Continue work on Swift port of Parser.
This commit is contained in:
Brent Simmons
parent
61825a6d88
commit
f86a5f8314
@ -49,6 +49,20 @@
|
||||
ReferencedContainer = "container:">
|
||||
</BuildableReference>
|
||||
</BuildActionEntry>
|
||||
<BuildActionEntry
|
||||
buildForTesting = "YES"
|
||||
buildForRunning = "YES"
|
||||
buildForProfiling = "YES"
|
||||
buildForArchiving = "YES"
|
||||
buildForAnalyzing = "YES">
|
||||
<BuildableReference
|
||||
BuildableIdentifier = "primary"
|
||||
BlueprintIdentifier = "FeedParser"
|
||||
BuildableName = "FeedParser"
|
||||
BlueprintName = "FeedParser"
|
||||
ReferencedContainer = "container:">
|
||||
</BuildableReference>
|
||||
</BuildActionEntry>
|
||||
</BuildActionEntries>
|
||||
</BuildAction>
|
||||
<TestAction
|
||||
@ -88,6 +102,16 @@
|
||||
ReferencedContainer = "container:">
|
||||
</BuildableReference>
|
||||
</TestableReference>
|
||||
<TestableReference
|
||||
skipped = "NO">
|
||||
<BuildableReference
|
||||
BuildableIdentifier = "primary"
|
||||
BlueprintIdentifier = "DateParserTests"
|
||||
BuildableName = "DateParserTests"
|
||||
BlueprintName = "DateParserTests"
|
||||
ReferencedContainer = "container:">
|
||||
</BuildableReference>
|
||||
</TestableReference>
|
||||
</Testables>
|
||||
</TestAction>
|
||||
<LaunchAction
|
||||
|
||||
@ -19,10 +19,14 @@ let package = Package(
|
||||
.library(
|
||||
name: "OPMLParser",
|
||||
type: .dynamic,
|
||||
targets: ["OPMLParser"])
|
||||
|
||||
targets: ["OPMLParser"]),
|
||||
.library(
|
||||
name: "DateParser",
|
||||
type: .dynamic,
|
||||
targets: ["DateParser"])
|
||||
],
|
||||
dependencies: [
|
||||
.package(path: "../FoundationExtras"),
|
||||
],
|
||||
targets: [
|
||||
// Targets are the basic building blocks of a package. A target can define a module or a test suite.
|
||||
@ -38,7 +42,9 @@ let package = Package(
|
||||
.target(
|
||||
name: "FeedParser",
|
||||
dependencies: [
|
||||
"SAX"
|
||||
"SAX",
|
||||
"FoundationExtras",
|
||||
"DateParser"
|
||||
],
|
||||
swiftSettings: [
|
||||
.enableExperimentalFeature("StrictConcurrency")
|
||||
@ -49,6 +55,12 @@ let package = Package(
|
||||
swiftSettings: [
|
||||
.enableExperimentalFeature("StrictConcurrency")
|
||||
]),
|
||||
.target(
|
||||
name: "DateParser",
|
||||
dependencies: [],
|
||||
swiftSettings: [
|
||||
.enableExperimentalFeature("StrictConcurrency")
|
||||
]),
|
||||
.testTarget(
|
||||
name: "FeedParserTests",
|
||||
dependencies: ["FeedParser"],
|
||||
@ -58,6 +70,9 @@ let package = Package(
|
||||
name: "OPMLParserTests",
|
||||
dependencies: ["OPMLParser"],
|
||||
resources: [.copy("Resources")]),
|
||||
.testTarget(
|
||||
name: "DateParserTests",
|
||||
dependencies: ["DateParser"])
|
||||
]
|
||||
)
|
||||
|
||||
|
||||
516
Modules/Parser/Sources/DateParser/DateParser.swift
Normal file
516
Modules/Parser/Sources/DateParser/DateParser.swift
Normal file
@ -0,0 +1,516 @@
|
||||
//
|
||||
// DateParser.swift
|
||||
//
|
||||
//
|
||||
// Created by Brent Simmons on 8/28/24.
|
||||
//
|
||||
|
||||
import Foundation
|
||||
|
||||
public final class DateParser {
|
||||
|
||||
typealias DateBuffer = UnsafeBufferPointer<UInt8>
|
||||
|
||||
// MARK: - Public API
|
||||
|
||||
/// Parse W3C and pubDate dates — used for feed parsing.
|
||||
/// This is a fast alternative to system APIs
|
||||
/// for parsing dates.
|
||||
public static func date(data: Data) -> Date? {
|
||||
|
||||
let numberOfBytes = data.count
|
||||
|
||||
// Make sure it’s in reasonable range for a date string.
|
||||
if numberOfBytes < 6 || numberOfBytes > 150 {
|
||||
return nil
|
||||
}
|
||||
|
||||
return data.withUnsafeBytes { bytes in
|
||||
let buffer = bytes.bindMemory(to: UInt8.self)
|
||||
|
||||
if dateIsW3CDate(buffer, numberOfBytes) {
|
||||
return parseW3CDate(buffer, numberOfBytes)
|
||||
}
|
||||
else if dateIsPubDate(buffer, numberOfBytes) {
|
||||
return parsePubDate(buffer, numberOfBytes)
|
||||
}
|
||||
|
||||
// Fallback, in case our detection fails.
|
||||
return parseW3CDate(buffer, numberOfBytes)
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// MARK: - Private
|
||||
|
||||
private extension DateParser {
|
||||
|
||||
struct DateCharacter {
|
||||
|
||||
static let space = Character(" ").asciiValue
|
||||
static let `return` = Character("\r").asciiValue
|
||||
static let newline = Character("\n").asciiValue
|
||||
static let tab = Character("\t").asciiValue
|
||||
static let hyphen = Character("-").asciiValue
|
||||
static let comma = Character(",").asciiValue
|
||||
static let dot = Character(".").asciiValue
|
||||
static let colon = Character(":").asciiValue
|
||||
static let plus = Character("+").asciiValue
|
||||
static let minus = Character("-").asciiValue
|
||||
static let Z = Character("Z").asciiValue
|
||||
static let z = Character("z").asciiValue
|
||||
static let F = Character("F").asciiValue
|
||||
static let f = Character("f").asciiValue
|
||||
static let S = Character("S").asciiValue
|
||||
static let s = Character("s").asciiValue
|
||||
static let O = Character("O").asciiValue
|
||||
static let o = Character("o").asciiValue
|
||||
static let N = Character("N").asciiValue
|
||||
static let n = Character("n").asciiValue
|
||||
static let D = Character("D").asciiValue
|
||||
static let d = Character("d").asciiValue
|
||||
}
|
||||
|
||||
enum Month: Int {
|
||||
|
||||
January = 1,
|
||||
February,
|
||||
March,
|
||||
April,
|
||||
May,
|
||||
June,
|
||||
July,
|
||||
August,
|
||||
September,
|
||||
October,
|
||||
November,
|
||||
December
|
||||
}
|
||||
|
||||
// MARK: - Standard Formats
|
||||
|
||||
static func dateIsW3CDate(_ bytes: DateBuffer, numberOfBytes: Int) -> Bool {
|
||||
|
||||
// Something like 2010-11-17T08:40:07-05:00
|
||||
// But might be missing T character in the middle.
|
||||
// Looks for four digits in a row followed by a -.
|
||||
|
||||
for i in 0..<numberOfBytes - 4 {
|
||||
|
||||
let ch = bytes[i]
|
||||
// Skip whitespace.
