|
| 1 | +import kotlinx.serialization.json.Json |
| 2 | +import kotlinx.serialization.json.jsonArray |
| 3 | +import kotlinx.serialization.json.jsonObject |
| 4 | +import kotlinx.serialization.json.jsonPrimitive |
| 5 | +import org.gradle.api.DefaultTask |
| 6 | +import org.gradle.api.file.DirectoryProperty |
| 7 | +import org.gradle.api.provider.Property |
| 8 | +import org.gradle.api.tasks.Input |
| 9 | +import org.gradle.api.tasks.Internal |
| 10 | +import org.gradle.api.tasks.OutputDirectory |
| 11 | +import org.gradle.api.tasks.TaskAction |
| 12 | +import java.io.File |
| 13 | +import java.net.URI |
| 14 | +import java.nio.file.Files |
| 15 | +import java.nio.file.StandardCopyOption |
| 16 | + |
| 17 | +abstract class GenerateEmojiList : DefaultTask() { |
| 18 | + |
| 19 | + @get:Input |
| 20 | + abstract val emojiUrl: Property<String> |
| 21 | + |
| 22 | + @get:Input |
| 23 | + abstract val emojiKeywordsUrl: Property<String> |
| 24 | + |
| 25 | + @get:Internal |
| 26 | + abstract val emojiCacheDir: DirectoryProperty |
| 27 | + |
| 28 | + @get:OutputDirectory |
| 29 | + abstract val outputDir: DirectoryProperty |
| 30 | + |
| 31 | + init { |
| 32 | + description = "Fetches Unicode emoji list and generates categorized Kotlin source file if needed" |
| 33 | + group = "emoji" |
| 34 | + |
| 35 | + onlyIf { |
| 36 | + !outputDir.get().asFile.resolve("Emojis.kt").exists() |
| 37 | + } |
| 38 | + } |
| 39 | + |
| 40 | + @TaskAction |
| 41 | + fun generate() { |
| 42 | + val outDir = outputDir.get().asFile |
| 43 | + val cacheDir = emojiCacheDir.get().asFile |
| 44 | + val emojiFile = File(cacheDir, "emoji-test.txt") |
| 45 | + val keywordsFile = File(cacheDir, "en-keywords.json") |
| 46 | + |
| 47 | + outDir.mkdirs() |
| 48 | + if (!emojiFile.exists()) { |
| 49 | + logger.lifecycle("Downloading emoji-test.txt") |
| 50 | + URI(emojiUrl.get()).toURL().openStream().use { input -> |
| 51 | + Files.copy(input, emojiFile.toPath(), StandardCopyOption.REPLACE_EXISTING) |
| 52 | + } |
| 53 | + } |
| 54 | + |
| 55 | + if (!keywordsFile.exists()) { |
| 56 | + logger.lifecycle("Downloading CLDR annotations") |
| 57 | + URI(emojiKeywordsUrl.get()).toURL().openStream().use { input -> |
| 58 | + Files.copy(input, keywordsFile.toPath(), StandardCopyOption.REPLACE_EXISTING) |
| 59 | + } |
| 60 | + } |
| 61 | + |
| 62 | + val json = Json { ignoreUnknownKeys = true } |
| 63 | + val cldrData = json.parseToJsonElement(keywordsFile.readText()).jsonObject |
| 64 | + val cldrAnnotations = cldrData["annotations"]?.jsonObject?.get("annotations")?.jsonObject |
| 65 | + |
| 66 | + val emojiText = emojiFile.readText() |
| 67 | + val emojiCategories = |
| 68 | + mutableMapOf<String, MutableMap<String, MutableList<MutableMap<String, Any>>>>() |
| 69 | + val emojiCategoriesNoSkinTones = |
| 70 | + mutableMapOf<String, MutableMap<String, MutableList<MutableMap<String, Any>>>>() |
| 71 | + var currentGroup = "Uncategorized" |
| 72 | + var currentSubgroup = "Uncategorized" |
| 73 | + |
| 74 | + emojiText.lines().forEach { line -> |
| 75 | + when { |
| 76 | + line.startsWith("# group:") -> { |
| 77 | + val groupName = line.removePrefix("# group:").trim() |
| 78 | + currentGroup = when (groupName) { |
| 79 | + "Smileys & Emotion", "People & Body" -> "Smileys & People" |
| 80 | + else -> groupName |
| 81 | + } |
| 82 | + emojiCategories.getOrPut(currentGroup) { mutableMapOf() } |
