Commonize Char.titlecaseChar() and Char.titlecase() #KT-44369

This commit is contained in:
Abduqodiri Qurbonzoda
2021-03-01 21:49:11 +03:00
parent 0c9205423f
commit 2266cd4631
15 changed files with 195 additions and 42 deletions
@@ -92,10 +92,8 @@ fun main(args: Array<String>) {
fun addOneToManyMappingsGenerators(generatedDir: File, target: KotlinTarget) {
val uppercase = OneToManyMappingsGenerator.forUppercase(generatedDir.resolve("_OneToManyUppercaseMappings.kt"), target, bmpUnicodeDataLines)
val lowercase = OneToManyMappingsGenerator.forLowercase(generatedDir.resolve("_OneToManyLowercaseMappings.kt"), target, bmpUnicodeDataLines)
val titlecase = OneToManyMappingsGenerator.forTitlecase(generatedDir.resolve("_OneToManyTitlecaseMappings.kt"), target, bmpUnicodeDataLines)
oneToManyMappingsGenerators.add(uppercase)
oneToManyMappingsGenerators.add(lowercase)
oneToManyMappingsGenerators.add(titlecase)
}
var categoryTestGenerator: CharCategoryTestGenerator? = null
@@ -112,11 +110,16 @@ fun main(args: Array<String>) {
val categoryTestFile = baseDir.resolve("libraries/stdlib/js/test/text/unicodeData/_CharCategoryTest.kt")
categoryTestGenerator = CharCategoryTestGenerator(categoryTestFile)
val commonGeneratedDir = baseDir.resolve("libraries/stdlib/common/src/generated")
oneToManyMappingsGenerators.add(OneToManyMappingsGenerator.forTitlecase(commonGeneratedDir.resolve("_OneToManyTitlecaseMappings.kt"), bmpUnicodeDataLines))
val jsGeneratedDir = baseDir.resolve("libraries/stdlib/js/src/generated/")
addRangesGenerators(jsGeneratedDir, KotlinTarget.JS)
oneToOneMappingsGenerators.add(MappingsGenerator.forTitlecase(jsGeneratedDir.resolve("_TitlecaseMappings.kt")))
val jsIrGeneratedDir = baseDir.resolve("libraries/stdlib/js-ir/src/generated/")
addRangesGenerators(jsIrGeneratedDir, KotlinTarget.JS_IR)
oneToOneMappingsGenerators.add(MappingsGenerator.forTitlecase(jsIrGeneratedDir.resolve("_TitlecaseMappings.kt")))
// For debugging. To see the file content
fun downloadFile(fromUrl: String) {
@@ -11,7 +11,7 @@ import generators.unicode.hexToInt
import java.util.TreeMap
internal abstract class OneToManyMappingsBuilder(bmpUnicodeDataLines: List<UnicodeDataLine>) {
private val bmpUnicodeDataLines = bmpUnicodeDataLines.associateBy { it.char.hexToInt() }
protected val bmpUnicodeDataLines = bmpUnicodeDataLines.associateBy { it.char.hexToInt() }
private val mappings = TreeMap<Int, List<String>>()
fun append(line: SpecialCasingLine) {
@@ -33,7 +33,7 @@ internal abstract class OneToManyMappingsBuilder(bmpUnicodeDataLines: List<Unico
}
private fun mapping(charCode: Int, line: SpecialCasingLine): List<String>? {
val mapping = line.mapping()
val mapping = line.mapping() ?: return null
check(mapping.isNotEmpty() && mapping.all { it.isNotEmpty() })
@@ -53,6 +53,6 @@ internal abstract class OneToManyMappingsBuilder(bmpUnicodeDataLines: List<Unico
return mapping
}
abstract fun SpecialCasingLine.mapping(): List<String>
abstract fun SpecialCasingLine.mapping(): List<String>?
