From 300986f4972f2973c3642e60952ac985814f9755 Mon Sep 17 00:00:00 2001
From: Peter Leitzen <pl@neopoly.de>
Date: Sun, 6 Mar 2016 14:14:58 +0100
Subject: [PATCH] Add support for crystal language

Heavily based on the Ruby lexer.
Add "guessing" specs for crystal lang
---
 lib/rouge/demos/crystal     |  45 ++++
 lib/rouge/lexers/crystal.rb | 416 ++++++++++++++++++++++++++++++++++++
 spec/lexers/crystal_spec.rb |  22 ++
 spec/visual/samples/crystal |  45 ++++
 4 files changed, 528 insertions(+)
 create mode 100644 lib/rouge/demos/crystal
 create mode 100644 lib/rouge/lexers/crystal.rb
 create mode 100644 spec/lexers/crystal_spec.rb
 create mode 100644 spec/visual/samples/crystal

diff --git a/lib/rouge/demos/crystal b/lib/rouge/demos/crystal
new file mode 100644
index 0000000000..988753e96d
--- /dev/null
+++ b/lib/rouge/demos/crystal
@@ -0,0 +1,45 @@
+lib LibC
+  WNOHANG = 0x00000001
+
+  @[ReturnsTwice]
+  fun fork : PidT
+  fun getpgid(pid : PidT) : PidT
+  fun kill(pid : PidT, signal : Int) : Int
+  fun getpid : PidT
+  fun getppid : PidT
+  fun exit(status : Int) : NoReturn
+
+  ifdef x86_64
+    alias ClockT = UInt64
+  else
+    alias ClockT = UInt32
+  end
+
+  SC_CLK_TCK = 3
+
+  struct Tms
+    utime : ClockT
+    stime : ClockT
+    cutime : ClockT
+    cstime : ClockT
+  end
+
+  fun times(buffer : Tms*) : ClockT
+  fun sysconf(name : Int) : Long
+end
+
+class Process
+  def self.exit(status = 0)
+    LibC.exit(status)
+  end
+
+  def self.pid
+    LibC.getpid
+  end
+
+  def self.getpgid(pid : Int32)
+    ret = LibC.getpgid(pid)
+    raise Errno.new(ret) if ret < 0
+    ret
+  end
+end
diff --git a/lib/rouge/lexers/crystal.rb b/lib/rouge/lexers/crystal.rb
new file mode 100644
index 0000000000..19029c375c
--- /dev/null
+++ b/lib/rouge/lexers/crystal.rb
@@ -0,0 +1,416 @@
+# -*- coding: utf-8 -*- #
+
+module Rouge
+  module Lexers
+    class Crystal < RegexLexer
+      title "Crystal"
+      desc "Crystal The Programming Language (crystal-lang.org)"
+      tag 'crystal'
+      aliases 'cr'
+      filenames '*.cr'
+
+      mimetypes 'text/x-crystal', 'application/x-crystal'
+
+      def self.analyze_text(text)
+        return 1 if text.shebang? 'crystal'
+      end
+
+      state :symbols do
+        # symbols
+        rule %r(
+          :  # initial :
+          @{0,2} # optional ivar, for :@foo and :@@foo
+          [a-z_]\w*[!?]? # the symbol
+        )xi, Str::Symbol
+
+        # special symbols
+        rule %r(:(?:\*\*|[-+]@|[/\%&\|^`~]|\[\]=?|<<|>>|<=?>|<=?|===?)),
+          Str::Symbol
+
+        rule /:'(\\\\|\\'|[^'])*'/, Str::Symbol
+        rule /:"/, Str::Symbol, :simple_sym
+      end
+
+      state :sigil_strings do
+        # %-sigiled strings
+        # %(abc), %[abc], %<abc>, %.abc., %r.abc., etc
+        delimiter_map = { '{' => '}', '[' => ']', '(' => ')', '<' => '>' }
+        rule /%([rqswQWxiI])?([^\w\s])/ do |m|
+          open = Regexp.escape(m[2])
+          close = Regexp.escape(delimiter_map[m[2]] || m[2])
+          interp = /[rQWxI]/ === m[1]
+          toktype = Str::Other
+
+          puts "    open: #{open.inspect}" if @debug
+          puts "    close: #{close.inspect}" if @debug
+
+          # regexes
+          if m[1] == 'r'
+            toktype = Str::Regex
+            push :regex_flags
+          end
+
+          token toktype
+
+          push do
+            rule /\\[##{open}#{close}\\]/, Str::Escape
+            # nesting rules only with asymmetric delimiters
+            if open != close
+              rule /#{open}/ do
+                token toktype
+                push
+              end
+            end
+            rule /#{close}/, toktype, :pop!