|
||||
if ch == DateCharacter.space || ch == DateCharacter.`return` || ch == DateCharacter.newline || ch == DateCharacter.tab {
|
||||
continue
|
||||
}
|
||||
|
||||
assert(i + 4 < numberOfBytes)
|
||||
// First non-whitespace character must be the beginning of the year, as in `2010-`
|
||||
return isdigit(ch) && isdigit(bytes[i + 1]) && isdigit(bytes[i + 2]) && isdigit(bytes[i + 3]) && bytes[i + 4] == DateCharacter.hyphen
|
||||
}
|
||||
|
||||
return false
|
||||
}
|
||||
|
||||
static func dateIsPubDate(_ bytes: DateBuffer, numberOfBytes: Int) -> Bool {
|
||||
|
||||
for ch in bytes {
|
||||
if ch == DateCharacter.space || ch == DateCharacter.comma {
|
||||
return true
|
||||
}
|
||||
}
|
||||
|
||||
return false
|
||||
}
|
||||
|
||||
static func parseW3CDate(_ bytes: DateBuffer, numberOfBytes: Int) -> Date {
|
||||
|
||||
/*@"yyyy'-'MM'-'dd'T'HH':'mm':'ss"
|
||||
@"yyyy-MM-dd'T'HH:mm:sszzz"
|
||||
@"yyyy-MM-dd'T'HH:mm:ss'.'SSSzzz"
|
||||
etc.*/
|
||||
|
||||
var finalIndex = 0
|
||||
|
||||
let year = nextNumericValue(bytes, numberOfBytes, 0, 4, &finalIndex)
|
||||
let month = nextNumericValue(bytes, numberOfBytes, finalIndex + 1, 2, &finalIndex)
|
||||
let day = nextNumericValue(bytes, numberOfBytes, finalIndex + 1, 2, &finalIndex)
|
||||
let hour = nextNumericValue(bytes, numberOfBytes, finalIndex + 1, 2, &finalIndex)
|
||||
let minute = nextNumericValue(bytes, numberOfBytes, finalIndex + 1, 2, &finalIndex)
|
||||
let second = nextNumericValue(bytes, numberOfBytes, finalIndex + 1, 2, &finalIndex)
|
||||
|
||||
let currentIndex = finalIndex + 1
|
||||
|
||||
let milliseconds = {
|
||||
var ms = 0
|
||||
let hasMilliseconds = (currentIndex < numberOfBytes) && (bytes[currentIndex] == DateCharacter.dot)
|
||||
if hasMilliseconds {
|
||||
ms = nextNumericValue(bytes, numberOfBytes, currentIndex, 3, &finalIndex)
|
||||
currentIndex = finalIndex + 1
|
||||
}
|
||||
return ms
|
||||
}()
|
||||
|
||||
let timeZoneOffset = parsedtimeZoneOffset(bytes, numberOfBytes, currentIndex)
|
||||
|
||||
return dateWithYearMonthDayHourMinuteSecondAndtimeZoneOffset(year, month, day, hour, minute, second, milliseconds, timeZoneOffset)
|
||||
}
|
||||
|
||||
static func parsePubDate(_ bytes: DateBuffer, numberOfBytes: Int) -> Date {
|
||||
|
||||
var finalIndex = 0
|
||||
|
||||
let day = nextNumericValue(bytes, numberOfBytes, 0, 2, &finalIndex) ?? 1
|
||||
let month = nextMonthValue(bytes, numberOfBytes, finalIndex + 1, &finalIndex)
|
||||
let year = nextNumericValue(bytes, numberOfBytes, finalIndex + 1, 4, &finalIndex)
|
||||
let hour = nextNumericValue(bytes, numberOfBytes, finalIndex + 1, 2, &finalIndex) ?? 0
|
||||
let minute = nextNumericValue(bytes, numberOfBytes, finalIndex + 1, 2, &finalIndex) ?? 0
|
||||
|
||||
var currentIndex = finalIndex + 1
|
||||
|
||||
let second = {
|
||||
var s = 0
|
||||
let hasSeconds = (currentIndex < numberOfBytes) && (bytes[currentIndex] == DateCharacter.colon)
|
||||
if hasSeconds {
|
||||
s = nextNumericValue(bytes, numberOfBytes, currentIndex, 2, &finalIndex)
|
||||
}
|
||||
return s
|
||||
}()
|
||||
|
||||
currentIndex = finalIndex + 1
|
||||
|
||||
let timeZoneOffset = {
|
||||
var offset = 0
|
||||
let hasTimeZone = (currentIndex < numberOfBytes) && (bytes[currentIndex] == DateCharacter.space)
|
||||
if hasTimeZone {
|
||||
offset = parsedtimeZoneOffset(bytes, numberOfBytes, currentIndex)
|
||||
}
|
||||
return offset
|
||||
}()
|
||||
|
||||
return dateWithYearMonthDayHourMinuteSecondAndtimeZoneOffset(year, month, day, hour, minute, second, 0, timeZoneOffset)
|
||||
}
|
||||
|
||||
// MARK: - Date Creation
|
||||
|
||||
static func dateWithYearMonthDayHourMinuteSecondAndtimeZoneOffset(_ year: Int, _ month: Int, _ day: Int, _ hour: Int, _ minute: Int, _ second: Int, _ milliseconds: Int, _ timeZoneOffset: Int) {
|
||||
|
||||
var timeInfo = tm()
|
||||
timeInfo.tm_sec = CInt(second)
|
||||
timeInfo.tm_min = CInt(minute)
|
||||
timeInfo.tm_hour = CInt(hour)
|
||||
timeInfo.tm_mday = CInt(day)
|
||||
timeInfo.tm_mon = CInt(month - 1) //It's 1-based coming in
|
||||
timeInfo.tm_year = CInt(year - 1900) //see time.h -- it's years since 1900
|
||||
timeInfo.tm_wday = -1
|
||||
timeInfo.tm_yday = -1
|
||||
timeInfo.tm_isdst = -1
|
||||
timeInfo.tm_gmtoff = timeZoneOffset;
|
||||
timeInfo.tm_zone = nil;
|
||||
|
||||
var rawTime = timegm(&timeInfo)
|
||||
if rawTime == time_t(UInt.max) {
|
||||
|
||||
// NSCalendar is super-amazingly slow (which is partly why this parser exists),
|
||||
// so this is used only when the date is far enough in the future
|
||||
// (19 January 2038 03:14:08Z on 32-bit systems) that timegm fails.
|
||||
// Hopefully by the time we consistently need dates that far in the future
|
||||
// the performance of NSCalendar won’t be an issue.
|
||||
|
||||
var dateComponents = DateComponents()
|
||||
|
||||
dateComponents.timeZone = TimeZone(forSecondsFromGMT: timeZoneOffset)
|
||||
dateComponents.year = year
|
||||
dateComponents.month = month
|
||||
dateComponents.day = day
|
||||
dateComponents.hour = hour
|
||||
dateComponents.minute = minute
|
||||
dateComponents.second = second + (milliseconds / 1000)
|
||||
|
||||
return Calendar.autoupdatingCurrent.date(from: dateComponents)
|
||||
}
|
||||
|
||||
if milliseconds > 0 {
|
||||
rawTime += Float(milliseconds) / 1000.0
|
||||
}
|
||||
|
||||
return Date(timeIntervalSince1970: rawTime)
|
||||
}
|
||||
|
||||
// MARK: - Time Zones and Offsets
|
||||
|
||||
static let kGMT = "GMT".utf8CString
|
||||
static let kUTC = "UTC".utf8CString
|
||||
|
||||
static func parsedTimeZoneOffset(_ bytes: DateBuffer, _ numberOfBytes: Int, _ startingIndex: Int) -> Int {
|
||||
|
||||
var timeZoneCharacters: [CChar] = [0, 0, 0, 0, 0, 0] // nil-terminated last character
|
||||
var numberOfCharactersFound = 0
|
||||
var hasAtLeastOneAlphaCharacter = false
|
||||
|
||||
for i in startingIndex..<numberOfBytes {
|
||||
let ch = bytes[i]
|
||||
if ch == DateCharacter.colon || ch == DateCharacter.space {
|
||||
continue
|
||||
}
|
||||
let isAlphaCharacter = isalpha(ch)
|
||||
if isAlphaCharacter {
|
||||
hasAtLeastOneAlphaCharacter = true
|
||||
}
|
||||
if isAlphaCharacter || isdigit(ch) || ch == DateCharacter.plus || ch == DateCharacter.minus {
|
||||
numberOfCharactersFound += 1
|
||||
timeZoneCharacters[numberOfCharactersFound - 1] = ch
|
||||
}
|
||||
if numberOfCharactersFound >= 5 {
|
||||
break
|
||||
}
|
||||
}
|
||||
|
||||
if numberOfCharactersFound < 1 || timeZoneCharacters[0] == DateCharacter.Z || timeZoneCharacters[0] == DateCharacter.z {
|
||||
return 0
|
||||
}
|
||||
if strcasestr(timeZoneCharacters, kGMT) != nil || strcasestr(timeZoneCharacters, kUTC) != nil {
|
||||
return 0
|
||||
}
|
||||
|
||||
if hasAtLeastOneAlphaCharacter {
|
||||
return offsetInSecondsForTimeZoneAbbreviation(timeZoneCharacters)
|
||||
}
|
||||
return offsetInSecondsForOffsetCharacters(timeZoneCharacters)
|
||||
}
|
||||
|
||||
static func offsetInSecondsForOffsetCharacters(_ timeZoneCharacters: DateBuffer) {
|
||||
|
||||
let isPlus = timeZoneCharacters[0] == DateCharacter.plus
|
||||
|
||||
var finalIndex = 0
|
||||
let numberOfCharacters = strlen(timeZoneCharacters)
|
||||
let hours = nextNumericValue(timeZoneCharacters, numberOfCharacters, 0, 2, &finalIndex) ?? 0
|
||||
let minutes = nextNumericValue(timeZoneCharacters, numberOfCharacters, finalIndex + 1, 2, &finalIndex) ?? 0
|
||||
|
||||
if hours == 0 && minutes == 0 {
|
||||
return 0
|
||||
}
|
||||
|
||||
var seconds = (hours * 60 * 60) + (minutes * 60)
|
||||
if !isPlus {
|
||||
seconds = 0 - seconds
|
||||
}
|
||||
|
||||
return seconds
|
||||
}
|
||||
|
||||
/// Returns offset in seconds.
