rouge_ecl 0.0.1

data/lib/rouge/lexers/haskell.rb

@@ -0,0 +1,182 @@
+# -*- coding: utf-8 -*- #
+
+module Rouge
+  module Lexers
+    class Haskell < RegexLexer
+      title "Haskell"
+      desc "The Haskell programming language (haskell.org)"
+
+      tag 'haskell'
+      aliases 'hs'
+      filenames '*.hs'
+      mimetypes 'text/x-haskell'
+
+      def self.detect?(text)
+        return true if text.shebang?('runhaskell')
+      end
+
+      reserved = %w(
+        _ case class data default deriving do else if in
+        infix[lr]? instance let newtype of then type where
+      )
+
+      ascii = %w(
+        NUL SOH [SE]TX EOT ENQ ACK BEL BS HT LF VT FF CR S[OI] DLE
+        DC[1-4] NAK SYN ETB CAN EM SUB ESC [FGRU]S SP DEL
+      )
+
+      state :basic do
+        rule /\s+/m, Text
+        rule /{-#/, Comment::Preproc, :comment_preproc
+        rule /{-/, Comment::Multiline, :comment
+        rule /^--\s+\|.*?$/, Comment::Doc
+        # this is complicated in order to support custom symbols
+        # like -->
+        rule /--(?![!#\$\%&*+.\/<=>?@\^\|_~]).*?$/, Comment::Single
+      end
+
+      # nested commenting
+      state :comment do
+        rule /-}/, Comment::Multiline, :pop!
+        rule /{-/, Comment::Multiline, :comment
+        rule /[^-{}]+/, Comment::Multiline
+        rule /[-{}]/, Comment::Multiline
+      end
+
+      state :comment_preproc do
+        rule /-}/, Comment::Preproc, :pop!
+        rule /{-/, Comment::Preproc, :comment
+        rule /[^-{}]+/, Comment::Preproc
+        rule /[-{}]/, Comment::Preproc
+      end
+
+      state :root do
+        mixin :basic
+
+        rule /\bimport\b/, Keyword::Reserved, :import
+        rule /\bmodule\b/, Keyword::Reserved, :module
+        rule /\b(?:#{reserved.join('|')})\b/, Keyword::Reserved
+        # not sure why, but ^ doesn't work here
+        # rule /^[_a-z][\w']*/, Name::Function
+        rule /[_a-z][\w']*/, Name
+        rule /[A-Z][\w']*/, Keyword::Type
+
+        # lambda operator
+        rule %r(\\(?![:!#\$\%&*+.\\/<=>?@^\|~-]+)), Name::Function
+        # special operators
+        rule %r((<-|::|->|=>|=)(?![:!#\$\%&*+.\\/<=>?@^\|~-]+)), Operator
+        # constructor/type operators
+        rule %r(:[:!#\$\%&*+.\\/<=>?@^\|~-]*), Operator
+        # other operators
+        rule %r([:!#\$\%&*+.\\/<=>?@^\|~-]+), Operator
+
+        rule /\d+e[+-]?\d+/i, Num::Float
+        rule /\d+\.\d+(e[+-]?\d+)?/i, Num::Float
+        rule /0o[0-7]+/i, Num::Oct
+        rule /0x[\da-f]+/i, Num::Hex
+        rule /\d+/, Num::Integer
+
+        rule /'/, Str::Char, :character
+        rule /"/, Str, :string
