rouge 3.19.0 → 3.26.0

data/lib/rouge/lexers/crystal.rb

@@ -25,8 +25,7 @@ module Rouge
         )xi, Str::Symbol
 
         # special symbols
-        rule %r(:(?:\*\*|[-+]@|[/\%&\|^`~]|\[\]=?|<<|>>|<=?>|<=?|===?)),
-          Str::Symbol
+        rule %r(:(?:===|=?~|\[\][=?]?|\*\*=?|\/\/=?|[=^*/+-]=?|&[&*+-]?=?|\|\|?=?|![=~]?|%=?|<=>|<<?=?|>>?=?|\.\.\.?)), Str::Symbol
 
         rule %r/:'(\\\\|\\'|[^'])*'/, Str::Symbol
         rule %r/:"/, Str::Symbol, :simple_sym
@@ -36,7 +35,7 @@ module Rouge
         # %-sigiled strings
         # %(abc), %[abc], %<abc>, %.abc., %r.abc., etc
         delimiter_map = { '{' => '}', '[' => ']', '(' => ')', '<' => '>' }
-        rule %r/%([rqswQWxiI])?([^\w\s])/ do |m|
+        rule %r/%([rqswQWxiI])?([^\w\s}])/ do |m|
           open = Regexp.escape(m[2])
           close = Regexp.escape(delimiter_map[m[2]] || m[2])
           interp = /[rQWxI]/ === m[1]
@@ -79,9 +78,11 @@ module Rouge
       state :strings do
         mixin :symbols
         rule %r/\b[a-z_]\w*?[?!]?:\s+/, Str::Symbol, :expr_start
-        rule %r/'(\\\\|\\'|[^'])*'/, Str::Single
         rule %r/"/, Str::Double, :simple_string
         rule %r/(?<!\.)`/, Str::Backtick, :simple_backtick
+        rule %r/(')(\\u[a-fA-F0-9]{4}|\\u\{[a-fA-F0-9]{1,6}\}|\\[abefnrtv])?(\\\\|\\'|[^'])*(')/ do
+          groups Str::Single, Str::Escape, Str::Single, Str::Single
+        end
       end
 
       state :regex_flags do
@@ -154,11 +155,13 @@ module Rouge
         mixin :whitespace
         rule %r/__END__/, Comment::Preproc, :end_part
 
-        rule %r/0_?[0-7]+(?:_[0-7]+)*/, Num::Oct
+        rule %r/0o[0-7]+(?:_[0-7]+)*/, Num::Oct
         rule %r/0x[0-9A-Fa-f]+(?:_[0-9A-Fa-f]+)*/, Num::Hex
         rule %r/0b[01]+(?:_[01]+)*/, Num::Bin
-        rule %r/\d+\.\d+(e[\+\-]?\d+)?/, Num::Float
-        rule %r/[\d]+(?:_\d+)*/, Num::Integer
+        rule %r/\d+\.\d+(e[\+\-]?\d+)?(_f32)?/i, Num::Float
+        rule %r/\d+(e[\+\-]?\d+)/i, Num::Float
+        rule %r/\d+_f32/i, Num::Float
+        rule %r/[\d]+(?:_\d+)*(_[iu]\d+)?/, Num::Integer
 
         rule %r/@\[([^\]]+)\]/, Name::Decorator
 
@@ -183,7 +186,7 @@ module Rouge
           groups Keyword, Text, Name::Namespace
         end
 
-        rule %r/(def\b)(\s*)/ do
+        rule %r/(def|macro\b)(\s*)/ do
           groups Keyword, Text
           push :funcname
         end
@@ -212,8 +215,9 @@ module Rouge
 
