plurimath 0.8.2 → 0.8.4

data/lib/plurimath/asciimath/constants.rb

@@ -16,232 +16,7 @@ module Plurimath
         "{": "}",
         "[": "]",
       }.freeze
-      SYMBOLS = {
-        twoheadrightarrowtail: :"⤖",
-        twoheadrightarrow: :"↠",
-        rightarrowtail: :"↣",
-        Leftrightarrow: :"⇔",
-        leftrightarrow: :"↔",
-        Rightarrow: :"⇒",
-        rightarrow: :"→",
-        varepsilon: :"ɛ",
-        Leftarrow: :"⇐",
-        leftarrow: :"←",
-        downarrow: :"↓",
-        therefore: :"∴",
-        backslash: :"\",
-        setminus: :"⧵",
-        triangle: :"△",
-        bigwedge: :"⋀",
-        rceiling: :"⌉",
-        lceiling: :"⌈",
-        supseteq: :"⊇",
-        subseteq: :"⊆",
-        vartheta: :"ϑ",
-        emptyset: :"∅",
-        diamond: :"⋄",
-        uparrow: :"↑",
-        implies: :"⇒",
-        partial: :"∂",
-        because: :"∵",
-        upsilon: :"υ",
-        epsilon: :"ε",
-        bigcap: :"⋂",
-        bigvee: :"⋁",
-        propto: :"∝",
-        approx: :"≈",
-        exists: :"∃",
-        forall: :"∀",
-        otimes: :"⊗",
-        ltimes: :"⋉",
-        bowtie: :"⋈",
-        rtimes: :"⋊",
-        models: :"⊨",
-        mapsto: :"↦",
-        bigcup: :"⋃",
-        succeq: :"⪰",
-        preceq: :"⪯",
-        rfloor: :"⌋",
-        lfloor: :"⌊",
-        square: :"□",
-        supset: :"⊃",
-        subset: :"⊂",
-        lambda: :"λ",
-        Lambda: :"Λ",
-        varphi: :"φ",
-        rangle: :"〉",
-        langle: :"〈",
-        ">->>": :"⤖",
-        "/_\\": :"△",
-        "|><|": :"&#x22c8;",
-        kappa: :"&#x3ba;",
-        Delta: :"&#x394;",
-        delta: :"&#x3b4;",
-        gamma: :"&#x3b3;",
-        Gamma: :"&#x393;",
-        Theta: :"&#x398;",
-        theta: :"&#x3b8;",
-        alpha: :"&#x3b1;",
-        aleph: :"&#x2135;",
-        infty: :"&#x221e;",
-        equiv: :"&#x2261;",
-        frown: :"&#x2322;",
-        notin: :"&#x2209;",
-        angle: :"&#x2220;",
-        prime: :"&#x2032;",
-        "!in": :"&#x2209;",
-        cdots: :"&#x22ef;",
-        vdash: :"&#x22a2;",
-        wedge: :"&#x2227;",
-        oplus: :"&#x2295;",
-        nabla: :"&#x2207;",
-        ddots: :"&#x22f1;",
-        vdots: :"&#x22ee;",
-        Sigma: :"&#x3a3;",
-        Omega: :"&#x3a9;",
-        omega: :"&#x3c9;",
-        sigma: :"&#x3c3;",
-        times: :"&#xd7;",
-        ldots: :"&#x2026;",
-        ">-=": :"&#x2ab0;",
-        "-<=": :"&#x2aaf;",
-        "><|": :"&#x22ca;",
-        "|==": :"&#x22a8;",
-        "|--": :"&#x22a2;",
-        "^^^": :"&#x22c0;",
-        "|->": :"&#x21a6;",
-        ">->": :"&#x21a3;",
-        "->>": :"&#x21a0;",
-        "__|": :"&#x230b;",
-        "|__": :"&#x230a;",
-        "|><": :"&#x22c9;",
-        "_|_": :"&#x22a5;",
-        "***": :"&#x22c6;",
-        "<=>": :"&#x21d4;",
-        "...": :"&#x2026;",
-        "(:": :"&#x2329;",
-        ":)": :"&#x232a;",
-        quad: :"&#x2001;",
-        star: :"&#x22c6;",
-        odot: :"&#x2299;",
-        cdot: :"&#x22c5;",
-        rarr: :"&#x2192;",
-        darr: :"&#x2193;",
-        prop: :"&#x221d;",
-        lArr: :"&#x21d0;",
-        rArr: :"&#x21d2;",
-        uarr: :"&#x2191;",
-        hArr: :"&#x21d4;",
-        harr: :"&#x2194;",
-        larr: :"&#x2190;",
-        grad: :"&#x2207;",
-        circ: :"&#x2218;",
-        sube: :"&#x2286;",
-        supe: :"&#x2287;",
-        succ: :"&#x227b;",
-        prec: :"&#x227a;",
-        cong: :"&#x2245;",
-        beta: :"&#x3b2;",
-        zeta: :"&#x3b6;",
-        iota: :"&#x3b9;",
