From 22f75d539dca7b6f33cbf86e4e9d1944bb22731f Mon Sep 17 00:00:00 2001 From: Manfred Stienstra Date: Sun, 21 Sep 2008 17:21:30 +0200 Subject: Simplify ActiveSupport::Multibyte and make it run on Ruby 1.9. * Unicode methods are now defined directly on Chars instead of a handler * Updated Unicode database to Unicode 5.1.0 * Improved documentation --- .../lib/active_support/multibyte/chars.rb | 782 +++++++++++++++++---- .../lib/active_support/multibyte/exceptions.rb | 7 + .../multibyte/generators/generate_tables.rb | 149 ---- .../multibyte/handlers/passthru_handler.rb | 9 - .../multibyte/handlers/utf8_handler.rb | 564 --------------- .../multibyte/handlers/utf8_handler_proc.rb | 43 -- .../active_support/multibyte/unicode_database.rb | 71 ++ 7 files changed, 733 insertions(+), 892 deletions(-) create mode 100644 activesupport/lib/active_support/multibyte/exceptions.rb delete mode 100755 activesupport/lib/active_support/multibyte/generators/generate_tables.rb delete mode 100644 activesupport/lib/active_support/multibyte/handlers/passthru_handler.rb delete mode 100644 activesupport/lib/active_support/multibyte/handlers/utf8_handler.rb delete mode 100644 activesupport/lib/active_support/multibyte/handlers/utf8_handler_proc.rb create mode 100644 activesupport/lib/active_support/multibyte/unicode_database.rb (limited to 'activesupport/lib/active_support/multibyte') diff --git a/activesupport/lib/active_support/multibyte/chars.rb b/activesupport/lib/active_support/multibyte/chars.rb index de2c83f8d1..c05419bfbf 100644 --- a/activesupport/lib/active_support/multibyte/chars.rb +++ b/activesupport/lib/active_support/multibyte/chars.rb @@ -1,142 +1,670 @@ -require 'active_support/multibyte/handlers/utf8_handler' -require 'active_support/multibyte/handlers/passthru_handler' - -# Encapsulates all the functionality related to the Chars proxy. -module ActiveSupport::Multibyte #:nodoc: - # Chars enables you to work transparently with multibyte encodings in the Ruby String class without having extensive - # knowledge about the encoding. A Chars object accepts a string upon initialization and proxies String methods in an - # encoding safe manner. All the normal String methods are also implemented on the proxy. - # - # String methods are proxied through the Chars object, and can be accessed through the +chars+ method. Methods - # which would normally return a String object now return a Chars object so methods can be chained. - # - # "The Perfect String ".chars.downcase.strip.normalize # => "the perfect string" - # - # Chars objects are perfectly interchangeable with String objects as long as no explicit class checks are made. - # If certain methods do explicitly check the class, call +to_s+ before you pass chars objects to them. - # - # bad.explicit_checking_method "T".chars.downcase.to_s - # - # The actual operations on the string are delegated to handlers. Theoretically handlers can be implemented for - # any encoding, but the default handler handles UTF-8. This handler is set during initialization, if you want to - # use you own handler, you can set it on the Chars class. Look at the UTF8Handler source for an example how to - # implement your own handler. If you your own handler to work on anything but UTF-8 you probably also - # want to override Chars#handler. - # - # ActiveSupport::Multibyte::Chars.handler = MyHandler - # - # Note that a few methods are defined on Chars instead of the handler because they are defined on Object or Kernel - # and method_missing can't catch them. - class Chars - - attr_reader :string # The contained string - alias_method :to_s, :string - - include Comparable - - # The magic method to make String and Chars comparable - def to_str - # Using any other ways of overriding the String itself will lead you all the way from infinite loops to - # core dumps. Don't go there. - @string - end +# encoding: utf-8 + +module ActiveSupport #:nodoc: + module Multibyte #:nodoc: + # Chars enables you to work transparently with multibyte encodings in the Ruby String class without having extensive + # knowledge about the encoding. A Chars object accepts a string upon initialization and proxies String methods in an + # encoding safe manner. All the normal String methods are also implemented on the proxy. + # + # String methods are proxied through the Chars object, and can be accessed through the +mb_chars+ method. Methods + # which would normally return a String object now return a Chars object so methods can be chained. + # + # "The Perfect String ".chars.downcase.strip.normalize #=> "the perfect string" + # + # Chars objects are perfectly interchangeable with String objects as long as no explicit class checks are made. + # If certain methods do explicitly check the class, call +to_s+ before you pass chars objects to them. + # + # bad.explicit_checking_method "T".chars.downcase.to_s + # + # The default Chars implementation assumes that the encoding of the string is UTF-8, if you want to handle different + # encodings you can write your own multibyte string handler and configure it through + # ActiveSupport::Multibyte.proxy_class. + # + # class CharsForUTF32 + # def size + # @wrapped_string.size / 4 + # end + # + # def self.accepts?(string) + # string.length % 4 == 0 + # end + # end + # + # ActiveSupport::Multibyte.proxy_class = CharsForUTF32 + class Chars + # Hangul character boundaries and properties + HANGUL_SBASE = 0xAC00 + HANGUL_LBASE = 0x1100 + HANGUL_VBASE = 0x1161 + HANGUL_TBASE = 0x11A7 + HANGUL_LCOUNT = 19 + HANGUL_VCOUNT = 21 + HANGUL_TCOUNT = 28 + HANGUL_NCOUNT = HANGUL_VCOUNT * HANGUL_TCOUNT + HANGUL_SCOUNT = 11172 + HANGUL_SLAST = HANGUL_SBASE + HANGUL_SCOUNT + HANGUL_JAMO_FIRST = 0x1100 + HANGUL_JAMO_LAST = 0x11FF + + # All the unicode whitespace + UNICODE_WHITESPACE = [ + (0x0009..0x000D).to_a, # White_Space # Cc [5] .. + 0x0020, # White_Space # Zs SPACE + 0x0085, # White_Space # Cc + 0x00A0, # White_Space # Zs NO-BREAK SPACE + 0x1680, # White_Space # Zs OGHAM SPACE MARK + 0x180E, # White_Space # Zs MONGOLIAN VOWEL SEPARATOR + (0x2000..0x200A).to_a, # White_Space # Zs [11] EN QUAD..HAIR SPACE + 0x2028, # White_Space # Zl LINE SEPARATOR + 0x2029, # White_Space # Zp PARAGRAPH SEPARATOR + 0x202F, # White_Space # Zs NARROW NO-BREAK SPACE + 0x205F, # White_Space # Zs MEDIUM MATHEMATICAL SPACE + 0x3000, # White_Space # Zs IDEOGRAPHIC SPACE + ].flatten.freeze + + # BOM (byte order mark) can also be seen as whitespace, it's a non-rendering character used to distinguish + # between little and big endian. This is not an issue in utf-8, so it must be ignored. + UNICODE_LEADERS_AND_TRAILERS = UNICODE_WHITESPACE + [65279] # ZERO-WIDTH NO-BREAK SPACE aka BOM + + # Returns a regular expression pattern that matches the passed Unicode codepoints + def self.codepoints_to_pattern(array_of_codepoints) #:nodoc: + array_of_codepoints.collect{ |e| [e].pack 'U*' }.join('|') + end + UNICODE_TRAILERS_PAT = /(#{codepoints_to_pattern(UNICODE_LEADERS_AND_TRAILERS)})+\Z/ + UNICODE_LEADERS_PAT = /\A(#{codepoints_to_pattern(UNICODE_LEADERS_AND_TRAILERS)})+/ + + # Borrowed from the Kconv library by Shinji KONO - (also as seen on the W3C site) + UTF8_PAT = /\A(?: + [\x00-\x7f] | + [\xc2-\xdf] [\x80-\xbf] | + \xe0 [\xa0-\xbf] [\x80-\xbf] | + [\xe1-\xef] [\x80-\xbf] [\x80-\xbf] | + \xf0 [\x90-\xbf] [\x80-\xbf] [\x80-\xbf] | + [\xf1-\xf3] [\x80-\xbf] [\x80-\xbf] [\x80-\xbf] | + \xf4 [\x80-\x8f] [\x80-\xbf] [\x80-\xbf] + )*\z/xn + + attr_reader :wrapped_string + alias to_s wrapped_string + alias to_str wrapped_string + + # Creates a new Chars instance. +string+ is the wrapped string. + if '1.9'.respond_to?