         rule %r/[a-zA-Z_]\w*[?!]/, Name, :expr_start
         rule %r/[a-zA-Z_]\w*/, Name, :method_call
-        rule %r/\*\*|<<?|>>?|>=|<=|<=>|=~|={3}|!~|&&?|\|\||\./,
+        rule %r/\*\*|\/\/|>=|<=|<=>|<<?|>>?|=~|={3}|!~|&&?|\|\||\./,
           Operator, :expr_start
+        rule %r/{%|%}/, Punctuation
         rule %r/[-+\/*%=<>&!^|~]=?/, Operator, :expr_start
         rule(/[?]/) { token Punctuation; push :ternary; push :expr_start }
         rule %r<[\[({,:\\;/]>, Punctuation, :expr_start
@@ -346,6 +350,7 @@ module Rouge
         mixin :string_intp
         rule %r/\\([\\abefnrstv#"']|x[a-fA-F0-9]{1,2}|[0-7]{1,3})/,
           Str::Escape
+        rule %r/\\u([a-fA-F0-9]{4}|\{[^}]+\})/, Str::Escape
         rule %r/\\./, Str::Escape
       end
 

data/lib/rouge/lexers/css.rb

@@ -11,7 +11,9 @@ module Rouge
       filenames '*.css'
       mimetypes 'text/css'
 
-      identifier = /[a-zA-Z0-9_-]+/
+      # Documentation: https://www.w3.org/TR/CSS21/syndata.html#characters
+
+      identifier = /[\p{L}_-][\p{Word}\p{Cf}-]*/
       number = /-?(?:[0-9]+(\.[0-9]+)?|\.[0-9]+)/
 
       def self.attributes

data/lib/rouge/lexers/diff.rb

@@ -14,7 +14,7 @@ module Rouge
       def self.detect?(text)
         return true if text.start_with?('Index: ')
         return true if text =~ %r(\Adiff[^\n]*?\ba/[^\n]*\bb/)
-        return true if text =~ /(---|[+][+][+]).*?\n(---|[+][+][+])/
+        return true if text =~ /---.*?\n[+][+][+]/ || text =~ /[+][+][+].*?\n---/
       end
 
       state :root do

data/lib/rouge/lexers/docker.rb

@@ -7,8 +7,8 @@ module Rouge
       title "Docker"
       desc "Dockerfile syntax"
       tag 'docker'
-      aliases 'dockerfile'
-      filenames 'Dockerfile', '*.Dockerfile', '*.docker', 'Dockerfile_*'
+      aliases 'dockerfile', 'Dockerfile'
+      filenames 'Dockerfile', '*.Dockerfile', '*.docker'
       mimetypes 'text/x-dockerfile-config'
 
       KEYWORDS = %w(

data/lib/rouge/lexers/elm.rb

@@ -78,13 +78,13 @@ module Rouge
         rule %r/"/, Str::Double, :pop!
       end
 
-      # Multiple line string with tripple double quotes, e.g. """ multi """
+      # Multiple line string with triple double quotes, e.g. """ multi """
       state :multiline_string do
-        rule %r/\s*"""/, Str, :pop!
-        rule %r/.*/, Str
-        rule %r/\s*/, Str
+        rule %r/\\"/, Str::Escape
+        rule %r/"""/, Str, :pop!
+        rule %r/[^"]+/, Str
+        rule %r/"/, Str
       end
-
     end
   end
 end

data/lib/rouge/lexers/email.rb

@@ -0,0 +1,39 @@
+# -*- coding: utf-8 -*- #
+# frozen_string_literal: true
+
+module Rouge
+  module Lexers
+    class Email < RegexLexer
+      tag 'email'
+      aliases 'eml', 'e-mail'
+      filenames '*.eml'
+      mimetypes 'message/rfc822'
+
+      title "Email"
+      desc "An email message"
+
+      start do
+        push :fields
+      end
+
+      state :fields do
+        rule %r/[:]/, Operator, :field_body
+        rule %r/[^\n\r:]+/, Name::Tag
+        rule %r/[\n\r]/, Name::Tag
+      end
+
+      state :field_body do
+        rule(/(\r?\n){2}/) { token Text; pop!(2) }
+        rule %r/\r?\n(?![ \v\t\f])/, Text, :pop!
+        rule %r/[^\n\r]+/, Name::Attribute
+        rule %r/[\n\r]/, Name::Attribute
+      end
+
+      state :root do
+        rule %r/\n/, Text
+        rule %r/^>.*/, Comment
+        rule %r/.+/, Text
+      end
+    end
+  end
+end

data/lib/rouge/lexers/ghc_core.rb

@@ -27,8 +27,9 @@ module Rouge
         mixin :function
 
         # rest is Text
+        # TODO: this is really inefficient
         rule %r/\s/m, Text
-        rule %r/.*/, Text
+        rule %r/./, Text
       end
 
