Make searching for timezone databases on Linux more robust

Using the official implementation of tzinfo as a reference.

Author: dkhalanskyjb

core/darwin/src/internal/TimeZoneNative.kt

@@ -12,7 +12,9 @@ import kotlinx.datetime.toKotlinInstant
 import kotlinx.datetime.internal.*
 import platform.Foundation.*
 
-internal actual val systemTzdb: TimeZoneDatabase get() = tzdbOnFilesystem
+internal actual val systemTzdb: TimeZoneDatabase = TzdbOnFilesystem(Path.fromString(defaultTzdbPath()))
+
+internal expect fun defaultTzdbPath(): String
 
 internal actual fun currentSystemDefaultZone(): Pair<String, TimeZoneRules?> {
     /* The framework has its own cache of the system timezone. Calls to
@@ -69,7 +71,3 @@ internal actual fun currentSystemDefaultZone(): Pair<String, TimeZoneRules?> {
 }
 
 internal actual fun currentTime(): Instant = NSDate.date().toKotlinInstant()
-
-private val tzdbOnFilesystem = TzdbOnFilesystem(Path.fromString(defaultTzdbPath()))
-
-internal expect fun defaultTzdbPath(): String

core/linux/src/internal/TimeZoneNative.kt

@@ -4,16 +4,17 @@
  */
 
 @file:OptIn(ExperimentalForeignApi::class)
+
 package kotlinx.datetime.internal
 
 import kotlinx.cinterop.*
 import kotlinx.datetime.Instant
 import platform.posix.*
 
-internal actual val systemTzdb: TimeZoneDatabase get() = tzdbOnFilesystem
+internal actual val systemTzdb: TimeZoneDatabase = TzdbOnFilesystem()
 
 internal actual fun currentSystemDefaultZone(): Pair<String, TimeZoneRules?> {
-    val zoneId = tzdbOnFilesystem.currentSystemDefault()?.second?.toString()
+    val zoneId = pathToSystemDefault()?.second?.toString()
         ?: throw IllegalStateException("Failed to get the system timezone")
     return zoneId to null
 }
@@ -35,5 +36,3 @@ internal actual fun currentTime(): Instant = memScoped {
         throw IllegalStateException("The readings from the system clock are not representable as an Instant")
     }
 }
-
-private val tzdbOnFilesystem: TzdbOnFilesystem = TzdbOnFilesystem(Path.fromString("/usr/share/zoneinfo"))

core/nix/src/internal/TzdbOnFilesystem.kt

@@ -5,7 +5,11 @@
 
 package kotlinx.datetime.internal
 
-internal class TzdbOnFilesystem(defaultTzdbPath: Path): TimeZoneDatabase {
+internal class TzdbOnFilesystem(defaultTzdbPath: Path? = null): TimeZoneDatabase {
+
+    private val tzdbPath = tzdbPaths(defaultTzdbPath).find {
+        it.chaseSymlinks().check()?.isDirectory == true
+    } ?: throw IllegalStateException("Could not find the path to the timezone database")
 
     override fun rulesForId(id: String): TimeZoneRules =
         readTzFile(tzdbPath.resolve(Path.fromString(id)).readBytes()).toTimeZoneRules()
@@ -14,31 +18,44 @@ internal class TzdbOnFilesystem(defaultTzdbPath: Path): TimeZoneDatabase {
         tzdbPath.traverseDirectory(exclude = tzdbUnneededFiles) { add(it.toString()) }
     }
 
-    internal fun currentSystemDefault(): Pair<Path, Path>? {
-        val info = Path(true, listOf("etc", "localtime")).readLink() ?: return null
-        val i = info.components.indexOf("zoneinfo")
-        if (!info.isAbsolute || i == -1 || i == info.components.size - 1) return null
-        return Pair(
-            Path(true, info.components.subList(0, i + 1)),
-            Path(false, info.components.subList(i + 1, info.components.size))
-        )
-    }
-
-    private val tzdbPath = defaultTzdbPath.check()?.let { defaultTzdbPath }
-        ?: currentSystemDefault()?.first ?: throw IllegalStateException("Could not find the path to the timezone database")
-
 }
 
+/** The files that sometimes lie in the `zoneinfo` directory but aren't actually time zones. */
 private val tzdbUnneededFiles = setOf(
+    // taken from https://github.com/tzinfo/tzinfo/blob/9953fc092424d55deaea2dcdf6279943f3495724/lib/tzinfo/data_sources/zoneinfo_data_source.rb#L88C29-L97C21
+    "+VERSION",
+    "leapseconds",
+    "localtime",
     "posix",
     "posixrules",
+    "right",
+    "SECURITY",
+    "src",
+    "timeconfig",
+    // taken from https://github.com/HowardHinnant/date/blob/ab37c362e35267d6dee02cb47760f9e9c669d3be/src/tz.cpp#L2863-L2874
     "Factory",
     "iso3166.tab",
-    "right",
-    "+VERSION",
     "zone.tab",
     "zone1970.tab",
     "tzdata.zi",
-    "leapseconds",
     "leap-seconds.list"
 )
+
+/** The directories checked for a valid timezone database. */
+internal fun tzdbPaths(defaultTzdbPath: Path?) = sequence {
+    defaultTzdbPath?.let { yield(it) }
+    // taken from https://github.com/tzinfo/tzinfo/blob/9953fc092424d55deaea2dcdf6279943f3495724/lib/tzinfo/data_sources/zoneinfo_data_source.rb#L70
+    yieldAll(listOf("/usr/share/zoneinfo", "/usr/share/lib/zoneinfo", "/etc/zoneinfo").map { Path.fromString(it) })
+    pathToSystemDefault()?.first?.let { yield(it) }
+}
+
+// taken from https://github.com/HowardHinnant/date/blob/ab37c362e35267d6dee02cb47760f9e9c669d3be/src/tz.cpp#L3951-L3952
+internal fun pathToSystemDefault(): Pair<Path, Path>? {
+    val info = Path(true, listOf("etc", "localtime")).chaseSymlinks()
+    val i = info.components.indexOf("zoneinfo")
+    if (!info.isAbsolute || i == -1 || i == info.components.size - 1) return null
+    return Pair(
+        Path(true, info.components.subList(0, i + 1)),
+        Path(false, info.components.subList(i + 1, info.components.size))
+    )
+}

core/nix/src/internal/filesystem.kt

@@ -53,6 +53,16 @@ internal class Path(val isAbsolute: Boolean, val components: List<String>) {
     }
 }
 
+internal fun Path.chaseSymlinks(maxDepth: Int = 100): Path {
+    var realPath = this
+    var depth = maxDepth
+    while (true) {
+        realPath = realPath.readLink() ?: break
+        if (depth-- == 0) throw RuntimeException("Too many levels of symbolic links")
+    }
+    return realPath
+}
+
 // `stat(2)` lists the other available fields
 internal interface PathInfo {
     val isDirectory: Boolean