Merge pull request #131 from barucden/hyphen

Replace hyphen for minus sign only inside math

Author: Kolaru

src/parser/commands_registration.jl

@@ -110,10 +110,6 @@ for symbol in spaced_symbols
     symbol_to_canonical[symbol] = TeXExpr(:spaced, symbol_expr)
 end
 
-# Special case for hyphen that must be replaced by a minus sign
-# TODO Make sure it is not replaced outside of math mode and when starting a group
-symbol_to_canonical['-'] = TeXExpr(:spaced, TeXExpr(:symbol, '−'))
-
 for com_str in spaced_commands
     symbol = get_symbol_char(com_str)
     symbol_expr = TeXExpr(:symbol, symbol)

src/parser/parser.jl

@@ -62,16 +62,16 @@ function push_down!(stack)
 
     if head(first(stack)) in [:subscript, :superscript]
         decoration = pop!(stack)
-        decorated = pop!(first(stack)) 
+        decorated = pop!(first(stack))
         decorated.args[subsuperindex(head(decoration))] = first(decoration.args)
         push!(first(stack).args, decorated)
-    end 
+    end
 
     conclude_command!!(stack)
 end
 
 function conclude_command!!(stack)
-    com = first(stack) 
+    com = first(stack)
     head(com) != :command && return false
     nargs = length(com.args) - 1
 
@@ -111,7 +111,10 @@ arguments.
 Setting `showdebug` to `true` show a very verbose break down of the parsing.
 """
 function texparse(tex ; root = TeXExpr(:lines), showdebug = false)
+    contains_math = occursin(raw"$", tex)
+
     stack = Stack{Any}()
+    inside_math = false
     push!(stack, root)
     push!(stack, TeXExpr(:line))
 
@@ -125,12 +128,14 @@ function texparse(tex ; root = TeXExpr(:lines), showdebug = false)
         try
             if token == dollar
                 if head(first(stack)) == :inline_math
+                    inside_math = false
                     push_down!(stack)
                 else
+                    inside_math = true
                     push!(stack, TeXExpr(:inline_math))
                 end
             elseif token == newline
-                if length(stack) > 2                
+                if length(stack) > 2
                     throw(TeXParseError("unexpected new line", stack, length(tex), tex))
                 end
 
@@ -188,7 +193,20 @@ function texparse(tex ; root = TeXExpr(:lines), showdebug = false)
                     push!(stack, TeXExpr(dec))
                 end
             elseif token == char
-                push!(stack, canonical_expr(tex[pos])) 
+                c = tex[pos]
+
+                # hyphen replaced by minus sign if expression does not contain
+                # inline math at all, or the hyphen is inside inline math
+                if c == '-'
+                    if !contains_math || inside_math
+                        expr = TeXExpr(:spaced, TeXExpr(:symbol, '−'))
+                    else
+                        expr = TeXExpr(:char, '-')
+                    end
+                else
+                    expr = canonical_expr(c)
+                end
+                push!(stack, expr)
                 push_down!(stack)
             end
         catch err
@@ -210,4 +228,4 @@ function texparse(tex ; root = TeXExpr(:lines), showdebug = false)
     else
         return lines
     end
-end
+end

src/parser/texexpr.jl

@@ -125,7 +125,7 @@ end
 function Base.:(==)(tex1::TeXExpr, tex2::TeXExpr)
     childs1 = children(tex1)
     childs2 = children(tex2)
-    
+
     length(childs1) != length(childs2) && return false
 
     return all(childs1 .== childs2)
@@ -141,7 +141,7 @@ function leafmap(f, texexpr::TeXExpr)
     isleaf(texexpr) && return f(texexpr)
 
     args = map(texexpr.args) do arg
-        isnothing(arg) && return nothing
+        arg isa TeXExpr || return arg
         return leafmap(f, arg)
     end
 

test/parser.jl

@@ -56,14 +56,14 @@ end
     @testset "Fonts" begin
         test_parse(raw"\mathrm{u}", (:font, :rm, (:char, 'u')))
         test_parse(raw"\text{u}", (:text, :rm, (:char, 'u')))
-        
-        test_parse(raw"\mathrm{u v}", (:text, :rm, 
+
+        test_parse(raw"\mathrm{u v}", (:text, :rm,
             (:group,
                 (:char, 'u'),
                 (:char, ' '),
                 (:char, 'v')
             )))
-        test_parse(raw"\text{u v}", (:text, :rm, 
+        test_parse(raw"\text{u v}", (:text, :rm,
             (:group,
                 (:char, 'u'),
                 (:char, ' '),
@@ -177,6 +177,14 @@ end
         )
         # Hyphen must be replaced by a minus sign
         test_parse(raw"-", (:spaced, (:symbol, '−')))
+
+        test_parse(raw"a-b $c-d$",
+                   (:char, 'a'), (:char, '-'), (:char, 'b'),
+                   (:char, ' '),
+                   (:inline_math,
+                    (:char, 'c'),
+                    (:spaced, (:symbol, '−')),
+                    (:char, 'd')))
     end
 
     @testset "Subscript and superscript" begin
@@ -205,4 +213,4 @@ end
             (:decorated, (:symbol, 'ω'), 'k', nothing)
         )
     end
-end
+end