+
+        rule /\[\s*\]/, Keyword::Type
+        rule /\(\s*\)/, Name::Builtin
+        rule /[\[\](),;`{}]/, Punctuation
+      end
+
+      state :import do
+        rule /\s+/, Text
+        rule /"/, Str, :string
+        rule /\bqualified\b/, Keyword
+        # import X as Y
+        rule /([A-Z][\w.]*)(\s+)(as)(\s+)([A-Z][a-zA-Z0-9_.]*)/ do
+          groups(
+            Name::Namespace, # X
+            Text, Keyword, # as
+            Text, Name # Y
+          )
+          pop!
+        end
+
+        # import X hiding (functions)
+        rule /([A-Z][\w.]*)(\s+)(hiding)(\s+)(\()/ do
+          groups(
+            Name::Namespace, # X
+            Text, Keyword, # hiding
+            Text, Punctuation # (
+          )
+          goto :funclist
+        end
+
+        # import X (functions)
+        rule /([A-Z][\w.]*)(\s+)(\()/ do
+          groups(
+            Name::Namespace, # X
+            Text,
+            Punctuation # (
+          )
+          goto :funclist
+        end
+
+        rule /[\w.]+/, Name::Namespace, :pop!
+      end
+
+      state :module do
+        rule /\s+/, Text
+        # module Foo (functions)
+        rule /([A-Z][\w.]*)(\s+)(\()/ do
+          groups Name::Namespace, Text, Punctuation
+          push :funclist
+        end
+
+        rule /\bwhere\b/, Keyword::Reserved, :pop!
+
+        rule /[A-Z][a-zA-Z0-9_.]*/, Name::Namespace, :pop!
+      end
+
+      state :funclist do
+        mixin :basic
+        rule /[A-Z]\w*/, Keyword::Type
+        rule /(_[\w\']+|[a-z][\w\']*)/, Name::Function
+        rule /,/, Punctuation
+        rule /[:!#\$\%&*+.\\\/<=>?@^\|~-]+/, Operator
+        rule /\(/, Punctuation, :funclist
+        rule /\)/, Punctuation, :pop!
+      end
+
+      state :character do
+        rule /\\/ do
+          token Str::Escape
+          goto :character_end
+          push :escape
+        end
+
+        rule /./ do
+          token Str::Char
+          goto :character_end
+        end
+      end
+
+      state :character_end do
+        rule /'/, Str::Char, :pop!
+        rule /./, Error, :pop!
+      end
+
+      state :string do
+        rule /"/, Str, :pop!
+        rule /\\/, Str::Escape, :escape
+        rule /[^\\"]+/, Str
+      end
+
+      state :escape do
+        rule /[abfnrtv"'&\\]/, Str::Escape, :pop!
+        rule /\^[\]\[A-Z@\^_]/, Str::Escape, :pop!
+        rule /#{ascii.join('|')}/, Str::Escape, :pop!
+        rule /o[0-7]+/i, Str::Escape, :pop!
+        rule /x[\da-f]+/i, Str::Escape, :pop!
+        rule /\d+/, Str::Escape, :pop!
+        rule /\s+\\/, Str::Escape, :pop!
+      end
+    end
+  end
+end