|
||||
static func timeZoneOffset(_ hours: Int, _ minutes: Int) -> Int {
|
||||
|
||||
if hours < 0 {
|
||||
return (hours * 60 * 60) - (minutes * 60)
|
||||
}
|
||||
return (hours * 60 * 60) + (minutes * 60)
|
||||
}
|
||||
|
||||
// See http://en.wikipedia.org/wiki/List_of_time_zone_abbreviations for list
|
||||
private let timeZoneTable: [String: Int] = [
|
||||
|
||||
"GMT": timeZoneOffset(0, 0),
|
||||
"PDT": timeZoneOffset(-7, 0),
|
||||
"PST": timeZoneOffset(-8, 0),
|
||||
"EST": timeZoneOffset(-5, 0),
|
||||
"EDT": timeZoneOffset(-4, 0),
|
||||
"MDT": timeZoneOffset(-6, 0),
|
||||
"MST": timeZoneOffset(-7, 0),
|
||||
"CST": timeZoneOffset(-6, 0),
|
||||
"CDT": timeZoneOffset(-5, 0),
|
||||
"ACT": timeZoneOffset(-8, 0),
|
||||
"AFT": timeZoneOffset(4, 30),
|
||||
"AMT": timeZoneOffset(4, 0),
|
||||
"ART": timeZoneOffset(-3, 0),
|
||||
"AST": timeZoneOffset(3, 0),
|
||||
"AZT": timeZoneOffset(4, 0),
|
||||
"BIT": timeZoneOffset(-12, 0),
|
||||
"BDT": timeZoneOffset(8, 0),
|
||||
"ACST": timeZoneOffset(9, 30),
|
||||
"AEST": timeZoneOffset(10, 0),
|
||||
"AKST": timeZoneOffset(-9, 0),
|
||||
"AMST": timeZoneOffset(5, 0),
|
||||
"AWST": timeZoneOffset(8, 0),
|
||||
"AZOST": timeZoneOffset(-1, 0),
|
||||
"BIOT": timeZoneOffset(6, 0),
|
||||
"BRT": timeZoneOffset(-3, 0),
|
||||
"BST": timeZoneOffset(6, 0),
|
||||
"BTT": timeZoneOffset(6, 0),
|
||||
"CAT": timeZoneOffset(2, 0),
|
||||
"CCT": timeZoneOffset(6, 30),
|
||||
"CET": timeZoneOffset(1, 0),
|
||||
"CEST": timeZoneOffset(2, 0),
|
||||
"CHAST": timeZoneOffset(12, 45),
|
||||
"ChST": timeZoneOffset(10, 0),
|
||||
"CIST": timeZoneOffset(-8, 0),
|
||||
"CKT": timeZoneOffset(-10, 0),
|
||||
"CLT": timeZoneOffset(-4, 0),
|
||||
"CLST": timeZoneOffset(-3, 0),
|
||||
"COT": timeZoneOffset(-5, 0),
|
||||
"COST": timeZoneOffset(-4, 0),
|
||||
"CVT": timeZoneOffset(-1, 0),
|
||||
"CXT": timeZoneOffset(7, 0),
|
||||
"EAST": timeZoneOffset(-6, 0),
|
||||
"EAT": timeZoneOffset(3, 0),
|
||||
"ECT": timeZoneOffset(-4, 0),
|
||||
"EEST": timeZoneOffset(3, 0),
|
||||
"EET": timeZoneOffset(2, 0),
|
||||
"FJT": timeZoneOffset(12, 0),
|
||||
"FKST": timeZoneOffset(-4, 0),
|
||||
"GALT": timeZoneOffset(-6, 0),
|
||||
"GET": timeZoneOffset(4, 0),
|
||||
"GFT": timeZoneOffset(-3, 0),
|
||||
"GILT": timeZoneOffset(7, 0),
|
||||
"GIT": timeZoneOffset(-9, 0),
|
||||
"GST": timeZoneOffset(-2, 0),
|
||||
"GYT": timeZoneOffset(-4, 0),
|
||||
"HAST": timeZoneOffset(-10, 0),
|
||||
"HKT": timeZoneOffset(8, 0),
|
||||
"HMT": timeZoneOffset(5, 0),
|
||||
"IRKT": timeZoneOffset(8, 0),
|
||||
"IRST": timeZoneOffset(3, 30),
|
||||
"IST": timeZoneOffset(2, 0),
|
||||
"JST": timeZoneOffset(9, 0),
|
||||
"KRAT": timeZoneOffset(7, 0),
|
||||
"KST": timeZoneOffset(9, 0),
|
||||
"LHST": timeZoneOffset(10, 30),
|
||||
"LINT": timeZoneOffset(14, 0),
|
||||
"MAGT": timeZoneOffset(11, 0),
|
||||
"MIT": timeZoneOffset(-9, 30),
|
||||
"MSK": timeZoneOffset(3, 0),
|
||||
"MUT": timeZoneOffset(4, 0),
|
||||
"NDT": timeZoneOffset(-2, 30),
|
||||
"NFT": timeZoneOffset(11, 30),
|
||||
"NPT": timeZoneOffset(5, 45),
|
||||
"NT": timeZoneOffset(-3, 30),
|
||||
"OMST": timeZoneOffset(6, 0),
|
||||
"PETT": timeZoneOffset(12, 0),
|
||||
"PHOT": timeZoneOffset(13, 0),
|
||||
"PKT": timeZoneOffset(5, 0),
|
||||
"RET": timeZoneOffset(4, 0),
|
||||
"SAMT": timeZoneOffset(4, 0),
|
||||
"SAST": timeZoneOffset(2, 0),
|
||||
"SBT": timeZoneOffset(11, 0),
|
||||
"SCT": timeZoneOffset(4, 0),
|
||||
"SLT": timeZoneOffset(5, 30),
|
||||
"SST": timeZoneOffset(8, 0),
|
||||
"TAHT": timeZoneOffset(-10, 0),
|
||||
"THA": timeZoneOffset(7, 0),
|
||||
"UYT": timeZoneOffset(-3, 0),
|
||||
"UYST": timeZoneOffset(-2, 0),
|
||||
"VET": timeZoneOffset(-4, 30),
|
||||
"VLAT": timeZoneOffset(10, 0),
|
||||
"WAT": timeZoneOffset(1, 0),
|
||||
"WET": timeZoneOffset(0, 0),
|
||||
"WEST": timeZoneOffset(1, 0),
|
||||
"YAKT": timeZoneOffset(9, 0),
|
||||
"YEKT": timeZoneOffset(5, 0)
|
||||
]
|
||||
|
||||
static func offsetInSecondsForTimeZoneAbbreviation(_ abbreviation: DateBuffer) -> Int? {
|
||||
|
||||
let name = String(cString: abbreviation)
|
||||
return timeZoneTable[name]
|
||||
}
|
||||
|
||||
// MARK: - Parser
|
||||
|
||||
static func nextMonthValue(_ buffer: DateBuffer, _ numberOfBytes: Int, _ startingIndex: Int, _ finalIndex: inout Int) -> DateParser.Month? {
|
||||
|
||||
// Lots of short-circuits here. Not strict.
|
||||
|
||||
var numberOfAlphaCharactersFound = 0
|
||||
var monthCharacters: [CChar] = [0, 0, 0]
|
||||
|
||||
for i in startingIndex..<numberOfBytes {
|
||||
|
||||
finalIndex = i
|
||||
let ch = bytes[i]
|
||||
|
||||
let isAlphaCharacter = isalpha(ch)
|
||||
if !isAlphaCharacter {
|
||||
if numberOfAlphaCharactersFound < 1 {
|
||||
continue
|
||||
}
|
||||
if numberOfAlphaCharactersFound > 0 {
|
||||
break
|
||||
}
|
||||
}
|
||||
|
||||
numberOfAlphaCharactersFound +=1
|
||||
if numberOfAlphaCharactersFound == 1 {
|
||||
if ch == DateCharacter.F || ch == DateCharacter.f {
|
||||
return February
|
||||
}
|
||||
if ch == DateCharacter.S || ch == DateCharacter.s {
|
||||
return September
|
||||
}
|
||||
if ch == DateCharacter.O || ch == DateCharacter.o {
|
||||
return October
|
||||
}
|
||||
if ch == DateCharacter.N || ch == DateCharacter.n {
|
||||
return November
|
||||
}
|
||||
if ch == DateCharacter.D || ch == DateCharacter.d {
|
||||
return December
|
||||
}
|
||||
}
|
||||
|
||||
monthCharacters[numberOfAlphaCharactersFound - 1] = character
|
||||
if numberOfAlphaCharactersFound >=3
|
||||
break
|
||||
}
|
||||
|
||||
if numberOfAlphaCharactersFound < 2 {
|
||||
return nil
|
||||
}
|
||||
|
||||
if monthCharacters[0] == DateCharater.J || monthCharacters[0] == DateCharacter.j { // Jan, Jun, Jul
|
||||
if monthCharacters[1] == DateCharacter.A || monthCharacters[1] == DateCharacter.a {
|
||||
return Month.January
|
||||
}
|
||||
if monthCharacters[1] = DateCharacter.U || monthCharacters[1] == DateCharacter.u {
|
||||
if monthCharacters[2] == DateCharacter.N || monthCharacters[2] == DateCharacter.n {
|
||||
return June
|
||||
}
|
||||
return July
|
||||
}
|
||||
return January
|
||||
}
|
||||
|
||||
if monthCharacters[0] == DateCharacter.M || monthCharacters[0] == DateCharacter.m { // March, May
|
||||
if monthCharacters[2] == DateCharacter.Y || monthCharacters[2] == DateCharacter.y {
|
||||
return May
|
||||
}
|
||||
return March
|
||||
}
|
||||
|
||||
if monthCharacters[0] == DateCharacter.A || monthCharacters[0] == DateCharacter.a { // April, August
|
||||
if monthCharacters[1] == DateCharacter.U || monthCharacters[1] == DateCharacter.u {
|
||||
return August
|
||||
}
|
||||
return April
|
||||
}
|
||||
|
||||
return January // Should never get here (but possibly do)
|
||||
}
|
||||
|
||||
static func nextNumericValue(_ bytes: DateBuffer, numberOfBytes: Int, startingIndex: Int, maximumNumberOfDigits: Int, finalIndex: inout Int) -> Int? {
|
||||
|
||||
// Maximum for the maximum is 4 (for time zone offsets and years)
|
||||
assert(maximumNumberOfDigits > 0 && maximumNumberOfDigits <= 4)
|
||||
|
||||
var numberOfDigitsFound = 0
|
||||
var digits = [0, 0, 0, 0]
|
||||
|
||||
for i in startingIndex..<numberOfBytes {
|
||||
|
||||
finalIndex = i
|
||||
|
||||
let isDigit = isDigit(
|
||||
}
|
||||
|
||||
}
|
||||
}
|
||||
354
Modules/Parser/Sources/DateParser/DateParserx.swift
Normal file
354
Modules/Parser/Sources/DateParser/DateParserx.swift
Normal file
@ -0,0 +1,354 @@
|
||||
////
|
||||
//// DateParser.swift
|
||||
////
|
||||
////
|
||||
//// Created by Brent Simmons on 8/28/24.
