Implement FSA-based timezone parser for DateTimeComponents.Format

DmitryNekrasov · web-flow · commit 209428921f37 · 2025-05-26T15:57:07.000+04:00
- Introduce a finite state automaton to parse timezone identifiers character by character. - Support named identifiers: UTC, GMT, UT, Z and lowercase z. - Parse fixed-offset formats in various styles: - Single‐digit offsets (+1, -5) - Full offsets with optional colons (+010203, -05:30:15) - Handle combined identifiers (e.g., UTC+01:00, GMT-05:30:45, UT+3). - Add comprehensive tests covering valid, invalid, edge‐case, and boundary scenarios. Fixes #444
diff --git a/core/common/src/format/DateTimeComponents.kt b/core/common/src/format/DateTimeComponents.kt
@@ -10,6 +10,8 @@ import kotlinx.datetime.DayOfWeek
 import kotlinx.datetime.internal.*
 import kotlinx.datetime.internal.format.*
 import kotlinx.datetime.internal.format.parser.Copyable
+import kotlinx.datetime.internal.format.parser.ParserStructure
+import kotlinx.datetime.internal.format.parser.TimeZoneParserOperation
 import kotlinx.datetime.internal.safeMultiply
 import kotlin.reflect.*
 
@@ -570,6 +572,18 @@ internal class TimeZoneIdDirective(private val knownZones: Set<String>) :
         get() =
             "${DateTimeFormatBuilder.WithDateTimeComponents::timeZoneId.name}()"
 
+    override fun parser(): ParserStructure<DateTimeComponentsContents> =
+        ParserStructure(
+            emptyList(),
+            listOf(
+                super.parser(),
+                ParserStructure(
+                    listOf(TimeZoneParserOperation(timeZoneField.accessor)),
+                    emptyList()
+                )
+            )
+        )
+
     override fun equals(other: Any?): Boolean = other is TimeZoneIdDirective && other.knownZones == knownZones
     override fun hashCode(): Int = knownZones.hashCode()
 }
diff --git a/core/common/src/internal/format/parser/ParserOperation.kt b/core/common/src/internal/format/parser/ParserOperation.kt
@@ -139,6 +139,118 @@ internal class UnconditionalModification<Output>(
     }
 }
 
+internal class TimeZoneParserOperation<Output>(
+    private val setter: AssignableField<Output, String>
+) : ParserOperation<Output> {
+
+    override fun consume(storage: Output, input: CharSequence, startIndex: Int): ParseResult {
+        val lastMatch = validateTimezone(input, startIndex)
+        return if (lastMatch > startIndex) {
+            setter.setWithoutReassigning(storage, input.substring(startIndex, lastMatch), startIndex, lastMatch)
+            ParseResult.Ok(lastMatch)
+        } else {
+            ParseResult.Error(startIndex) { "Invalid timezone format" }
+        }
+    }
+
+    companion object {
+        private enum class State {
+            START,
+            AFTER_PREFIX,
+            AFTER_SIGN,
+            AFTER_HOUR,
+            AFTER_MINUTE,
+            AFTER_COLON_MINUTE,
+            END,
+            INVALID
+        }
+
+        private fun validateTimezone(input: CharSequence, startIndex: Int): Int {
+            var index = startIndex
+            var lastValidIndex = startIndex
+
+            fun validatePrefix(validValues: List<String>): Boolean =
+                validValues.firstOrNull { input.startsWith(it) }?.let {
+                    index += it.length
+                    lastValidIndex = index
+                    true
+                } ?: false
+
+            fun validateTimeComponent(length: Int): Boolean {
+                if ((index..<(index + length)).all { input.getOrNull(it)?.isAsciiDigit() ?: false }) {
+                    index += length
+                    lastValidIndex = index
+                    return true
+                }
+                return false
+            }
+
+            var state = State.START
+            while (index < input.length) {
+                state = when (state) {
+                    State.START -> when {
+                        input[index] == 'Z' || input[index] == 'z' -> {
+                            index++
+                            State.END
+                        }
+
+                        input[index] in listOf('+', '-') -> {
+                            index++
+                            State.AFTER_SIGN
+                        }
+
+                        validatePrefix(listOf("UTC", "GMT", "UT")) -> State.AFTER_PREFIX
+                        else -> State.INVALID
+                    }
+
+                    State.AFTER_PREFIX -> when {
+                        input[index] in listOf('+', '-') -> {
+                            index++
+                            State.AFTER_SIGN
+                        }
+
+                        else -> State.INVALID
+                    }
+
+                    State.AFTER_SIGN -> when {
+                        validateTimeComponent(2) -> State.AFTER_HOUR
+                        validateTimeComponent(1) -> State.END
+                        else -> State.INVALID
+                    }
+
+                    State.AFTER_HOUR -> when {
+                        input[index] == ':' -> {
+                            index++
+                            if (validateTimeComponent(2)) State.AFTER_COLON_MINUTE else State.INVALID
+                        }
+
+                        validateTimeComponent(2) -> State.AFTER_MINUTE
+                        else -> State.INVALID
+                    }
+
+                    State.AFTER_MINUTE -> when {
+                        validateTimeComponent(2) -> State.END
+                        else -> State.INVALID
+                    }
+
+                    State.AFTER_COLON_MINUTE -> when {
+                        input[index] == ':' -> {
+                            index++
+                            if (validateTimeComponent(2)) State.END else State.INVALID
+                        }
+
+                        else -> State.INVALID
+                    }
+
+                    State.END, State.INVALID -> break
+                }
+            }
+
+            return if (state == State.END) index else lastValidIndex
+        }
+    }
+}
+
 /**
  * Matches the longest suitable string from `strings` and calls [consume] with the matched string.
  */
diff --git a/core/common/test/format/DateTimeComponentsFormatTest.kt b/core/common/test/format/DateTimeComponentsFormatTest.kt
@@ -1,5 +1,5 @@
 /*
- * Copyright 2019-2023 JetBrains s.r.o. and contributors.
+ * Copyright 2019-2025 JetBrains s.r.o. and contributors.
  * Use of this source code is governed by the Apache 2.0 License that can be found in the LICENSE.txt file.
  */
 