(:force_encoding) + def initialize(string) + @wrapped_string = string + @wrapped_string.force_encoding(Encoding::UTF_8) unless @wrapped_string.frozen? + end + else + def initialize(string) + @wrapped_string = string + end + end + + # Forward all undefined methods to the wrapped string. + def method_missing(method, *args, &block) + if method.to_s =~ /!$/ + @wrapped_string.__send__(method, *args, &block) + self + else + chars(@wrapped_string.__send__(method, *args, &block)) + end + end + + # Returns +true+ if _obj_ responds to the given method. Private methods are included in the search + # only if the optional second parameter evaluates to +true+. + def respond_to?(method, include_private=false) + super || @wrapped_string.respond_to?(method, include_private) || false + end + + # Enable more predictable duck-typing on String-like classes. See Object#acts_like?. + def acts_like_string? + true + end + + # Returns +true+ if the Chars class can and should act as a proxy for the string +string+. Returns + # +false+ otherwise. + def self.wants?(string) + RUBY_VERSION < '1.9' && $KCODE == 'UTF8' && consumes?(string) + end - # Make duck-typing with String possible - def respond_to?(method, include_priv = false) - super || @string.respond_to?(method, include_priv) || - handler.respond_to?(method, include_priv) || - (method.to_s =~ /(.*)!/ && handler.respond_to?($1, include_priv)) || + # Returns +true+ when the proxy class can handle the string. Returns +false+ otherwise. + def self.consumes?(string) + # Unpack is a little bit faster than regular expressions. + string.unpack('U*') + true + rescue ArgumentError false - end + end - # Enable more predictable duck-typing on String-like classes. See Object#acts_like?. - def acts_like_string? - true - end + include Comparable - # Create a new Chars instance. - def initialize(str) - @string = str.respond_to?(:string) ? str.string : str - end - - # Returns -1, 0 or +1 depending on whether the Chars object is to be sorted before, equal or after the - # object on the right side of the operation. It accepts any object that implements +to_s+. See String.<=> - # for more details. - def <=>(other); @string <=> other.to_s; end - - # Works just like String#split, with the exception that the items in the resulting list are Chars - # instances instead of String. This makes chaining methods easier. - def split(*args) - @string.split(*args).map { |i| i.chars } - end - - # Gsub works exactly the same as gsub on a normal string. - def gsub(*a, &b); @string.gsub(*a, &b).chars; end - - # Like String.=~ only it returns the character offset (in codepoints) instead of the byte offset. - def =~(other) - handler.translate_offset(@string, @string =~ other) - end - - # Try to forward all undefined methods to the handler, when a method is not defined on the handler, send it to - # the contained string. Method_missing is also responsible for making the bang! methods destructive. - def method_missing(m, *a, &b) - begin - # Simulate methods with a ! at the end because we can't touch the enclosed string from the handlers. - if m.to_s =~ /^(.*)\!$/ && handler.respond_to?($1) - result = handler.send($1, @string, *a, &b) - if result == @string - result = nil + # Returns -1, 0 or +1 depending on whether the Chars object is to be sorted before, equal or after the + # object on the right side of the operation. It accepts any object that implements +to_s+. See String.<=> + # for more details. + # + # Example: + # 'é'.mb_chars <=> 'ü'.mb_chars #=> -1 + def <=>(other) + @wrapped_string <=> other.to_s + end + + # Returns a new Chars object containing the other object concatenated to the string. + # + # Example: + # ('Café'.mb_chars + ' périferôl').to_s #=> "Café périferôl" + def +(other) + self << other + end + + # Like String.=~ only it returns the character offset (in codepoints) instead of the byte offset. + # + # Example: + # 'Café périferôl'.mb_chars =~ /ô/ #=> 12 + def =~(other) + translate_offset(@wrapped_string =~ other) + end + + # Works just like String#split, with the exception that the items in the resulting list are Chars + # instances instead of String. This makes chaining methods easier. + # + # Example: + # 'Café périferôl'.mb_chars.split(/é/).map { |part| part.upcase.to_s } #=> ["CAF", " P", "RIFERÔL"] + def split(*args) + @wrapped_string.split(*args).map { |i| i.mb_chars } + end + + # Inserts the passed string at specified codepoint offsets + # + # Example: + # 'Café'.mb_chars.insert(4, ' périferôl').to_s #=> "Café périferôl" + def insert(offset, fragment) + unpacked = self.class.u_unpack(@wrapped_string) + unless offset > unpacked.length + @wrapped_string.replace( + self.class.u_unpack(@wrapped_string).insert(offset, *self.class.u_unpack(fragment)).pack('U*') + ) + else + raise IndexError, "index #{offset} out of string" + end + self + end + + # Returns true if contained string contains +other+. Returns false otherwise. + # + # Example: + # 'Café'.mb_chars.include?('é') #=> true + def include?(other) + # We have to redefine this method because Enumerable defines it. + @wrapped_string.include?(other) + end + + # Returns the position of the passed argument in the string, counting in codepoints + # + # Example: + # 'Café périferôl'.mb_chars.index('ô') #=> 12 + def index(*args) + index = @wrapped_string.index(*args) + index ? (self.class.u_unpack(@wrapped_string.slice(0...index)).size) : nil + end + + # Works just like the indexed replace method on string, except instead of byte offsets you specify + # character offsets. + # + # Example: + # + # s = "Müller" + # s.chars[2] = "e" # Replace character with offset 2 + # s + # #=> "Müeler" + # + # s = "Müller" + # s.chars[1, 2] = "ö" # Replace 2 characters at character offset 1 + # s + # #=> "Möler" + def []=(*args) + replace_by = args.pop + # Indexed replace with regular expressions already works + if args.first.is_a?(Regexp) + @wrapped_string[*args] = replace_by + else + result = self.class.u_unpack(@wrapped_string) + if args[0].is_a?