asciimath2unitsml 0.3.0 → 0.4.0

data/lib/asciimath2unitsml/read.rb

@@ -0,0 +1,42 @@
+module Asciimath2UnitsML
+  class Conv
+    def read_yaml(path)
+      validate_yaml(symbolize_keys(YAML
+        .load_file(File.join(File.join(File.dirname(__FILE__), path)))), path)
+    end
+
+    def symbolize_keys(hash)
+      return hash if hash.is_a? String
+
+      hash.inject({}) do |result, (key, value)|
+        new_key = case key
+                  when String then key.to_sym
+                  else key
+                  end
+        new_value = case value
+                    when Hash then symbolize_keys(value)
+                    when Array then value.map { |m| symbolize_keys(m) }
+                    else value
+                    end
+        result[new_key] = new_value
+        result
+      end
+    end
+
+    def flip_name_and_symbol(hash)
+      hash.each_with_object({}) do |(_k, v), m|
+        next if v.name.nil? || v.name.empty?
+
+        m[v.symbolid] = v
+      end
+    end
+
+    def flip_name_and_symbols(hash)
+      hash.each_with_object({}) do |(_k, v), m|
+        next if v.name.nil? || v.name.empty?
+
+        v.symbolids.each { |s| m[s] = v }
+      end
+    end
+  end
+end

data/lib/asciimath2unitsml/render.rb

@@ -1,13 +1,14 @@
 module Asciimath2UnitsML
   class Conv
-    def multiplier(x)
-      case x
+    def multiplier(val)
+      case val
       when :space
         { html: "&#xA0;", mathml: "<mo rspace='thickmathspace'>&#x2062;</mo>" }
       when :nospace
         { html: "", mathml: "<mo>&#x2062;</mo>" }
       else
-        { html: HTMLEntities.new.encode(x), mathml: "<mo>#{HTMLEntities.new.encode(x)}</mo>" }
+        { html: HTMLEntities.new.encode(val),
+          mathml: "<mo>#{HTMLEntities.new.encode(val)}</mo>" }
       end
     end
 
@@ -15,28 +16,31 @@ module Asciimath2UnitsML
       @symbols[unit][style] || unit
     end
 
-    def htmlent(x)
-      HTMLEntities.new.decode(x).split(/([<>&])/)
-        .map { |c| /[<>'"]/.match(c) ? c : HTMLEntities.new.encode(c, :hexadecimal) }.join
+    def htmlent(xml)
+      HTMLEntities.new.decode(xml).split(/([<>&])/).map do |c|
+        /[<>'"]/.match?(c) ? c : HTMLEntities.new.encode(c, :hexadecimal)
+      end.join
     end
 
     def htmlsymbol(units, normalise)
       units.map do |u|
-        if u[:multiplier] then u[:multiplier] == "*" ? @multiplier[:html] : u[:multiplier]
+        if u[:multiplier]
+          u[:multiplier] == "*" ? @multiplier[:html] : u[:multiplier]
         elsif u[:unit].nil? && u[:prefix]
           @prefixes[u[:prefix]].html
         else
-          base = (u[:prefix] || "") + render(normalise ? @units[u[:unit]].symbolid : u[:unit], :html)
+          base = (u[:prefix] || "") +
+            render(normalise ? @units[u[:unit]].symbolid : u[:unit], :html)
           htmlsymbol_exponent(u, base)
         end
-      end.join("")
+      end.join
     end
 