@@ -8,7 +8,6 @@ package kotlinx.datetime.test.format
 import kotlinx.datetime.*
 import kotlinx.datetime.format.*
 import kotlin.reflect.KMutableProperty1
-import kotlin.reflect.KProperty
 import kotlin.test.*
 
 class DateTimeComponentsFormatTest {
@@ -268,4 +267,138 @@ class DateTimeComponentsFormatTest {
             }
         }
     }
+
+    private object TimezoneTestData {
+        val correctParsableOffsets = listOf(
+            // Single digit hours (H format)
+            "1", "9", "0",
+            // Two-digit hours (HH format)
+            "09", "11", "18",
+            // Hours and minutes without a separator (HHMM format)
+            "0110", "0230", "0930",
+            // Hours, minutes, and seconds without a separator (HHMMSS format)
+            "010000", "000100", "012345",
+            // Hours and minutes with colon separator (HH:MM format)
+            "01:15", "02:35", "09:35",
+            // Hours, minutes, and seconds with colon separators (HH:MM:SS format)
+            "01:10:32", "15:51:00", "17:54:32"
+        )
+
+        val incorrectParsableOffsets = listOf(
+            // Invalid hours (exceeding typical timezone ranges)
+            "19", "99", "20",
+            // HHMM format with invalid minutes (>59) or hours (>18)
+            "2010", "0260", "0999", "9999",
+            // HHMMSS format with invalid hours, minutes, or seconds
+            "180001", "006000", "000099", "999999",
+            // HH:MM format with invalid hours or minutes
+            "30:10", "02:70", "99:99",
+            // HH:MM:SS format with invalid hours, minutes, or seconds
+            "19:00:00", "00:60:00", "99:99:99",
+        )
+
+        val incorrectUnparsableOffsets = listOf(
+            // Single non-digit characters
+            "a", "_", "+",
+            // Two characters: letter+digit, letter+symbol, digit+symbol
+            "a9", "y!", "1#",
+            // Three digits (invalid length - not 2 or 4 digits)
+            "110", "020",
+            // Five digits (invalid length - not 4 or 6 digits)
+            "18000", "02300",
+            // HH:MM format violations: single digit hour, missing minute, missing hour
+            "3:10", "2:70", "99:", ":20",
+            // Invalid colon-separated formats: too many digits in an hour/minute component
+            "12:3456", "1234:56",
+            // HH:MM:SS format violations: single digit hour, single digit minute, single digit second
+            "1:00:00", "00:6:00", "09:99:9",
+            // Colon placement errors
+            ":00:00", "00::00", "09:99:", "::00", "00::", "::",
+            // HH:MM:SS format violations: 3-digit hour, 3-digit minute, 3-digit second
+            "180:00:00", "00:610:00", "99:99:199"
+        )
+
+        val tzPrefixes = listOf("UTC", "GMT", "UT")
+
+        val timezoneDbIdentifiers = listOf(
+            "America/New_York", "Europe/London", "Asia/Tokyo", "Australia/Sydney",
