rouge 3.6.0 → 3.8.0

data/lib/rouge/lexers/ada.rb

@@ -0,0 +1,162 @@
+# -*- coding: utf-8 -*- #
+# frozen_string_literal: true
+
+module Rouge
+  module Lexers
+    class Ada < RegexLexer
+      tag 'ada'
+      filenames '*.ada', '*.ads', '*.adb', '*.gpr'
+      mimetypes 'text/x-ada'
+
+      title 'Ada'
+      desc 'The Ada 2012 programming language'
+
+      # Ada identifiers are Unicode with underscores only allowed as separators.
+      ID = /\b[[:alpha:]](?:\p{Pc}?[[:alnum:]])*\b/
+
+      # Numerals can also contain underscores.
+      NUM = /\d(_?\d)*/
+      XNUM = /\h(_?\h)*/
+      EXP = /(E[-+]?#{NUM})?/i
+
+      # Return a hash mapping lower-case identifiers to token classes.
+      def self.idents
+        @idents ||= Hash.new(Name).tap do |h|
+          %w(
+            abort abstract accept access aliased all array at begin body
+            case constant declare delay delta digits do else elsif end
+            exception exit for generic goto if in interface is limited
+            loop new null of others out overriding pragma private
+            protected raise range record renames requeue return reverse
+            select separate some synchronized tagged task terminate then
+            until use when while with
+          ).each {|w| h[w] = Keyword}
+
+          %w(abs and mod not or rem xor).each {|w| h[w] = Operator::Word}
+
+          %w(
+            entry function package procedure subtype type
+          ).each {|w| h[w] = Keyword::Declaration}
+
+          %w(
+            boolean character constraint_error duration float integer
+            natural positive long_float long_integer long_long_float
+            long_long_integer program_error short_float short_integer
+            short_short_integer storage_error string tasking_error
+            wide_character wide_string wide_wide_character
+            wide_wide_string
+          ).each {|w| h[w] = Name::Builtin}
+        end
+      end
+
+      state :whitespace do
+        rule %r{\s+}m, Text
+        rule %r{--.*$}, Comment::Single
+      end
+
+      state :dquote_string do
+        rule %r{[^"\n]+}, Literal::String::Double
+        rule %r{""}, Literal::String::Escape
+        rule %r{"}, Literal::String::Double, :pop!
+        rule %r{\n}, Error, :pop!
+      end
+
+      state :attr do
+        mixin :whitespace
+        rule ID, Name::Attribute, :pop!
+        rule %r{}, Text, :pop!
+      end
+
+      # Handle a dotted name immediately following a declaration keyword.
+      state :decl_name do
+        mixin :whitespace
+        rule %r{body\b}i, Keyword::Declaration  # package body Foo.Bar is...
+        rule %r{(#{ID})(\.)} do
+          groups Name::Namespace, Punctuation
+        end
+        # function "<=" (Left, Right: Type) is ...
+        rule %r{#{ID}|"(and|or|xor|/?=|<=?|>=?|\+|–|&\|/|mod|rem|\*?\*|abs|not)"},
+             Name::Function, :pop!
+        rule %r{}, Text, :pop!
+      end
+
+      # Handle a sequence of library unit names: with Ada.Foo, Ada.Bar;
+      #
+      # There's a chance we entered this state mistakenly since 'with'
+      # has multiple other uses in Ada (none of which are likely to
+      # appear at the beginning of a line). Try to bail as soon as
+      # possible if we see something suspicious like keywords.
+      #
+      # See ada_spec.rb for some examples.
+      state :libunit_name do
+        mixin :whitespace
+
+        rule ID do |m|
+          t = self.class.idents[m[0].downcase]
+          if t <= Name
+            # Convert all kinds of Name to namespaces in this context.
+            token Name::Namespace
+          else
+            # Yikes, we're not supposed to get a keyword in a library unit name!
+            # We probably entered this state by mistake, so try to fix it.
+            token t
+            if t == Keyword::Declaration
+              goto :decl_name
+            else
+              pop!
+            end
+          end
+        end
+
+        rule %r{[.,]}, Punctuation
+        rule %r{}, Text, :pop!
+      end
+
+      state :root do
+        mixin :whitespace
+
+        # String literals.
