Chris@0: module CodeRay
Chris@0: module Scanners
Chris@0:   
Chris@0:   # YAML Scanner
Chris@0:   #
Chris@0:   # Based on the YAML scanner from Syntax by Jamis Buck.
Chris@0:   class YAML < Scanner
Chris@0:     
Chris@0:     register_for :yaml
Chris@0:     file_extension 'yml'
Chris@0:     
Chris@0:     KINDS_NOT_LOC = :all
Chris@0:     
Chris@0:     def scan_tokens tokens, options
Chris@0:       
Chris@0:       value_expected = nil
Chris@0:       state = :initial
Chris@0:       key_indent = indent = 0
Chris@0:       
Chris@0:       until eos?
Chris@0:         
Chris@0:         kind = nil
Chris@0:         match = nil
Chris@0:         key_indent = nil if bol?
Chris@0:         
Chris@0:         if match = scan(/ +[\t ]*/)
Chris@0:           kind = :space
Chris@0:           
Chris@0:         elsif match = scan(/\n+/)
Chris@0:           kind = :space
Chris@0:           state = :initial if match.index(?\n)
Chris@0:           
Chris@0:         elsif match = scan(/#.*/)
Chris@0:           kind = :comment
Chris@0:           
Chris@0:         elsif bol? and case
Chris@0:           when match = scan(/---|\.\.\./)
Chris@0:             tokens << [:open, :head]
Chris@0:             tokens << [match, :head]
Chris@0:             tokens << [:close, :head]
Chris@0:             next
Chris@0:           when match = scan(/%.*/)
Chris@0:             tokens << [match, :doctype]
Chris@0:             next
Chris@0:           end
Chris@0:         
Chris@0:         elsif state == :value and case
Chris@0:           when !check(/(?:"[^"]*")(?=: |:$)/) && scan(/"/)
Chris@0:             tokens << [:open, :string]
Chris@0:             tokens << [matched, :delimiter]
Chris@0:             tokens << [matched, :content] if scan(/ [^"\\]* (?: \\. [^"\\]* )* /mx)
Chris@0:             tokens << [matched, :delimiter] if scan(/"/)
Chris@0:             tokens << [:close, :string]
Chris@0:             next
Chris@0:           when match = scan(/[|>][-+]?/)
Chris@0:             tokens << [:open, :string]
Chris@0:             tokens << [match, :delimiter]
Chris@0:             string_indent = key_indent || column(pos - match.size - 1)
Chris@0:             tokens << [matched, :content] if scan(/(?:\n+ {#{string_indent + 1}}.*)+/)
Chris@0:             tokens << [:close, :string]
Chris@0:             next
Chris@0:           when match = scan(/(?![!"*&]).+?(?=$|\s+#)/)
Chris@0:             tokens << [match, :string]
Chris@0:             string_indent = key_indent || column(pos - match.size - 1)
Chris@0:             tokens << [matched, :string] if scan(/(?:\n+ {#{string_indent + 1}}.*)+/)
Chris@0:             next
Chris@0:           end
Chris@0:           
Chris@0:         elsif case
Chris@0:           when match = scan(/[-:](?= |$)/)
Chris@0:             state = :value if state == :colon && (match == ':' || match == '-')
Chris@0:             state = :value if state == :initial && match == '-'
Chris@0:             kind = :operator
Chris@0:           when match = scan(/[,{}\[\]]/)
Chris@0:             kind = :operator
Chris@0:           when state == :initial && match = scan(/[\w.() ]*\S(?=: |:$)/)
Chris@0:             kind = :key
Chris@0:             key_indent = column(pos - match.size - 1)
Chris@0:             # tokens << [key_indent.inspect, :debug]
Chris@0:             state = :colon
Chris@0:           when match = scan(/(?:"[^"\n]*"|'[^'\n]*')(?=: |:$)/)
Chris@0:             tokens << [:open, :key]
Chris@0:             tokens << [match[0,1], :delimiter]
Chris@0:             tokens << [match[1..-2], :content]
Chris@0:             tokens << [match[-1,1], :delimiter]
Chris@0:             tokens << [:close, :key]
Chris@0:             key_indent = column(pos - match.size - 1)
Chris@0:             # tokens << [key_indent.inspect, :debug]
Chris@0:             state = :colon
Chris@0:             next
Chris@0:           when scan(/(![\w\/]+)(:([\w:]+))?/)
Chris@0:             tokens << [self[1], :type]
Chris@0:             if self[2]
Chris@0:               tokens << [':', :operator]
Chris@0:               tokens << [self[3], :class]
Chris@0:             end
Chris@0:             next
Chris@0:           when scan(/&\S+/)
Chris@0:             kind = :variable
Chris@0:           when scan(/\*\w+/)
Chris@0:             kind = :global_variable
Chris@0:           when scan(/<</)
Chris@0:             kind = :class_variable
Chris@0:           when scan(/\d\d:\d\d:\d\d/)
Chris@0:             kind = :oct
Chris@0:           when scan(/\d\d\d\d-\d\d-\d\d\s\d\d:\d\d:\d\d(\.\d+)? [-+]\d\d:\d\d/)
Chris@0:             kind = :oct
Chris@0:           when scan(/:\w+/)
Chris@0:             kind = :symbol
Chris@0:           when scan(/[^:\s]+(:(?! |$)[^:\s]*)* .*/)
Chris@0:             kind = :error
Chris@0:           when scan(/[^:\s]+(:(?! |$)[^:\s]*)*/)
Chris@0:             kind = :error
Chris@0:           end
Chris@0:           
Chris@0:         else
Chris@0:           getch
Chris@0:           kind = :error
Chris@0:           
Chris@0:         end
Chris@0:         
Chris@0:         match ||= matched
Chris@0:         
Chris@0:         if $CODERAY_DEBUG and not kind
Chris@0:           raise_inspect 'Error token %p in line %d' %
Chris@0:             [[match, kind], line], tokens, state
Chris@0:         end
Chris@0:         raise_inspect 'Empty token', tokens, state unless match
Chris@0:         
Chris@0:         tokens << [match, kind]
Chris@0:         
Chris@0:       end
Chris@0:       
Chris@0:       tokens
Chris@0:     end
Chris@0:     
Chris@0:   end
Chris@0:   
Chris@0: end
Chris@0: end