+            "Pacific/Auckland", "Africa/Cairo", "America/Los_Angeles", "Europe/Paris",
+            "Asia/Singapore", "Australia/Melbourne", "Africa/Johannesburg", "Europe/Isle_of_Man"
+        )
+
+        val invalidTimezoneIds = listOf("INVALID", "XYZ", "ABC/DEF", "NOT_A_TIMEZONE", "SYSTEM")
+    }
+
+    @Test
+    fun testZuluTimeZone() {
+        // Replace it to:
+        // listOf("z", "Z").forEach(::assertParseableAsTimeZone)
+        // when TimeZone.of("z") works correctly
+        assertParseableAsTimeZone("Z")
+        assertIncorrectlyParseableAsTimeZone("z")
+    }
+
+    @Test
+    fun testSpecialNamedTimezones() {
+        TimezoneTestData.tzPrefixes.forEach(::assertParseableAsTimeZone)
+    }
+
+    @Test
+    fun testPrefixWithCorrectParsableOffset() {
+        val timezoneIds =
+            generateTimezoneIds(TimezoneTestData.tzPrefixes + "", TimezoneTestData.correctParsableOffsets)
+        timezoneIds.forEach(::assertParseableAsTimeZone)
+    }
+
+    @Test
+    fun testPrefixWithIncorrectParsableOffset() {
+        val timezoneIds =
+            generateTimezoneIds(TimezoneTestData.tzPrefixes + "", TimezoneTestData.incorrectParsableOffsets)
+        timezoneIds.forEach(::assertIncorrectlyParseableAsTimeZone)
+    }
+
+    @Test
+    fun testPrefixWithIncorrectUnparsableOffset() {
+        val timezoneIds =
+            generateTimezoneIds(TimezoneTestData.tzPrefixes + "", TimezoneTestData.incorrectUnparsableOffsets)
+        timezoneIds.forEach(::assertNonParseableAsTimeZone)
+    }
+
+    @Test
+    fun testTimezoneDBIdentifiers() {
+        TimezoneTestData.timezoneDbIdentifiers.forEach(::assertParseableAsTimeZone)
+    }
+
+    @Test
+    fun testInvalidTimezoneIds() {
+        TimezoneTestData.invalidTimezoneIds.forEach(::assertNonParseableAsTimeZone)
+    }
+
+    private fun generateTimezoneIds(prefixes: List<String>, offsets: List<String>): List<String> = buildList {
+        for (prefix in prefixes) {
+            for (sign in listOf('+', '-')) {
+                for (offset in offsets) {
+                    add("$prefix$sign$offset")
+                }
+            }
+        }
+    }
+
+    private fun assertParseableAsTimeZone(zoneId: String) {
+        TimeZone.of(zoneId)
+        val result = DateTimeComponents.Format { timeZoneId() }.parse(zoneId)
+        assertEquals(zoneId, result.timeZoneId)
+    }
+
+    private fun assertIncorrectlyParseableAsTimeZone(zoneId: String) {
+        assertFailsWith<IllegalTimeZoneException> { TimeZone.of(zoneId) }
+        val result = DateTimeComponents.Format { timeZoneId() }.parse(zoneId)
+        assertEquals(zoneId, result.timeZoneId)
+    }
+
+    private fun assertNonParseableAsTimeZone(zoneId: String) {
+        assertFailsWith<DateTimeFormatException> {
+            DateTimeComponents.Format { timeZoneId() }.parse(zoneId)
+        }
+    }
 }
