plurimath 0.1.0 → 0.2.1

data/lib/plurimath/math/number.rb

@@ -0,0 +1,37 @@
+# frozen_string_literal: true
+
+module Plurimath
+  module Math
+    class Number
+      attr_accessor :value
+
+      def initialize(value)
+        @value = value
+      end
+
+      def ==(object)
+        object.value == value
+      end
+
+      def to_asciimath
+        value
+      end
+
+      def to_mathml_without_math_tag
+        "<mn>#{value}</mn>"
+      end
+
+      def to_latex
+        value
+      end
+
+      def to_html
+        value
+      end
+
+      def class_name
+        self.class.name.split("::").last.downcase
+      end
+    end
+  end
+end

data/lib/plurimath/math/symbol.rb

@@ -0,0 +1,43 @@
+# frozen_string_literal: true
+
+module Plurimath
+  module Math
+    class Symbol
+      attr_accessor :value
+
+      def initialize(sym)
+        @value = sym
+      end
+
+      def ==(object)
+        object.value == value
+      end
+
+      def to_asciimath
+        symbol = Asciimath::Constants::SYMBOLS.invert[value.to_sym].to_s
+        symbol.empty? ? value : symbol
+      end
+
+      def to_mathml_without_math_tag
+        "<mo>#{mathml_symbol_value}</mo>"
+      end
+
+      def mathml_symbol_value
+        Mathml::Constants::UNICODE_SYMBOLS.invert[value] || value
+      end
+
+      def to_latex
+        symbols = Latex::Constants::SYMBOLS.invert
+        symbols.key?(value) ? "\\#{symbols[value]}" : value
+      end
+
+      def to_html
+        value
+      end
+
+      def class_name
+        self.class.name.split("::").last.downcase
+      end
+    end
+  end
+end

data/lib/plurimath/math.rb

@@ -0,0 +1,56 @@
+# frozen_string_literal: true
+
+require "ox"
+require_relative "unicode"
+require_relative "asciimath"
+require_relative "omml"
+require_relative "mathml"
+require_relative "html"
+require_relative "latex"
+require_relative "unitsml"
+require_relative "math/formula"
+require_relative "math/function"
+require_relative "math/number"
+require_relative "math/symbol"
+require_relative "asciimath/parser"
+require_relative "mathml/parser"
+require_relative "latex/parser"
+require_relative "html/parser"
+require_relative "omml/parser"
+require_relative "utility"
+module Plurimath
+  module Math
+    class Error < StandardError; end
+
+    VALID_TYPES = {
+      omml: Omml,
+      html: Html,
+      latex: Latex,
+      mathml: Mathml,
+      unitsml: Unitsml,
+      unicode: Unicode,
+      asciimath: Asciimath,
+    }.freeze
+
+    def parse(text, type)
+      raise_error! unless valid_type?(type)
+
+      klass = VALID_TYPES[type.to_sym]
+      klass.new(text).to_formula
+    end
+
+    private
+
+    def raise_error!
+      raise Plurimath::Math::Error, Error.new("Type is not valid, "\
+                                              "please enter string or symbol")
+    end
+
+    def valid_type?(type)
+      (type.is_a?(::Symbol) || type.is_a?(String)) &&
+        VALID_TYPES.key?(type.to_sym)
+    end
+
+    module_function :parse, :raise_error!, :valid_type?
+  end
+end