|
||||
////
|
||||
//
|
||||
//import Foundation
|
||||
//
|
||||
//private struct TimeZoneSpecifier {
|
||||
// let abbreviation: String
|
||||
// let offsetHours: Int
|
||||
// let offsetMinutes: Int
|
||||
//
|
||||
// init(_ abbreviation: String, _ offsetHours: Int, _ offsetMinutes: Int) {
|
||||
// self.abbreviation = abbreviation
|
||||
// self.offsetHours = offsetHours
|
||||
// self.offsetMinutes = offsetMinutes
|
||||
// }
|
||||
//}
|
||||
//
|
||||
//// See http://en.wikipedia.org/wiki/List_of_time_zone_abbreviations for list
|
||||
//private let timeZoneTable: [TimeZoneAbbreviationAndOffset] = [
|
||||
// // Most common at top for performance
|
||||
// TimeZoneSpecifier("GMT", 0, 0),
|
||||
// TimeZoneSpecifier("PDT", -7, 0),
|
||||
// TimeZoneSpecifier("PST", -8, 0),
|
||||
// TimeZoneSpecifier("EST", -5, 0),
|
||||
// TimeZoneSpecifier("EDT", -4, 0),
|
||||
// TimeZoneSpecifier("MDT", -6, 0),
|
||||
// TimeZoneSpecifier("MST", -7, 0),
|
||||
// TimeZoneSpecifier("CST", -6, 0),
|
||||
// TimeZoneSpecifier("CDT", -5, 0),
|
||||
// TimeZoneSpecifier("ACT", -8, 0),
|
||||
// TimeZoneSpecifier("AFT", 4, 30),
|
||||
// TimeZoneSpecifier("AMT", 4, 0),
|
||||
// TimeZoneSpecifier("ART", -3, 0),
|
||||
// TimeZoneSpecifier("AST", 3, 0),
|
||||
// TimeZoneSpecifier("AZT", 4, 0),
|
||||
// TimeZoneSpecifier("BIT", -12, 0),
|
||||
// TimeZoneSpecifier("BDT", 8, 0),
|
||||
// TimeZoneSpecifier("ACST", 9, 30),
|
||||
// TimeZoneSpecifier("AEST", 10, 0),
|
||||
// TimeZoneSpecifier("AKST", -9, 0),
|
||||
// TimeZoneSpecifier("AMST", 5, 0),
|
||||
// TimeZoneSpecifier("AWST", 8, 0),
|
||||
// TimeZoneSpecifier("AZOST", -1, 0),
|
||||
// TimeZoneSpecifier("BIOT", 6, 0),
|
||||
// TimeZoneSpecifier("BRT", -3, 0),
|
||||
// TimeZoneSpecifier("BST", 6, 0),
|
||||
// TimeZoneSpecifier("BTT", 6, 0),
|
||||
// TimeZoneSpecifier("CAT", 2, 0),
|
||||
// TimeZoneSpecifier("CCT", 6, 30),
|
||||
// TimeZoneSpecifier("CET", 1, 0),
|
||||
// TimeZoneSpecifier("CEST", 2, 0),
|
||||
// TimeZoneSpecifier("CHAST", 12, 45),
|
||||
// TimeZoneSpecifier("ChST", 10, 0),
|
||||
// TimeZoneSpecifier("CIST", -8, 0),
|
||||
// TimeZoneSpecifier("CKT", -10, 0),
|
||||
// TimeZoneSpecifier("CLT", -4, 0),
|
||||
// TimeZoneSpecifier("CLST", -3, 0),
|
||||
// TimeZoneSpecifier("COT", -5, 0),
|
||||
// TimeZoneSpecifier("COST", -4, 0),
|
||||
// TimeZoneSpecifier("CVT", -1, 0),
|
||||
// TimeZoneSpecifier("CXT", 7, 0),
|
||||
// TimeZoneSpecifier("EAST", -6, 0),
|
||||
// TimeZoneSpecifier("EAT", 3, 0),
|
||||
// TimeZoneSpecifier("ECT", -4, 0),
|
||||
// TimeZoneSpecifier("EEST", 3, 0),
|
||||
// TimeZoneSpecifier("EET", 2, 0),
|
||||
// TimeZoneSpecifier("FJT", 12, 0),
|
||||
// TimeZoneSpecifier("FKST", -4, 0),
|
||||
// TimeZoneSpecifier("GALT", -6, 0),
|
||||
// TimeZoneSpecifier("GET", 4, 0),
|
||||
// TimeZoneSpecifier("GFT", -3, 0),
|
||||
// TimeZoneSpecifier("GILT", 7, 0),
|
||||
// TimeZoneSpecifier("GIT", -9, 0),
|
||||
// TimeZoneSpecifier("GST", -2, 0),
|
||||
// TimeZoneSpecifier("GYT", -4, 0),
|
||||
// TimeZoneSpecifier("HAST", -10, 0),
|
||||
// TimeZoneSpecifier("HKT", 8, 0),
|
||||
// TimeZoneSpecifier("HMT", 5, 0),
|
||||
// TimeZoneSpecifier("IRKT", 8, 0),
|
||||
// TimeZoneSpecifier("IRST", 3, 30),
|
||||
// TimeZoneSpecifier("IST", 2, 0),
|
||||
// TimeZoneSpecifier("JST", 9, 0),
|
||||
// TimeZoneSpecifier("KRAT", 7, 0),
|
||||
// TimeZoneSpecifier("KST", 9, 0),
|
||||
// TimeZoneSpecifier("LHST", 10, 30),
|
||||
// TimeZoneSpecifier("LINT", 14, 0),
|
||||
// TimeZoneSpecifier("MAGT", 11, 0),
|
||||
// TimeZoneSpecifier("MIT", -9, 30),
|
||||
// TimeZoneSpecifier("MSK", 3, 0),
|
||||
// TimeZoneSpecifier("MUT", 4, 0),
|
||||
// TimeZoneSpecifier("NDT", -2, 30),
|
||||
// TimeZoneSpecifier("NFT", 11, 30),
|
||||
// TimeZoneSpecifier("NPT", 5, 45),
|
||||
// TimeZoneSpecifier("NT", -3, 30),
|
||||
// TimeZoneSpecifier("OMST", 6, 0),
|
||||
// TimeZoneSpecifier("PETT", 12, 0),
|
||||
// TimeZoneSpecifier("PHOT", 13, 0),
|
||||
// TimeZoneSpecifier("PKT", 5, 0),
|
||||
// TimeZoneSpecifier("RET", 4, 0),
|
||||
// TimeZoneSpecifier("SAMT", 4, 0),
|
||||
// TimeZoneSpecifier("SAST", 2, 0),
|
||||
// TimeZoneSpecifier("SBT", 11, 0),
|
||||
// TimeZoneSpecifier("SCT", 4, 0),
|
||||
// TimeZoneSpecifier("SLT", 5, 30),
|
||||
// TimeZoneSpecifier("SST", 8, 0),
|
||||
// TimeZoneSpecifier("TAHT", -10, 0),
|
||||
// TimeZoneSpecifier("THA", 7, 0),
|
||||
// TimeZoneSpecifier("UYT", -3, 0),
|
||||
// TimeZoneSpecifier("UYST", -2, 0),
|
||||
// TimeZoneSpecifier("VET", -4, 30),
|
||||
// TimeZoneSpecifier("VLAT", 10, 0),
|
||||
// TimeZoneSpecifier("WAT", 1, 0),
|
||||
// TimeZoneSpecifier("WET", 0, 0),
|
||||
// TimeZoneSpecifier("WEST", 1, 0),
|
||||
// TimeZoneSpecifier("YAKT", 9, 0),
|
||||
// TimeZoneSpecifier("YEKT", 5, 0)
|
||||
//]
|
||||
//
|
||||
//private enum Month: Int {
|
||||
// case January = 1, February, March, April, May, June, July, August, September, October, November, December
|
||||
//}
|
||||
//
|
||||
//private func nextMonthValue(bytes: String, startingIndex: Int, finalIndex: inout Int) -> Int? {
|
||||
//
|
||||
// // Months are 1-based -- January is 1, Dec is 12.