-        ":'": :"&#x2235;",
-        "^^": :"&#x2227;",
-        "o+": :"&#x2295;",
-        "o.": :"&#x2299;",
-        "**": :"&#x2217;",
-        "~~": :"&#x2248;",
-        "O/": :"&#x2205;",
-        "->": :"&#x2192;",
-        "=>": :"&#x21d2;",
-        ">>": :"&#x232a;",
-        "<<": :"&#x2329;",
-        "~|": :"&#x2309;",
-        "!=": :"&#x2260;",
-        ">-": :"&#x227b;",
-        "-<": :"&#x227a;",
-        "~=": :"&#x2245;",
-        "-=": :"&#x2261;",
-        ":.": :"&#x2234;",
-        ">=": :"&#x2265;",
-        "<=": :"&#x2264;",
-        "|~": :"&#x2308;",
-        "/_": :"&#x2220;",
-        "''": :"&#x2033;",
-        "+-": :"&#xb1;",
-        "-:": :"&#xf7;",
-        "\\ ": :"&#xa0;",
-        "\\": :"\\",
-        "//": :/,
-        sup: :"&#x2283;",
-        sub: :"&#x2282;",
-        top: :"&#x22a4;",
-        vvv: :"&#x22c1;",
-        vee: :"&#x2228;",
-        nnn: :"&#x22c2;",
-        cap: :"&#x2229;",
-        ast: :"&#x2217;",
-        bot: :"&#x22a5;",
-        del: :"&#x2202;",
-        uuu: :"&#x22c3;",
-        cup: :"&#x222a;",
-        iff: :"&#x21d4;",
-        eta: :"&#x3b7;",
-        Phi: :"&#x3a6;",
-        Psi: :"&#x3a8;",
-        psi: :"&#x3c8;",
-        chi: :"&#x3c7;",
-        phi: :"&#x3d5;",
-        rho: :"&#x3c1;",
-        tau: :"&#x3c4;",
-        div: :"&#xf7;",
-        neg: :"&#xac;",
-        not: :"&#xac;",
-        "*": :"&#x22c5;",
-        "@": :"&#x2218;",
-        "<": :"&#x3c;",
-        ">": :"&#x3e;",
-        "/": :"&#x2f;",
-        ":": :"&#x3a;",
-        "!": :"&#x21;",
-        ",": :"&#x2c;",
-        ";": :"&#x3b;",
-        "?": :"&#x3f;",
-        "$": :"&#x24;",
-        "~": :"&#x7e;",
-        "|": :"&#x7c;",
-        "%": :"&#x25;",
-        "'": :"&#x2032;",
-        "&": :"&#x26;",
-        "#": :"&#x23;",
-        "=": :"=",
-        "-": :"-",
-        "+": :"+",
-        nn: :"&#x2229;",
-        vv: :"&#x2228;",
-        TT: :"&#x22a4;",
-        EE: :"&#x2203;",
-        ox: :"&#x2297;",
-        to: :"&#x2192;",
-        AA: :"&#x2200;",
-        uu: :"&#x222a;",
-        ne: :"&#x2260;",
-        oo: :"&#x221e;",
-        ge: :"&#x2265;",
-        le: :"&#x2264;",
-        in: :"&#x2208;",
-        nu: :"&#x3bd;",
-        mu: :"&#x3bc;",
-        pi: :"&#x3c0;",
-        Pi: :"&#x3a0;",
-        xi: :"&#x3be;",
-        Xi: :"&#x39e;",
-        xx: :"&#xd7;",
-        pm: :"&#xb1;",
-        gt: :"&#x3e;",
-        lt: :"&#x3c;",
-        if: :if,
-      }.freeze
-      UNARY_CLASSES = %i[
+      UNARY_CLASSES = %w[
         underbrace
         overbrace
         underline
@@ -287,14 +62,7 @@ module Plurimath
         ul
         ln
       ].freeze
-      BINARY_CLASSES = %i[
-        underset
-        stackrel
-        overset
-        frac
-        root
-      ].freeze
-      FONT_STYLES = %i[
+      FONT_STYLES = %w[
         mathfrak
         mathcal
         mathbb
@@ -310,29 +78,17 @@ module Plurimath
         tt
         ii
       ].freeze
-      TERNARY_CLASSES = %w[
-        prod
-        oint
-        sum
-        int
-      ].freeze
-      SUB_SUP_CLASSES = %w[
-        lim
-        log
-      ].freeze
-      SPECIAL_BOLD_ALPHABETS = %w[
-        ZZ
-        RR
-        QQ
-        NN
-        CC
-      ].freeze
+      SUB_SUP_CLASSES = %w[lim log].freeze
+      TERNARY_CLASSES = %w[prod oint sum int].freeze
+      SPECIAL_BOLD_ALPHABETS = %w[ZZ RR QQ NN CC].freeze
+      BINARY_CLASSES = %w[underset stackrel overset frac root].freeze
+      SKIP_INPUT_PARENS = ["[", "]", "{", "}", "(", ")", "(:", ":)"].freeze
 