| 83 | + emojiCategoriesNoSkinTones.getOrPut(currentGroup) { mutableMapOf() } |
| 84 | + } |
| 85 | + |
| 86 | + line.startsWith("# subgroup:") -> { |
| 87 | + currentSubgroup = line.removePrefix("# subgroup:").trim() |
| 88 | + emojiCategories[currentGroup]?.getOrPut(currentSubgroup) { mutableListOf() } |
| 89 | + emojiCategoriesNoSkinTones[currentGroup]?.getOrPut(currentSubgroup) { mutableListOf() } |
| 90 | + } |
| 91 | + |
| 92 | + line.isNotBlank() && !line.startsWith("#") -> { |
| 93 | + val parts = line.split(";").map { it.trim() } |
| 94 | + if (parts.size > 1 && parts[1].contains("fully-qualified")) { |
| 95 | + val codePoints = parts[0].split(" ").map { it.toInt(16) } |
| 96 | + val unicode = codePoints.map { codePoint -> |
| 97 | + if (codePoint <= 0xFFFF) { |
| 98 | + codePoint.toChar().toString() |
| 99 | + } else { |
| 100 | + String(Character.toChars(codePoint)) |
| 101 | + } |
| 102 | + }.joinToString("") |
| 103 | + val nameParts = line.split("#")[1].trim().split(" ") |
| 104 | + val name = nameParts.drop(2).joinToString(" ") |
| 105 | + val baseKeywords = name.split(" ").filter { it.isNotBlank() } |
| 106 | + |
| 107 | + val cldrAnnotation = cldrAnnotations?.get(unicode)?.jsonObject |
| 108 | + val cldrKeywords = cldrAnnotation?.get("default")?.jsonArray |
| 109 | + ?.mapNotNull { it.jsonPrimitive.toString().removeSurrounding("\"") } |
| 110 | + .orEmpty() |
| 111 | + val allKeywords = (baseKeywords + cldrKeywords).distinct() |
| 112 | + |
| 113 | + val emojiEntry = mutableMapOf( |
| 114 | + "unicode" to unicode, |
| 115 | + "name" to name, |
| 116 | + "keywords" to allKeywords |
| 117 | + ) |
| 118 | + emojiCategories[currentGroup]?.get(currentSubgroup)?.add(emojiEntry) |
| 119 | + |
| 120 | + val hasSkinTone = codePoints.any { it in 0x1F3FB..0x1F3FF } |
| 121 | + if (!hasSkinTone) { |
| 122 | + emojiCategoriesNoSkinTones[currentGroup]?.get(currentSubgroup) |
| 123 | + ?.add(emojiEntry) |
| 124 | + } |
| 125 | + } |
| 126 | + } |
| 127 | + } |
| 128 | + } |
| 129 | + |
| 130 | + val nonEmptyCategories = emojiCategories.filter { (_, subgroups) -> |
| 131 | + subgroups.any { (_, emojis) -> emojis.isNotEmpty() } |
| 132 | + } |
| 133 | + val nonEmptyCategoriesNoSkinTones = |
| 134 | + emojiCategoriesNoSkinTones.filter { (_, subgroups) -> |
| 135 | + subgroups.any { (_, emojis) -> emojis.isNotEmpty() } |
| 136 | + } |
| 137 | + |
| 138 | + val mainFile = File(outDir, "Emojis.kt") |
| 139 | + val mainCode = buildString { |
| 140 | + appendLine("// Generated file - Do not edit manually") |
| 141 | + appendLine("package com.getcode.libs.emojis.generated") |
| 142 | + appendLine() |
| 143 | + appendLine("data class Emoji(") |
| 144 | + appendLine(" val unicode: String,") |
| 145 | + appendLine(" val name: String,") |
| 146 | + appendLine(" val keywords: List<String>") |
| 147 | + appendLine(")") |
| 148 | + appendLine() |
| 149 | + appendLine("enum class Category(val displayName: String) {") |
| 150 | + nonEmptyCategories.keys.forEach { group -> |
| 151 | + val enumName = group.replace("[^A-Za-z0-9]".toRegex(), "").uppercase() |
| 152 | + appendLine(" $enumName(\"$group\"),") |
| 153 | + } |
| 154 | + appendLine(" FREQUENT(\"Frequently Used\"),") |
| 155 | + appendLine("}") |
| 156 | + appendLine() |
| 157 | + appendLine("object Emojis {") |
| 158 | + appendLine(" val categorized = mapOf(") |
| 159 | + nonEmptyCategories.forEach { (group, subgroups) -> |