|
||||
// // Lots of short-circuits here. Not strict. GIGO
|
||||
//
|
||||
// var i = startingIndex
|
||||
// var numberOfBytes = bytes.count
|
||||
// var numberOfAlphaCharactersFound = 0
|
||||
// var monthCharacters = [Character]()
|
||||
//
|
||||
// while index < bytes.count {
|
||||
//
|
||||
//
|
||||
// }
|
||||
//
|
||||
//
|
||||
// var index = startingIndex
|
||||
// var numberOfAlphaCharactersFound = 0
|
||||
// var monthCharacters: [Character] = []
|
||||
//
|
||||
// while index < bytes.count {
|
||||
// let character = bytes[bytes.index(bytes.startIndex, offsetBy: index)]
|
||||
//
|
||||
// if !character.isLetter, numberOfAlphaCharactersFound < 1 {
|
||||
// index += 1
|
||||
// continue
|
||||
// }
|
||||
// if !character.isLetter, numberOfAlphaCharactersFound > 0 {
|
||||
// break
|
||||
// }
|
||||
//
|
||||
// numberOfAlphaCharactersFound += 1
|
||||
// if numberOfAlphaCharactersFound == 1 {
|
||||
// switch character.lowercased() {
|
||||
// case "f": return (.February.rawValue, index)
|
||||
// case "s": return (.September.rawValue, index)
|
||||
// case "o": return (.October.rawValue, index)
|
||||
// case "n": return (.November.rawValue, index)
|
||||
// case "d": return (.December.rawValue, index)
|
||||
// default: break
|
||||
// }
|
||||
// }
|
||||
//
|
||||
// monthCharacters.append(character)
|
||||
// if numberOfAlphaCharactersFound >= 3 {
|
||||
// break
|
||||
// }
|
||||
// index += 1
|
||||
// }
|
||||
//
|
||||
// if numberOfAlphaCharactersFound < 2 {
|
||||
// return (nil, index)
|
||||
// }
|
||||
//
|
||||
// if monthCharacters[0].lowercased() == "j" {
|
||||
// if monthCharacters[1].lowercased() == "a" {
|
||||
// return (.January.rawValue, index)
|
||||
// }
|
||||
// if monthCharacters[1].lowercased() == "u" {
|
||||
// if monthCharacters.count > 2 && monthCharacters[2].lowercased() == "n" {
|
||||
// return (.June.rawValue, index)
|
||||
// }
|
||||
// return (.July.rawValue, index)
|
||||
// }
|
||||
// return (.January.rawValue, index)
|
||||
// }
|
||||
//
|
||||
// if monthCharacters[0].lowercased() == "m" {
|
||||
// if monthCharacters.count > 2 && monthCharacters[2].lowercased() == "y" {
|
||||
// return (.May.rawValue, index)
|
||||
// }
|
||||
// return (.March.rawValue, index)
|
||||
// }
|
||||
//
|
||||
// if monthCharacters[0].lowercased() == "a" {
|
||||
// if monthCharacters[1].lowercased() == "u" {
|
||||
// return (.August.rawValue, index)
|
||||
// }
|
||||
// return (.April.rawValue, index)
|
||||
// }
|
||||
//
|
||||
// return (.January.rawValue, index)
|
||||
//}
|
||||
//
|
||||
//func nextNumericValue(bytes: String, startingIndex: Int, maximumNumberOfDigits: Int) -> (Int?, Int) {
|
||||
// let digits = bytes.dropFirst(startingIndex).prefix(maximumNumberOfDigits)
|
||||
// guard let value = Int(digits) else {
|
||||
// return (nil, startingIndex)
|
||||
// }
|
||||
// return (value, startingIndex + digits.count)
|
||||
//}
|
||||
//
|
||||
//func hasAtLeastOneAlphaCharacter(_ s: String) -> Bool {
|
||||
// return s.contains { $0.isLetter }
|
||||
//}
|
||||
//
|
||||
//func offsetInSeconds(forTimeZoneAbbreviation abbreviation: String) -> Int {
|
||||
// for zone in timeZoneTable {
|
||||
// if zone.abbreviation.caseInsensitiveCompare(abbreviation) == .orderedSame {
|
||||
// if zone.offsetHours < 0 {
|
||||
// return (zone.offsetHours * 3600) - (zone.offsetMinutes * 60)
|
||||
// }
|
||||
// return (zone.offsetHours * 3600) + (zone.offsetMinutes * 60)
|
||||
// }
|
||||
// }
|
||||
// return 0
|
||||
//}
|
||||
//
|
||||
//func offsetInSeconds(forOffsetCharacters timeZoneCharacters: String) -> Int {
|
||||
// let isPlus = timeZoneCharacters.hasPrefix("+")
|
||||
// let numericValue = timeZoneCharacters.filter { $0.isNumber || $0 == "-" }
|
||||
// let (hours, finalIndex) = nextNumericValue(bytes: numericValue, startingIndex: 0, maximumNumberOfDigits: 2)
|
||||
// let (minutes, _) = nextNumericValue(bytes: numericValue, startingIndex: finalIndex + 1, maximumNumberOfDigits: 2)
|
||||
//
|
||||
// let seconds = ((hours ?? 0) * 3600) + ((minutes ?? 0) * 60)
|
||||
// return isPlus ? seconds : -seconds
|
||||
//}
|
||||
//
|
||||
//func parsedTimeZoneOffset(bytes: String, startingIndex: Int) -> Int {
|
||||
// var timeZoneCharacters: String = ""
|
||||
// var numberOfCharactersFound = 0
|
||||
// var i = startingIndex
|
||||
//
|
||||
// while i < bytes.count, numberOfCharactersFound < 5 {
|
||||
// let character = bytes[bytes.index(bytes.startIndex, offsetBy: i)]
|
||||
// if character != ":" && character != " " {
|
||||
// timeZoneCharacters.append(character)
|
||||
// numberOfCharactersFound += 1
|
||||
// }
|
||||
// i += 1
|
||||
// }
|
||||
//
|
||||
// if numberOfCharactersFound < 1 || timeZoneCharacters.lowercased() == "z" {
|
||||
// return 0
|
||||
// }
|
||||
//
|
||||
// if timeZoneCharacters.range(of: "GMT", options: .caseInsensitive) != nil ||
|
||||
// timeZoneCharacters.range(of: "UTC", options: .caseInsensitive) != nil {
|
||||
// return 0
|
||||
// }
|
||||
//
|
||||
// if hasAtLeastOneAlphaCharacter(timeZoneCharacters) {
|
||||
// return offsetInSeconds(forTimeZoneAbbreviation: timeZoneCharacters)
|
||||
// }
|
||||
// return offsetInSeconds(forOffsetCharacters: timeZoneCharacters)
|
||||
//}
|
||||
//
|
||||
//func dateWithYearMonthDayHourMinuteSecondAndTimeZoneOffset(
|
||||
// year: Int, month: Int, day: Int,
|
||||
// hour: Int, minute: Int, second: Int,
|
||||
// milliseconds: Int, timeZoneOffset: Int) -> Date? {
|
||||
//
|
||||
// var dateComponents = DateComponents()
|
||||
// dateComponents.year = year
|
||||
// dateComponents.month = month
|
||||
// dateComponents.day = day
|
||||
// dateComponents.hour = hour
|
||||
// dateComponents.minute = minute
|
||||
// dateComponents.second = second
|
||||
// dateComponents.timeZone = TimeZone(secondsFromGMT: timeZoneOffset)
|
||||
//
|
||||
// let calendar = Calendar.current
|
||||
// return calendar.date(from: dateComponents)
|
||||
//}
|
||||
//
|
||||
//func parsePubDate(bytes: String) -> Date? {
|
||||
// let (day, finalIndex) = nextNumericValue(bytes: bytes, startingIndex: 0, maximumNumberOfDigits: 2)
|
||||
// let (month, finalIndex2) = nextMonthValue(bytes: bytes, startingIndex: finalIndex + 1)
|
||||
// let (year, finalIndex3) = nextNumericValue(bytes: bytes, startingIndex: finalIndex2 + 1, maximumNumberOfDigits: 4)
|
||||
// let (hour, finalIndex4) = nextNumericValue(bytes: bytes, startingIndex: finalIndex3 + 1, maximumNumberOfDigits: 2)
|
||||
// let (minute, finalIndex5) = nextNumericValue(bytes: bytes, startingIndex: finalIndex4 + 1, maximumNumberOfDigits: 2)
|
||||
//
|
||||
// var second = 0
|
||||
// let currentIndex = finalIndex5 + 1
|
||||
// if currentIndex < bytes.count, bytes[bytes.index(bytes.startIndex, offsetBy: currentIndex)] == ":" {
|
||||
// second = nextNumericValue(bytes: bytes, startingIndex: currentIndex, maximumNumberOfDigits: 2).0 ?? 0
|
||||
// }
|
||||
//
|
||||
// let timeZoneOffset = parsedTimeZoneOffset(bytes: bytes, startingIndex: currentIndex + 1)
|
||||
//
|
||||
// return dateWithYearMonthDayHourMinuteSecondAndTimeZoneOffset(
|
||||
// year: year ?? 1970,
|
||||
// month: month ?? RSMonth.January.rawValue,
|
||||
// day: day ?? 1,
|
||||
// hour: hour ?? 0,
|
||||
// minute: minute ?? 0,
|
||||
// second: second,
|
||||
// milliseconds: 0,
|
||||
// timeZoneOffset: timeZoneOffset
|
||||
// )
|
||||
//}
|
||||
//
|
||||
//func parseW3C(bytes: String) -> Date? {
|
||||
// let (year, finalIndex) = nextNumericValue(bytes: bytes, startingIndex: 0, maximumNumberOfDigits: 4)
|
||||
// let (month, finalIndex2) = nextNumericValue(bytes: bytes, startingIndex: finalIndex + 1, maximumNumberOfDigits: 2)
|
||||
// let (day, finalIndex3) = nextNumericValue(bytes: bytes, startingIndex: finalIndex2 + 1, maximumNumberOfDigits: 2)
|
||||
// let (hour, finalIndex4) = nextNumericValue(bytes: bytes, startingIndex: finalIndex3 + 1, maximumNumberOfDigits: 2)
|
||||
// let (minute, finalIndex5) = nextNumericValue(bytes: bytes, startingIndex: finalIndex4 + 1, maximumNumberOfDigits: 2)
|
||||
// let (second, finalIndex6) = nextNumericValue(bytes: bytes, startingIndex: finalIndex5 + 1, maximumNumberOfDigits: 2)
|
||||
//
|
||||
// var milliseconds = 0
|
||||
// let currentIndex = finalIndex6 + 1
|
||||
// if currentIndex < bytes.count, bytes[bytes.index(bytes.startIndex, offsetBy: currentIndex)] == "." {
|
||||
// milliseconds = nextNumericValue(bytes: bytes, startingIndex: currentIndex + 1, maximumNumberOfDigits: 3).0 ?? 0
|
||||
// }
|
||||
//
|
||||
// let timeZoneOffset = parsedTimeZoneOffset(bytes: bytes, startingIndex: currentIndex + 1)
|
||||
//
|
||||
// return dateWithYearMonthDayHourMinuteSecondAndTimeZoneOffset(
|
||||
// year: year ?? 1970,
|
||||
// month: month ?? RSMonth.January.rawValue,
|
||||
// day: day ?? 1,
|
||||
// hour: hour ?? 0,
|
||||
// minute: minute ?? 0,
|
||||
// second: second ?? 0,
|
||||
// milliseconds: milliseconds,
|
||||
// timeZoneOffset: timeZoneOffset
|
||||
// )
|
||||
//}
|
||||
//
|
||||
//func dateWithBytes(bytes: String) -> Date? {
|
||||
// guard !bytes.isEmpty else { return nil }
|
||||
//
|
||||
// if bytes.range(of: "-") != nil {
|
||||
// return parseW3C(bytes: bytes)
|
||||
// }
|
||||
// return parsePubDate(bytes: bytes)
|
||||
//}
|
||||
@ -7,6 +7,7 @@
|
||||
//
|
||||
|
||||
import Foundation
|
||||
import SAX
|
||||
|
||||
// FeedParser handles RSS, Atom, JSON Feed, and RSS-in-JSON.
