Chris@210: module CodeRay module Scanners
Chris@210:   
Chris@210:   # by Josh Goebel
Chris@210:   class SQL < Scanner
Chris@210: 
Chris@210:     register_for :sql
Chris@210:     
Chris@210:     RESERVED_WORDS = %w(
Chris@210:       create database table index trigger drop primary key set select
Chris@210:       insert update delete replace into
Chris@210:       on from values before and or if exists case when
Chris@210:       then else as group order by avg where
Chris@210:       join inner outer union engine not
Chris@210:       like end using collate show columns begin
Chris@210:     )
Chris@210:     
Chris@210:     PREDEFINED_TYPES = %w(
Chris@210:       char varchar enum binary text tinytext mediumtext
Chris@210:       longtext blob tinyblob mediumblob longblob timestamp
Chris@210:       date time datetime year double decimal float int
Chris@210:       integer tinyint mediumint bigint smallint unsigned bit
Chris@210:       bool boolean hex bin oct
Chris@210:     )
Chris@210:     
Chris@210:     PREDEFINED_FUNCTIONS = %w( sum cast abs pi count min max avg )
Chris@210:     
Chris@210:     DIRECTIVES = %w( auto_increment unique default charset )
Chris@210: 
Chris@210:     PREDEFINED_CONSTANTS = %w( null true false )
Chris@210:     
Chris@210:     IDENT_KIND = CaseIgnoringWordList.new(:ident).
Chris@210:       add(RESERVED_WORDS, :reserved).
Chris@210:       add(PREDEFINED_TYPES, :pre_type).
Chris@210:       add(PREDEFINED_CONSTANTS, :pre_constant).
Chris@210:       add(PREDEFINED_FUNCTIONS, :predefined).
Chris@210:       add(DIRECTIVES, :directive)
Chris@210:     
Chris@210:     ESCAPE = / [rbfntv\n\\\/'"] | x[a-fA-F0-9]{1,2} | [0-7]{1,3} | . /mx
Chris@210:     UNICODE_ESCAPE =  / u[a-fA-F0-9]{4} | U[a-fA-F0-9]{8} /x
Chris@210:     
Chris@210:     STRING_PREFIXES = /[xnb]|_\w+/i
Chris@210:     
Chris@210:     def scan_tokens tokens, options
Chris@210:       
Chris@210:       state = :initial
Chris@210:       string_type = nil
Chris@210:       string_content = ''
Chris@210:       
Chris@210:       until eos?
Chris@210:         
Chris@210:         kind = nil
Chris@210:         match = nil
Chris@210:         
Chris@210:         if state == :initial
Chris@210:           
Chris@210:           if scan(/ \s+ | \\\n /x)
Chris@210:             kind = :space
Chris@210:           
Chris@210:           elsif scan(/(?:--\s?|#).*/)
Chris@210:             kind = :comment
Chris@210:             
Chris@210:           elsif scan(%r! /\* (?: .*? \*/ | .* ) !mx)
Chris@210:             kind = :comment
Chris@210:             
Chris@210:           elsif scan(/ [-+*\/=<>;,!&^|()\[\]{}~%] | \.(?!\d) /x)
Chris@210:             kind = :operator
Chris@210:             
Chris@210:           elsif scan(/(#{STRING_PREFIXES})?([`"'])/o)
Chris@210:             prefix = self[1]
Chris@210:             string_type = self[2]
Chris@210:             tokens << [:open, :string]
Chris@210:             tokens << [prefix, :modifier] if prefix
Chris@210:             match = string_type
Chris@210:             state = :string
Chris@210:             kind = :delimiter
Chris@210:             
Chris@210:           elsif match = scan(/ @? [A-Za-z_][A-Za-z_0-9]* /x)
Chris@210:             kind = match[0] == ?@ ? :variable : IDENT_KIND[match.downcase]
Chris@210:             
Chris@210:           elsif scan(/0[xX][0-9A-Fa-f]+/)
Chris@210:             kind = :hex
Chris@210:             
Chris@210:           elsif scan(/0[0-7]+(?![89.eEfF])/)
Chris@210:             kind = :oct
Chris@210:             
Chris@210:           elsif scan(/(?>\d+)(?![.eEfF])/)
Chris@210:             kind = :integer
Chris@210:             
Chris@210:           elsif scan(/\d[fF]|\d*\.\d+(?:[eE][+-]?\d+)?|\d+[eE][+-]?\d+/)
Chris@210:             kind = :float
Chris@210:             
Chris@210:           else
Chris@210:             getch
Chris@210:             kind = :error
Chris@210:             
Chris@210:           end
Chris@210:           
Chris@210:         elsif state == :string
Chris@210:           if match = scan(/[^\\"'`]+/)
Chris@210:             string_content << match
Chris@210:             next
Chris@210:           elsif match = scan(/["'`]/)
Chris@210:             if string_type == match
Chris@210:               if peek(1) == string_type  # doubling means escape
Chris@210:                 string_content << string_type << getch
Chris@210:                 next
Chris@210:               end
Chris@210:               unless string_content.empty?
Chris@210:                 tokens << [string_content, :content]
Chris@210:                 string_content = ''
Chris@210:               end
Chris@210:               tokens << [matched, :delimiter]
Chris@210:               tokens << [:close, :string]
Chris@210:               state = :initial
Chris@210:               string_type = nil
Chris@210:               next
Chris@210:             else
Chris@210:               string_content << match
Chris@210:             end
Chris@210:             next
Chris@210:           elsif scan(/ \\ (?: #{ESCAPE} | #{UNICODE_ESCAPE} ) /mox)
Chris@210:             unless string_content.empty?
Chris@210:               tokens << [string_content, :content]
Chris@210:               string_content = ''
Chris@210:             end
Chris@210:             kind = :char
Chris@210:           elsif match = scan(/ \\ . /mox)
Chris@210:             string_content << match
Chris@210:             next
Chris@210:           elsif scan(/ \\ | $ /x)
Chris@210:             unless string_content.empty?
Chris@210:               tokens << [string_content, :content]
Chris@210:               string_content = ''
Chris@210:             end
Chris@210:             kind = :error
Chris@210:             state = :initial
Chris@210:           else
Chris@210:             raise "else case \" reached; %p not handled." % peek(1), tokens
Chris@210:           end
Chris@210:           
Chris@210:         else
Chris@210:           raise 'else-case reached', tokens
Chris@210:           
Chris@210:         end
Chris@210:         
Chris@210:         match ||= matched
Chris@210:         unless kind
Chris@210:           raise_inspect 'Error token %p in line %d' %
Chris@210:             [[match, kind], line], tokens, state
Chris@210:         end
Chris@210:         raise_inspect 'Empty token', tokens unless match
Chris@210:         
Chris@210:         tokens << [match, kind]
Chris@210:         
Chris@210:       end
Chris@210:       tokens
Chris@210:       
Chris@210:     end
Chris@210:     
Chris@210:   end
Chris@210:   
Chris@210: end end