Chris@909: module CodeRay module Scanners
Chris@909:   
Chris@909:   # by Josh Goebel
Chris@909:   class SQL < Scanner
Chris@909: 
Chris@909:     register_for :sql
Chris@909:     
Chris@909:     KEYWORDS = %w(
Chris@909:       all and any as before begin between by case check collate
Chris@909:       each else end exists
Chris@909:       for foreign from full group having if in inner is join
Chris@909:       like not of on or order outer over references
Chris@909:       then to union using values when where
Chris@909:       left right distinct
Chris@909:     )
Chris@909:     
Chris@909:     OBJECTS = %w(
Chris@909:       database databases table tables column columns fields index constraint
Chris@909:       constraints transaction function procedure row key view trigger
Chris@909:     )
Chris@909:     
Chris@909:     COMMANDS = %w(
Chris@909:       add alter comment create delete drop grant insert into select update set
Chris@909:       show prompt begin commit rollback replace truncate
Chris@909:     )
Chris@909:     
Chris@909:     PREDEFINED_TYPES = %w(
Chris@909:       char varchar varchar2 enum binary text tinytext mediumtext
Chris@909:       longtext blob tinyblob mediumblob longblob timestamp
Chris@909:       date time datetime year double decimal float int
Chris@909:       integer tinyint mediumint bigint smallint unsigned bit
Chris@909:       bool boolean hex bin oct
Chris@909:     )
Chris@909:     
Chris@909:     PREDEFINED_FUNCTIONS = %w( sum cast substring abs pi count min max avg now )
Chris@909:     
Chris@909:     DIRECTIVES = %w( 
Chris@909:       auto_increment unique default charset initially deferred
Chris@909:       deferrable cascade immediate read write asc desc after
Chris@909:       primary foreign return engine
Chris@909:     )
Chris@909:     
Chris@909:     PREDEFINED_CONSTANTS = %w( null true false )
Chris@909:     
Chris@909:     IDENT_KIND = WordList::CaseIgnoring.new(:ident).
Chris@909:       add(KEYWORDS, :keyword).
Chris@909:       add(OBJECTS, :type).
Chris@909:       add(COMMANDS, :class).
Chris@909:       add(PREDEFINED_TYPES, :predefined_type).
Chris@909:       add(PREDEFINED_CONSTANTS, :predefined_constant).
Chris@909:       add(PREDEFINED_FUNCTIONS, :predefined).
Chris@909:       add(DIRECTIVES, :directive)
Chris@909:     
Chris@909:     ESCAPE = / [rbfntv\n\\\/'"] | x[a-fA-F0-9]{1,2} | [0-7]{1,3} | . /mx
Chris@909:     UNICODE_ESCAPE =  / u[a-fA-F0-9]{4} | U[a-fA-F0-9]{8} /x
Chris@909:     
Chris@909:     STRING_PREFIXES = /[xnb]|_\w+/i
Chris@909:     
Chris@909:     def scan_tokens encoder, options
Chris@909:       
Chris@909:       state = :initial
Chris@909:       string_type = nil
Chris@909:       string_content = ''
Chris@909:       name_expected = false
Chris@909:       
Chris@909:       until eos?