(Fixnum) + raise IndexError, "index #{args[0]} out of string" if args[0] >= result.length + min = args[0] + max = args[1].nil? ? min : (min + args[1] - 1) + range = Range.new(min, max) + replace_by = [replace_by].pack('U') if replace_by.is_a?(Fixnum) + elsif args.first.is_a?(Range) + raise RangeError, "#{args[0]} out of range" if args[0].min >= result.length + range = args[0] else - @string.replace result + needle = args[0].to_s + min = index(needle) + max = min + self.class.u_unpack(needle).length - 1 + range = Range.new(min, max) end - elsif handler.respond_to?(m) - result = handler.send(m, @string, *a, &b) - else - result = @string.send(m, *a, &b) + result[range] = self.class.u_unpack(replace_by) + @wrapped_string.replace(result.pack('U*')) end - rescue Handlers::EncodingError - @string.replace handler.tidy_bytes(@string) - retry + self end - - if result.kind_of?(String) - result.chars - else - result + + # Works just like String#rjust, only integer specifies characters instead of bytes. + # + # Example: + # + # "¾ cup".chars.rjust(8).to_s + # #=> " ¾ cup" + # + # "¾ cup".chars.rjust(8, " ").to_s # Use non-breaking whitespace + # #=> "   ¾ cup" + def rjust(integer, padstr=' ') + justify(integer, :right, padstr) end - end - - # Set the handler class for the Char objects. - def self.handler=(klass) - @@handler = klass - end - # Returns the proper handler for the contained string depending on $KCODE and the encoding of the string. This - # method is used internally to always redirect messages to the proper classes depending on the context. - def handler - if utf8_pragma? - @@handler - else - ActiveSupport::Multibyte::Handlers::PassthruHandler + # Works just like String#ljust, only integer specifies characters instead of bytes. + # + # Example: + # + # "¾ cup".chars.rjust(8).to_s + # #=> "¾ cup " + # + # "¾ cup".chars.rjust(8, " ").to_s # Use non-breaking whitespace + # #=> "¾ cup   " + def ljust(integer, padstr=' ') + justify(integer, :left, padstr) + end + + # Works just like String#center, only integer specifies characters instead of bytes. + # + # Example: + # + # "¾ cup".chars.center(8).to_s + # #=> " ¾ cup " + # + # "¾ cup".chars.center(8, " ").to_s # Use non-breaking whitespace + # #=> " ¾ cup  " + def center(integer, padstr=' ') + justify(integer, :center, padstr) end - end - private + # Strips entire range of Unicode whitespace from the right of the string. + def rstrip + chars(@wrapped_string.gsub(UNICODE_TRAILERS_PAT, '')) + end + + # Strips entire range of Unicode whitespace from the left of the string. + def lstrip + chars(@wrapped_string.gsub(UNICODE_LEADERS_PAT, '')) + end + + # Strips entire range of Unicode whitespace from the right and left of the string. + def strip + rstrip.lstrip + end + + # Returns the number of codepoints in the string + def size + self.class.u_unpack(@wrapped_string).size + end + alias_method :length, :size - # +utf8_pragma+ checks if it can send this string to the handlers. It makes sure @string isn't nil and $KCODE is - # set to 'UTF8'. - def utf8_pragma? - !@string.nil? && ($KCODE == 'UTF8') + # Reverses all characters in the string + # + # Example: + # 'Café'.mb_chars.reverse.to_s #=> 'éfaC' + def reverse + chars(self.class.u_unpack(@wrapped_string).reverse.pack('U*')) end + + # Implements Unicode-aware slice with codepoints. Slicing on one point returns the codepoints for that + # character. + # + # Example: + # 'こにちわ'.mb_chars.slice(2..3).to_s #=> "ちわ" + def slice(*args) + if args.size > 2 + raise ArgumentError, "wrong number of arguments (#{args.size} for 1)" # Do as if we were native + elsif (args.size == 2 && !(args.first.is_a?(Numeric) || args.first.is_a?(Regexp))) + raise TypeError, "cannot convert #{args.first.class} into Integer" # Do as if we were native + elsif (args.size == 2 && !args[1].is_a?(Numeric)) + raise TypeError, "cannot convert #{args[1].class} into Integer" # Do as if we were native + elsif args[0].kind_of? Range + cps = self.class.u_unpack(@wrapped_string).slice(*args) + result = cps.nil? ? nil : cps.pack('U*') + elsif args[0].kind_of? Regexp + result = @wrapped_string.slice(*args) + elsif args.size == 1 && args[0].kind_of?(Numeric) + character = self.class.u_unpack(@wrapped_string)[args[0]] + result = character.nil? ? nil : [character].pack('U') + else + result = self.class.u_unpack(@wrapped_string).slice(*args).pack('U*') + end + result.nil? ? nil : chars(result) + end + alias_method :[], :slice + + # Convert characters in the string to uppercase + # + # Example: + # 'Laurent, òu sont les tests?'.mb_chars.upcase.to_s #=> "LAURENT, ÒU SONT LES TESTS?" + def upcase + apply_mapping :uppercase_mapping + end + + # Convert characters in the string to lowercase + # + # Example: + # 'VĚDA A VÝZKUM'.mb_chars.downcase.to_s #=> "věda a výzkum" + def downcase + apply_mapping :lowercase_mapping + end + + # Converts the first character to uppercase and the remainder to lowercase + # + # Example: + # 'über'.mb_chars.capitalize.to_s #=> "Über" + def capitalize + (slice(0) || '').upcase + (slice(1..-1) || '').downcase + end + + # Returns the KC normalization of the string by default. NFKC is considered the best normalization form for + # passing strings to databases and validations. + # + # * str - The string to perform normalization on. + # * form - The form you want to normalize in. Should be one of the following: + # :c, :kc, :d, or :kd. Default is + # ActiveSupport::Multibyte.default_normalization_form + def normalize(form=ActiveSupport::Multibyte.default_normalization_form) + # See http://www.unicode.org/reports/tr15, Table 1 + codepoints = self.class.u_unpack(@wrapped_string) + chars(case form + when :d + self.class.reorder_characters(self.class.decompose_codepoints(:canonical, codepoints)) + when :c + self.class.compose_codepoints(self.class.reorder_characters(self.class.decompose_codepoints(:canonical, codepoints))) + when :kd + self.class.reorder_characters(self.class.decompose_codepoints(:compatability, codepoints)) + when :kc + self.class.compose_codepoints(self.class.reorder_characters(self.class.decompose_codepoints(:compatability, codepoints))) + else + raise ArgumentError, "#{form} is not a valid normalization variant", caller + end.pack('U*')) + end + + # Performs canonical decomposition on all the characters. + # + # Example: + # 'é'.length #=> 2 + # 'é'.mb_chars.decompose.to_s.length #=> 3 + def decompose + chars(self.class.decompose_codepoints(:canonical, self.class.u_unpack(@wrapped_string)).pack('U*')) + end + + # Performs composition on all the characters. + # + # Example: + # 'é'.length #=> 3 + # 'é'.mb_chars.compose.to_s.length #=> 2 + def compose + chars(self.class.compose_codepoints(self.class.u_unpack(@wrapped_string)).pack('U*')) + end + + # Returns the number of grapheme clusters in the string. + # + # Example: + # 'क्षि'.mb_chars.length #=> 4 + # 'क्षि'.mb_chars.g_length #=> 3 + def g_length + self.class.g_unpack(@wrapped_string).length + end + + def tidy_bytes + chars(self.class.tidy_bytes(@wrapped_string)) + end + + %w(lstrip rstrip strip reverse upcase downcase slice tidy_bytes capitalize).each do |method| + define_method("#{method}!") do |*args| + unless args.nil? + @wrapped_string = send(method, *args).to_s + else + @wrapped_string = send(method).to_s + end + self + end + end + + class << self + + # Unpack the string at codepoints boundaries + def u_unpack(str) + begin + str.unpack 'U*' + rescue ArgumentError + raise EncodingError.new('malformed UTF-8 character') + end + end + + # Detect whether the codepoint is in a certain character class. Primarily used by the + # grapheme cluster support. + def in_char_class?