|
||||
// You don’t need to know the type of feed.
|
||||
|
||||
@ -7,6 +7,7 @@
|
||||
//
|
||||
|
||||
import Foundation
|
||||
import SAX
|
||||
|
||||
public enum FeedType: Sendable {
|
||||
case rss
|
||||
|
||||
@ -7,6 +7,7 @@
|
||||
//
|
||||
|
||||
import Foundation
|
||||
import SAX
|
||||
|
||||
// See https://jsonfeed.org/version/1.1
|
||||
|
||||
|
||||
@ -7,6 +7,7 @@
|
||||
//
|
||||
|
||||
import Foundation
|
||||
import SAX
|
||||
|
||||
// See https://github.com/scripting/Scripting-News/blob/master/rss-in-json/README.md
|
||||
// Also: http://cyber.harvard.edu/rss/rss.html
|
||||
|
||||
@ -8,7 +8,7 @@
|
||||
|
||||
import Foundation
|
||||
|
||||
public struct ParsedItem: Hashable, Sendable {
|
||||
public final class ParsedItem: Hashable, Sendable {
|
||||
|
||||
public let syncServiceID: String? //Nil when not syncing
|
||||
public let uniqueID: String //RSS guid, for instance; may be calculated
|
||||
|
||||
@ -7,6 +7,7 @@
|
||||
//
|
||||
|
||||
import Foundation
|
||||
import SAX
|
||||
|
||||
// RSSParser wraps the Objective-C RSAtomParser.
|
||||
//
|
||||
|
||||
111
Modules/Parser/Sources/FeedParser/Feeds/XML/RSSArticle.swift
Normal file
111
Modules/Parser/Sources/FeedParser/Feeds/XML/RSSArticle.swift
Normal file
@ -0,0 +1,111 @@
|
||||
//
|
||||
// RSSArticle.swift
|
||||
//
|
||||
//
|
||||
// Created by Brent Simmons on 8/27/24.
|
||||
//
|
||||
|
||||
import Foundation
|
||||
import FoundationExtras
|
||||
|
||||
final class RSSArticle {
|
||||
|
||||
var feedURL: String
|
||||
|
||||
/// An RSS guid, if present, or calculated from other attributes.
|
||||
/// Should be unique to the feed, but not necessarily unique
|
||||
/// across different feeds. (Not suitable for a database ID.)
|
||||
lazy var articleID: String = {
|
||||
if let guid {
|
||||
return guid
|
||||
}
|
||||
return calculatedArticleID()
|
||||
}()
|
||||
|
||||
var guid: String?
|
||||
var title: String?
|
||||
var body: String?
|
||||
var link: String?
|
||||
var permalink: String?
|
||||
var authors: [RSSAuthor]?
|
||||
var enclosures: [RSSEnclosure]?
|
||||
var datePublished: Date?
|
||||
var dateModified: Date?
|
||||
var dateParsed: Date
|
||||
var language: String?
|
||||
|
||||
init(_ feedURL: String) {
|
||||
self.feedURL = feedURL
|
||||
self.dateParsed = Date()
|
||||
}
|
||||
|
||||
func addEnclosure(_ enclosure: RSSEnclosure) {
|
||||
|
||||
if enclosures == nil {
|
||||
enclosures = [RSSEnclosure]()
|
||||
}
|
||||
enclosures!.append(enclosure)
|
||||
}
|
||||
|
||||
func addAuthor(_ author: RSSAuthor) {
|
||||
|
||||
if authors == nil {
|
||||
authors = [RSSAuthor]()
|
||||
}
|
||||
authors!.append(author)
|
||||
}
|
||||
}
|
||||
|
||||
private extension RSSArticle {
|
||||
|
||||
func calculatedArticleID() -> String {
|
||||
|
||||
// Concatenate a combination of properties when no guid. Then hash the result.
|
||||
// In general, feeds should have guids. When they don't, re-runs are very likely,
|
||||
// because there's no other 100% reliable way to determine identity.
|
||||
// This is intended to create an ID unique inside a feed, but not globally unique.
|
||||
// Not suitable for a database ID, in other words.
|
||||
|
||||
var s = ""
|
||||
|
||||
let datePublishedTimeStampString: String? = {
|
||||
guard let datePublished else {
|
||||
return nil
|
||||
}
|
||||
return String(format: "%.0f", datePublished.timeIntervalSince1970)
|
||||
}()
|
||||
|
||||
// Ideally we have a permalink and a pubDate.
|
||||
// Either one would probably be a good guid, but together they should be rock-solid.
|
||||
// (In theory. Feeds are buggy, though.)
|
||||
if let permalink, !permalink.isEmpty, let datePublishedTimeStampString {
|
||||
s.append(permalink)
|
||||
s.append(datePublishedTimeStampString)
|
||||
}
|
||||
else if let link, !link.isEmpty, let datePublishedTimeStampString {
|
||||
s.append(link)
|
||||
s.append(datePublishedTimeStampString)
|
||||
}
|
||||
else if let title, !title.isEmpty, let datePublishedTimeStampString {
|
||||
s.append(title)
|
||||
s.append(datePublishedTimeStampString)
|
||||
}
|
||||
else if let datePublishedTimeStampString {
|
||||
s.append(datePublishedTimeStampString)
|
||||
}
|
||||
else if let permalink, !permalink.isEmpty {
|
||||
s.append(permalink)
|
||||
}
|
||||
else if let link, !link.isEmpty {
|
||||
s.append(link)
|
||||
}
|
||||
else if let title, !title.isEmpty {
|
||||
s.append(title)
|
||||
}
|
||||
else if let body, !body.isEmpty {
|
||||
s.append(body)
|
||||
}
|
||||
|
||||
return s.md5String
|
||||
}
|
||||
}
|
||||
35
Modules/Parser/Sources/FeedParser/Feeds/XML/RSSAuthor.swift
Normal file
35
Modules/Parser/Sources/FeedParser/Feeds/XML/RSSAuthor.swift
Normal file
@ -0,0 +1,35 @@
|
||||
//
|
||||
// RSSAuthor.swift
|
||||
//
|
||||
//
|
||||
// Created by Brent Simmons on 8/27/24.
|
||||
//
|
||||
|
||||
import Foundation
|
||||
|
||||
final class RSSAuthor {
|
||||
|
||||
var name: String?
|
||||
var url: String?
|
||||
var avatarURL: String?
|
||||
var emailAddress: String?
|
||||
|
||||
init(name: String?, url: String?, avatarURL: String?, emailAddress: String?) {
|
||||
self.name = name
|
||||
self.url = url
|
||||
self.avatarURL = avatarURL
|
||||
self.emailAddress = emailAddress
|
||||
}
|
||||
|
||||
/// Use when the actual property is unknown. Guess based on contents of the string. (This is common with RSS.)
