plurimath 0.2.0 → 0.2.1

data/lib/plurimath/asciimath/constants.rb

@@ -3,233 +3,226 @@
 module Plurimath
   class Asciimath
     class Constants
-      PARENTHESIS = {
+      TABLE_PARENTHESIS = {
+        "[": "]",
+        "{": "}",
         "(": ")",
+        "|": "|",
+      }.freeze
+      PARENTHESIS = {
         "(:": ":)",
-        "{": "}",
         "{:": ":}",
+        "(": ")",
+        "{": "}",
         "[": "]",
       }.freeze
-      SYMBOLS = %i[
-        twoheadrightarrowtail
-        twoheadrightarrow
-        rightarrowtail
-        Leftrightarrow
-        leftrightarrow
-        varepsilon
-        Rightarrow
-        rightarrow
-        Leftarrow
-        leftarrow
-        backslash
-        downarrow
-        therefore
-        triangle
-        vartheta
-        bigwedge
-        rceiling
-        lceiling
-        setminus
-        supseteq
-        subseteq
-        emptyset
-        "|\\ |"
-        upsilon
-        diamond
-        uparrow
-        implies
-        partial
-        epsilon
-        because
-        |cdots|
-        |ldots|
-        bigcap
-        bigvee
-        propto
-        approx
-        exists
-        forall
-        otimes
-        ltimes
-        bowtie
-        rtimes
-        models
-        mapsto
-        bigcup
-        lambda
-        Lambda
-        "(x:}"
-        "{:x)"
-        succeq
-        preceq
-        varphi
-        rfloor
-        lfloor
-        square
-        |quad|
-        supset
-        subset
-        vdash
-        times
-        wedge
-        oplus
-        alpha
-        nabla
-        delta
-        theta
-        gamma
-        Gamma
-        Delta
-        Theta
-        kappa
-        ddots
-        vdots
-        |...|
-        equiv
-        Sigma
-        Omega
-        omega
-        aleph
-        infty
-        sigma
-        frown
-        notin
-        angle
-        succ
-        prec
-        cong
-        \\\\
-        star
-        odot
-        cdot
-        rarr
-        darr
-        prop
-        lArr
-        rArr
-        uarr
-        beta
-        hArr
-        harr
-        larr
-        grad
-        circ
-        >->>
-        zeta
-        iota
-        /_\\
-        sube
-        sup
-        sub
-        !in
-        eta
-        top
-        |><|
-        __|
-        |__
-        ":'"
-        Phi
-        supe
-        Psi
-        psi
-        chi
-        >-=
-        -<=
-        mgt
-        mlt
-        ><|
-        |==
-        |--
-        vvv
-        vee
-        nnn
-        cap
-        ^^^
-        div
-        ast
-        bot
-        del
-        and
-        neg
-        not
-        uuu
-        cup
-        iff
-        phi
-        rho
-        tau
-        |->
-        >->
-        <=>
-        ->>
-        ^^
-        nn
-        vv
-        TT
-        EE
-        ox
-        xx
-        o+
-        o.
-        |><
-        _|_
-        ***
-        **
-        -:
-        ~~
-        //
-        pm
-        O/
-        ->
-        =>
-        +-
-        to
-        Xi
-        AA
-        if
-        uu
-        or
-        lt
-        ne
-        nu
-        mu
-        ZZ
-        RR
-        QQ
-        NN
-        CC
-        >>
-        <<
-        ~|
-        !=
-        >-
-        -<
-        ~=
-        -=
-        pi
-        xi
-        oo
-        Pi
-        :.
-        gg
-        ll
-        ge
-        >=
-        le
-        <=
-        gt
-        |~
-        in
-        /_
-        >
-        @
-        /
-        *
-        -
-        <
-        =
-        +
-      ].freeze
+      SYMBOLS = {
+        twoheadrightarrowtail: :"&#x2916;",
+        twoheadrightarrow: :"&#x21A0;",
+        rightarrowtail: :"&#x21A3;",
+        Leftrightarrow: :"&#x21D4;",
+        leftrightarrow: :"&#x2194;",