abstract fun UnicodeDataLine.mapping(): String
}
@@ -7,8 +7,28 @@ package generators.unicode.mappings.oneToMany.builders
import generators.unicode.SpecialCasingLine
import generators.unicode.UnicodeDataLine
import generators.unicode.hexToInt
internal class OneToManyTitlecaseMappingsBuilder(bmpUnicodeDataLines: List<UnicodeDataLine>) : OneToManyMappingsBuilder(bmpUnicodeDataLines) {
override fun SpecialCasingLine.mapping(): List<String> = titlecaseMapping
private fun lower(char: Int): Int {
return bmpUnicodeDataLines[char]?.lowercaseMapping?.takeIf { it.isNotEmpty() }?.hexToInt() ?: char
}
override fun SpecialCasingLine.mapping(): List<String>? {
val title = titlecaseMapping.map { it.hexToInt().toChar() }.joinToString(separator = "")
val upperFirst = uppercaseMapping[0].hexToInt().toChar()
// The version of Unicode in JVM executing this code is likely different than the version of the Unicode being parsed.
// Thus use the Unicode data being parsed to convert chars to lowercase.
val lowercasedTail = uppercaseMapping.drop(1).map { lower(it.hexToInt()).toChar() }.joinToString(separator = "")
// Skip titlecase mappings that can be derived by lowercasing tail characters
if (title == upperFirst + lowercasedTail) {
return null
}
return titlecaseMapping
}
override fun UnicodeDataLine.mapping(): String = titlecaseMapping
}
@@ -5,27 +5,31 @@
package generators.unicode.mappings.oneToMany.writers
import generators.unicode.ranges.RangesWritingStrategy
import generators.unicode.writeMappings
import generators.unicode.toHexCharLiteral
import java.io.FileWriter
internal class OneToManyTitlecaseMappingsWriter(private val strategy: RangesWritingStrategy) : OneToManyMappingsWriter {
internal class OneToManyTitlecaseMappingsWriter : OneToManyMappingsWriter {
override fun write(mappings: Map<Int, List<String>>, writer: FileWriter) {
strategy.beforeWritingRanges(writer)
writer.writeMappings(mappings, strategy)
strategy.afterWritingRanges(writer)
writer.appendLine()
writer.appendLine(titlecaseImpl())
// We have decided to ignore GREEK EXTENDED block due to their rare usage.
// It also leads to decreased js code size and simplified implementation.
val nonGreekExtended = mappings.filterNot { it.key in 0x1f00..0x1fff }
check(nonGreekExtended.size == 1)
val n = nonGreekExtended.keys.single()
check(n == 0x0149) // LATIN SMALL LETTER N PRECEDED BY APOSTROPHE
// one-to-many titlecase equivalent of the char is "\u02BC\u004E", the same as the uppercase equivalent
writer.appendLine(titlecaseImpl(n.toHexCharLiteral()))
}
private fun titlecaseImpl(): String = """
private fun titlecaseImpl(apostropheN: String): String = """
@OptIn(ExperimentalStdlibApi::class)
internal fun Char.titlecaseImpl(): String {
val code = this.toInt()
val index = binarySearchRange(keys, code)
if (index >= 0 && keys[index] == code) {
return values[index]
val uppercase = uppercase()
if (uppercase.length > 1) {
return if (this == $apostropheN) uppercase else uppercase[0] + uppercase.substring(1).lowercase()
}
return titlecaseCharImpl().toString()
return titlecaseChar().toString()
}
""".trimIndent()
}
@@ -22,15 +22,18 @@ internal class TitlecaseMappingsWriter : MappingsWriter {
writer.append(
"""
@OptIn(ExperimentalStdlibApi::class)
internal fun Char.titlecaseCharImpl(): Char {
val code = this.toInt()
// Letters repeating <Lu, Lt, Ll> sequence and code of the Lt is a multiple of 3, e.g. <DŽ, Dž, dž>
if (${rangeChecks(LuLtLlMappings, "code")}) {
return (3 * ((code + 1) / 3)).toChar()
}
// Lower case letters whose title case mapping equivalent is equal to the original letter
if (${rangeChecks(zeroMappings, "code")}) {
return this
}
return uppercaseCharImpl()
return uppercaseChar()
}
""".trimIndent()
)