| 160 | + val enumName = group.replace("[^A-Za-z0-9]".toRegex(), "").uppercase() |
| 161 | + appendLine(" Category.$enumName to mapOf(") |
| 162 | + subgroups.forEach { (subgroup, _) -> |
| 163 | + val safeGroupName = group.replace("[^A-Za-z0-9]".toRegex(), "") |
| 164 | + val safeSubgroupName = subgroup.replace("[^A-Za-z0-9]".toRegex(), "") |
| 165 | + appendLine(" \"$subgroup\" to ${safeGroupName}${safeSubgroupName}Emojis.categorized,") |
| 166 | + } |
| 167 | + appendLine(" ),") |
| 168 | + } |
| 169 | + appendLine(" )") |
| 170 | + appendLine() |
| 171 | + appendLine(" val categorizedNoSkinTones = mapOf(") |
| 172 | + nonEmptyCategoriesNoSkinTones.forEach { (group, subgroups) -> |
| 173 | + val enumName = group.replace("[^A-Za-z0-9]".toRegex(), "").uppercase() |
| 174 | + appendLine(" Category.$enumName to mapOf(") |
| 175 | + subgroups.forEach { (subgroup, _) -> |
| 176 | + val safeGroupName = group.replace("[^A-Za-z0-9]".toRegex(), "") |
| 177 | + val safeSubgroupName = subgroup.replace("[^A-Za-z0-9]".toRegex(), "") |
| 178 | + appendLine(" \"$subgroup\" to ${safeGroupName}${safeSubgroupName}Emojis.categorizedNoSkinTones,") |
| 179 | + } |
| 180 | + appendLine(" ),") |
| 181 | + } |
| 182 | + appendLine(" )") |
| 183 | + appendLine("}") |
| 184 | + }.trimIndent() |
| 185 | + mainFile.writeText(mainCode) |
| 186 | + |
| 187 | + emojiCategories.forEach { (group, subgroups) -> |
| 188 | + subgroups.forEach { (subgroup, emojis) -> |
| 189 | + val safeGroupName = group.replace("[^A-Za-z0-9]".toRegex(), "") |
| 190 | + val safeSubgroupName = subgroup.replace("[^A-Za-z0-9]".toRegex(), "") |
| 191 | + val subgroupFile = |
| 192 | + File(outDir, "${safeGroupName}${safeSubgroupName}Emojis.kt") |
| 193 | + @Suppress("UNCHECKED_CAST") |
| 194 | + val subgroupCode = buildString { |
| 195 | + appendLine("// Generated file - Do not edit manually") |
| 196 | + appendLine("package com.getcode.libs.emojis.generated") |
| 197 | + appendLine() |
| 198 | + appendLine("object ${safeGroupName}${safeSubgroupName}Emojis {") |
| 199 | + appendLine(" val categorized = ${if (emojis.isEmpty()) "emptyList<Emoji>()" else "listOf("}") |
| 200 | + if (emojis.isNotEmpty()) { |
| 201 | + appendLine(" ${emojis.joinToString(",\n ") { "Emoji(\"${it["unicode"]}\", \"${it["name"]}\", listOf(${(it["keywords"] as List<String>).joinToString { "\"$it\"" }}))" }}") |
| 202 | + appendLine(" )") |
| 203 | + } |
| 204 | + appendLine() |
| 205 | + appendLine( |
| 206 | + " val categorizedNoSkinTones = ${ |
| 207 | + if (emojiCategoriesNoSkinTones[group]?.get( |
| 208 | + subgroup |
| 209 | + )?.isEmpty() != false |
| 210 | + ) "emptyList<Emoji>()" else "listOf(" |
| 211 | + }" |
| 212 | + ) |
| 213 | + val noSkinTones = |
| 214 | + emojiCategoriesNoSkinTones[group]?.get(subgroup) ?: emptyList() |
| 215 | + if (noSkinTones.isNotEmpty()) { |
| 216 | + appendLine(" ${noSkinTones.joinToString(",\n ") { "Emoji(\"${it["unicode"]}\", \"${it["name"]}\", listOf(${(it["keywords"] as List<String>).joinToString { "\"$it\"" }}))" }}") |
| 217 | + appendLine(" )") |
| 218 | + } |
| 219 | + appendLine("}") |
| 220 | + }.trimIndent() |
| 221 | + subgroupFile.writeText(subgroupCode) |
| 222 | + } |
| 223 | + } |
| 224 | + val totalEmojis = emojiCategories.values.sumOf { it.values.sumOf { e -> e.size } } |
| 225 | + logger.lifecycle("Generated $totalEmojis emojis across ${emojiCategories.size} categories") |
| 226 | + } |
| 227 | +} |
0 commit comments