(codepoint, classes) + classes.detect { |c| UCD.boundary[c] === codepoint } ? true : false + end + + # Unpack the string at grapheme boundaries + def g_unpack(str) + codepoints = u_unpack(str) + unpacked = [] + pos = 0 + marker = 0 + eoc = codepoints.length + while(pos < eoc) + pos += 1 + previous = codepoints[pos-1] + current = codepoints[pos] + if ( + # CR X LF + one = ( previous == UCD.boundary[:cr] and current == UCD.boundary[:lf] ) or + # L X (L|V|LV|LVT) + two = ( UCD.boundary[:l] === previous and in_char_class?(current, [:l,:v,:lv,:lvt]) ) or + # (LV|V) X (V|T) + three = ( in_char_class?(previous, [:lv,:v]) and in_char_class?(current, [:v,:t]) ) or + # (LVT|T) X (T) + four = ( in_char_class?(previous, [:lvt,:t]) and UCD.boundary[:t] === current ) or + # X Extend + five = (UCD.boundary[:extend] === current) + ) + else + unpacked << codepoints[marker..pos-1] + marker = pos + end + end + unpacked + end + + # Reverse operation of g_unpack + def g_pack(unpacked) + (unpacked.flatten).pack('U*') + end + + # Generates a padding string of a certain size. + def padding(padsize, padstr=' ') + if padsize != 0 + new(padstr * ((padsize / u_unpack(padstr).size) + 1)).slice(0, padsize) + else + '' + end + end + + # Re-order codepoints so the string becomes canonical + def reorder_characters(codepoints) + length = codepoints.length- 1 + pos = 0 + while pos < length do + cp1, cp2 = UCD.codepoints[codepoints[pos]], UCD.codepoints[codepoints[pos+1]] + if (cp1.combining_class > cp2.combining_class) && (cp2.combining_class > 0) + codepoints[pos..pos+1] = cp2.code, cp1.code + pos += (pos > 0 ? -1 : 1) + else + pos += 1 + end + end + codepoints + end + + # Decompose composed characters to the decomposed form + def decompose_codepoints(type, codepoints) + codepoints.inject([]) do |decomposed, cp| + # if it's a hangul syllable starter character + if HANGUL_SBASE <= cp and cp < HANGUL_SLAST + sindex = cp - HANGUL_SBASE + ncp = [] # new codepoints + ncp << HANGUL_LBASE + sindex / HANGUL_NCOUNT + ncp << HANGUL_VBASE + (sindex % HANGUL_NCOUNT) / HANGUL_TCOUNT + tindex = sindex % HANGUL_TCOUNT + ncp << (HANGUL_TBASE + tindex) unless tindex == 0 + decomposed.concat ncp + # if the codepoint is decomposable in with the current decomposition type + elsif (ncp = UCD.codepoints[cp].decomp_mapping) and (!UCD.codepoints[cp].decomp_type || type == :compatability) + decomposed.concat decompose_codepoints(type, ncp.dup) + else + decomposed << cp + end + end + end + + # Compose decomposed characters to the composed form + def compose_codepoints(codepoints) + pos = 0 + eoa = codepoints.length - 1 + starter_pos = 0 + starter_char = codepoints[0] + previous_combining_class = -1 + while pos < eoa + pos += 1 + lindex = starter_char - HANGUL_LBASE + # -- Hangul + if 0 <= lindex and lindex < HANGUL_LCOUNT + vindex = codepoints[starter_pos+1] - HANGUL_VBASE rescue vindex = -1 + if 0 <= vindex and vindex < HANGUL_VCOUNT + tindex = codepoints[starter_pos+2] - HANGUL_TBASE rescue tindex = -1 + if 0 <= tindex and tindex < HANGUL_TCOUNT + j = starter_pos + 2 + eoa -= 2 + else + tindex = 0 + j = starter_pos + 1 + eoa -= 1 + end + codepoints[starter_pos..j] = (lindex * HANGUL_VCOUNT + vindex) * HANGUL_TCOUNT + tindex + HANGUL_SBASE + end + starter_pos += 1 + starter_char = codepoints[starter_pos] + # -- Other characters + else + current_char = codepoints[pos] + current = UCD.codepoints[current_char] + if current.combining_class > previous_combining_class + if ref = UCD.composition_map[starter_char] + composition = ref[current_char] + else + composition = nil + end + unless composition.nil? + codepoints[starter_pos] = composition + starter_char = composition + codepoints.delete_at pos + eoa -= 1 + pos -= 1 + previous_combining_class = -1 + else + previous_combining_class = current.combining_class + end + else + previous_combining_class = current.combining_class + end + if current.combining_class == 0 + starter_pos = pos + starter_char = codepoints[pos] + end + end + end + codepoints + end + + # Replaces all the non-UTF-8 bytes by their iso-8859-1 or cp1252 equivalent resulting in a valid UTF-8 string + def tidy_bytes(str) + str.split(//u).map do |c| + if !UTF8_PAT.match(c) + n = c.unpack('C')[0] + n < 128 ? n.chr : + n < 160 ? [UCD.cp1252[n] || n].pack('U') : + n < 192 ? "\xC2" + n.chr : "\xC3" + (n-64).chr + else + c + end + end.join + end + end + + protected + + # Translate a byte offset in the wrapped string to a character offset by looking for the character boundary + def translate_offset(byte_offset) + return nil if byte_offset.nil? + return 0 if @wrapped_string == '' + chunk = @wrapped_string[0..byte_offset] + begin + begin + chunk.unpack('U*').length - 1 + rescue ArgumentError => e + chunk = @wrapped_string[0..(byte_offset+=1)] + # Stop retrying at the end of the string + raise e unless byte_offset < chunk.length + # We damaged a character, retry + retry + end + # Catch the ArgumentError so we can throw our own + rescue ArgumentError + raise EncodingError, 'malformed UTF-8 character' + end + end + + # Justifies a string in a certain way. Valid values for way are :right, :left and + # :center. + def justify(integer, way, padstr=' ') + raise ArgumentError, "zero width padding" if padstr.length == 0 + padsize = integer - size + padsize = padsize > 0 ? padsize : 0 + case way + when :right + result = @wrapped_string.dup.insert(0, self.class.padding(padsize, padstr)) + when :left + result = @wrapped_string.dup.insert(-1, self.class.padding(padsize, padstr)) + when :center + lpad = self.class.padding((padsize / 2.0).floor, padstr) + rpad = self.class.padding((padsize / 2.0).ceil, padstr) + result = @wrapped_string.dup.insert(0, lpad).insert(-1, rpad) + end + chars(result) + end + + # Map codepoints to one of it's attributes. + def apply_mapping(mapping) + chars(self.class.u_unpack(@wrapped_string).map do |codepoint| + cp = UCD.codepoints[codepoint] + if cp and (ncp = cp.send(mapping)) and ncp > 0 + ncp + else + codepoint + end + end.pack('U*')) + end + + # Creates a new instance + def