise 'ᵀ' would parse as (|'| |'ᵀ|)
+                                       (and postfix? (eqv? c0 #\') sufchar?))
                                    (begin (read-char port)
                                           (loop newop (peek-char port) sufchar?))
                                    str))
@@ -1244,7 +1247,10 @@
              (if (not (ts:space? s))
                  (begin
                    (take-token s)
-                   (loop (list t ex)))
+                   (let ((t (read-operator (ts:port s) #\' #t)))
+                     (loop (if (eq? t '|'|)
+                               (list t ex)
+                               (list 'call t ex)))))
                  ex))
             ((|.'|) (error "the \".'\" operator is discontinued"))
             ((#\{ )

test/show.jl

@@ -2028,3 +2028,12 @@ end
 
 @test sprint(show, :(./)) == ":((./))"
 @test sprint(show, :((.|).(.&, b))) == ":((.|).((.&), b))"
+
+@test sprint(show, :(a'ᵀ)) == ":(a'ᵀ)"
+@test sprint(show, :((+)')) == ":((+)')"
+for s in (Symbol("'"), Symbol("'⁻¹"))
+    @test Base.isoperator(s)
+    @test !Base.isunaryoperator(s)
+    @test !Base.isbinaryoperator(s)
+    @test Base.ispostfixoperator(s)
+end

test/syntax.jl

@@ -2445,3 +2445,10 @@ end
 import .TestImportAs.Mod2 as M2
 @test !@isdefined(Mod2)
 @test M2 === TestImportAs.Mod2
+
+@testset "unicode modifiers after '" begin
+    @test Meta.parse("a'ᵀ") == Expr(:call, Symbol("'ᵀ"), :a)
+    @test Meta.parse("a'⁻¹") == Expr(:call, Symbol("'⁻¹"), :a)
+    @test Meta.parse("a'ᵀb") == Expr(:call, :*, Expr(:call, Symbol("'ᵀ"), :a), :b)
+    @test Meta.parse("a'⁻¹b") == Expr(:call, :*, Expr(:call, Symbol("'⁻¹"), :a), :b)
+end