UTF-16 encoded string

# UTF-16 encoded string
class U16String
	super Finalizable
	super Text

	# Pointer to a `UChar *` string
	private var uchar_string: UCharString

	# Number of code units (aka UTF-16 encoded code units or `UChar`) allocated to `uchar_string`
	private var capacity = 0

	# Number of code units actually in `uchar_string`.
	# `code_units` <= `capacity`.
	private var code_units = 0

	redef fun length: Int do return uchar_string.code_points(code_units)

	# Returns an empty `U16String` of capacity `cap` or a NULL `U16String` if no `cap` parameter is provided.
	# The `cap` argument is the number of code units (aka UTF-16 encoded characters or `UChar`) allocated to `uchar_string`.
	# If the number of code units is known in advance, it can be provided with the `units` parameter.
	init (cap: nullable Int, units: nullable Int) do
		if cap == null then
			uchar_string = new UCharString.nul
		else
			assert cap >= 0

			if not units == null then
				assert units <= cap
				code_units = units
			end

			uchar_string = new UCharString.empty(cap)
			capacity = cap
		end
	end

	# Returns a converted `U16String` from a `String`
	init from_string(source: String) do
		var csource = source.to_cstring
		var csource_length = source.byte_length

		uchar_string = new UCharString.nul
		var required_length = uchar_string.from_cstring(0, csource, csource_length)

		uchar_string = new UCharString.empty(required_length)
		uchar_string.from_cstring(required_length, csource, csource_length)

		capacity = required_length
		code_units = source.u16_length
	end

	# Copies the characters of `source` to `self`.
	# A maximum of `self.capacity` code units will be copied to `self`.
	# If a code point >0xFFFF has to be divided, it will not be copied.
	fun copy_from(source: String) do
		uchar_string.from_cstring(capacity, source.to_cstring, source.byte_length)

		code_units = source.u16_length
	end

	redef fun chars do return new U16StringCharView(self)

	redef fun[](index: Int): Char do
		assert index >= 0 and index < length
		var offset = 0
		var c = '\0'

		for i in [0..index] do
			c = uchar_string.char_at_offset(offset, code_units)
			if c.to_i > 0xFFFF then offset += 2 else offset +=1
		end
		return c
	end

	redef fun to_cstring: CString do
		var cself = new CString.nul
		var required_length = uchar_string.to_cstring(cself, 0, code_units)

		cself = new CString(required_length + 1)
		uchar_string.to_cstring(cself, required_length + 1, code_units)

		return cself
	end

	# Returns the number of UTF-8 code units (bytes) in `self`
	redef fun byte_length: Int do
		var offset = 0
		var l = 0
		var c = '\0'

		for i in chars do
			c = uchar_string.char_at_offset(offset, code_units)
			var b = c.to_i
			if b > 0xFFFF then offset += 2 else offset +=1

			if b <= 0x7F then
				l += 1
			else if b <= 0x7FF then
				l += 2
			else if b <= 0xD7FF or b > 0x10FFFF then
				l += 0
			else if b <= 0xFFFF then
				l += 3
			else
				l += 4
			end
		end
		return l
	end

	redef fun to_s: String do return to_cstring.to_s_with_length(byte_length)
	redef fun finalize do uchar_string.free
end

lib/core/text/u16_string.nit:30,1--142,3

class U16String

Summary

UTF-16 encoded string

Introduced properties

_capacity

_code_units

_uchar_string

capacity

capacity=

code_units

code_units=

copy_from

defaultinit

from_string

uchar_string

uchar_string=

Redefined properties

SELF

[]

byte_length

chars

finalize

length

to_cstring

to_s

UTF-16 encoded string

Introduced properties

private var _capacity: Int

private var _code_units: Int

private var _uchar_string: UCharString

private fun capacity: Int

private fun capacity=(capacity: Int)

private fun code_units: Int

private fun code_units=(code_units: Int)

fun copy_from(source: String)

init defaultinit(cap: nullable Int, units: nullable Int)

init from_string(source: String)

private fun uchar_string: UCharString

private fun uchar_string=(uchar_string: UCharString)

Redefined properties

redef type SELF: U16String

redef fun [](index: Int): Char

redef fun byte_length: Int

redef fun chars: SequenceRead[Char]

redef fun finalize

redef fun length: Int

redef fun to_cstring: CString

redef fun to_s: String

Summary

All properties

!=

*

+

/

<

<=

<=>

==

>

>=

CLASS

OTHER

SELF

SELFTYPE

[]

_capacity

_code_units

_hash_cache

_uchar_string

a_to

accept_inspect_serializer

accept_inspect_serializer_core

accept_json_serializer

accept_msgpack_attribute_counter

accept_msgpack_serializer

add_to_bundle

alert

append_to_bytes

apply_format

basename