       state :expression do

data/lib/rouge/lexers/graphql.rb

@@ -20,7 +20,7 @@ module Rouge
         end
 
         rule %r/\bfragment\b/, Keyword, :fragment_definition
-        
+
         rule %r/\bscalar\b/, Keyword, :value
 
         rule %r/\b(?:type|interface|enum)\b/, Keyword, :type_definition

data/lib/rouge/lexers/hack.rb

@@ -32,7 +32,7 @@ module Rouge
         )
       end
 
-      prepend :template do
+      prepend :root do
         rule %r/<\?hh(\s*\/\/\s*(strict|decl|partial))?$/, Comment::Preproc, :php
       end
 

data/lib/rouge/lexers/haskell.rb

@@ -17,8 +17,8 @@ module Rouge
       end
 
       reserved = %w(
-        _ case class data default deriving do else if in
-        infix[lr]? instance let newtype of then type where
+        _ case class data default deriving do else if in infix infixl infixr
+        instance let newtype of then type where
       )
 
       ascii = %w(
@@ -54,14 +54,31 @@ module Rouge
       state :root do
         mixin :basic
 
-        rule %r/\bimport\b/, Keyword::Reserved, :import
-        rule %r/\bmodule\b/, Keyword::Reserved, :module
-        rule %r/\b(?:#{reserved.join('|')})\b/, Keyword::Reserved
-        # not sure why, but ^ doesn't work here
-        # rule %r/^[_a-z][\w']*/, Name::Function
-        rule %r/[_a-z][\w']*/, Name
-        rule %r/[A-Z][\w']*/, Keyword::Type
-        rule %r/'[A-Z]\w+'?/, Keyword::Type  # promoted data constructor
+        rule %r/'(?=(?:.|\\\S+)')/, Str::Char, :character
+        rule %r/"/, Str, :string
+
+        rule %r/\d+e[+-]?\d+/i, Num::Float
+        rule %r/\d+\.\d+(e[+-]?\d+)?/i, Num::Float
+        rule %r/0o[0-7]+/i, Num::Oct
+        rule %r/0x[\da-f]+/i, Num::Hex
+        rule %r/\d+/, Num::Integer
+
+        rule %r/[\w']+/ do |m|
+          match = m[0]
+          if match == "import"
+            token Keyword::Reserved
+            push :import
+          elsif match == "module"
+            token Keyword::Reserved
+            push :module
+          elsif reserved.include?(match)
+            token Keyword::Reserved
+          elsif match =~ /\A'?[A-Z]/
+            token Keyword::Type
+          else
+            token Name
+          end
+        end
 
         # lambda operator
         rule %r(\\(?![:!#\$\%&*+.\\/<=>?@^\|~-]+)), Name::Function
@@ -72,15 +89,6 @@ module Rouge
         # other operators
         rule %r([:!#\$\%&*+.\\/<=>?@^\|~-]+), Operator
 
-        rule %r/\d+e[+-]?\d+/i, Num::Float
-        rule %r/\d+\.\d+(e[+-]?\d+)?/i, Num::Float
-        rule %r/0o[0-7]+/i, Num::Oct
-        rule %r/0x[\da-f]+/i, Num::Hex
-        rule %r/\d+/, Num::Integer
-
-        rule %r/'/, Str::Char, :character
-        rule %r/"/, Str, :string
-
         rule %r/\[\s*\]/, Keyword::Type
         rule %r/\(\s*\)/, Name::Builtin
 