+
+            if interp
+              mixin :string_intp_escaped
+              rule /#/, toktype
+            else
+              rule /[\\#]/, toktype
+            end
+
+            rule /[^##{open}#{close}\\]+/m, toktype
+          end
+        end
+      end
+
+      state :strings do
+        mixin :symbols
+        rule /\b[a-z_]\w*?:\s+/, Str::Symbol, :expr_start
+        rule /'(\\\\|\\'|[^'])*'/, Str::Single
+        rule /"/, Str::Double, :simple_string
+        rule /(?<!\.)`/, Str::Backtick, :simple_backtick
+      end
+
+      state :regex_flags do
+        rule /[mixounse]*/, Str::Regex, :pop!
+      end
+
+      # double-quoted string and symbol
+      [[:string, Str::Double, '"'],
+       [:sym, Str::Symbol, '"'],
+       [:backtick, Str::Backtick, '`']].each do |name, tok, fin|
+        state :"simple_#{name}" do
+          mixin :string_intp_escaped
+          rule /[^\\#{fin}#]+/m, tok
+          rule /[\\#]/, tok
+          rule /#{fin}/, tok, :pop!
+        end
+      end
+
+      keywords = %w(
+        BEGIN END alias begin break case defined\? do else elsif end
+        ensure for if ifdef in next redo rescue raise retry return super then
+        undef unless until when while yield lib fun type of as
+      )
+
+      keywords_pseudo = %w(
+        initialize new loop include extend raise attr_reader attr_writer
+        attr_accessor alias_method attr catch throw private module_function
+        public protected true false nil __FILE__ __LINE__
+        getter getter? getter! property property? property! struct record
+      )
+
+      builtins_g = %w(
+        __id__ __send__ abort ancestors at_exit autoload binding callcc
+        caller catch chomp chop class_eval class_variables clone
+        const_defined\? const_get const_missing const_set constants
+        display dup eval exec exit extend fail fork format freeze
+        getc gets global_variables gsub hash id included_modules
+        inspect instance_eval instance_method instance_methods
+        instance_variable_get instance_variable_set instance_variables
+        lambda load local_variables loop method method_missing
+        methods module_eval name object_id open p print printf
+        private_class_method private_instance_methods private_methods proc
+        protected_instance_methods protected_methods public_class_method
+        public_instance_methods public_methods putc puts raise rand
+        readline readlines require require_relative scan select self send set_trace_func
+        singleton_methods sleep split sprintf srand sub syscall system
+        taint test throw to_a to_s trace_var trap untaint untrace_var warn
+      )
+
+      builtins_q = %w(
+        autoload block_given const_defined eql equal frozen
+        include instance_of is_a iterator kind_of method_defined
+        nil private_method_defined protected_method_defined
+        public_method_defined respond_to tainted
+      )
+
+      builtins_b = %w(chomp chop exit gsub sub)
+
+      start do
+        push :expr_start
+        @heredoc_queue = []
+      end
+
+      state :whitespace do
+        mixin :inline_whitespace
+        rule /\n\s*/m, Text, :expr_start
+        rule /#.*$/, Comment::Single
+
+        rule %r(=begin\b.*?end\b)m, Comment::Multiline
+      end
+
+      state :inline_whitespace do
+        rule /[ \t\r]+/, Text
+      end
+
+      state :root do
+        mixin :whitespace
+        rule /__END__/, Comment::Preproc, :end_part
+
+        rule /0_?[0-7]+(?:_[0-7]+)*/, Num::Oct
+        rule /0x[0-9A-Fa-f]+(?:_[0-9A-Fa-f]+)*/, Num::Hex
+        rule /0b[01]+(?:_[01]+)*/, Num::Bin
+        rule /[\d]+(?:_\d+)*/, Num::Integer
+
+        rule /@\[([^\]]+)\]/, Name::Decorator
+
+        # names
+        rule /@@[a-z_]\w*/i, Name::Variable::Class
+        rule /@[a-z_]\w*/i, Name::Variable::Instance
+        rule /\$\w+/, Name::Variable::Global
+        rule %r(\$[!@&`'+~=/\\,;.<>_*\$?:"]), Name::Variable::Global
+        rule /\$-[0adFiIlpvw]/, Name::Variable::Global
+        rule /::/, Operator
+
+        mixin :strings
+