-    def htmlsymbol_exponent(u, base)
-      if u[:display_exponent] == "0.5"
+    def htmlsymbol_exponent(unit, base)
+      if unit[:display_exponent] == "0.5"
         base = "&#x221a;#{base}"
-      elsif u[:display_exponent]
-        exp = "<sup>#{u[:display_exponent].sub(/-/, "&#x2212;")}</sup>"
+      elsif unit[:display_exponent]
+        exp = "<sup>#{unit[:display_exponent].sub(/-/, '&#x2212;')}</sup>"
         base += exp
       end
       base
@@ -44,41 +48,54 @@ module Asciimath2UnitsML
 
     def mathmlsymbol(units, normalise, multiplier = nil)
       multiplier = multiplier ? "<mo>#{multiplier}</mo>" : @multiplier[:mathml]
-      exp = units.map do |u|
-        if u[:multiplier] then u[:multiplier] == "*" ? multiplier : "<mo>#{u[:multiplier]}</mo>"
+      units.map do |u|
+        if u[:multiplier]
+          u[:multiplier] == "*" ? multiplier : "<mo>#{u[:multiplier]}</mo>"
         elsif u[:unit].nil? && u[:prefix]
           %(<mi mathvariant='normal'>#{htmlent(@prefixes[u[:prefix]].html)}</mi>)
         else
           mathmlsymbol1(u, normalise)
         end
-      end.join("")
+      end.join
+    end
+
+    def mathmlsymbol1(unit, normalise)
+      base = if unit[:dim]
+               render(normalise ? @dimensions[unit[:dim]].symbolid : unit[:dim],
+                      :mathml)
+             else
+               render(normalise ? @units[unit[:unit]].symbolid : unit[:unit],
+                      :mathml)
+             end
+      unit[:prefix] and base = mathmlsymbol1_prefixed(unit, base)
+      mathmlsymbol_exponent(unit, base)
     end
 
-    def mathmlsymbol1(u, normalise)
-      base = render(normalise ? @units[u[:unit]].symbolid : u[:unit], :mathml)
-      if u[:prefix]
-        prefix = htmlent(@prefixes[u[:prefix]].html)
-        base = base.match(/<mi mathvariant='normal'>/) ?
-          base.sub(/<mi mathvariant='normal'>/, "<mi mathvariant='normal'>#{prefix}") :
-          "<mrow><mi mathvariant='normal'>#{prefix}#{base}</mrow>"
+    def mathmlsymbol1_prefixed(unit, base)
+      prefix = htmlent(@prefixes[unit[:prefix]].html)
+      if /<mi mathvariant='normal'>/.match?(base)
+        base.sub(/<mi mathvariant='normal'>/,
+                 "<mi mathvariant='normal'>#{prefix}")
+      else
+        "<mrow><mi mathvariant='normal'>#{prefix}#{base}</mrow>"
       end
-      mathmlsymbol_exponent(u, base)
     end
 
-    def mathmlsymbol_exponent(u, base)
-      if u[:display_exponent] == "0.5"
+    def mathmlsymbol_exponent(unit, base)
+      if unit[:display_exponent] == "0.5"
         base = "<msqrt>#{base}</msqrt>"
-      elsif u[:display_exponent]
-        exp = "<mn>#{u[:display_exponent]}</mn>".sub(/<mn>-/, "<mo>&#x2212;</mo><mn>")
+      elsif unit[:display_exponent]
+        exp = "<mn>#{unit[:display_exponent]}</mn>"
+          .sub(/<mn>-/, "<mo>&#x2212;</mo><mn>")
         base = "<msup><mrow>#{base}</mrow><mrow>#{exp}</mrow></msup>"
       end
       base
     end
 
     def mathmlsymbolwrap(units, normalise)
-      <<~END
-      <math xmlns='#{MATHML_NS}'><mrow>#{mathmlsymbol(units, normalise)}</mrow></math>
-      END
+      <<~XML
+        <math xmlns='#{MATHML_NS}'><mrow>#{mathmlsymbol(units, normalise)}</mrow></math>
+      XML
     end
   end
 end

data/lib/asciimath2unitsml/unit.rb

@@ -7,21 +7,22 @@ module Asciimath2UnitsML
     def unit_id(text)
       text = text.gsub(/[()]/, "")
       /-$/.match(text) and return @prefixes[text.sub(/-$/, "")].id
-      "U_" + (@units[text] ? @units[text].id.gsub(/'/, "_") : text.gsub(/\*/, ".").gsub(/\^/, ""))
+      "U_#{@units[text] ? @units[text].id.gsub(/'/, '_') : text.gsub(/\*/, '.').gsub(/\^/, '')}"
     end
 