data/lib/rouge/lexers/html.rb

@@ -0,0 +1,139 @@
+# -*- coding: utf-8 -*- #
+
+module Rouge
+  module Lexers
+    class HTML < RegexLexer
+      title "HTML"
+      desc "HTML, the markup language of the web"
+      tag 'html'
+      filenames '*.htm', '*.html', '*.xhtml'
+      mimetypes 'text/html', 'application/xhtml+xml'
+
+      def self.detect?(text)
+        return true if text.doctype?(/\bhtml\b/i)
+        return true if text =~ /<\s*html\b/
+      end
+
+      start do
+        @javascript = Javascript.new(options)
+        @css = CSS.new(options)
+      end
+
+      state :root do
+        rule /[^<&]+/m, Text
+        rule /&\S*?;/, Name::Entity
+        rule /<!DOCTYPE .*?>/im, Comment::Preproc
+        rule /<!\[CDATA\[.*?\]\]>/m, Comment::Preproc
+        rule /<!--/, Comment, :comment
+        rule /<\?.*?\?>/m, Comment::Preproc # php? really?
+
+        rule /<\s*script\s*/m do
+          token Name::Tag
+          @javascript.reset!
+          push :script_content
+          push :tag
+        end
+
+        rule /<\s*style\s*/m do
+          token Name::Tag
+          @css.reset!
+          @lang = @css
+          push :style_content
+          push :tag
+        end
+
+        rule /<\//, Name::Tag, :tag_end
+        rule /</, 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
+      end
+
+      state :tag_end do
+        mixin :tag_end_end
+        rule /[a-zA-Z0-9:-]+/ do
+          token Name::Tag
+          goto :tag_end_end
+        end
+      end
+
+      state :tag_end_end do
+        rule /\s+/, Text
+        rule />/, Name::Tag, :pop!
+      end
+
+      state :tag_start do
+        rule /\s+/, Text
+
+        rule /[a-zA-Z0-9:-]+/ do
+          token Name::Tag
+          goto :tag
+        end
+
+        rule(//) { goto :tag }
+      end
+
+      state :comment do
+        rule /[^-]+/, Comment
+        rule /-->/, Comment, :pop!
+        rule /-/, Comment
+      end
+
+      state :tag do
+        rule /\s+/m, Text
+        rule /[a-zA-Z0-9_:-]+\s*=/m, Name::Attribute, :attr
+        rule /[a-zA-Z0-9_:-]+/, Name::Attribute
+        rule %r(/?\s*>)m, Name::Tag, :pop!
+      end
+
+      state :attr do
+        # TODO: are backslash escapes valid here?
+        rule /"/ do
+          token Str
+          goto :dq
+        end
+
+        rule /'/ do
+          token Str
+          goto :sq
+        end
+
+        rule /[^\s>]+/, Str, :pop!
+      end
+
+      state :dq do
+        rule /"/, Str, :pop!
+        rule /[^"]+/, Str
+      end
+
+      state :sq do
+        rule /'/, Str, :pop!
+        rule /[^']+/, Str
+      end
+
+      state :script_content do
+        rule %r([^<]+) do
+          delegate @javascript
+        end
+
+        rule %r(<\s*/\s*script\s*>)m, Name::Tag, :pop!
+
+        rule %r(<) do
+          delegate @javascript
+        end
+      end
+
+      state :style_content do
+        rule /[^<]+/ do
+          delegate @lang
+        end
+
+        rule %r(<\s*/\s*style\s*>)m, Name::Tag, :pop!
+
+        rule /</ do
+          delegate @lang
+        end
+      end
+    end
+  end
+end

data/lib/rouge/lexers/http.rb

@@ -0,0 +1,80 @@
+# -*- coding: utf-8 -*- #
+
+module Rouge
+  module Lexers
+    class HTTP < RegexLexer
+      tag 'http'
+      title "HTTP"
+      desc 'http requests and responses'
+
+      def self.http_methods
+        @http_methods ||= %w(GET POST PUT DELETE HEAD OPTIONS TRACE PATCH)
+      end
+
+      def content_lexer
+        return Lexers::PlainText unless @content_type
+
+        @content_lexer ||= Lexer.guess_by_mimetype(@content_type)
+      rescue Lexer::AmbiguousGuess
+        @content_lexer = Lexers::PlainText
+      end
+
+      start { @content_type = 'text/plain' }
+
+      state :root do
+        # request
+        rule %r(
+          (#{HTTP.http_methods.join('|')})([ ]+) # method
+          ([^ ]+)([ ]+)                     # path
+          (HTTPS?)(/)(1[.][01])(\r?\n|$)  # http version
+        )ox do
+          groups(
+            Name::Function, Text,
+            Name::Namespace, Text,
+            Keyword, Operator, Num, Text
+          )
+
+          push :headers
+        end
+
+        # response
+        rule %r(
+          (HTTPS?)(/)(1[.][01])([ ]+) # http version
+          (\d{3})([ ]+)               # status
+          ([^\r\n]+)(\r?\n|$)       # status message
+        )x do
+          groups(
+            Keyword, Operator, Num, Text,
+            Num, Text,
+            Name::Exception, Text
+          )
+          push :headers
+        end
+      end
+
+      state :headers do
+        rule /([^\s:]+)( *)(:)( *)([^\r\n]+)(\r?\n|$)/ do |m|
+          key = m[1]
+          value = m[5]
+          if key.strip.casecmp('content-type').zero?
+            @content_type = value.split(';')[0].downcase
+          end
+
+          groups Name::Attribute, Text, Punctuation, Text, Str, Text
+        end
+
+        rule /([^\r\n]+)(\r?\n|$)/ do
+          groups Str, Text
+        end
+
+        rule /\r?\n/, Text, :content
+      end
+
+      state :content do
+        rule /.+/m do |m|
+          delegate(content_lexer)
+        end
+      end
+    end
+  end
+end