chars(str) + self.class.new(str) + end + end end -end - -# When we can load the utf8proc library, override normalization with the faster methods -begin - require 'utf8proc_native' - require 'active_support/multibyte/handlers/utf8_handler_proc' - ActiveSupport::Multibyte::Chars.handler = ActiveSupport::Multibyte::Handlers::UTF8HandlerProc -rescue LoadError - ActiveSupport::Multibyte::Chars.handler = ActiveSupport::Multibyte::Handlers::UTF8Handler -end +end \ No newline at end of file diff --git a/activesupport/lib/active_support/multibyte/exceptions.rb b/activesupport/lib/active_support/multibyte/exceptions.rb new file mode 100644 index 0000000000..af760cc561 --- /dev/null +++ b/activesupport/lib/active_support/multibyte/exceptions.rb @@ -0,0 +1,7 @@ +# encoding: utf-8 + +module ActiveSupport #:nodoc: + module Multibyte #:nodoc: + class EncodingError < StandardError; end + end +end \ No newline at end of file diff --git a/activesupport/lib/active_support/multibyte/generators/generate_tables.rb b/activesupport/lib/active_support/multibyte/generators/generate_tables.rb deleted file mode 100755 index 7f807585c5..0000000000 --- a/activesupport/lib/active_support/multibyte/generators/generate_tables.rb +++ /dev/null @@ -1,149 +0,0 @@ -#!/usr/bin/env ruby -begin - require File.dirname(__FILE__) + '/../../../active_support' -rescue IOError -end -require 'open-uri' -require 'tmpdir' - -module ActiveSupport::Multibyte::Handlers #:nodoc: - class UnicodeDatabase #:nodoc: - def self.load - [Hash.new(Codepoint.new),[],{},{}] - end - end - - class UnicodeTableGenerator #:nodoc: - BASE_URI = "http://www.unicode.org/Public/#{ActiveSupport::Multibyte::UNICODE_VERSION}/ucd/" - SOURCES = { - :codepoints => BASE_URI + 'UnicodeData.txt', - :composition_exclusion => BASE_URI + 'CompositionExclusions.txt', - :grapheme_break_property => BASE_URI + 'auxiliary/GraphemeBreakProperty.txt', - :cp1252 => 'http://unicode.org/Public/MAPPINGS/VENDORS/MICSFT/WINDOWS/CP1252.TXT' - } - - def initialize - @ucd = UnicodeDatabase.new - - default = Codepoint.new - default.combining_class = 0 - default.uppercase_mapping = 0 - default.lowercase_mapping = 0 - @ucd.codepoints = Hash.new(default) - - @ucd.composition_exclusion = [] - @ucd.composition_map = {} - @ucd.boundary = {} - @ucd.cp1252 = {} - end - - def parse_codepoints(line) - codepoint = Codepoint.new - raise "Could not parse input." unless line =~ /^ - ([0-9A-F]+); # code - ([^;]+); # name - ([A-Z]+); # general category - ([0-9]+); # canonical combining class - ([A-Z]+); # bidi class - (<([A-Z]*)>)? # decomposition type - ((\ ?[0-9A-F]+)*); # decompomposition mapping - ([0-9]*); # decimal digit - ([0-9]*); # digit - ([^;]*); # numeric - ([YN]*); # bidi mirrored - ([^;]*); # unicode 1.0 name - ([^;]*); # iso comment - ([0-9A-F]*); # simple uppercase mapping - ([0-9A-F]*); # simple lowercase mapping - ([0-9A-F]*)$/ix # simple titlecase mapping - codepoint.code = $1.hex - #codepoint.name = $2 - #codepoint.category = $3 - codepoint.combining_class = Integer($4) - #codepoint.bidi_class = $5 - codepoint.decomp_type = $7 - codepoint.decomp_mapping = ($8=='') ? nil : $8.split.collect { |element| element.hex } - #codepoint.bidi_mirrored = ($13=='Y') ? true : false - codepoint.uppercase_mapping = ($16=='') ? 0 : $16.hex - codepoint.lowercase_mapping = ($17=='') ? 0 : $17.hex - #codepoint.titlecase_mapping = ($18=='') ? nil : $18.hex - @ucd.codepoints[codepoint.code] = codepoint - end - - def parse_grapheme_break_property(line) - if line =~ /^([0-9A-F\.]+)\s*;\s*([\w]+)\s*#/ - type = $2.downcase.intern - @ucd.boundary[type] ||= [] - if $1.include? '..' - parts = $1.split '..' - @ucd.boundary[type] << (parts[0].hex..parts[1].hex) - else - @ucd.boundary[type] << $1.hex - end - end - end - - def parse_composition_exclusion(line) - if line =~ /^([0-9A-F]+)/i - @ucd.composition_exclusion << $1.hex - end - end - - def parse_cp1252(line) - if line =~ /^([0-9A-Fx]+)\s([0-9A-Fx]+)/i - @ucd.cp1252[$1.hex] = $2.hex - end - end - - def create_composition_map - @ucd.codepoints.each do |_, cp| - if !cp.nil? and cp.combining_class == 0 and cp.decomp_type.nil? and !cp.decomp_mapping.nil? and cp.decomp_mapping.length == 2 and @ucd[cp.decomp_mapping[0]].combining_class == 0 and !@ucd.composition_exclusion.include?(cp.code) - @ucd.composition_map[cp.decomp_mapping[0]] ||= {} - @ucd.composition_map[cp.decomp_mapping[0]][cp.decomp_mapping[1]] = cp.code - end - end - end - - def normalize_boundary_map - @ucd.boundary.each do |k,v| - if [:lf, :cr].include? k - @ucd.boundary[k] = v[0] - end - end - end - - def parse - SOURCES.each do |type, url| - filename = File.join(Dir.tmpdir, "#{url.split('/').last}") - unless File.exist?(filename) - $stderr.puts "Downloading #{url.split('/').last}" - File.open(filename, 'wb') do |target| - open(url) do |source| - source.each_line { |line| target.write line } - end - end - end - File.open(filename) do |file| - file.each_line { |line| send "parse_#{type}".intern, line } - end - end - create_composition_map - normalize_boundary_map - end - - def dump_to(filename) - File.open(filename, 'wb') do |f| - f.write Marshal.dump([@ucd.codepoints, @ucd.composition_exclusion, @ucd.composition_map, @ucd.boundary, @ucd.cp1252]) - end - end - end -end - -if __FILE__ == $0 - filename = ActiveSupport::Multibyte::Handlers::UnicodeDatabase.filename - generator = ActiveSupport::Multibyte::Handlers::UnicodeTableGenerator.new - generator.parse - print "Writing to: #{filename}" - generator.dump_to filename - puts " (#{File.size(filename)} bytes)" -end diff --git a/activesupport/lib/active_support/multibyte/handlers/passthru_handler.rb b/activesupport/lib/active_support/multibyte/handlers/passthru_handler.rb deleted file mode 100644 index 916215c2ce..0000000000 --- a/activesupport/lib/active_support/multibyte/handlers/passthru_handler.rb +++ /dev/null @@ -1,9 +0,0 @@ -# Chars uses this handler when $KCODE is not set to 'UTF8'. Because this handler doesn't define any methods all call -# will be forwarded to String. -class ActiveSupport::Multibyte::Handlers::PassthruHandler #:nodoc: - - # Return the original byteoffset - def self.translate_offset(string, byte_offset) #:nodoc: - byte_offset - end -end \ No newline at end of file diff --git a/activesupport/lib/active_support/multibyte/handlers/utf8_handler.rb b/activesupport/lib/active_support/multibyte/handlers/utf8_handler.rb deleted file mode 100644 index aa9c16f575..0000000000 --- a/activesupport/lib/active_support/multibyte/handlers/utf8_handler.rb +++ /dev/null @@ -1,564 +0,0 @@ -# Contains all the handlers and helper classes -module ActiveSupport::Multibyte::Handlers #:nodoc: - class EncodingError < ArgumentError #:nodoc: - end - - class Codepoint #:nodoc: - attr_accessor :code, :combining_class, :decomp_type, :decomp_mapping, :uppercase_mapping, :lowercase_mapping - end - - class UnicodeDatabase #:nodoc: - attr_writer :codepoints, :composition_exclusion, :composition_map, :boundary, :cp1252 - - # self-expiring methods that lazily load the Unicode database and then return the value. - [:codepoints, :composition_exclusion, :composition_map, :boundary, :cp1252].each do |attr_name| - class_eval(<<-EOS, __FILE__, __LINE__) - def #{attr_name} - load - @#{attr_name} - end - EOS - end - - # Shortcut to ucd.codepoints[] - def [](index); codepoints[index]; end - - # Returns the directory in which the data files are stored - def self.dirname - File.dirname(__FILE__) + '/../../values/' - end - - # Returns the filename for the data file for this version - def self.filename - File.expand_path File.join(dirname, "unicode_tables.dat") - end - - # Loads the unicode database and returns all the internal objects of UnicodeDatabase - # Once the values have been loaded, define attr_reader methods for the instance variables. - def load - begin - @codepoints, @composition_exclusion, @composition_map, @boundary, @cp1252 = File.open(self.class.filename, 'rb') { |f| Marshal.load f.read } - rescue Exception => e - raise IOError.new("Couldn't load the unicode tables for UTF8Handler (#{e.message}), handler is unusable") - end - @codepoints ||= Hash.new(Codepoint.new) - @composition_exclusion ||= [] - @composition_map ||= {} - @boundary ||= {} - @cp1252 ||= {} - - # Redefine the === method so we can write shorter rules for grapheme cluster breaks - @boundary.each do |k,_| - @boundary[k].instance_eval do - def ===(other) - detect { |i| i === other } ? true : false - end - end if @boundary[k].kind_of?(Array) - end - - # define attr_reader methods for the instance variables - class << self - attr_reader :codepoints, :composition_exclusion, :composition_map, :boundary, :cp1252 - end - end - end - - # UTF8Handler implements Unicode aware operations for strings, these operations will be used by the Chars - # proxy when $KCODE is set to 'UTF8'. - class UTF8Handler - # Hangul character boundaries and properties - HANGUL_SBASE = 0xAC00 - HANGUL_LBASE = 0x1100 - HANGUL_VBASE = 0x1161 - HANGUL_TBASE = 0x11A7 - HANGUL_LCOUNT = 19 - HANGUL_VCOUNT = 21 - HANGUL_TCOUNT = 28 - HANGUL_NCOUNT = HANGUL_VCOUNT * HANGUL_TCOUNT - HANGUL_SCOUNT = 11172 - HANGUL_SLAST = HANGUL_SBASE + HANGUL_SCOUNT - HANGUL_JAMO_FIRST = 0x1100 - HANGUL_JAMO_LAST = 0x11FF - - # All the unicode whitespace - UNICODE_WHITESPACE = [ - (0x0009..0x000D).to_a, # White_Space # Cc [5] .. - 0x0020, # White_Space # Zs SPACE - 0x0085, # White_Space # Cc - 0x00A0, # White_Space # Zs NO-BREAK SPACE - 0x1680, # White_Space # Zs OGHAM SPACE MARK - 0x180E, # White_Space # Zs MONGOLIAN VOWEL SEPARATOR - (0x2000..0x200A).to_a, # White_Space # Zs [11] EN QUAD..HAIR SPACE - 0x2028, # White_Space # Zl LINE SEPARATOR - 0x2029, # White_Space # Zp PARAGRAPH SEPARATOR - 0x202F, # White_Space # Zs NARROW NO-BREAK SPACE - 0x205F, # White_Space # Zs MEDIUM MATHEMATICAL SPACE - 0x3000, # White_Space # Zs IDEOGRAPHIC SPACE - ].flatten.freeze - - # BOM (byte order mark) can also be seen as whitespace, it's a non-rendering character used to distinguish - # between little and big endian. This is not an issue in utf-8, so it must be ignored. - UNICODE_LEADERS_AND_TRAILERS = UNICODE_WHITESPACE + [65279] # ZERO-WIDTH NO-BREAK SPACE aka BOM - - # Borrowed from the Kconv library by Shinji KONO - (also as seen on the W3C site) - UTF8_PAT = /\A(?: - [\x00-\x7f] | - [\xc2-\xdf] [\x80-\xbf] | - \xe0 [\xa0-\xbf] [\x80-\xbf] | - [\xe1-\xef] [\x80-\xbf] [\x80-\xbf] | - \xf0 [\x90-\xbf] [\x80-\xbf] [\x80-\xbf] | - [\xf1-\xf3] [\x80-\xbf] [\x80-\xbf] [\x80-\xbf] | - \xf4 [\x80-\x8f] [\x80-\xbf] [\x80-\xbf] - )*\z/xn - - # Returns a regular expression pattern that matches the passed Unicode codepoints - def self.codepoints_to_pattern(array_of_codepoints) #:nodoc: - array_of_codepoints.collect{ |e| [e].pack 'U*' }.join('|') - end - UNICODE_TRAILERS_PAT = /(#{codepoints_to_pattern(UNICODE_LEADERS_AND_TRAILERS)})+\Z/ - UNICODE_LEADERS_PAT = /\A(#{codepoints_to_pattern(UNICODE_LEADERS_AND_TRAILERS)})+/ - - class << self - - # /// - # /// BEGIN String method overrides - # /// - - # Inserts the passed string at specified codepoint offsets - def insert(str, offset, fragment) - str.replace( - u_unpack(str).insert( - offset, - u_unpack(fragment) - ).flatten.pack('U*') - ) - end - - # Returns the position of the passed argument in the string, counting in codepoints - def index(str, *args) - bidx = str.index(*args) - bidx ? (u_unpack(str.slice(0...bidx)).size) : nil - end - - # Works just like the indexed replace method on string, except instead of byte offsets you specify - # character offsets. - # - # Example: - # - # s = "Müller" - # s.chars[2] = "e" # Replace character with offset 2 - # s # => "Müeler" - # - # s = "Müller" - # s.chars[1, 2] = "ö" # Replace 2 characters at character offset 1 - # s # => "Möler" - def []=(str, *args) - replace_by = args.pop - # Indexed replace with regular expressions already works - return str[*args] = replace_by if args.first.is_a?(Regexp) - result = u_unpack(str) - if args[0].is_a?(Fixnum) - raise IndexError, "index #{args[0]} out of string" if args[0] >= result.length - min = args[0] - max = args[1].nil? ? min : (min + args[1] - 1) - range = Range.new(min, max) - replace_by = [replace_by].pack('U') if replace_by.is_a?(Fixnum) - elsif args.first.is_a?(Range) - raise RangeError, "#{args[0]} out of range" if args[0].min >= result.length - range = args[0] - else - needle = args[0].to_s - min = index(str, needle) - max = min + length(needle) - 1 - range = Range.new(min, max) - end - result[range] = u_unpack(replace_by) - str.replace(result.pack('U*')) - end - - # Works just like String#rjust, only integer specifies characters instead of bytes. - # - # Example: - # - # "¾ cup".chars.rjust(8).to_s - # # => " ¾ cup" - # - # "¾ cup".chars.rjust(8, " ").to_s # Use non-breaking whitespace - # # => "   ¾ cup" - def rjust(str, integer, padstr=' ') - justify(str, integer, :right, padstr) - end - - # Works just like String#ljust, only integer specifies characters instead of bytes. - # - # Example: - # - # "¾ cup".chars.rjust(8).to_s - # # => "¾ cup " - # - # "¾ cup".chars.rjust(8, " ").to_s # Use non-breaking whitespace - # # => "¾ cup   " - def ljust(str, integer, padstr=' ') - justify(str, integer, :left, padstr) - end - - # Works just like String#center, only integer specifies characters instead of bytes. - # - # Example: - # - # "¾ cup".chars.center(8).to_s - # # => " ¾ cup " - # - # "¾ cup".chars.center(8, " ").to_s # Use non-breaking whitespace - # # => " ¾ cup  " - def center(str, integer, padstr=' ') - justify(str, integer, :center, padstr) - end - - # Does Unicode-aware rstrip - def rstrip(str) - str.gsub(UNICODE_TRAILERS_PAT, '') - end - - # Does Unicode-aware lstrip - def lstrip(str) - str.gsub(UNICODE_LEADERS_PAT, '') - end - - # Removed leading and trailing whitespace - def strip(str) - str.gsub(UNICODE_LEADERS_PAT, '').gsub(UNICODE_TRAILERS_PAT, '') - end - - # Returns the number of codepoints in the string - def size(str) - u_unpack(str).size - end - alias_method :length, :size - - # Reverses codepoints in the string. - def reverse(str) - u_unpack(str).reverse.pack('U*') - end - - # Implements Unicode-aware slice with codepoints. Slicing on one point returns the codepoints for that - # character. - def slice(str, *args) - if args.size > 2 - raise ArgumentError, "wrong number of arguments (#{args.size} for 1)" # Do as if we were native - elsif (args.size == 2 && !