|
||||
convenience init(singleString: String) {
|
||||
|
||||
if singleString.contains("@") {
|
||||
self.init(name: nil, url: nil, avatarURL: nil, emailAddress: singleString)
|
||||
} else if singleString.lowercased().hasPrefix("http") {
|
||||
self.init(name: nil, url: singleString, avatarURL: nil, emailAddress: nil)
|
||||
} else {
|
||||
self.init(name: singleString, url: nil, avatarURL: nil, emailAddress: nil)
|
||||
}
|
||||
}
|
||||
}
|
||||
@ -0,0 +1,20 @@
|
||||
//
|
||||
// RSSEnclosure.swift
|
||||
//
|
||||
//
|
||||
// Created by Brent Simmons on 8/27/24.
|
||||
//
|
||||
|
||||
import Foundation
|
||||
|
||||
final class RSSEnclosure {
|
||||
|
||||
var url: String
|
||||
var length: Int?
|
||||
var mimeType: String?
|
||||
var title: String?
|
||||
|
||||
init(url: String) {
|
||||
self.url = url
|
||||
}
|
||||
}
|
||||
22
Modules/Parser/Sources/FeedParser/Feeds/XML/RSSFeed.swift
Normal file
22
Modules/Parser/Sources/FeedParser/Feeds/XML/RSSFeed.swift
Normal file
@ -0,0 +1,22 @@
|
||||
//
|
||||
// RSSFeed.swift
|
||||
//
|
||||
//
|
||||
// Created by Brent Simmons on 8/27/24.
|
||||
//
|
||||
|
||||
import Foundation
|
||||
|
||||
final class RSSFeed {
|
||||
|
||||
var urlString: String
|
||||
var title: String?
|
||||
var link: String?
|
||||
var language: String?
|
||||
|
||||
var articles: [RSSArticle]?
|
||||
|
||||
init(urlString: String) {
|
||||
self.urlString = urlString
|
||||
}
|
||||
}
|
||||
@ -11,12 +11,205 @@ import SAX
|
||||
|
||||
public final class RSSParser {
|
||||
|
||||
private var parseFeed: ParsedFeed?
|
||||
private var parserData: ParserData
|
||||
private var feedURL: String {
|
||||
parserData.url
|
||||
}
|
||||
private var data: Data {
|
||||
parserData.data
|
||||
}
|
||||
|
||||
private let feed: RSSFeed
|
||||
private var articles = [RSSArticle]()
|
||||
private var currentArticle: RSSArticle? {
|
||||
articles.last
|
||||
}
|
||||
|
||||
public static func parsedFeed(with parserData: ParserData) -> ParsedFeed? {
|
||||
private var endRSSFound = false
|
||||
private var isRDF = false
|
||||
private var parsingArticle = false
|
||||
private var parsingChannelImage = false
|
||||
private var parsingAuthor = false
|
||||
private var currentAttributes: XMLAttributesDictionary?
|
||||
|
||||
public static func parsedFeed(with parserData: ParserData) -> RSSFeed {
|
||||
|
||||
let parser = RSSParser(parserData)
|
||||
parser.parse()
|
||||
return parser.parsedFeed
|
||||
return parser.feed
|
||||
}
|
||||
|
||||
init(_ parserData: ParserData) {
|
||||
self.parserData = parserData
|
||||
self.feed = RSSFeed(urlString: parserData.url)
|
||||
}
|
||||
}
|
||||
|
||||
private extension RSSParser {
|
||||
|
||||
private struct XMLName {
|
||||
static let uppercaseRDF = "RDF".utf8CString
|
||||
static let item = "item".utf8CString
|
||||
static let guid = "guid".utf8CString
|
||||
static let enclosure = "enclosure".utf8CString
|
||||
static let rdfAbout = "rdf:about".utf8CString
|
||||
static let image = "image".utf8CString
|
||||
static let author = "author".utf8CString
|
||||
static let rss = "rss".utf8CString
|
||||
static let link = "link".utf8CString
|
||||
static let title = "title".utf8CString
|
||||
static let language = "language".utf8CString
|
||||
static let dc = "dc".utf8CString
|
||||
static let content = "content".utf8CString
|
||||
static let encoded = "encoded".utf8CString
|
||||
}
|
||||
|
||||
func addFeedElement(_ localName: XMLPointer, _ prefix: XMLPointer?) {
|
||||
|
||||
guard prefix == nil else {
|
||||
return
|
||||
}
|
||||
|
||||
if SAXEqualTags(localName, XMLName.link) {
|
||||
if feed.link == nil {
|
||||
feed.link = currentString
|
||||
}
|
||||
}
|
||||
else if SAXEqualTags(localName, XMLName.title) {
|
||||
feed.title = currentString
|
||||
}
|
||||
else if SAXEqualTags(localName, XMLName.language) {
|
||||
feed.language = currentString
|
||||
}
|
||||
}
|
||||
|
||||
func addArticle() {
|
||||
let article = RSSArticle(feedURL)
|
||||
articles.append(article)
|
||||
}
|
||||
|
||||
func addArticleElement(_ localName: XMLPointer, _ prefix: XMLPointer?) {
|
||||
|
||||
if SAXEqualTags(prefix, XMLName.dc) {
|
||||
addDCElement(localName)
|
||||
return;
|
||||
}
|
||||
|
||||
if SAXEqualTags(prefix, XMLName.content) && SAXEqualTags(localName, XMLName.encoded) {
|
||||
if let currentString, !currentString.isEmpty {
|
||||
currentArticle.body = currentString
|
||||
}
|
||||
return
|
||||
}
|
||||
|
||||
guard prefix == nil else {
|
||||
return
|
||||
}
|
||||
|
||||
if SAXEqualTags(localName, XMLName.guid) {
|
||||
addGuid()
|
||||
}
|
||||
else if SAXEqualTags(localName, XMLName.pubDate) {
|
||||
currentArticle.datePublished = currentDate
|
||||
}
|
||||
else if SAXEqualTags(localName, XMLName.author) {
|
||||
addAuthorWithString(currentString)
|
||||
}
|
||||
else if SAXEqualTags(localName, XMLName.link) {
|
||||
currentArticle.link = urlString(currentString)
|
||||
}
|
||||
else if SAXEqualTags(localName, XMLName.description) {
|
||||
if currentArticle.body == nil {
|
||||
currentArticle.body = currentString
|
||||
}
|
||||
}
|
||||
else if !parsingAuthor && SAXEqualTags(localName, XMLName.title) {
|
||||
if let currentString {
|
||||
currentArticle.title = currentString
|
||||
}
|
||||
}
|
||||
else if SAXEqualTags(localName, XMLName.enclosure) {
|
||||
addEnclosure()
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
extension RSSParser: SAXParserDelegate {
|
||||
|
||||
public func saxParser(_ saxParser: SAXParser, xmlStartElement localName: XMLPointer, prefix: XMLPointer?, uri: XMLPointer?, namespaceCount: Int, namespaces: UnsafePointer<XMLPointer?>?, attributeCount: Int, attributesDefaultedCount: Int, attributes: UnsafePointer<XMLPointer?>?) {
|
||||
|
||||
if endRSSFound {
|
||||
return
|
||||
}
|
||||
|
||||
if SAXEqualTags(localName, XMLName.uppercaseRDF) {
|
||||
isRDF = true
|
||||
return
|
||||
}
|
||||
|
||||
var xmlAttributes: XMLAttributesDictionary? = nil
|
||||
if (isRDF && SAXEqualTags(localName, XMLName.item)) || SAXEqualTags(localName, XMLName.guid) || SAXEqualTags(enclosure, XMLName.enclosure) {
|
||||
xmlAttributes = saxParser.attributesDictionary(attributes, attributeCount: attributeCount)
|
||||
}
|
||||
if currentAttributes != xmlAttributes {
|
||||
currentAttributes = xmlAttributes
|
||||
}
|
||||
|
||||
if prefix == nil && SAXEqualTags(localName, XMLName.item) {
|
||||
addArticle()
|
||||
parsingArticle = true
|
||||
|
||||
if isRDF && let rdfGuid = xmlAttributes?[XMLName.rdfAbout], let currentArticle { // RSS 1.0 guid
|
||||
currentArticle.guid = rdfGuid
|
||||
currentArticle.permalink = rdfGuid
|
||||
}
|
||||
}
|
||||
else if prefix == nil && SAXEqualTags(localName, XMLName.image) {
|
||||
parsingChannelImage = true
|
||||
}
|
||||
else if prefix == nil && SAXEqualTags(localName, XMLName.author) {
|
||||
if parsingArticle {
|
||||
parsingAuthor = true
|
||||
}
|
||||
}
|
||||
|
||||
if !parsingChannelImage {
|
||||
saxParser.beginStoringCharacters()
|
||||
}
|
||||
}
|
||||
|
||||
public func saxParser(_ saxParser: SAXParser, xmlEndElement localName: XMLPointer, prefix: XMLPointer?, uri: XMLPointer?) {
|
||||
|
||||
if endRSSFound {
|
||||
return
|
||||
}
|
||||
|
||||
if isRDF && SAXEqualTags(localName, XMLName.uppercaseRDF) {
|
||||
endRSSFound = true
|
||||
}
|
||||
else if SAXEqualTags(localName, XMLName.rss) {
|
||||
endRSSFound = true
|
||||
}
|
||||
else if SAXEqualTags(localName, XMLName.image) {
|
||||
parsingChannelImage = false
|
||||
}
|
||||
else if SAXEqualTags(localName, XMLName.item) {
|
||||
parsingArticle = false
|
||||
}
|
||||
else if parsingArticle {
|
||||
addArticleElement(localName, prefix)
|
||||
if SAXEqualTags(localName, XMLName.author) {
|
||||
parsingAuthor = false
|
||||
}
|
||||
}
|
||||
else if !parsingChannelImage {
|
||||
addFeedElement(localName, prefix)
|
||||
}
|
||||
}
|
||||
|
||||
public func saxParser(_ saxParser: SAXParser, xmlCharactersFound: XMLPointer, count: Int) {
|
||||
|
||||
// Required method.
