deferred class
	EL_WRITEABLE_ZSTRING

inherit
	EL_ZSTRING_IMPLEMENTATION

feature -- Write to other

	write_latin (writeable: EL_WRITABLE)
		-- write `area' sequence as raw characters to `writeable'
		local
			i, l_count: INTEGER; l_area: like area
		do
			l_area := area; l_count := count
			from i := 0 until i = l_count loop
				writeable.write_encoded_character_8 (l_area [i])
				i := i + 1
			end
		end

feature -- Append to other

	append_to (other: EL_ZSTRING)
		do
			other.append (current_readable)
		end

	append_to_general (general: STRING_GENERAL)
		do
			if general.is_string_8 then
				if attached {STRING_8} general as other_8 then
					append_to_string_8 (other_8)
				end

			elseif same_type (general) and then attached {ZSTRING} general as z_str then
				z_str.append (current_readable)

			else
				if attached {STRING_32} general as other_32 then
					append_to_string_32 (other_32)
				end
			end
		end

	append_to_string_32 (other: STRING_32)
		local
			offset: INTEGER
		do
			offset := other.count
			other.grow (offset + count)
			internal_append_to_string_32 (other, offset)
		end

	append_to_string_8 (other: STRING_8)
		local
			i, o_first_index, i_upper, block_index: INTEGER; already_latin_1: BOOLEAN
			c_i: CHARACTER; uc_i: CHARACTER_32; iter: EL_COMPACT_SUBSTRINGS_32_ITERATION
		do
			other.grow (other.count + count)
			o_first_index := other.count
			i_upper := count - 1

			already_latin_1 := Codec.encoded_as_latin (1)

			if already_latin_1 and then not has_mixed_encoding then
				other.area.copy_data (area, 0, o_first_index, count)

			elseif attached unicode_table as l_unicode_table and then attached unencoded_area as area_32
				and then attached other.area as o_area and then attached area as l_area
			then
				from i := area_lower until i > i_upper loop
					c_i := l_area [i]
					inspect character_8_band (c_i)
						when Substitute then
							uc_i := iter.item ($block_index, area_32, i + 1)
							if uc_i.code <= Max_8_bit_code then
								o_area [o_first_index + i] := uc_i.to_character_8
							else
								o_area [o_first_index + i] := Substitute
							end
						when Ascii_range then
							o_area [o_first_index + i] := c_i
					else
						if already_latin_1 then
							o_area [o_first_index + i] := c_i
						else
							uc_i := l_unicode_table [c_i.code]
							if uc_i.code <= Max_8_bit_code then
								o_area [o_first_index + i] := uc_i.to_character_8
							else
								o_area [o_first_index + i] := Substitute
							end
						end
					end
					i := i + 1
				end
			end
			other.set_count (other.count + count)
		end

	append_to_utf_8 (a_utf_8: STRING_8)
		local
			i, i_upper, block_index, offset, ascii_count: INTEGER; code_i: NATURAL; c_i: CHARACTER
			iter: EL_COMPACT_SUBSTRINGS_32_ITERATION
		do
			a_utf_8.grow (a_utf_8.count + utf_8_byte_count)

			ascii_count := leading_ascii_count (area, area_lower, area_upper)
			if ascii_count > 0 then
				offset := a_utf_8.count
				a_utf_8.area.copy_data (area, area_lower, offset, ascii_count)
				a_utf_8.area [ascii_count + offset] := '%U'
				a_utf_8.set_count (a_utf_8.count + ascii_count)
			end
			if ascii_count < count and then attached area as l_area and then attached unencoded_area as area_32
				and then attached Unicode_table as uc_table and then attached Utf_8_sequence as utf_8
			then
				i_upper := area_upper
				from i := area_lower + ascii_count until i > i_upper loop
					c_i := l_area [i]
					inspect character_8_band (c_i)
						when Substitute then
							utf_8.set (iter.item ($block_index, area_32, i + 1))
							utf_8.append_to_string (a_utf_8)