+        Rightarrow: :"&#x21D2;",
+        rightarrow: :"&#x2192;",
+        varepsilon: :"&#x25B;",
+        Leftarrow: :"&#x21D0;",
+        leftarrow: :"&#x2190;",
+        downarrow: :"&#x2193;",
+        therefore: :"&#x2234;",
+        backslash: :"\\",
+        setminus: :"\\",
+        triangle: :"&#x25B3;",
+        bigwedge: :"&#x22C0;",
+        rceiling: :"&#x2309;",
+        lceiling: :"&#x2308;",
+        supseteq: :"&#x2287;",
+        subseteq: :"&#x2286;",
+        vartheta: :"&#x3D1;",
+        emptyset: :"&#x2205;",
+        diamond: :"&#x22C4;",
+        uparrow: :"&#x2191;",
+        implies: :"&#x21D2;",
+        partial: :"&#x2202;",
+        because: :"&#x2235;",
+        upsilon: :"&#x3C5;",
+        epsilon: :"&#x3B5;",
+        bigcap: :"&#x22C2;",
+        bigvee: :"&#x22C1;",
+        propto: :"&#x221D;",
+        approx: :"&#x2248;",
+        exists: :"&#x2203;",
+        forall: :"&#x2200;",
+        otimes: :"&#x2297;",
+        ltimes: :"&#x22C9;",
+        bowtie: :"&#x22C8;",
+        rtimes: :"&#x22CA;",
+        models: :"&#x22A8;",
+        mapsto: :"&#x21A6;",
+        bigcup: :"&#x22C3;",
+        succeq: :"&#x2AB0;",
+        preceq: :"&#x2AAF;",
+        rfloor: :"&#x230B;",
+        lfloor: :"&#x230A;",
+        square: :"&#x25A1;",
+        supset: :"&#x2283;",
+        subset: :"&#x2282;",
+        lambda: :"&#x3BB;",
+        Lambda: :"&#x39B;",
+        varphi: :"&#x3C6;",
+        ">->>": :"&#x2916;",
+        "/_\\": :"&#x25B3;",
+        "|><|": :"&#x22C8;",
+        kappa: :"&#x3BA;",
+        Delta: :"&#x394;",
+        delta: :"&#x3B4;",
+        gamma: :"&#x3B3;",
+        Gamma: :"&#x393;",
+        Theta: :"&#x398;",
+        theta: :"&#x3B8;",
+        alpha: :"&#x3B1;",
+        aleph: :"&#2135;",
+        infty: :"&#221E;",
+        equiv: :"&#2261;",
+        frown: :"&#2322;",
+        notin: :"&#2209;",
+        angle: :"&#2220;",
+        "!in": :"&#x2209;",
+        cdots: :"&#x22EF;",
+        vdash: :"&#x22A2;",
+        wedge: :"&#x2227;",
+        oplus: :"&#x2295;",
+        nabla: :"&#x2207;",
+        ddots: :"&#x22F1;",
+        vdots: :"&#x22EE;",
+        Sigma: :"&#3A3;",
+        Omega: :"&#3A9;",
+        omega: :"&#3C9;",
+        sigma: :"&#3C3;",
+        times: :"&#xD7;",
+        ldots: :"...",
+        ">-=": :"&#x2AB0;",
+        "-<=": :"&#x2AAF;",
+        "><|": :"&#x22CA;",
+        "|==": :"&#x22A8;",
+        "|--": :"&#x22A2;",
+        "^^^": :"&#x22C0;",
+        "|->": :"&#x21A6;",
+        ">->": :"&#x21A3;",
+        "->>": :"&#x21A0;",
+        "__|": :"&#x230B;",
+        "|__": :"&#x230A;",
+        "|><": :"&#x22C9;",
+        "_|_": :"&#x22A5;",
+        "***": :"&#x22C6;",
+        "<=>": :"&#x21D4;",
+        quad: :"&#xA0;&#xA0;",
+        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: :"&#227B;",
+        prec: :"&#227A;",
+        cong: :"&#2245;",
+        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;",
+        "+-": :"&#xB1;",
+        "-:": :"&#xF7;",
+        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;",
+        "@": :"&#x2218;",
+        "*": :"&#x22C5;",
+        "<": :"&lt;",
+        ">": :"&gt;",
+        "-": :"-",
+        "=": :"=",
+        "+": :"+",
+        "/": :"/",
+        nn: :"&#x2229;",
+        vv: :"&#x2228;",
+        TT: :"&#x22A4;",
+        EE: :"&#x2203;",
+        ox: :"&#x2297;",
+        to: :"&#x2192;",
+        AA: :"&#x2200;",
+        uu: :"&#x222A;",
+        ne: :"&#x2260;",
+        ZZ: :"&#x2124;",
+        RR: :"&#x211D;",
+        QQ: :"&#x211A;",
+        NN: :"&#x2115;",
+        CC: :"&#x2102;",
+        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: :"&gt;",
+        lt: :"&lt;",
+      }.freeze
       UNARY_CLASSES = %i[
         arccos
         arcsin