+        rule %r{'.'}, Literal::String::Char
+        rule %r{"[^"\n]*}, Literal::String::Double, :dquote_string
+
+        # Real literals.
+        rule %r{#{NUM}\.#{NUM}#{EXP}}, Literal::Number::Float
+        rule %r{#{NUM}##{XNUM}\.#{XNUM}##{EXP}}, Literal::Number::Float
+
+        # Integer literals.
+        rule %r{2#[01](_?[01])*##{EXP}}, Literal::Number::Bin
+        rule %r{8#[0-7](_?[0-7])*##{EXP}}, Literal::Number::Oct
+        rule %r{16##{XNUM}*##{EXP}}, Literal::Number::Hex
+        rule %r{#{NUM}##{XNUM}##{EXP}}, Literal::Number::Integer
+        rule %r{#{NUM}#\w+#}, Error
+        rule %r{#{NUM}#{EXP}}, Literal::Number::Integer
+
+        # Special constructs.
+        rule %r{'}, Punctuation, :attr
+        rule %r{<<#{ID}>>}, Name::Label
+
+        # Context clauses are tricky because the 'with' keyword is used
+        # for many purposes. Detect at beginning of the line only.
+        rule %r{^(?:(limited)(\s+))?(?:(private)(\s+))?(with)\b}i do
+          groups Keyword::Namespace, Text, Keyword::Namespace, Text, Keyword::Namespace
+          push :libunit_name
+        end
+
+        # Operators and punctuation characters.
+        rule %r{[+*/&<=>|]|-|=>|\.\.|\*\*|[:></]=|<<|>>|<>}, Operator
+        rule %r{[.,:;()]}, Punctuation
+
+        rule ID do |m|
+          t = self.class.idents[m[0].downcase]
+          token t
+          if t == Keyword::Declaration
+            push :decl_name
+          end
+        end
+
+        # Flag word-like things that don't match the ID pattern.
+        rule %r{\b(\p{Pc}|[[alpha]])\p{Word}*}, Error
+      end
+    end
+  end
+end

data/lib/rouge/lexers/armasm.rb

@@ -0,0 +1,145 @@
+# -*- coding: utf-8 -*- #
+# frozen_string_literal: true
+
+module Rouge
+  module Lexers
+    class ArmAsm < RegexLexer
+      title "ArmAsm"
+      desc "Arm assembly syntax"
+      tag 'armasm'
+      filenames '*.s'
+
+      def self.preproc_keyword
+        @preproc_keyword ||= %w(
+          define elif else endif error if ifdef ifndef include line pragma undef warning
+        )
+      end
+
+      def self.file_directive
+        @file_directive ||= %w(
+          BIN GET INCBIN INCLUDE LNK
+        )
+      end
+
+      def self.general_directive
+        @general_directive ||= %w(
+          ALIAS ALIGN AOF AOUT AREA ARM ASSERT ATTR CN CODE16 CODE32 COMMON CP
+          DATA DCB DCD DCDO DCDU DCFD DCFDU DCFH DCFHU DCFS DCFSU DCI DCI.N DCI.W
+          DCQ DCQU DCW DCWU DN ELIF ELSE END ENDFUNC ENDIF ENDP ENTRY EQU EXPORT
+          EXPORTAS EXTERN FIELD FILL FN FRAME FUNCTION GBLA GBLL GBLS GLOBAL IF
+          IMPORT INFO KEEP LCLA LCLL LCLS LEADR LEAF LTORG MACRO MAP MEND MEXIT
+          NOFP OPT ORG PRESERVE8 PROC QN RELOC REQUIRE REQUIRE8 RLIST RN ROUT
+          SETA SETL SETS SN SPACE STRONG SUBT THUMB THUMBX TTL WEND WHILE
+          \[ \] [|!#*=%&^]
+        )
+      end
+
+      def self.shift_or_condition
+        @shift_or_condition ||= %w(
+          ASR LSL LSR ROR RRX AL CC CS EQ GE GT HI HS LE LO LS LT MI NE PL VC VS
+          asr lsl lsr ror rrx al cc cs eq ge gt hi hs le lo ls lt mi ne pl vc vs
+        )
+      end
+
+      def self.builtin
+        @builtin ||= %w(