-    def unit(units, origtext, normtext, dims, name)
+    def unit(units, _origtext, normtext, dims, name)
       return if units_only(units).any? { |x| x[:unit].nil? }
+
       dimid = dim_id(dims)
       norm_units = normalise_units(units)
-      <<~END
-      <Unit xmlns='#{UNITSML_NS}' xml:id='#{unit_id(normtext)}'#{dimid ? " dimensionURL='##{dimid}'" : ""}>
-      #{unitsystem(units)}
-      #{unitname(norm_units, normtext, name)}
-      #{unitsymbol(norm_units)}
-      #{rootunits(units)}
-      </Unit>
-      END
+      <<~XML
+        <Unit xmlns='#{UNITSML_NS}' xml:id='#{unit_id(normtext)}'#{dimid ? " dimensionURL='##{dimid}'" : ''}>
+        #{unitsystem(units)}
+        #{unitname(norm_units, normtext, name)}
+        #{unitsymbol(norm_units)}
+        #{rootunits(units)}
+        </Unit>
+      XML
     end
 
     def normalise_units(units)
@@ -36,14 +37,17 @@ module Asciimath2UnitsML
     # kg exception
     def unitsystem(units)
       return if units_only(units).any? { |x| x[:unit].nil? }
+
       ret = []
       units = units_only(units)
       units.any? { |x| @units[x[:unit]].system_name != "SI" } and
         ret << "<UnitSystem name='not_SI' type='not_SI' xml:lang='en-US'/>"
       if units.any? { |x| @units[x[:unit]].system_name == "SI" }
-        base = units.size == 1 && @units[units[0][:unit]].system_type == "SI-base"
-        base = true if units.size == 1 && units[0][:unit] == "g" && units[0][:prefix] == "k"
-        ret << "<UnitSystem name='SI' type='#{base ? "SI_base" : "SI_derived"}' xml:lang='en-US'/>"
+        base = units.size == 1 &&
+          @units[units[0][:unit]].system_type == "SI-base"
+        base = true if units.size == 1 && units[0][:unit] == "g" &&
+          units[0][:prefix] == "k"
+        ret << "<UnitSystem name='SI' type='#{base ? 'SI_base' : 'SI_derived'}' xml:lang='en-US'/>"
       end
       ret.join("\n")
     end
@@ -54,28 +58,30 @@ module Asciimath2UnitsML
     end
 
     # TODO: compose name from the component units
-    def compose_name(units, text)
+    def compose_name(_units, text)
       text
     end
 
     def unitsymbol(units)
-      <<~END
-      <UnitSymbol type="HTML">#{htmlsymbol(units, true)}</UnitSymbol>
-      <UnitSymbol type="MathML">#{mathmlsymbolwrap(units, true)}</UnitSymbol>
-      END
+      <<~XML
+        <UnitSymbol type="HTML">#{htmlsymbol(units, true)}</UnitSymbol>
+        <UnitSymbol type="MathML">#{mathmlsymbolwrap(units, true)}</UnitSymbol>
+      XML
     end
 
     def rootunits(units)
       return if units_only(units).any? { |x| x[:unit].nil? }
       return if units.size == 1 && !units[0][:prefix]
+
       exp = units_only(units).map do |u|
         prefix = " prefix='#{u[:prefix]}'" if u[:prefix]
-        exponent = " powerNumerator='#{u[:exponent]}'" if u[:exponent] && u[:exponent] != "1"
-        "<EnumeratedRootUnit unit='#{@units[u[:unit]].name}'#{prefix}#{exponent}/>"
+        u[:exponent] && u[:exponent] != "1" and
+          arg = " powerNumerator='#{u[:exponent]}'"
+        "<EnumeratedRootUnit unit='#{@units[u[:unit]].name}'#{prefix}#{arg}/>"
       end.join("\n")
-      <<~END
-      <RootUnits>#{exp}</RootUnits>
-      END
+      <<~XML
+        <RootUnits>#{exp}</RootUnits>
+      XML
     end
   end
 end