data/lib/plurimath/asciimath/parse.rb

@@ -4,17 +4,32 @@ require "parslet"
 module Plurimath
   class Asciimath
     class Parse < Parslet::Parser
-      rule(:base)    { str("_").as(:_) }
+      rule(:space) { str(" ") }
+      rule(:base)  { str("_").as(:_) }
+      rule(:power) { str("^").as(:^) }
+      rule(:comma) { (str(",") >> space) | str(",") }
 
-      rule(:power)   { str("^").as(:^) }
+      rule(:symbols)    { arr_to_expression(Constants::SYMBOLS.keys, :symbol) }
+      rule(:font_style) { arr_to_expression(Constants::FONT_STYLES, :fonts) }
 
-      rule(:symbols) { arr_to_expression(Constants::SYMBOLS, :symbol) }
+      rule(:unary_functions)  { arr_to_expression(Constants::UNARY_CLASSES) }
+      rule(:binary_functions) { arr_to_expression(Constants::BINARY_CLASSES) }
 
-      rule(:unary_functions) { arr_to_expression(Constants::UNARY_CLASSES) }
+      rule(:left_right_open_paren)  { str("(") | str("[") }
+      rule(:left_right_close_paren) { str(")") | str("]") }
 
-      rule(:font_style) { arr_to_expression(Constants::FONT_STYLES, :fonts) }
+      rule(:left_right) do
+        (str("left") >> left_right_open_paren.as(:left) >> iteration.as(:left_right_value) >> str("right") >> left_right_close_paren.as(:right)) |
+          table
+      end
 
-      rule(:binary_functions) { arr_to_expression(Constants::BINARY_CLASSES) }
+      rule(:open_table) do
+        arr_to_expression(Constants::TABLE_PARENTHESIS.keys, :table_left)
+      end
+
+      rule(:close_table) do
+        arr_to_expression(Constants::TABLE_PARENTHESIS.values, :table_right)
+      end
 
       rule(:lparen) do
         Constants::PARENTHESIS.keys.reduce do |expression, parenthesis|
@@ -43,6 +58,21 @@ module Plurimath
           match("[0-9]").repeat(1).as(:number)
       end
 
+      rule(:table) do
+        (open_table.as(:table_left) >> tr >> close_table.as(:table_right)) |
+          (str("left") >> left_right_open_paren.as(:left) >> tr >> str("right") >> left_right_close_paren.as(:right))
+      end
+
+      rule(:tr) do
+        ((str("[").as(:open_tr) >> td.as(:tds_list) >> str("]")).as(:table_row) >> comma >> tr.as(:expr)) |
+          (str("[").as(:open_tr) >> td.as(:tds_list) >> str("]")).as(:table_row)
+      end
+
+      rule(:td) do
+        (sequence.as(:td) >> str(",") >> sequence.as(:tds)) |
+          sequence.as(:td)
+      end
+
       rule(:sequence) do
         (lparen >> expression >> rparen).as(:intermediate_exp) |
           (binary_functions >> sequence.as(:base) >> sequence.maybe.as(:exponent)).as(:binary) |
@@ -53,16 +83,18 @@ module Plurimath
       end
 
       rule(:iteration) do
-        (sequence.as(:dividend) >> str("mod").as(:mod) >> sequence.as(:divisor)).as(:mod) |
+        table.as(:table) |
+          (sequence.as(:dividend) >> str("mod").as(:mod) >> sequence.as(:divisor)).as(:mod) |
           (sequence >> base >> sequence.as(:base) >> power >> sequence.as(:exponent)).as(:power_base) |
           (sequence >> base >> sequence).as(:base) |
           (sequence >> power >> sequence).as(:power) |
           sequence.as(:sequence) |
-          str(" ")
+          space
       end
 
       rule(:expression) do
-        (iteration >> expression).as(:expr) |
+        left_right.as(:left_right) |
+          (iteration >> expression).as(:expr) |
           (iteration >> str("/").as(:/) >> iteration).as(:expr) |
           str("")
       end
@@ -70,9 +102,11 @@ module Plurimath
       root :expression
 
       def arr_to_expression(arr, name = nil)
+        type = arr.first.class
         arr.reduce do |expression, expr_string|
-          expression = str(expression).as(name || expression) if expression.is_a?(Symbol)
-          expression | str(expr_string).as(name || expr_string)
+          as_value = name.nil? ? expr_string || expression : name
+          expression = str(expression).as(as_value) if expression.is_a?(type)
+          expression | str(expr_string).as(as_value)
         end
       end
 

data/lib/plurimath/asciimath/parser.rb

@@ -14,10 +14,11 @@ module Plurimath
 
       def parse
         nodes = Parse.new.parse(text)
-        transformed_tree = Plurimath::Asciimath::Transform.new.apply(nodes)
-        return transformed_tree if transformed_tree.is_a?(Plurimath::Math::Formula)
+        nodes = JSON.parse(nodes.to_json, symbolize_names: true)
+        transformed_tree = Transform.new.apply(nodes)
+        return transformed_tree if transformed_tree.is_a?(Math::Formula)
 
-        Plurimath::Math::Formula.new(transformed_tree)
+        Math::Formula.new(transformed_tree)
       end
     end
   end