Port PerCharacterEscaper

[nedtwigg]

• Take the next step in Epic: .ss file read/write #7.

The goal is to implement this interface

selfie-python-wip/jvm/selfie-lib/src/commonMain/kotlin/com/diffplug/selfie/PerCharacterEscaper.kt

Lines 23 to 31 in 757e0fe

	/**
	* If your escape policy is `'123`, it means this: <br>
	*
	* ```
	* abc->abc
	* 123->'1'2'3
	* I won't->I won''t
	* ```
	*/

Using this code

selfie-python-wip/jvm/selfie-lib/src/jvmMain/kotlin/com/diffplug/selfie/PerCharacterEscaper.jvm.kt

Lines 18 to 153 in 757e0fe

	/**
	* If your escape policy is "'123", it means this:
	* ```
	* abc->abc
	* 123->'1'2'3
	* I won't->I won''t
	* ```
	*/
	actual class PerCharacterEscaper
	/**
	* The first character in the string will be uses as the escape character, and all characters will
	* be escaped.
	*/
	private constructor(
	private val escapeCodePoint: Int,
	private val escapedCodePoints: IntArray,
	private val escapedByCodePoints: IntArray
	) {
	private fun firstOffsetNeedingEscape(input: String): Int {
	val length = input.length
	var firstOffsetNeedingEscape = -1
	var offset = 0
	outer@ while (offset < length) {
	val codepoint = input.codePointAt(offset)
	for (escaped in escapedCodePoints) {
	if (codepoint == escaped) {
	firstOffsetNeedingEscape = offset
	break@outer
	}
	}
	offset += Character.charCount(codepoint)
	}
	return firstOffsetNeedingEscape
	}
	actual fun escape(input: String): String {
	val noEscapes = firstOffsetNeedingEscape(input)
	return if (noEscapes == -1) {
	input
	} else {
	val length = input.length
	val needsEscapes = length - noEscapes
	val builder = StringBuilder(noEscapes + 4 + needsEscapes * 5 / 4)
	builder.append(input, 0, noEscapes)
	var offset = noEscapes
	while (offset < length) {
	val codepoint = input.codePointAt(offset)
	offset += Character.charCount(codepoint)
	val idx = indexOf(escapedCodePoints, codepoint)
	if (idx == -1) {
	builder.appendCodePoint(codepoint)
	} else {
	builder.appendCodePoint(escapeCodePoint)
	builder.appendCodePoint(escapedByCodePoints[idx])
	}
	}
	builder.toString()
	}
	}
	private fun firstOffsetNeedingUnescape(input: String): Int {
	val length = input.length
	var firstOffsetNeedingEscape = -1
	var offset = 0
	while (offset < length) {
	val codepoint = input.codePointAt(offset)
	if (codepoint == escapeCodePoint) {
	firstOffsetNeedingEscape = offset
	break
	}
	offset += Character.charCount(codepoint)
	}
	return firstOffsetNeedingEscape
	}
	actual fun unescape(input: String): String {
	val noEscapes = firstOffsetNeedingUnescape(input)
	return if (noEscapes == -1) {
	input
	} else {
	val length = input.length
	val needsEscapes = length - noEscapes
	val builder = StringBuilder(noEscapes + 4 + needsEscapes * 5 / 4)
	builder.append(input, 0, noEscapes)
	var offset = noEscapes
	while (offset < length) {
	var codepoint = input.codePointAt(offset)
	offset += Character.charCount(codepoint)
	// if we need to escape something, escape it
	if (codepoint == escapeCodePoint) {
	if (offset < length) {
	codepoint = input.codePointAt(offset)
	val idx = indexOf(escapedByCodePoints, codepoint)
	if (idx != -1) {
	codepoint = escapedCodePoints[idx]
	}
	offset += Character.charCount(codepoint)
	} else {
	throw IllegalArgumentException(
	"Escape character '" +
	String(intArrayOf(escapeCodePoint), 0, 1) +
	"' can't be the last character in a string.")
	}
	}
	// we didn't escape it, append it raw
	builder.appendCodePoint(codepoint)
	}
	builder.toString()
	}
	}
	actual companion object {
	private fun indexOf(arr: IntArray, target: Int): Int {
	for ((index, value) in arr.withIndex()) {
	if (value == target) {
	return index
	}
	}
	return -1
	}
	actual fun selfEscape(escapePolicy: String): PerCharacterEscaper {
	val escapedCodePoints = escapePolicy.codePoints().toArray()
	val escapeCodePoint = escapedCodePoints[0]
	return PerCharacterEscaper(escapeCodePoint, escapedCodePoints, escapedCodePoints)
	}
	actual fun specifiedEscape(escapePolicy: String): PerCharacterEscaper {
	val codePoints = escapePolicy.codePoints().toArray()
	require(codePoints.size % 2 == 0)
	val escapeCodePoint = codePoints[0]
	val escapedCodePoints = IntArray(codePoints.size / 2)
	val escapedByCodePoints = IntArray(codePoints.size / 2)
	for (i in escapedCodePoints.indices) {
	escapedCodePoints[i] = codePoints[2 * i]
	escapedByCodePoints[i] = codePoints[2 * i + 1]
	}
	return PerCharacterEscaper(escapeCodePoint, escapedCodePoints, escapedByCodePoints)
	}
	}
	}