Chris@909:         
Chris@909:         if state == :initial
Chris@909:           
Chris@909:           if match = scan(/ \s+ | \\\n /x)
Chris@909:             encoder.text_token match, :space
Chris@909:           
Chris@909:           elsif match = scan(/(?:--\s?|#).*/)
Chris@909:             encoder.text_token match, :comment
Chris@909:             
Chris@909:           elsif match = scan(%r( /\* (!)? (?: .*? \*/ | .* ) )mx)
Chris@909:             encoder.text_token match, self[1] ? :directive : :comment
Chris@909:             
Chris@909:           elsif match = scan(/ [*\/=<>:;,!&^|()\[\]{}~%] | [-+\.](?!\d) /x)
Chris@909:             name_expected = true if match == '.' && check(/[A-Za-z_]/)
Chris@909:             encoder.text_token match, :operator
Chris@909:             
Chris@909:           elsif match = scan(/(#{STRING_PREFIXES})?([`"'])/o)
Chris@909:             prefix = self[1]
Chris@909:             string_type = self[2]
Chris@909:             encoder.begin_group :string
Chris@909:             encoder.text_token prefix, :modifier if prefix
Chris@909:             match = string_type
Chris@909:             state = :string
Chris@909:             encoder.text_token match, :delimiter
Chris@909:             
Chris@909:           elsif match = scan(/ @? [A-Za-z_][A-Za-z_0-9]* /x)
Chris@909:             encoder.text_token match, name_expected ? :ident : (match[0] == ?@ ? :variable : IDENT_KIND[match])
Chris@909:             name_expected = false
Chris@909:             
Chris@909:           elsif match = scan(/0[xX][0-9A-Fa-f]+/)
Chris@909:             encoder.text_token match, :hex
Chris@909:             
Chris@909:           elsif match = scan(/0[0-7]+(?![89.eEfF])/)
Chris@909:             encoder.text_token match, :octal
Chris@909:             
Chris@909:           elsif match = scan(/[-+]?(?>\d+)(?![.eEfF])/)
Chris@909:             encoder.text_token match, :integer
Chris@909:             
Chris@909:           elsif match = scan(/[-+]?(?:\d[fF]|\d*\.\d+(?:[eE][+-]?\d+)?|\d+[eE][+-]?\d+)/)
Chris@909:             encoder.text_token match, :float
Chris@909:           
Chris@909:           elsif match = scan(/\\N/)
Chris@909:             encoder.text_token match, :predefined_constant
Chris@909:             
Chris@909:           else
Chris@909:             encoder.text_token getch, :error
Chris@909:             
Chris@909:           end
Chris@909:           
Chris@909:         elsif state == :string
Chris@909:           if match = scan(/[^\\"'`]+/)
Chris@909:             string_content << match
Chris@909:             next
Chris@909:           elsif match = scan(/["'`]/)
Chris@909:             if string_type == match
Chris@909:               if peek(1) == string_type  # doubling means escape
Chris@909:                 string_content << string_type << getch
Chris@909:                 next
Chris@909:               end
Chris@909:               unless string_content.empty?
Chris@909:                 encoder.text_token string_content, :content
Chris@909:                 string_content = ''
Chris@909:               end
Chris@909:               encoder.text_token match, :delimiter
Chris@909:               encoder.end_group :string
Chris@909:               state = :initial
Chris@909:               string_type = nil
Chris@909:             else
Chris@909:               string_content << match
Chris@909:             end
Chris@909:           elsif match = scan(/ \\ (?: #{ESCAPE} | #{UNICODE_ESCAPE} ) /mox)
Chris@909:             unless string_content.empty?
Chris@909:               encoder.text_token string_content, :content
Chris@909:               string_content = ''
Chris@909:             end
Chris@909:             encoder.text_token match, :char
Chris@909:           elsif match = scan(/ \\ . /mox)
Chris@909:             string_content << match
Chris@909:             next
Chris@909:           elsif match = scan(/ \\ | $ /x)
Chris@909:             unless string_content.empty?
Chris@909:               encoder.text_token string_content, :content
Chris@909:               string_content = ''
Chris@909:             end
Chris@909:             encoder.text_token match, :error
Chris@909:             state = :initial
Chris@909:           else
Chris@909:             raise "else case \" reached; %p not handled." % peek(1), encoder
Chris@909:           end
Chris@909:           
Chris@909:         else
Chris@909:           raise 'else-case reached', encoder
Chris@909:           
Chris@909:         end
Chris@909:         
Chris@909:       end
Chris@909:       
Chris@909:       if state == :string
Chris@909:         encoder.end_group state
Chris@909:       end
Chris@909:       
Chris@909:       encoder
Chris@909:       
Chris@909:     end
Chris@909:     
Chris@909:   end
Chris@909:   
Chris@909: end end