data/lib/rouge/lexers/hylang.rb

@@ -0,0 +1,93 @@
+# -*- coding: utf-8 -*- #
+
+module Rouge
+  module Lexers
+    class HyLang < RegexLexer
+      title "HyLang"
+      desc "The HyLang programming language (hylang.org)"
+
+      tag 'hylang'
+      aliases 'hy'
+
+      filenames '*.hy'
+
+      mimetypes 'text/x-hy', 'application/x-hy'
+
+      def self.keywords
+        @keywords ||= Set.new %w(
+          False None True and as assert break class continue def
+          del elif else except finally for from global if import
+          in is lambda nonlocal not or pass raise return try
+        )
+      end
+
+      def self.builtins
+        @builtins ||= Set.new %w(
+          != % %= & &= * ** **= *= *map
+          + += , - -= -> ->> . / //
+          //= /= < << <<= <= = > >= >>
+          >>= @ @= ^ ^= accumulate apply as-> assoc butlast
+          calling-module-name car cdr chain coll? combinations comp complement compress cond
+          cons cons? constantly count cut cycle dec defclass defmacro defmacro!
+          defmacro/g! defmain defn defreader dict-comp disassemble dispatch-reader-macro distinct do doto
+          drop drop-last drop-while empty? eval eval-and-compile eval-when-compile even? every? filter
+          first flatten float? fn for* fraction genexpr gensym get group-by
+          identity if* if-not if-python2 inc input instance? integer integer-char? integer?
+          interleave interpose islice iterable? iterate iterator? juxt keyword keyword? last
+          let lif lif-not list* list-comp macro-error macroexpand macroexpand-1 map merge-with
+          multicombinations name neg? none? not-in not? nth numeric? odd? partition
+          permutations pos? product quasiquote quote range read read-str reduce remove
+          repeat repeatedly require rest second set-comp setv some string string?
+          symbol? take take-nth take-while tee unless unquote unquote-splicing when with*
+          with-decorator with-gensyms xor yield-from zero? zip zip-longest | |= ~
+        )
+      end
+
+      identifier = %r([\w!$%*+,<=>?/.-]+)
+      keyword = %r([\w!\#$%*+,<=>?/.-]+)
+
+      def name_token(name)
+        return Keyword if self.class.keywords.include?(name)
+        return Name::Builtin if self.class.builtins.include?(name)
+        nil
+      end
+
+      state :root do
+        rule /;.*?$/, Comment::Single
+        rule /\s+/m, Text::Whitespace
+
+        rule /-?\d+\.\d+/, Num::Float
+        rule /-?\d+/, Num::Integer
+        rule /0x-?[0-9a-fA-F]+/, Num::Hex
+
+        rule /"(\\.|[^"])*"/, Str
+        rule /'#{keyword}/, Str::Symbol
+        rule /::?#{keyword}/, Name::Constant
+        rule /\\(.|[a-z]+)/i, Str::Char
+
+
+        rule /~@|[`\'#^~&@]/, Operator
+
+        rule /(\()(\s*)(#{identifier})/m do |m|
+          token Punctuation, m[1]
+          token Text::Whitespace, m[2]
+          token(name_token(m[3]) || Name::Function, m[3])
+        end
+
+        rule identifier do |m|
+          token name_token(m[0]) || Name
+        end
+
+        # vectors
+        rule /[\[\]]/, Punctuation
+
+        # maps
+        rule /[{}]/, Punctuation
+
+        # parentheses
+        rule /[()]/, Punctuation
+      end
+    end
+  end
+end
+