[Trickybrain]

Test Part of PerCharacterEscaper

selfie-python-wip/jvm/selfie-lib/src/commonTest/kotlin/com/diffplug/selfie/PerCharacterEscaperTest.kt

Lines 25 to 91 in 5e98465

	class PerCharacterEscaperTest {
	@Test
	fun performanceOptimizationSelf() {
	val escaper = selfEscape("`123")
	// if nothing gets changed, it should return the exact same value
	val abc = "abc"
	assertSame(abc, escaper.escape(abc))
	assertSame(abc, escaper.unescape(abc))
	// otherwise it should have the normal behavior
	assertEquals("`1", escaper.escape("1"))
	assertEquals("``", escaper.escape("`"))
	assertEquals("abc`1`2`3``def", escaper.escape("abc123`def"))
	// in both directions
	assertEquals("1", escaper.unescape("`1"))
	assertEquals("`", escaper.unescape("``"))
	assertEquals("abc123`def", escaper.unescape("abc`1`2`3``def"))
	}
	@Test
	fun performanceOptimizationSpecific() {
	val escaper = specifiedEscape("`a1b2c3d")
	// if nothing gets changed, it should return the exact same value
	val abc = "abc"
	assertSame(abc, escaper.escape(abc))
	assertSame(abc, escaper.unescape(abc))
	// otherwise it should have the normal behavior
	assertEquals("`b", escaper.escape("1"))
	assertEquals("`a", escaper.escape("`"))
	assertEquals("abc`b`c`d`adef", escaper.escape("abc123`def"))
	// in both directions
	assertEquals("1", escaper.unescape("`b"))
	assertEquals("`", escaper.unescape("`a"))
	assertEquals("abc123`def", escaper.unescape("abc`1`2`3``def"))
	}
	@Test
	fun cornerCasesSelf() {
	val escaper = selfEscape("`123")
	// cornercase - escape character without follow-on will throw an error
	val exception = assertFails { escaper.unescape("`") }
	assertEquals("Escape character '`' can't be the last character in a string.", exception.message)
	// escape character followed by non-escape character is fine
	assertEquals("a", escaper.unescape("`a"))
	}
	@Test
	fun cornerCasesSpecific() {
	val escaper = specifiedEscape("`a1b2c3d")
	// cornercase - escape character without follow-on will throw an error
	val exception = assertFails { escaper.unescape("`") }
	assertEquals("Escape character '`' can't be the last character in a string.", exception.message)
	// escape character followed by non-escape character is fine
	assertEquals("e", escaper.unescape("`e"))
	}
	@Test
	fun roundtrip() {
	val escaper = selfEscape("`<>")
	val roundtrip = { str: String? -> assertEquals(str, escaper.unescape(escaper.escape(str!!))) }
	roundtrip("")
	roundtrip("<local>~`/")
	}
	}