(args.first.is_a?(Numeric) || args.first.is_a?(Regexp))) - raise TypeError, "cannot convert #{args.first.class} into Integer" # Do as if we were native - elsif (args.size == 2 && !args[1].is_a?(Numeric)) - raise TypeError, "cannot convert #{args[1].class} into Integer" # Do as if we were native - elsif args[0].kind_of? Range - cps = u_unpack(str).slice(*args) - cps.nil? ? nil : cps.pack('U*') - elsif args[0].kind_of? Regexp - str.slice(*args) - elsif args.size == 1 && args[0].kind_of?(Numeric) - u_unpack(str)[args[0]] - else - u_unpack(str).slice(*args).pack('U*') - end - end - alias_method :[], :slice - - # Convert characters in the string to uppercase - def upcase(str); to_case :uppercase_mapping, str; end - - # Convert characters in the string to lowercase - def downcase(str); to_case :lowercase_mapping, str; end - - # Returns a copy of +str+ with the first character converted to uppercase and the remainder to lowercase - def capitalize(str) - upcase(slice(str, 0..0)) + downcase(slice(str, 1..-1) || '') - end - - # /// - # /// Extra String methods for unicode operations - # /// - - # Returns the KC normalization of the string by default. NFKC is considered the best normalization form for - # passing strings to databases and validations. - # - # * str - The string to perform normalization on. - # * form - The form you want to normalize in. Should be one of the following: - # :c, :kc, :d, or :kd. Default is - # ActiveSupport::Multibyte::DEFAULT_NORMALIZATION_FORM. - def normalize(str, form=ActiveSupport::Multibyte::DEFAULT_NORMALIZATION_FORM) - # See http://www.unicode.org/reports/tr15, Table 1 - codepoints = u_unpack(str) - case form - when :d - reorder_characters(decompose_codepoints(:canonical, codepoints)) - when :c - compose_codepoints reorder_characters(decompose_codepoints(:canonical, codepoints)) - when :kd - reorder_characters(decompose_codepoints(:compatability, codepoints)) - when :kc - compose_codepoints reorder_characters(decompose_codepoints(:compatability, codepoints)) - else - raise ArgumentError, "#{form} is not a valid normalization variant", caller - end.pack('U*') - end - - # Perform decomposition on the characters in the string - def decompose(str) - decompose_codepoints(:canonical, u_unpack(str)).pack('U*') - end - - # Perform composition on the characters in the string - def compose(str) - compose_codepoints u_unpack(str).pack('U*') - end - - # /// - # /// BEGIN Helper methods for unicode operation - # /// - - # Used to translate an offset from bytes to characters, for instance one received from a regular expression match - def translate_offset(str, byte_offset) - return nil if byte_offset.nil? - return 0 if str == '' - chunk = str[0..byte_offset] - begin - begin - chunk.unpack('U*').length - 1 - rescue ArgumentError => e - chunk = str[0..(byte_offset+=1)] - # Stop retrying at the end of the string - raise e unless byte_offset < chunk.length - # We damaged a character, retry - retry - end - # Catch the ArgumentError so we can throw our own - rescue ArgumentError - raise EncodingError.new('malformed UTF-8 character') - end - end - - # Checks if the string is valid UTF8. - def consumes?(str) - # Unpack is a little bit faster than regular expressions - begin - str.unpack('U*') - true - rescue ArgumentError - false - end - end - - # Returns the number of grapheme clusters in the string. This method is very likely to be moved or renamed - # in future versions. - def g_length(str) - g_unpack(str).length - end - - # Replaces all the non-utf-8 bytes by their iso-8859-1 or cp1252 equivalent resulting in a valid utf-8 string - def tidy_bytes(str) - str.split(//u).map do |c| - if !UTF8_PAT.match(c) - n = c.unpack('C')[0] - n < 128 ? n.chr : - n < 160 ? [UCD.cp1252[n] || n].pack('U') : - n < 192 ? "\xC2" + n.chr : "\xC3" + (n-64).chr - else - c - end - end.join - end - - protected - - # Detect whether the codepoint is in a certain character class. Primarily used by the - # grapheme cluster support. - def in_char_class?(codepoint, classes) - classes.detect { |c| UCD.boundary[c] === codepoint } ? true : false - end - - # Unpack the string at codepoints boundaries - def u_unpack(str) - begin - str.unpack 'U*' - rescue ArgumentError - raise EncodingError.new('malformed UTF-8 character') - end - end - - # Unpack the string at grapheme boundaries instead of codepoint boundaries - def g_unpack(str) - codepoints = u_unpack(str) - unpacked = [] - pos = 0 - marker = 0 - eoc = codepoints.length - while(pos < eoc) - pos += 1 - previous = codepoints[pos-1] - current = codepoints[pos] - if ( - # CR X LF - one = ( previous == UCD.boundary[:cr] and current == UCD.boundary[:lf] ) or - # L X (L|V|LV|LVT) - two = ( UCD.boundary[:l] === previous and in_char_class?(current, [:l,:v,:lv,:lvt]) ) or - # (LV|V) X (V|T) - three = ( in_char_class?(previous, [:lv,:v]) and in_char_class?(current, [:v,:t]) ) or - # (LVT|T) X (T) - four = ( in_char_class?(previous, [:lvt,:t]) and UCD.boundary[:t] === current ) or - # X Extend - five = (UCD.boundary[:extend] === current) - ) - else - unpacked << codepoints[marker..pos-1] - marker = pos - end - end - unpacked - end - - # Reverse operation of g_unpack - def g_pack(unpacked) - unpacked.flatten - end - - # Justifies a string in a certain way. Valid values for way are :right, :left and - # :center. Is primarily used as a helper method by rjust, ljust and center. - def justify(str, integer, way, padstr=' ') - raise ArgumentError, "zero width padding" if padstr.length == 0 - padsize = integer - size(str) - padsize = padsize > 0 ? padsize : 0 - case way - when :right - str.dup.insert(0, padding(padsize, padstr)) - when :left - str.dup.insert(-1, padding(padsize, padstr)) - when :center - lpad = padding((padsize / 2.0).floor, padstr) - rpad = padding((padsize / 2.0).ceil, padstr) - str.dup.insert(0, lpad).insert(-1, rpad) - end - end - - # Generates a padding string of a certain size. - def padding(padsize, padstr=' ') - if padsize != 0 - slice(padstr * ((padsize / size(padstr)) + 1), 0, padsize) - else - '' - end - end - - # Convert characters to a different case - def