data/lib/plurimath/mathml/constants.rb

@@ -0,0 +1,295 @@
+# frozen_string_literal: true
+
+module Plurimath
+  class Mathml
+    class Constants
+      UNICODE_SYMBOLS = {
+        "α": "alpha",
+        "β": "beta",
+        "γ": "gamma",
+        "Γ": "Gamma",
+        "δ": "delta",
+        "Δ": "Delta",
+        "∆": "Delta",
+        "ε": "epsilon",
+        "ɛ": "varepsilon",
+        "ζ": "zeta",
+        "η": "eta",
+        "θ": "theta",
+        "Θ": "Theta",
+        "ϑ": "vartheta",
+        "ι": "iota",
+        "κ": "kappa",
+        "λ": "lambda",
+        "Λ": "Lambda",
+        "μ": "mu",
+        "ν": "nu",
+        "ξ": "xi",
+        "Ξ": "Xi",
+        "π": "pi",
+        "Π": "Pi",
+        "ρ": "rho",
+        "ς": "beta",
+        "σ": "sigma",
+        "Σ": "Sigma",
+        "τ": "tau",
+        "υ": "upsilon",
+        "φ": "phi",
+        "Φ": "Phi",
+        "ϕ": "varphi",
+        "χ": "chi",
+        "ψ": "psi",
+        "Ψ": "Psi",
+        "ω": "omega",
+        "Ω": "omega",
+        "⋅": "dot",
+        "∙": "*",
+        "·": ".",
+        "∗": "**",
+        "⋆": "***",
+        "×": "xx",
+        "&#x22C9;": "|><",
+        "&#x22CA;": "><|",
+        "&#x22C8;": "|><|",
+        "&#xF7;": "-:",
+        "&#x2218;": "@",
+        "&#x2295;": "o+",
+        "&#x2A01;": "o+",
+        "&#x2297;": "ox",
+        "&#x2299;": " ",
+        "&#x2211;": "sum",
+        "&#x220F;": "prod",
+        "&#x220f;": "prod",
+        "&#x2227;": "^^",
+        "&#x22C0;": "^^^",
+        "&#x2228;": "vv",
+        "&#x22c1;": "vvv",
+        "&#x2229;": "nn",
+        "&#x22C2;": "nnn",
+        "&#x222A;": "cup",
+        "&#x22C3;": "uuu",
+        "&#x2260;": "!=",
+        "&#x2264;": "<=",
+        "&#x2265;": ">=",
+        "&#x227A;": "-<",
+        "&#x227B;": ">-",
+        "&#x2AAF;": "-<=",
+        "&#x2AB0;": " >-=",
+        "&#x2208;": "in",
+        "&#x2209;": "!in",
+        "&#x2282;": "sub",
+        "&#x2283;": "sup",
+        "&#x2286;": "sube",
+        "&#x2287;": "supe",
+        "&#x2261;": "-=",
+        "&#x2245;": "~=",
+        "&#x2248;": "~~",
+        "&#x221D;": "prop",
+        "&#xAC;": "not",
+        "&#x2200;": "AA",
+        "&#x2203;": "EE",
+        "&#x22A5;": "_|_",
+        "&#x22A4;": "TT",
+        "&#x22A2;": "|--",
+        "&#x22A8;": "|==",
+        "&#x2329;": "(:",
+        "&#x232A;": ":)",
+        "&#x27E8;": "<<",
+        "&#x27E9;": ">>",
+        "&#x222B;": "int",
+        "&#x222E;": "oint",
+        "&#x2202;": "del",
+        "&#x2207;": "grad",
+        "&#xB1;": "+-",
+        "&#x2205;": "O/",
+        "&#x221E;": "oo",
+        "&#x2135;": "aleph",
+        "&#x2234;": ":.",
+        "&#x2235;": ":'",
+        "&#x2220;": "/_",
+        "&#x25B3;": "/_\\",
+        "&#x2032;": "'",
+        "&#xA0;&#xA0;": "quad",
+        "&#xA0;&#xA0;&#xA0;&#xA0;": "qquad",
+        "&#x2322;": "frown",
+        "&#x22EF;": "cdots",
+        "&#x22EE;": "vdots",
+        "&#x22F1;": "ddots",
+        "&#x22C4;": "diamond",
+        "&#x25A1;": "square",
+        "&#x230A;": "|__",
+        "&#x230B;": "__|",
+        "&#x2308;": "|~",
+        "&#x2309;": "~|",
+        "&#x2102;": "CC",
+        "&#x2115;": "NN",
+        "&#x211A;": "QQ",
+        "&#x211D;": "RR",
+        "&#x2124;": "ZZ",
+        "&#x2191;": "uarr",
+        "&#x2193;": "darr",
+        "&#x2190;": "larr",
+        "&#x2194;": "harr",
+        "&#x21D2;": "rArr",
+        "&#x21D0;": "lArr",
+        "&#x21D4;": "hArr",
+        "&#x2192;": "->",
+        "&#x21A3;": ">->",
+        "&#x21A0;": "->>",
+        "&#x2916;": ">->>",
+        "&#x21A6;": "|->",
+        "&#x2026;": "...",
+        "&#x2212;": "-",
+        "&#x2061;": "",
+        "&#x23DE;": "obrace",
+        "&#x23DF;": "ubrace",
+      }.freeze
+      SYMBOLS = {
+        "|": "|",
+        "/": "//",
+        "\\": "\\\\",
+        "~": "tilde",
+        "(": "(",
+        ")": ")",
+        "(:": "(:",
+        ":)": ":)",
+        "{": "{",
+        "}": "}",
+        "{:": "{:",
+        ":}": ":}",
+        "]": "]",
+        "[": "[",
+        "=": "=",
+        "+": "+",
+        "-": "-",
+      }.freeze
+      CLASSES = %w[
+        mathfrak
+        underset
+        stackrel
+        overset
+        mathcal
+        arccos
+        arcsin
+        arctan
+        mathsf
+        mathbb
+        mathbf
+        mathtt
+        ubrace
+        obrace
+        cancel
+        tilde
+        floor
+        color
+        frac
+        root
+        oint
+        ceil
+        ddot
+        coth
+        csch
+        sech
+        sinh
+        tanh
+        cosh
+        sqrt
+        norm
+        text
+        prod
+        sec
+        int
+        sin
+        tan
+        cos
+        sum
+        exp
+        gcd
+        glb
+        lcm
+        lub
+        cot
+        csc
+        det
+        dim
+        max
+        min
+        abs
+        bar
+        dot
+        hat
+        vec
+        mod
+        log
+        ul
+        ln
+        f
+        g
+      ].freeze
+      TAGS = %i[
+        annotation-xml
+        annotation_xml
+        mmultiscripts
+        maligngroup
+        malignmark
+        annotation
+        munderover
+        mscarries
+        semantics
+        mphantom
+        mlongdiv
+        menclose
+        mscarry
+        msubsup
+        mpadded
+        maction
+        msgroup
+        mfenced
+        merror
+        munder
+        mtable
+        mstyle
+        mstack
+        mspace
+        msline
+        mfrac
+        mover
+        msrow
+        mroot
+        msqrt
+        msup
+        msub
+        mrow
+        math
+        mtr
+        mtd
+        ms
+        mi
+        mo
+        mn
+      ].freeze
+      BINARY_CLASSES = %i[
+        underset
+        stackrel
+        overset
+        color
+        prod
+        frac
+        root
+        oint
+        int
+        sum
+        mod
+        log
+      ].freeze
+      FONT_CLASSES = %i[
+        double-struck
+        sans-serif
+        monospace
+        fraktur
+        script
+        bold
+      ].freeze
+    end
+  end
+end