data/lib/rouge/lexers/hlsl.rb

@@ -0,0 +1,166 @@
+# -*- coding: utf-8 -*- #
+# frozen_string_literal: true
+
+module Rouge
+  module Lexers
+    load_lexer 'c.rb'
+
+    class HLSL < C
+      title "HLSL"
+      desc "HLSL, the High Level Shading Language for DirectX (docs.microsoft.com/en-us/windows/win32/direct3dhlsl/dx-graphics-hlsl)"
+      tag 'hlsl'
+      filenames '*.hlsl', '*.hlsli'
+      mimetypes 'text/x-hlsl'
+
+      def self.keywords
+        @keywords ||= Set.new %w(
+          asm asm_fragment break case cbuffer centroid class column_major
+          compile compile_fragment const continue default discard do else export
+          extern for fxgroup globallycoherent groupshared if in inline inout
+          interface line lineadj linear namespace nointerpolation noperspective
+          NULL out packoffset pass pixelfragment point precise return register
+          row_major sample sampler shared stateblock stateblock_state static
+          struct switch tbuffer technique technique10 technique11 texture
+          typedef triangle uniform vertexfragment volatile while
+        )
+      end
+
+      def self.keywords_type
+        @keywords_type ||= Set.new %w(
+          dword matrix snorm string unorm unsigned void vector BlendState Buffer
+          ByteAddressBuffer ComputeShader DepthStencilState DepthStencilView
+          DomainShader GeometryShader HullShader InputPatch LineStream
+          OutputPatch PixelShader PointStream RasterizerState RenderTargetView
+          RasterizerOrderedBuffer RasterizerOrderedByteAddressBuffer
+          RasterizerOrderedStructuredBuffer RasterizerOrderedTexture1D
+          RasterizerOrderedTexture1DArray RasterizerOrderedTexture2D
+          RasterizerOrderedTexture2DArray RasterizerOrderedTexture3D RWBuffer
+          RWByteAddressBuffer RWStructuredBuffer RWTexture1D RWTexture1DArray
+          RWTexture2D RWTexture2DArray RWTexture3D SamplerState
+          SamplerComparisonState StructuredBuffer Texture1D Texture1DArray
+          Texture2D Texture2DArray Texture2DMS Texture2DMSArray Texture3D
+          TextureCube TextureCubeArray TriangleStream VertexShader
+
+          bool1 bool2 bool3 bool4 BOOL1 BOOL2 BOOL3 BOOL4
+          int1 int2 int3 int4
+          half1 half2 half3 half4
+          float1 float2 float3 float4
+          double1 double2 double3 double4
+
+          bool1x1 bool1x2 bool1x3 bool1x4 bool2x1 bool2x2 bool2x3 bool2x4
+          bool3x1 bool3x2 bool3x3 bool3x4 bool4x1 bool4x2 bool4x3 bool4x4
+          BOOL1x1 BOOL1x2 BOOL1x3 BOOL1x4 BOOL2x1 BOOL2x2 BOOL2x3 BOOL2x4
+          BOOL3x1 BOOL3x2 BOOL3x3 BOOL3x4 BOOL4x1 BOOL4x2 BOOL4x3 BOOL4x4
+          half1x1 half1x2 half1x3 half1x4 half2x1 half2x2 half2x3 half2x4
+          half3x1 half3x2 half3x3 half3x4 half4x1 half4x2 half4x3 half4x4
+          int1x1 int1x2 int1x3 int1x4 int2x1 int2x2 int2x3 int2x4
+          int3x1 int3x2 int3x3 int3x4 int4x1 int4x2 int4x3 int4x4
+          float1x1 float1x2 float1x3 float1x4 float2x1 float2x2 float2x3 float2x4
+          float3x1 float3x2 float3x3 float3x4 float4x1 float4x2 float4x3 float4x4
+          double1x1 double1x2 double1x3 double1x4 double2x1 double2x2 double2x3 double2x4
+          double3x1 double3x2 double3x3 double3x4 double4x1 double4x2 double4x3 double4x4
+        )
+      end
+
+      def self.reserved
+        @reserved ||= Set.new %w(
+          auto catch char const_cast delete dynamic_cast enum explicit friend
+          goto long mutable new operator private protected public
+          reinterpret_cast short signed sizeof static_cast template this throw
+          try typename union unsigned using virtual
+        )
+      end
+
+      def self.builtins
+        @builtins ||= Set.new %w(
+          abort abs acos all AllMemoryBarrier AllMemoryBarrierWithGroupSync any
+          AppendStructuredBuffer asdouble asfloat asin asint asuint asuint atan
+          atan2 ceil CheckAccessFullyMapped clamp clip CompileShader
+          ConsumeStructuredBuffer cos cosh countbits cross D3DCOLORtoUBYTE4 ddx
+          ddx_coarse ddx_fine ddy ddy_coarse ddy_fine degrees determinant
+          DeviceMemoryBarrier DeviceMemoryBarrierWithGroupSync distance dot dst