+        rule /(?:#{keywords.join('|')})\b/, Keyword, :expr_start
+        rule /(?:#{keywords_pseudo.join('|')})\b/, Keyword::Pseudo, :expr_start
+
+        rule %r(
+          (module)
+          (\s+)
+          ([a-zA-Z_][a-zA-Z0-9_]*(::[a-zA-Z_][a-zA-Z0-9_]*)*)
+        )x do
+          groups Keyword, Text, Name::Namespace
+        end
+
+        rule /(def\b)(\s*)/ do
+          groups Keyword, Text
+          push :funcname
+        end
+
+        rule /(class\b)(\s*)/ do
+          groups Keyword, Text
+          push :classname
+        end
+
+        rule /(?:#{builtins_q.join('|')})[?]/, Name::Builtin, :expr_start
+        rule /(?:#{builtins_b.join('|')})!/,  Name::Builtin, :expr_start
+        rule /(?<!\.)(?:#{builtins_g.join('|')})\b/,
+          Name::Builtin, :method_call
+
+        # char operator.  ?x evaulates to "x", unless there's a digit
+        # beforehand like x>=0?n[x]:""
+        rule %r(
+          [?](\\[MC]-)*     # modifiers
+          (\\([\\abefnrstv\#"']|x[a-fA-F0-9]{1,2}|[0-7]{1,3})|\S)
+          (?!\w)
+        )x, Str::Char
+
+        mixin :has_heredocs
+
+        rule /[A-Z][a-zA-Z0-9_]*/, Name::Constant, :method_call
+        rule /(\.|::)(\s*)([a-z_]\w*[!?]?|[*%&^`~+-\/\[<>=])/ do
+          groups Punctuation, Text, Name::Function
+          push :method_call
+        end
+
+        rule /[a-zA-Z_]\w*[?!]/, Name, :expr_start
+        rule /[a-zA-Z_]\w*/, Name, :method_call
+        rule /\*\*|<<?|>>?|>=|<=|<=>|=~|={3}|!~|&&?|\|\||\.{1,3}/,
+          Operator, :expr_start
+        rule /[-+\/*%=<>&!^|~]=?/, Operator, :expr_start
+        rule %r<[\[({,?:\\;/]>, Punctuation, :expr_start
+        rule %r<[\])}]>, Punctuation
+      end
+
+      state :has_heredocs do
+        rule /(?<!\w)(<<-?)(["`']?)([a-zA-Z_]\w*)(\2)/ do |m|
+          token Operator, m[1]
+          token Name::Constant, "#{m[2]}#{m[3]}#{m[4]}"
+          @heredoc_queue << [m[1] == '<<-', m[3]]
+          push :heredoc_queue unless state? :heredoc_queue
+        end
+
+        rule /(<<-?)(["'])(\2)/ do |m|
+          token Operator, m[1]
+          token Name::Constant, "#{m[2]}#{m[3]}#{m[4]}"
+          @heredoc_queue << [m[1] == '<<-', '']
+          push :heredoc_queue unless state? :heredoc_queue
+        end
+      end
+
+      state :heredoc_queue do
+        rule /(?=\n)/ do
+          goto :resolve_heredocs
+        end
+
+        mixin :root
+      end
+
+      state :resolve_heredocs do
+        mixin :string_intp_escaped
+
+        rule /(\n)([^#\\\n]*)$/ do |m|
+          tolerant, heredoc_name = @heredoc_queue.first
+          check = tolerant ? m[2].strip : m[2].rstrip
+
+          # check if we found the end of the heredoc
+          line_tok = if check == heredoc_name
+            @heredoc_queue.shift
+            # if there's no more, we're done looking.
+            pop! if @heredoc_queue.empty?
+            Name::Constant
+          else
+            Str::Heredoc
+          end
+
+          groups(Str::Heredoc, line_tok)
+        end
+
+        rule /[#\\\n]/, Str::Heredoc
+        rule /[^#\\\n]+/, Str::Heredoc
+      end
+
+      state :funcname do
+        rule /\s+/, Text
+        rule /\(/, Punctuation, :defexpr
+        rule %r(
+          (?:([a-zA-Z_][\w_]*)(\.))?
+          (
+            [a-zA-Z_][\w_]*[!?]? |
+            \*\*? | [-+]@? | [/%&\|^`~] | \[\]=? |
+            <<? | >>? | <=>? | >= | ===?
+          )
+        )x do |m|
+          puts "matches: #{[m[0], m[1], m[2], m[3]].inspect}" if @debug
+          groups Name::Class, Operator, Name::Function
+          pop!
+        end
+
+        rule(//) { pop! }
+      end
+
+      state :classname do
+        rule /\s+/, Text
+        rule /\(/ do
+          token Punctuation
+          push :defexpr
+          push :expr_start
+        end
+
+        # class << expr
+        rule /<</ do
+          token Operator
+          goto :expr_start
+        end
+
+        rule /[A-Z_]\w*/, Name::Class
+
+        rule(//) { pop! }
+      end
+
+      state :defexpr do
+        rule /(\))(\.|::)?/ do
+          groups Punctuation, Operator
+          pop!