+          ARCHITECTURE AREANAME ARMASM_VERSION CODESIZE COMMANDLINE CONFIG CPU
+          ENDIAN FALSE FPIC FPU INPUTFILE INTER LINENUM LINENUMUP LINENUMUPPER
+          OBJASM_VERSION OPT PC PCSTOREOFFSET REENTRANT ROPI RWPI TRUE VAR
+        )
+      end
+
+      def self.operator
+        @operator ||= %w(
+          AND BASE CC CC_ENCODING CHR DEF EOR FATTR FEXEC FLOAD FSIZE INDEX LAND
+          LEFT LEN LEOR LNOT LOR LOWERCASE MOD NOT OR RCONST REVERSE_CC RIGHT ROL
+          ROR SHL SHR STR TARGET_ARCH_[0-9A-Z_]+ TARGET_FEATURE_[0-9A-Z_]+
+          TARGET_FPU_[A-Z_] TARGET_PROFILE_[ARM] UAL UPPERCASE
+        )
+      end
+
+      state :root do
+        rule %r/\n/, Text
+        rule %r/^([ \t]*)(#[ \t]*(?:(?:#{ArmAsm.preproc_keyword.join('|')})(?:[ \t].*)?)?)(\n)/ do
+          groups Text, Comment::Preproc, Text
+        end
+        rule %r/[ \t]+/, Text, :command
+        rule %r/;.*/, Comment
+        rule %r/\$[a-z_]\w*\.?/i, Name::Namespace # variable substitution or macro argument
+        rule %r/\w+|\|[^|\n]+\|/, Name::Label
+      end
+
+      state :command do
+        rule %r/\n/, Text, :pop!
+        rule %r/[ \t]+/ do |m|
+          token Text
+          goto :args
+        end
+        rule %r/;.*/, Comment, :pop!
+        rule %r/(?:#{ArmAsm.file_directive.join('|')})\b/ do |m|
+          token Keyword
+          goto :filespec
+        end
+        rule %r/(?:#{ArmAsm.general_directive.join('|')})(?=[; \t\n])/, Keyword
+        rule %r/(?:[A-Z][\dA-Z]*|[a-z][\da-z]*)(?:\.[NWnw])?(?:\.[DFIPSUdfipsu]?(?:8|16|32|64)?){,3}\b/, Name::Builtin # rather than attempt to list all opcodes, rely on all-uppercase or all-lowercase rule
+        rule %r/[a-z_]\w*|\|[^|\n]+\|/i, Name::Function # probably a macro name
+        rule %r/\$[a-z]\w*\.?/i, Name::Namespace
+      end
+
+      state :args do
+        rule %r/\n/, Text, :pop!
+        rule %r/[ \t]+/, Text
+        rule %r/;.*/, Comment, :pop!
+        rule %r/(?:#{ArmAsm.shift_or_condition.join('|')})\b/, Name::Builtin
+        rule %r/[a-z_]\w*|\|[^|\n]+\|/i, Name::Variable # various types of symbol
+        rule %r/%[bf]?[at]?\d+(?:[a-z_]\w*)?/i, Name::Label
+        rule %r/(?:&|0x)\h+(?!p)/i, Literal::Number::Hex
+        rule %r/(?:&|0x)[.\h]+(?:p[-+]?\d+)?/i, Literal::Number::Float
+        rule %r/0f_\h{8}|0d_\h{16}/i, Literal::Number::Float
+        rule %r/(?:2_[01]+|3_[0-2]+|4_[0-3]+|5_[0-4]+|6_[0-5]+|7_[0-6]+|8_[0-7]+|9_[0-8]+|\d+)(?!e)/i, Literal::Number::Integer
+        rule %r/(?:2_[.01]+|3_[.0-2]+|4_[.0-3]+|5_[.0-4]+|6_[.0-5]+|7_[.0-6]+|8_[.0-7]+|9_[.0-8]+|[.\d]+)(?:e[-+]?\d+)?/i, Literal::Number::Float
+        rule %r/[@:](?=[ \t]*(?:8|16|32|64|128|256)[^\d])/, Operator
+        rule %r/[.@]|\{(?:#{ArmAsm.builtin.join('|')})\}/, Name::Constant
+        rule %r/[-!#%&()*+,\/<=>?^{|}]|\[|\]|!=|&&|\/=|<<|<=|<>|==|><|>=|>>|\|\||:(?:#{ArmAsm.operator.join('|')}):/, Operator
+        rule %r/\$[a-z]\w*\.?/i, Name::Namespace
+        rule %r/'/ do |m|
+          token Literal::String::Char
+          goto :singlequoted
+        end
+        rule %r/"/ do |m|
+          token Literal::String::Double
+          goto :doublequoted
+        end
+      end
+
+      state :singlequoted do
+        rule %r/\n/, Text, :pop!