+          errorf EvaluateAttributeAtCentroid EvaluateAttributeAtSample
+          EvaluateAttributeSnapped exp exp2 f16tof32 f32tof16 faceforward
+          firstbithigh firstbitlow floor fma fmod frac frexp fwidth
+          GetRenderTargetSampleCount GetRenderTargetSamplePosition
+          GlobalOrderedCountIncrement GroupMemoryBarrier
+          GroupMemoryBarrierWithGroupSync InterlockedAdd InterlockedAnd
+          InterlockedCompareExchange InterlockedCompareStore InterlockedExchange
+          InterlockedMax InterlockedMin InterlockedOr InterlockedXor isfinite
+          isinf isnan ldexp length lerp lit log log10 log2 mad max min modf
+          msad4 mul noise normalize pow printf Process2DQuadTessFactorsAvg
+          Process2DQuadTessFactorsMax Process2DQuadTessFactorsMin
+          ProcessIsolineTessFactors ProcessQuadTessFactorsAvg
+          ProcessQuadTessFactorsMax ProcessQuadTessFactorsMin
+          ProcessTriTessFactorsAvg ProcessTriTessFactorsMax
+          ProcessTriTessFactorsMin QuadReadLaneAt QuadSwapX QuadSwapY radians
+          rcp reflect refract reversebits round rsqrt saturate sign sin sincos
+          sinh smoothstep sqrt step tan tanh tex1D tex1D tex1Dbias tex1Dgrad
+          tex1Dlod tex1Dproj tex2D tex2D tex2Dbias tex2Dgrad tex2Dlod tex2Dproj
+          tex3D tex3D tex3Dbias tex3Dgrad tex3Dlod tex3Dproj texCUBE texCUBE
+          texCUBEbias texCUBEgrad texCUBElod texCUBEproj transpose trunc
+          WaveAllBitAnd WaveAllMax WaveAllMin WaveAllBitOr WaveAllBitXor
+          WaveAllEqual WaveAllProduct WaveAllSum WaveAllTrue WaveAnyTrue
+          WaveBallot WaveGetLaneCount WaveGetLaneIndex WaveGetOrderedIndex
+          WaveIsHelperLane WaveOnce WavePrefixProduct WavePrefixSum
+          WaveReadFirstLane WaveReadLaneAt
+
+          SV_CLIPDISTANCE SV_CLIPDISTANCE0 SV_CLIPDISTANCE1 SV_CULLDISTANCE
+          SV_CULLDISTANCE0 SV_CULLDISTANCE1 SV_COVERAGE SV_DEPTH
+          SV_DEPTHGREATEREQUAL SV_DEPTHLESSEQUAL SV_DISPATCHTHREADID
+          SV_DOMAINLOCATION SV_GROUPID SV_GROUPINDEX SV_GROUPTHREADID
+          SV_GSINSTANCEID SV_INNERCOVERAGE SV_INSIDETESSFACTOR SV_INSTANCEID
+          SV_ISFRONTFACE SV_OUTPUTCONTROLPOINTID SV_POSITION SV_PRIMITIVEID
+          SV_RENDERTARGETARRAYINDEX SV_SAMPLEINDEX SV_STENCILREF SV_TESSFACTOR
+          SV_VERTEXID SV_VIEWPORTARRAYINDEX
+
+          allow_uav_condition branch call domain earlydepthstencil fastopt
+          flatten forcecase instance loop maxtessfactor numthreads
+          outputcontrolpoints outputtopology partitioning patchconstantfunc
+          unroll
+
+          BINORMAL BINORMAL0 BINORMAL1 BINORMAL2 BINORMAL3 BINORMAL4
+          BLENDINDICES0 BLENDINDICES1 BLENDINDICES2 BLENDINDICES3 BLENDINDICES4
+          BLENDWEIGHT0 BLENDWEIGHT1 BLENDWEIGHT2 BLENDWEIGHT3 BLENDWEIGHT4 COLOR
+          COLOR0 COLOR1 COLOR2 COLOR3 COLOR4 NORMAL NORMAL0 NORMAL1 NORMAL2
+          NORMAL3 NORMAL4 POSITION POSITION0 POSITION1 POSITION2 POSITION3
+          POSITION4 POSITIONT PSIZE0 PSIZE1 PSIZE2 PSIZE3 PSIZE4 TANGENT
+          TANGENT0 TANGENT1 TANGENT2 TANGENT3 TANGENT4 TESSFACTOR0 TESSFACTOR1
+          TESSFACTOR2 TESSFACTOR3 TESSFACTOR4 TEXCOORD0 TEXCOORD1 TEXCOORD2
+          TEXCOORD3 TEXCOORD4
+
+          FOG PSIZE
+
+          VFACE VPOS
+
+          DEPTH0 DEPTH1 DEPTH2 DEPTH3 DEPTH4
+        )
+      end
+
+      ws = %r((?:\s|//.*?\n|/[*].*?[*]/)+)
+      id = /[a-zA-Z_][a-zA-Z0-9_]*/
+
+      state :root do
+        mixin :expr_whitespace
+        rule %r(
+          ([\w*\s]+?[\s*])                  # return arguments
+          (#{id})                           # function name
+          (\s*\([^;]*?\)(?:\s*:\s+#{id})?)  # signature
+          (#{ws}?)({|;)                     # open brace or semicolon
+        )mx do |m|
+          # This is copied from the C lexer
+          recurse m[1]
+          token Name::Function, m[2]
+          recurse m[3]
+          recurse m[4]
+          token Punctuation, m[5]
+          if m[5] == ?{
+            push :function
+          end
+        end
+        rule %r/\{/, Punctuation, :function
+        mixin :statements
+      end
+    end
+  end
+end