						when Ascii_range then
							a_utf_8.extend (c_i)
					else
						code_i := uc_table [c_i.code].natural_32_code
						if code_i <= 0x7F then
							a_utf_8.extend (code_i.to_character_8)
						else
							utf_8.set_area (code_i)
							utf_8.append_to_string (a_utf_8)
						end
					end
					i := i + 1
				end
			end
		ensure
			size_agrees: utf_8_byte_count = (a_utf_8.count - old a_utf_8.count)
		end

feature -- Basic operations

	fill_with_z_code (str: STRING_32)
		-- fill `str' with z_code characters
		local
			i, l_count, block_index: INTEGER; c_i: CHARACTER; z_code_i: NATURAL
			iter: EL_COMPACT_SUBSTRINGS_32_ITERATION
		do
			l_count := count
			str.grow (l_count)
			str.set_count (l_count)
			if attached str.area as str_area_32 and then attached area as l_area
				and then attached unencoded_area as area_32
			then
				from i := 0 until i = l_count loop
					c_i := l_area [i]
					inspect c_i
						when Substitute then
							z_code_i := unicode_to_z_code (iter.code ($block_index, area_32, i + 1))
							str_area_32 [i] := z_code_i.to_character_32
					else
						str_area_32 [i] := c_i
					end
					i := i + 1
				end
				str_area_32 [i] := '%U'
			end
		end

	write_utf_8_to (writable: EL_WRITABLE)
		local
			i, i_upper, block_index: INTEGER; code_i: NATURAL; c_i: CHARACTER
			iter: EL_COMPACT_SUBSTRINGS_32_ITERATION
		do
			if attached area as l_area  and then attached unencoded_area as area_32
				and then attached Unicode_table as uc_table and then attached Utf_8_sequence as utf_8
			then
				from i_upper := count - 1 until i > i_upper loop
					c_i := l_area [i]
					inspect character_8_band (c_i)
						when Substitute then
							utf_8.set (iter.item ($block_index, area_32, i + 1))
							utf_8.write (writable)

						when Ascii_range then
							writable.write_encoded_character_8 (l_area [i])
					else
						code_i := uc_table [c_i.code].natural_32_code
						if code_i <= 0x7F then
							writable.write_encoded_character_8 (code_i.to_character_8)
						else
							utf_8.set_area (code_i)
							utf_8.write (writable)
						end
					end
					i := i + 1
				end
			end
		end

feature {NONE} -- Implementation

	internal_append_to_string_32 (other: STRING_32; offset: INTEGER)
		require
			sufficient_capacity: count <= other.capacity - other.count
		local
			i, j, i_upper: INTEGER; c_i: CHARACTER
		do
			if attached other.area as area_out then
				if has_mixed_encoding then
					Codec.decode (count, area, area_out, offset)
					write_unencoded (area_out, offset, count, False)

				elseif attached area as l_area then
					i_upper := count - 1
					inspect Codec.id
						when 1 then
							from j := offset until i > i_upper loop
								area_out [j] := l_area [i]
								i := i + 1; j := j + 1
							end
						when 15 then
							from j := offset until i > i_upper loop
								c_i := l_area [i]
								inspect c_i
								-- 8 points that differ from Latin-1
									when '¤' then area_out [j] := '€'
									when '¦' then area_out [j] := 'Š'
									when '¨' then area_out [j] := 'š'
									when '´' then area_out [j] := 'Ž'
									when '¸' then area_out [j] := 'ž'
									when '¼' then area_out [j] := 'Œ'
									when '½' then area_out [j] := 'œ'
									when '¾' then area_out [j] := 'Ÿ'
								else
									area_out [j] := c_i
								end
								i := i + 1; j := j + 1
							end
					else
						Codec.decode (count, area, area_out, offset)
						write_unencoded (area_out, offset, count, False)
					end
				end
				area_out [offset + count] := '%U'
				other.set_count (offset + count)
			end
		end

end