+        rule %r/\$\$/, Literal::String::Char
+        rule %r/\$[a-z]\w*\.?/i, Name::Namespace
+        rule %r/'/ do |m|
+          token Literal::String::Char
+          goto :args
+        end
+        rule %r/[^$'\n]+/, Literal::String::Char
+      end
+
+      state :doublequoted do
+        rule %r/\n/, Text, :pop!
+        rule %r/\$\$/, Literal::String::Double
+        rule %r/\$[a-z]\w*\.?/i, Name::Namespace
+        rule %r/"/ do |m|
+          token Literal::String::Double
+          goto :args
+        end
+        rule %r/[^$"\n]+/, Literal::String::Double
+      end
+
+      state :filespec do
+        rule %r/\n/, Text, :pop!
+        rule %r/\$\$/, Literal::String::Other
+        rule %r/\$[a-z]\w*\.?/i, Name::Namespace
+        rule %r/[^$\n]+/, Literal::String::Other
+      end
+    end
+  end
+end

data/lib/rouge/lexers/batchfile.rb

@@ -0,0 +1,164 @@
+# -*- coding: utf-8 -*- #
+# frozen_string_literal: true
+
+module Rouge
+  module Lexers
+    class Batchfile < RegexLexer
+      title "Batchfile"
+      desc "Windows Batch File"
+
+      tag 'batchfile'
+      aliases 'bat', 'batch', 'dosbatch', 'winbatch'
+      filenames '*.bat', '*.cmd'
+
+      mimetypes 'application/bat', 'application/x-bat', 'application/x-msdos-program'
+
+      def self.keywords
+        @keywords ||= %w(
+          if else for in do goto call exit
+        )
+      end
+
+      def self.operator_words
+        @operator_words ||= %w(
+          exist defined errorlevel cmdextversion not equ neq lss leq gtr geq
+        )
+      end
+
+      def self.devices
+        @devices ||= %w(
+          con prn aux nul com1 com2 com3 com4 com5 com6 com7 com8 com9 lpt1 lpt2
+          lpt3 lpt4 lpt5 lpt6 lpt7 lpt8 lpt9
+        )
+      end
+
+      def self.builtin_commands
+        @builtin_commands ||= %w(
+          assoc attrib break bcdedit cacls cd chcp chdir chkdsk chkntfs choice
+          cls cmd color comp compact convert copy date del dir diskpart doskey
+          dpath driverquery echo endlocal erase fc find findstr format fsutil
+          ftype gpresult graftabl help icacls label md mkdir mklink mode more
+          move openfiles path pause popd print prompt pushd rd recover ren
+          rename replace rmdir robocopy setlocal sc schtasks shift shutdown sort
+          start subst systeminfo takeown tasklist taskkill time timeout title
+          tree type ver verify vol xcopy waitfor wmic
+        )
+      end
+
+      def self.other_commands
+        @other_commands ||= %w(
+          addusers admodcmd ansicon arp at bcdboot bitsadmin browstat certreq
+          certutil change cidiag cipher cleanmgr clip cmdkey compress convertcp
+          coreinfo csccmd csvde cscript curl debug defrag delprof deltree devcon
+          diamond dirquota diruse diskshadow diskuse dism dnscmd dsacls dsadd
+          dsget dsquery dsmod dsmove dsrm dsmgmt dsregcmd edlin eventcreate
+          expand extract fdisk fltmc forfiles freedisk ftp getmac gpupdate
+          hostname ifmember inuse ipconfig kill lgpo lodctr logman logoff
+          logtime makecab mapisend mbsacli mem mountvol moveuser msg mshta
+          msiexec msinfo32 mstsc nbtstat net net1 netdom netsh netstat nlsinfo
+          nltest now nslookup ntbackup ntdsutil ntoskrnl ntrights nvspbind
+          pathping perms ping portqry powercfg pngout pnputil printbrm prncnfg