data/lib/asciimath2unitsml/validate.rb

@@ -0,0 +1,59 @@
+module Asciimath2UnitsML
+  class Conv
+    def validate_yaml(hash, path)
+      return hash if path == "../unitsdb/quantities.yaml"
+      return hash if path == "../unitsdb/dimensions.yaml"
+
+      hash.each_with_object({}) do |(k, v), m|
+        path == "../unitsdb/units.yaml" and validate_unit(v)
+        m = validate_symbols(m, v)
+        v[:unit_symbols]&.each { |s| validate_unit_symbol_cardinality(s, k) }
+      end
+      hash
+    end
+
+    def validate_unit(unit)
+      if unit[:quantity_reference]
+        unit[:quantity_reference].is_a?(Array) or
+          raise StandardError
+            .new "No quantity_reference array provided for unit: #{unit}"
+      end
+      if unit[:unit_name]
+        unit[:unit_name].is_a?(Array) or
+          raise StandardError
+            .new "No unit_name array provided for unit: #{unit}"
+      end
+    end
+
+    def validate_symbols(acc, val)
+      symbol = symbol_key(val)
+      !symbol.nil? or
+        raise StandardError.new "No symbol provided for unit: #{val}"
+      Array(symbol)&.each do |s|
+        acc[s] && s != "1" and
+          raise StandardError.new "symbol #{s} is not unique in #{val}: "\
+          "already used for #{acc[s]}"
+        acc[s] = val
+      end
+      acc
+    end
+
+    def validate_unit_symbol_cardinality(sym, key)
+      return true if sym.nil?
+
+      !sym[:id].nil? && !sym[:ascii].nil? && !sym[:html].nil? &&
+        !sym[:mathml].nil? && !sym[:latex].nil? &&
+        !sym[:unicode].nil? and return true
+      raise StandardError.new "malformed unit_symbol for #{key}: #{sym}"
+    end
+
+    def symbol_key(val)
+      symbol = val[:unit_symbols]&.each_with_object([]) do |s, m|
+        m << (s["id"] || s[:id])
+      end || val.dig(:symbol, :ascii) || val[:symbol] # || val[:short]
+      !symbol.nil? && val[:unit_symbols] && !symbol.is_a?(Array) and
+        symbol = [symbol]
+      symbol
+    end
+  end
+end

data/lib/asciimath2unitsml/version.rb

@@ -1,3 +1,3 @@
 module Asciimath2UnitsML
-  VERSION = '0.3.0'.freeze
+  VERSION = "0.4.0".freeze
 end

data/lib/unitsdb/dimensions.yaml

@@ -3,36 +3,86 @@ NISTd1:
   length:
     powerNumerator: 1
     symbol: L
+    dim_symbols:
+    - id: "dim_L"
+      ascii: "L"
+      html: "𝖫"
+      mathml: "<mi mathvariant='sans-serif'>L</mi>"
+      latex: \ensuremath{\mathsf{L}}
+      unicode: "𝖫"
+
 
 NISTd2:
   mass:
     powerNumerator: 1
     symbol: M
+    dim_symbols:
+    - id: "dim_M"
+      ascii: "M"
+      html: "&#x1D5AC;"
+      mathml: "<mi mathvariant='sans-serif'>M</mi>"
+      latex: \ensuremath{\mathsf{M}}
+      unicode: "𝖬"
 