to_case(way, str) - u_unpack(str).map do |codepoint| - cp = UCD[codepoint] - unless cp.nil? - ncp = cp.send(way) - ncp > 0 ? ncp : codepoint - else - codepoint - end - end.pack('U*') - end - - # Re-order codepoints so the string becomes canonical - def reorder_characters(codepoints) - length = codepoints.length- 1 - pos = 0 - while pos < length do - cp1, cp2 = UCD[codepoints[pos]], UCD[codepoints[pos+1]] - if (cp1.combining_class > cp2.combining_class) && (cp2.combining_class > 0) - codepoints[pos..pos+1] = cp2.code, cp1.code - pos += (pos > 0 ? -1 : 1) - else - pos += 1 - end - end - codepoints - end - - # Decompose composed characters to the decomposed form - def decompose_codepoints(type, codepoints) - codepoints.inject([]) do |decomposed, cp| - # if it's a hangul syllable starter character - if HANGUL_SBASE <= cp and cp < HANGUL_SLAST - sindex = cp - HANGUL_SBASE - ncp = [] # new codepoints - ncp << HANGUL_LBASE + sindex / HANGUL_NCOUNT - ncp << HANGUL_VBASE + (sindex % HANGUL_NCOUNT) / HANGUL_TCOUNT - tindex = sindex % HANGUL_TCOUNT - ncp << (HANGUL_TBASE + tindex) unless tindex == 0 - decomposed.concat ncp - # if the codepoint is decomposable in with the current decomposition type - elsif (ncp = UCD[cp].decomp_mapping) and (!UCD[cp].decomp_type || type == :compatability) - decomposed.concat decompose_codepoints(type, ncp.dup) - else - decomposed << cp - end - end - end - - # Compose decomposed characters to the composed form - def compose_codepoints(codepoints) - pos = 0 - eoa = codepoints.length - 1 - starter_pos = 0 - starter_char = codepoints[0] - previous_combining_class = -1 - while pos < eoa - pos += 1 - lindex = starter_char - HANGUL_LBASE - # -- Hangul - if 0 <= lindex and lindex < HANGUL_LCOUNT - vindex = codepoints[starter_pos+1] - HANGUL_VBASE rescue vindex = -1 - if 0 <= vindex and vindex < HANGUL_VCOUNT - tindex = codepoints[starter_pos+2] - HANGUL_TBASE rescue tindex = -1 - if 0 <= tindex and tindex < HANGUL_TCOUNT - j = starter_pos + 2 - eoa -= 2 - else - tindex = 0 - j = starter_pos + 1 - eoa -= 1 - end - codepoints[starter_pos..j] = (lindex * HANGUL_VCOUNT + vindex) * HANGUL_TCOUNT + tindex + HANGUL_SBASE - end - starter_pos += 1 - starter_char = codepoints[starter_pos] - # -- Other characters - else - current_char = codepoints[pos] - current = UCD[current_char] - if current.combining_class > previous_combining_class - if ref = UCD.composition_map[starter_char] - composition = ref[current_char] - else - composition = nil - end - unless composition.nil? - codepoints[starter_pos] = composition - starter_char = composition - codepoints.delete_at pos - eoa -= 1 - pos -= 1 - previous_combining_class = -1 - else - previous_combining_class = current.combining_class - end - else - previous_combining_class = current.combining_class - end - if current.combining_class == 0 - starter_pos = pos - starter_char = codepoints[pos] - end - end - end - codepoints - end - - # UniCode Database - UCD = UnicodeDatabase.new - end - end -end diff --git a/activesupport/lib/active_support/multibyte/handlers/utf8_handler_proc.rb b/activesupport/lib/active_support/multibyte/handlers/utf8_handler_proc.rb deleted file mode 100644 index f10eecc622..0000000000 --- a/activesupport/lib/active_support/multibyte/handlers/utf8_handler_proc.rb +++ /dev/null @@ -1,43 +0,0 @@ -# Methods in this handler call functions in the utf8proc ruby extension. These are significantly faster than the -# pure ruby versions. Chars automatically uses this handler when it can load the utf8proc extension. For -# documentation on handler methods see UTF8Handler. -class ActiveSupport::Multibyte::Handlers::UTF8HandlerProc < ActiveSupport::Multibyte::Handlers::UTF8Handler #:nodoc: - class << self - def normalize(str, form=ActiveSupport::Multibyte::DEFAULT_NORMALIZATION_FORM) #:nodoc: - codepoints = str.unpack('U*') - case form - when :d - utf8map(str, :stable) - when :c - utf8map(str, :stable, :compose) - when :kd - utf8map(str, :stable, :compat) - when :kc - utf8map(str, :stable, :compose, :compat) - else - raise ArgumentError, "#{form} is not a valid normalization variant", caller - end - end - - def decompose(str) #:nodoc: - utf8map(str, :stable) - end - - def downcase(str) #:nodoc:c - utf8map(str, :casefold) - end - - protected - - def utf8map(str, *option_array) #:nodoc: - options = 0 - option_array.each do |option| - flag = Utf8Proc::Options[option] - raise ArgumentError, "Unknown argument given to utf8map." unless - flag - options |= flag - end - return Utf8Proc::utf8map(str, options) - end - end -end diff --git a/activesupport/lib/active_support/multibyte/unicode_database.rb b/activesupport/lib/active_support/multibyte/unicode_database.rb new file mode 100644 index 0000000000..3b8cf8f9eb --- /dev/null +++ b/activesupport/lib/active_support/multibyte/unicode_database.rb @@ -0,0 +1,71 @@ +# encoding: utf-8 + +module ActiveSupport #:nodoc: + module Multibyte #:nodoc: + # Holds data about a codepoint in the Unicode database + class Codepoint + attr_accessor :code, :combining_class, :decomp_type, :decomp_mapping, :uppercase_mapping, :lowercase_mapping + end + + # Holds static data from the Unicode database + class UnicodeDatabase + ATTRIBUTES = :codepoints, :composition_exclusion, :composition_map, :boundary, :cp1252 + + attr_writer(*ATTRIBUTES) + + def initialize + @codepoints = Hash.new(Codepoint.new) + @composition_exclusion = [] + @composition_map = {} + @boundary = {} + @cp1252 = {} + end + + # Lazy load the Unicode database so it's only loaded when it's actually used + ATTRIBUTES.each do |attr_name| + class_eval(<<-EOS, __FILE__, __LINE__) + def #{attr_name} + load + @#{attr_name} + end + EOS + end + + # Loads the Unicode database and returns all the internal objects of UnicodeDatabase. + def load + begin + @codepoints, @composition_exclusion, @composition_map, @boundary, @cp1252 = File.open(self.class.filename, 'rb') { |f| Marshal.load f.read } + rescue Exception => e + raise IOError.new("Couldn't load the Unicode tables for UTF8Handler (#{e.message}), ActiveSupport::Multibyte is unusable") + end + + # Redefine the === method so we can write shorter rules for grapheme cluster breaks + @boundary.each do |k,_| + @boundary[k].instance_eval do + def ===(other) + detect { |i| i === other } ? true : false + end + end if @boundary[k].kind_of?(Array) + end + + # define attr_reader methods for the instance variables + class << self + attr_reader(*ATTRIBUTES) + end + end + + # Returns the directory in which the data files are stored + def self.dirname + File.dirname(__FILE__) + '/../values/' + end + + # Returns the filename for the data file for this version + def self.filename + File.expand_path File.join(dirname, "unicode_tables.dat") + end + end + + # UniCode Database + UCD = UnicodeDatabase.new + end +end \ No newline at end of file -- cgit v1.2.3