+        end
+        rule /\(/ do
+          token Punctuation
+          push :defexpr
+          push :expr_start
+        end
+
+        mixin :root
+      end
+
+      state :in_interp do
+        rule /}/, Str::Interpol, :pop!
+        mixin :root
+      end
+
+      state :string_intp do
+        rule /[#][{]/, Str::Interpol, :in_interp
+        rule /#(@@?|\$)[a-z_]\w*/i, Str::Interpol
+      end
+
+      state :string_intp_escaped do
+        mixin :string_intp
+        rule /\\([\\abefnrstv#"']|x[a-fA-F0-9]{1,2}|[0-7]{1,3})/,
+          Str::Escape
+        rule /\\./, Str::Escape
+      end
+
+      state :method_call do
+        rule %r(/) do
+          token Operator
+          goto :expr_start
+        end
+
+        rule(//) { goto :method_call_spaced }
+      end
+
+      state :method_call_spaced do
+        mixin :whitespace
+
+        rule %r([%/]=) do
+          token Operator
+          goto :expr_start
+        end
+
+        rule %r((/)(?=\S|\s*/)) do
+          token Str::Regex
+          goto :slash_regex
+        end
+
+        mixin :sigil_strings
+
+        rule(%r((?=\s*/))) { pop! }
+
+        rule(/\s+/) { token Text; goto :expr_start }
+        rule(//) { pop! }
+      end
+
+      state :expr_start do
+        mixin :inline_whitespace
+
+        rule %r(/) do
+          token Str::Regex
+          goto :slash_regex
+        end
+
+        # special case for using a single space.  Ruby demands that
+        # these be in a single line, otherwise it would make no sense.
+        rule /(\s*)(%[rqswQWxiI]? \S* )/ do
+          groups Text, Str::Other
+          pop!
+        end
+
+        mixin :sigil_strings
+
+        rule(//) { pop! }
+      end
+
+      state :slash_regex do
+        mixin :string_intp
+        rule %r(\\\\), Str::Regex
+        rule %r(\\/), Str::Regex
+        rule %r([\\#]), Str::Regex
+        rule %r([^\\/#]+)m, Str::Regex
+        rule %r(/) do
+          token Str::Regex
+          goto :regex_flags
+        end
+      end
+
+      state :end_part do
+        # eat up the rest of the stream as Comment::Preproc
+        rule /.+/m, Comment::Preproc, :pop!
+      end
+    end
+  end
+end
diff --git a/spec/lexers/crystal_spec.rb b/spec/lexers/crystal_spec.rb
new file mode 100644
index 0000000000..6c1031b62f
--- /dev/null
+++ b/spec/lexers/crystal_spec.rb
@@ -0,0 +1,22 @@
+# -*- coding: utf-8 -*- #
+
+describe Rouge::Lexers::Crystal do
+  let(:subject) { Rouge::Lexers::Crystal.new }
+
+  describe 'guessing' do
+    include Support::Guessing
+
+    it 'guesses by filename' do
+      assert_guess :filename => 'foo.cr'
+    end
+
+    it 'guesses by mimetype' do
+      assert_guess :mimetype => 'text/x-crystal'
+      assert_guess :mimetype => 'application/x-crystal'
+    end
+
+    it 'guesses by source' do
+      assert_guess :source => '#!/usr/local/bin/crystal'
+    end
+  end
+end
diff --git a/spec/visual/samples/crystal b/spec/visual/samples/crystal
new file mode 100644
index 0000000000..988753e96d
--- /dev/null
+++ b/spec/visual/samples/crystal
@@ -0,0 +1,45 @@
+lib LibC
+  WNOHANG = 0x00000001
+
+  @[ReturnsTwice]
+  fun fork : PidT
+  fun getpgid(pid : PidT) : PidT
+  fun kill(pid : PidT, signal : Int) : Int
+  fun getpid : PidT
+  fun getppid : PidT
+  fun exit(status : Int) : NoReturn
+
+  ifdef x86_64
+    alias ClockT = UInt64
+  else
+    alias ClockT = UInt32
+  end
+
+  SC_CLK_TCK = 3
+
+  struct Tms
+    utime : ClockT
+    stime : ClockT
+    cutime : ClockT
+    cstime : ClockT
+  end
+
+  fun times(buffer : Tms*) : ClockT
+  fun sysconf(name : Int) : Long
+end
+
+class Process
+  def self.exit(status = 0)
+    LibC.exit(status)
+  end
+
+  def self.pid
+    LibC.getpid
+  end
+
+  def self.getpgid(pid : Int32)
+    ret = LibC.getpgid(pid)
+    raise Errno.new(ret) if ret < 0
+    ret
+  end
+end