 NISTd3:
   time:
     powerNumerator: 1
     symbol: T
+    dim_symbols:
+    - id: "dim_T"
+      ascii: "T"
+      html: "&#x1D5B3;"
+      mathml: "<mi mathvariant='sans-serif'>T</mi>"
+      latex: \ensuremath{\mathsf{T}}
+      unicode: "𝖳"
 
 NISTd4:
   electric_current:
     powerNumerator: 1
     symbol: I
+    dim_symbols:
+    - id: "dim_T"
+      ascii: "T"
+      html: "&#x1D5A8;"
+      mathml: "<mi mathvariant='sans-serif'>T</mi>"
+      latex: \ensuremath{\mathsf{T}}
+      unicode: "𝖨"
 
 NISTd5:
   thermodynamic_temperature:
     powerNumerator: 1
-    symbol:
+    symbol: Theta
+    dim_symbols:
+    - id: "dim_Theta"
+      ascii: "Theta"
+      html: "&#x1D760;"
+      mathml: "<mi mathvariant='sans-serif'>&#x398;</mi>"
+      latex: \ensuremath{\mathsf{\Theta}}
+      unicode: "𝝧"
 
 NISTd6:
   amount_of_substance:
     powerNumerator: 1
     symbol: N
+    dim_symbols:
+    - id: "dim_N"
+      ascii: "N"
+      html: "&#x1D5AD;"
+      mathml: "<mi mathvariant='sans-serif'>N</mi>"
+      latex: \ensuremath{\mathsf{N}}
+      unicode: "𝖭"
 
 NISTd7:
   luminous_intensity:
     powerNumerator: 1
     symbol: J
+    dim_symbols:
+    - id: "dim_J"
+      ascii: "J"
+      html: "&#x1D5A9;"
+      mathml: "<mi mathvariant='sans-serif'>J</mi>"
+      latex: \ensuremath{\mathsf{J}}
+      unicode: "𝖩"
 
 NISTd8:
   length:
@@ -44,6 +94,13 @@ NISTd9:
   plane_angle:
     powerNumerator: 1
     symbol: phi
+    dim_symbols:
+    - id: "dim_phi"
+      ascii: "phi"
+      html: "&#x1D785;"
+      mathml: "<mi mathvariant='sans-serif'>&#x3c6;</mi>"
+      latex: \ensuremath{\mathsf{\phi}}
+      unicode: "𝞅"
 
 NISTd10:
   length:
@@ -312,7 +369,7 @@ NISTd39:
     symbol: T
   thermodynamic_temperature:
     powerNumerator: -1
-    symbol:
+    symbol: Theta
 
 NISTd36:
   length:
@@ -367,7 +424,7 @@ NISTd40:
     symbol: T
   thermodynamic_temperature:
     powerNumerator: -1
-    symbol:
+    symbol: Theta
 
 NISTd41:
   length:
@@ -381,7 +438,7 @@ NISTd41:
     symbol: T
   thermodynamic_temperature:
     powerNumerator: -1
-    symbol:
+    symbol: Theta
 
 NISTd46:
   length:
@@ -475,7 +532,7 @@ NISTd48:
     symbol: T
   thermodynamic_temperature:
     powerNumerator: -1
-    symbol:
+    symbol: Theta
   amount_of_substance:
     powerNumerator: -1
     symbol: N
@@ -603,7 +660,7 @@ NISTd60:
 NISTd68:
   thermodynamic_temperature:
     powerNumerator: -1
-    symbol:
+    symbol: Theta
 
 NISTd69:
   length:
@@ -617,7 +674,7 @@ NISTd69:
     symbol: T
   thermodynamic_temperature:
     powerNumerator: -1
-    symbol:
+    symbol: Theta
 
 NISTd70:
   length:
@@ -639,7 +696,7 @@ NISTd71:
     symbol: T
   thermodynamic_temperature:
     powerNumerator: -1
-    symbol:
+    symbol: Theta
 
 NISTd64:
   dimensionless: true