       class << self
         def precompile_constants
           @values ||=
             named_hash(UNARY_CLASSES, :unary_class)
-              .merge(named_hash(SYMBOLS.keys, :symbol))
+              .merge(named_hash(symbols_array, :symbol))
               .merge(named_hash(FONT_STYLES, :fonts))
               .merge(named_hash(SPECIAL_BOLD_ALPHABETS, :special_fonts))
           @values.sort_by { |v, _| -v.length }.to_h
@@ -341,6 +97,18 @@ module Plurimath
         def named_hash(hash_or_array, name_key)
           hash_or_array.each_with_object({}) { |d, i| i[d] = name_key }
         end
+
+        def symbols_array
+          @@symbols ||= Utility.symbols_hash(:asciimath).keys.concat(parens_symbols)
+        end
+
+        def wrapper_symbols
+          symbols_array.grep(/__{/)
+        end
+
+        def parens_symbols
+          Utility.parens_hash(:asciimath).keys.delete_if { |sym| SKIP_INPUT_PARENS.include?(sym) }
+        end
       end
     end
   end

data/lib/plurimath/asciimath/parse.rb

@@ -5,7 +5,7 @@ module Plurimath
   class Asciimath
     class Parse < Parslet::Parser
       rule(:td)     { expression.as(:td) }
-      rule(:base)   { str("_") }
+      rule(:base)   { arr_to_expression(Constants.wrapper_symbols, :symbol).absent? >> str("_") }
       rule(:power)  { str("^") }
       rule(:space)  { match(/\s+/) }
       rule(:comma)  { (str(",") >> space?) }
@@ -79,7 +79,7 @@ module Plurimath
           binary_classes |
           ternary_classes |
           hash_to_expression(Constants.precompile_constants) |
-          (match(/[0-9]/).as(:number) >> comma.as(:comma)).repeat(1).as(:comma_separated) |
+          (match(/[0-9]/).as(:number) >> str(",").as(:comma)).repeat(1).as(:comma_separated) |
           quoted_text |
           (str("d").as(:d) >> str("x").as(:x)).as(:intermediate_exp) |
           ((str("left").absent? >> str("right").absent?) >> match["a-zA-Z"].as(:symbol)) |
@@ -88,11 +88,11 @@ module Plurimath
       end
 
       rule(:power_base) do
-        (base >> space? >> sequence.as(:base_value) >> power >> space? >> sequence.as(:power_value)) |
+        (base >> space? >> sequence.as(:base_value) >> power >> (power >> power.maybe).absent? >> space? >> sequence.as(:power_value)) |
           (space? >> base >> space? >> sequence.as(:base_value)).as(:base) |
-          (space? >> power >> space? >> sequence.as(:power_value)).as(:power) |
+          (space? >> power >> (power >> power.maybe).absent? >> space? >> sequence.as(:power_value)).as(:power) |
           (space? >> base >> space? >> power.as(:symbol).as(:base_value)).as(:base) |
-          (space? >> power >> space? >> base.as(:symbol).as(:power_value)).as(:power)
+          (space? >> power >> (power >> power.maybe).absent? >> space? >> base.as(:symbol).as(:power_value)).as(:power)
       end
 
       rule(:power_base_rules) do
@@ -127,7 +127,7 @@ module Plurimath
 
       rule(:frac) do
         (sequence.as(:numerator) >> space? >> match(/(?<!\/)\/(?!\/)/) >> space? >> iteration.as(:denominator)).as(:frac) |
-          ((power_base_rules | power_base).as(:numerator) >> match(/(?<!\/)\/(?!\/)/) >> iteration.as(:denominator)).as(:frac)
+          ((power_base_rules | power_base).as(:numerator) >> space? >> match(/(?<!\/)\/(?!\/)/) >> space? >> iteration.as(:denominator)).as(:frac)
       end
 
       rule(:mod) do
@@ -190,11 +190,21 @@ module Plurimath
         first_value = str(expr.first.to_s)
         case expr.last
         when :symbol then (str("\\").as(:slash) >> match("\s").repeat >> str("\n")) | first_value.as(:symbol)
-        when :unary_class then (first_value.as(:unary_class) >> space? >> sequence.maybe).as(:unary)
+        when :unary_class then unary_functions(first_value)
         when :fonts then first_value.as(:fonts_class) >> space? >> sequence.as(:fonts_value)
         when :special_fonts then first_value.as(:bold_fonts)
         end
       end
+
+      def unary_functions(first_value)
+        if ["'underbrace'", "'ubrace'"].include?(first_value.to_s)
+          (first_value.as(:unary_class) >> space? >> str("_").as(:symbol)).as(:unary) |
+            (first_value.as(:unary_class) >> space? >> sequence.maybe).as(:unary)
+        else
+          (first_value.as(:unary_class).as(:power_base) >> space? >> power_base).as(:unary) |
+            (first_value.as(:unary_class) >> space? >> sequence.maybe).as(:unary)
+        end
+      end
     end
   end
 end