data/lib/plurimath/mathml/parse.rb

@@ -0,0 +1,68 @@
+# frozen_string_literal: true
+
+require "parslet"
+module Plurimath
+  class Mathml
+    class Parse < Parslet::Parser
+      rule(:parse_record) do
+        array_to_expression(Constants::CLASSES).as(:class) |
+          array_to_expression(Constants::UNICODE_SYMBOLS.keys).as(:symbol) |
+          array_to_expression(Constants::SYMBOLS.keys).as(:symbol) |
+          match["a-zA-Z"].repeat(1).as(:text) |
+          match(/[0-9]/).repeat(1).as(:number) |
+          str("")
+      end
+
+      rule(:tag) do
+        (parse_tag(:open) >> iteration.as(:iteration) >> parse_tag(:close)).as(:tag) |
+          parse_text_tag.as(:tag)
+      end
+
+      rule(:sequence) { (tag >> sequence.as(:sequence)) | tag }
+
+      rule(:iteration) { (sequence >> iteration.as(:iteration)) | parse_record }
+
+      rule(:expression) do
+        parse_tag(:open) >> iteration.as(:iteration) >> parse_tag(:close)
+      end
+
+      root :expression
+
+      def array_to_expression(array, name = nil)
+        initial_type = array.first.class
+        array.reduce do |expr, tag|
+          expr = str_to_expression(expr, name) if expr.is_a?(initial_type)
+          expr | str_to_expression(tag, name)
+        end
+      end
+
+      def str_to_expression(string, name)
+        return str(string) if name.nil?
+
+        str(string).as(name)
+      end
+
+      def parse_tag(opts)
+        tag = str("<")
+        tag = tag >> str("/") if opts == :close
+        tag = tag >> array_to_expression(Constants::TAGS, opts)
+        tag = tag >> attributes.as(:attributes) if opts == :open
+        tag >> str(">")
+      end
+
+      def attributes
+        (match["a-zA-Z"].repeat.as(:name) >>
+          str("=") >> quoted_string).repeat
+      end
+
+      def quoted_string
+        (str('"') >> match("[^\"]").repeat.as(:value) >> str('"')) |
+          (str("'") >> match("[^\']").repeat.as(:value) >> str("'"))
+      end
+
+      def parse_text_tag
+        str("<mtext>") >> match("[^<]").repeat.as(:quoted_text) >> str("</mtext>")
+      end
+    end
+  end
+end

data/lib/plurimath/mathml/parser.rb

@@ -0,0 +1,26 @@
+# frozen_string_literal: true
+
+require_relative "parse"
+require_relative "constants"
+require_relative "transform"
+module Plurimath
+  class Mathml
+    class Parser
+      attr_accessor :text
+
+      def initialize(text)
+        @text = text.gsub(/\s/, "")
+      end
+
+      def parse
+        tree_t = Plurimath::Mathml::Parse.new.parse(text)
+        tree_t = JSON.parse(tree_t.to_json, symbolize_names: true)
+        formula = Plurimath::Mathml::Transform.new.apply(tree_t)
+        formula = [formula] unless formula.is_a?(Array) || formula.nil?
+        return if formula.nil?
+
+        Plurimath::Math::Formula.new(formula)
+      end
+    end
+  end
+end