data/lib/rouge/lexers/html.rb

@@ -7,7 +7,7 @@ module Rouge
       title "HTML"
       desc "HTML, the markup language of the web"
       tag 'html'
-      filenames '*.htm', '*.html', '*.xhtml'
+      filenames '*.htm', '*.html', '*.xhtml', '*.cshtml'
       mimetypes 'text/html', 'application/xhtml+xml'
 
       def self.detect?(text)
@@ -47,13 +47,13 @@ module Rouge
         rule %r(</), Name::Tag, :tag_end
         rule %r/</, Name::Tag, :tag_start
 
-        rule %r(<\s*[a-zA-Z0-9:-]+), Name::Tag, :tag # opening tags
-        rule %r(<\s*/\s*[a-zA-Z0-9:-]+\s*>), Name::Tag # closing tags
+        rule %r(<\s*[\p{L}:_-][\p{Word}\p{Cf}:.·-]*), Name::Tag, :tag # opening tags
+        rule %r(<\s*/\s*[\p{L}:_-][\p{Word}\p{Cf}:.·-]*\s*>), Name::Tag # closing tags
       end
 
       state :tag_end do
         mixin :tag_end_end
-        rule %r/[a-zA-Z0-9:-]+/ do
+        rule %r/[\p{L}:_-][\p{Word}\p{Cf}:.·-]*/ do
           token Name::Tag
           goto :tag_end_end
         end
@@ -67,7 +67,7 @@ module Rouge
       state :tag_start do
         rule %r/\s+/, Text
 
-        rule %r/[a-zA-Z0-9:-]+/ do
+        rule %r/[\p{L}:_-][\p{Word}\p{Cf}:.·-]*/ do
           token Name::Tag
           goto :tag
         end
@@ -83,8 +83,8 @@ module Rouge
 
       state :tag do
         rule %r/\s+/m, Text
-        rule %r/[a-zA-Z0-9_:\[\]()*.-]+\s*=\s*/m, Name::Attribute, :attr
-        rule %r/[a-zA-Z0-9_:#*-]+/, Name::Attribute
+        rule %r/[\p{L}:_\[\]()*.-][\p{Word}\p{Cf}:.·\[\]()*-]*\s*=\s*/m, Name::Attribute, :attr
+        rule %r/[\p{L}:_*#-][\p{Word}\p{Cf}:.·*#-]*/, Name::Attribute
         rule %r(/?\s*>)m, Name::Tag, :pop!
       end