+          prnmngr procdump psexec psfile psgetsid psinfo pskill pslist
+          psloggedon psloglist pspasswd psping psservice psshutdown pssuspend
+          qbasic qgrep qprocess query quser qwinsta rasdial reg reg1 regdump
+          regedt32 regsvr32 regini reset restore rundll32 rmtshare route rpcping
+          run runas scandisk setspn setx sfc share shellrunas shortcut sigcheck
+          sleep slmgr strings subinacl sysmon telnet tftp tlist touch tracerpt
+          tracert tscon tsdiscon tskill tttracer typeperf tzutil undelete
+          unformat verifier vmconnect vssadmin w32tm wbadmin wecutil wevtutil
+          wget where whoami windiff winrm winrs wpeutil wpr wusa wuauclt wscript
+        )
+      end
+
+      def self.attributes
+        @attributes ||= %w(
+          on off disable enableextensions enabledelayedexpansion
+        )
+      end
+
+      state :basic do
+        # Comments
+        rule %r/\brem\b.*$/i, Comment
+        # Empty Labels
+        rule %r/^::.*$/, Comment
+
+        # Labels
+        rule %r/:[a-z]+/i, Name::Label
+
+        rule %r/([a-z]\w*)(\.exe|com|bat|cmd|msi)?/i do |m|
+          if self.class.devices.include? m[1]
+            groups Keyword::Reserved, Error
+          elsif self.class.keywords.include? m[1]
+            groups Keyword, Error
+          elsif self.class.operator_words.include? m[1]
+            groups Operator::Word, Error
+          elsif self.class.builtin_commands.include? m[1]
+            token Name::Builtin
+          elsif self.class.other_commands.include? m[1]
+            token Name::Builtin
+          elsif self.class.attributes.include? m[1]
+            groups Name::Attribute, Error
+          elsif "set".casecmp m[1]
+            groups Keyword::Declaration, Error
+          else
+            token Text
+          end
+        end
+
+        rule %r/([\/\-+][a-z]+)\s*/i, Name::Attribute
+
+        mixin :expansions
+
+        rule %r/[<>&|(){}\[\]\-+=;,~?*]/, Operator
+      end
+
+      state :escape do
+        rule %r/\^./m, Str::Escape
+      end
+
+      state :expansions do
+        # Normal and Delayed expansion
+        rule %r/[%!]+([a-z_$@#]+)[%!]+/i, Name::Variable
+        # For Variables
+        rule %r/(\%+~?[a-z]+\d?)/i, Name::Variable::Magic
+      end
+
+      state :double_quotes do
+        mixin :escape
+        rule %r/["]/, Str::Double, :pop!
+        mixin :expansions
+        rule %r/[^\^"%!]+/, Str::Double
+      end
+
+      state :single_quotes do
+        mixin :escape
+        rule %r/[']/, Str::Single, :pop!
+        mixin :expansions
+        rule %r/[^\^'%!]+/, Str::Single
+      end
+
+      state :backtick do
+        mixin :escape
+        rule %r/[`]/, Str::Backtick, :pop!
+        mixin :expansions
+        rule %r/[^\^`%!]+/, Str::Backtick
+      end
+
+      state :data do
+        rule %r/\s+/, Text
+        rule %r/0x[0-9a-f]+/i, Literal::Number::Hex
+        rule %r/[0-9]/, Literal::Number
+        rule %r/["]/, Str::Double, :double_quotes
+        rule %r/[']/, Str::Single, :single_quotes
+        rule %r/[`]/, Str::Backtick, :backtick
+        rule %r/[^\s&|()\[\]{}\^=;!%+\-,"'`~?*]+/, Text
+        mixin :escape
+      end
+
+      state :root do
+        mixin :basic
+        mixin :data
+      end
+    end
+  end
+end