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
@ -22,11 +22,6 @@ public final class OPMLParser {
|
||||
itemStack.last
|
||||
}
|
||||
|
||||
struct XMLKey {
|
||||
static let title = "title".utf8CString
|
||||
static let outline = "outline".utf8CString
|
||||
}
|
||||
|
||||
/// Returns nil if data can’t be parsed (if it’s not OPML).
|
||||
public static func document(with parserData: ParserData) -> OPMLDocument? {
|
||||
|
||||
@ -36,7 +31,6 @@ public final class OPMLParser {
|
||||
}
|
||||
|
||||
init(_ parserData: ParserData) {
|
||||
|
||||
self.parserData = parserData
|
||||
}
|
||||
}
|
||||
@ -79,14 +73,19 @@ private extension OPMLParser {
|
||||
|
||||
extension OPMLParser: SAXParserDelegate {
|
||||
|
||||
private struct XMLName {
|
||||
static let title = "title".utf8CString
|
||||
static let outline = "outline".utf8CString
|
||||
}
|
||||
|
||||
public func saxParser(_ saxParser: SAXParser, xmlStartElement localName: XMLPointer, prefix: XMLPointer?, uri: XMLPointer?, namespaceCount: Int, namespaces: UnsafePointer<XMLPointer?>?, attributeCount: Int, attributesDefaultedCount: Int, attributes: UnsafePointer<XMLPointer?>?) {
|
||||
|
||||
if SAXEqualTags(localName, XMLKey.title) {
|
||||
if SAXEqualTags(localName, XMLName.title) {
|
||||
saxParser.beginStoringCharacters()
|
||||
return
|
||||
}
|
||||
|
||||
if !SAXEqualTags(localName, XMLKey.outline) {
|
||||
if !SAXEqualTags(localName, XMLName.outline) {
|
||||
return
|
||||
}
|
||||
|
||||
@ -99,7 +98,7 @@ extension OPMLParser: SAXParserDelegate {
|
||||
|
||||
public func saxParser(_ saxParser: SAXParser, xmlEndElement localName: XMLPointer, prefix: XMLPointer?, uri: XMLPointer?) {
|
||||
|
||||
if SAXEqualTags(localName, XMLKey.title) {
|
||||
if SAXEqualTags(localName, XMLName.title) {
|
||||
if let item = currentItem as? OPMLDocument {
|
||||
item.title = saxParser.currentStringWithTrimmedWhitespace
|
||||
}
|
||||
@ -107,7 +106,7 @@ extension OPMLParser: SAXParserDelegate {
|
||||
return
|
||||
}
|
||||
|
||||
if SAXEqualTags(localName, XMLKey.outline) {
|
||||
if SAXEqualTags(localName, XMLName.outline) {
|
||||
popItem()
|
||||
}
|
||||
}
|
||||
|
||||
@ -91,7 +91,9 @@ public final class SAXParser {
|
||||
characters.count = 0
|
||||
}
|
||||
|
||||
public func attributesDictionary(_ attributes: UnsafePointer<XMLPointer?>?, attributeCount: Int) -> [String: String]? {
|
||||
public typealias XMLAttributesDictionary = [String: String]
|
||||
|
||||
public func attributesDictionary(_ attributes: UnsafePointer<XMLPointer?>?, attributeCount: Int) -> XMLAttributesDictionary? {
|
||||
|
||||
guard attributeCount > 0, let attributes else {
|
||||
return nil
|
||||
|
||||
116
Modules/Parser/Tests/DateParserTests/DateParserTests.swift
Normal file
116
Modules/Parser/Tests/DateParserTests/DateParserTests.swift
Normal file
@ -0,0 +1,116 @@
|
||||
//
|
||||
// RSDateParserTests.swift
|
||||
//
|
||||
//
|
||||
// Created by Maurice Parker on 4/1/21.
|
||||
//
|
||||
|
||||
import Foundation
|
||||
import XCTest
|
||||
@testable import DateParser
|
||||
|
||||
class DateParserTests: XCTestCase {
|
||||
|
||||
func dateWithValues(_ year: Int, _ month: Int, _ day: Int, _ hour: Int, _ minute: Int, _ second: Int) -> Date {
|
||||
var dateComponents = DateComponents()
|
||||
dateComponents.calendar = Calendar.current
|
||||
dateComponents.timeZone = TimeZone(secondsFromGMT: 0)
|
||||
|
||||
dateComponents.year = year
|
||||
dateComponents.month = month
|
||||
dateComponents.day = day
|
||||
dateComponents.hour = hour
|
||||
dateComponents.minute = minute
|
||||
dateComponents.second = second
|
||||
|
||||
return dateComponents.date!
|
||||
}
|
||||
|
||||
func testDateWithString() {
|
||||
var expectedDateResult = dateWithValues(2010, 5, 28, 21, 3, 38)
|
||||
|
||||
var d = date("Fri, 28 May 2010 21:03:38 +0000")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
d = date("Fri, 28 May 2010 21:03:38 +00:00")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
d = date("Fri, 28 May 2010 21:03:38 -00:00")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
d = date("Fri, 28 May 2010 21:03:38 -0000")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
d = date("Fri, 28 May 2010 21:03:38 GMT")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
d = date("2010-05-28T21:03:38+00:00")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
d = date("2010-05-28T21:03:38+0000")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
d = date("2010-05-28T21:03:38-0000")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
d = date("2010-05-28T21:03:38-00:00")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
d = date("2010-05-28T21:03:38Z")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
expectedDateResult = dateWithValues(2010, 7, 13, 17, 6, 40)
|
||||
d = date("2010-07-13T17:06:40+00:00")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
expectedDateResult = dateWithValues(2010, 4, 30, 12, 0, 0)
|
||||
d = date("30 Apr 2010 5:00 PDT")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
expectedDateResult = dateWithValues(2010, 5, 21, 21, 22, 53)
|
||||
d = date("21 May 2010 21:22:53 GMT")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
expectedDateResult = dateWithValues(2010, 6, 9, 5, 0, 0)
|
||||
d = date("Wed, 09 Jun 2010 00:00 EST")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
expectedDateResult = dateWithValues(2010, 6, 23, 3, 43, 50)
|
||||
d = date("Wed, 23 Jun 2010 03:43:50 Z")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
expectedDateResult = dateWithValues(2010, 6, 22, 3, 57, 49)
|
||||
d = date("2010-06-22T03:57:49+00:00")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
expectedDateResult = dateWithValues(2010, 11, 17, 13, 40, 07)
|
||||
d = date("2010-11-17T08:40:07-05:00")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
}
|
||||
|
||||
func testAtomDateWithMissingTCharacter() {
|
||||
let expectedDateResult = dateWithValues(2010, 11, 17, 13, 40, 07)
|
||||
let d = date("2010-11-17 08:40:07-05:00")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
}
|
||||
|
||||
func testFeedbinDate() {
|
||||
let expectedDateResult = dateWithValues(2019, 9, 27, 21, 01, 48)
|
||||
let d = date("2019-09-27T21:01:48.000000Z")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
}
|
||||
|
||||
func testHighMillisecondDate() {
|
||||
let expectedDateResult = dateWithValues(2021, 03, 29, 10, 46, 56)
|
||||
let d = date("2021-03-29T10:46:56.516941+00:00")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
}
|
||||
}
|
||||
|
||||
private extension DateParserTests {
|
||||
|
||||
func date(_ string: String) -> Date? {
|
||||
let d = Data(string.utf8)
|
||||
return Date(data: d)
|
||||
}
|
||||
}
|
||||
@ -1,109 +0,0 @@
|
||||
//
|
||||
// RSDateParserTests.swift
|
||||
//
|
||||
//
|
||||
// Created by Maurice Parker on 4/1/21.
|
||||
//
|
||||
|
||||
import Foundation
|
||||
import XCTest
|
||||
import Parser
|
||||
import ParserObjC
|
||||
|
||||
class RSDateParserTests: XCTestCase {
|
||||
|
||||
static func dateWithValues(_ year: Int, _ month: Int, _ day: Int, _ hour: Int, _ minute: Int, _ second: Int) -> Date {
|
||||
var dateComponents = DateComponents()
|
||||
dateComponents.calendar = Calendar.current
|
||||
dateComponents.timeZone = TimeZone(secondsFromGMT: 0)
|
||||
|
||||
dateComponents.year = year
|
||||
dateComponents.month = month
|
||||
dateComponents.day = day
|
||||
dateComponents.hour = hour
|
||||
dateComponents.minute = minute
|
||||
dateComponents.second = second
|
||||
|
||||
return dateComponents.date!
|
||||
}
|
||||
|
||||
func testDateWithString() {
|
||||
var expectedDateResult = Self.dateWithValues(2010, 5, 28, 21, 3, 38)
|
||||
|
||||
var d = RSDateWithString("Fri, 28 May 2010 21:03:38 +0000")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
d = RSDateWithString("Fri, 28 May 2010 21:03:38 +00:00")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
d = RSDateWithString("Fri, 28 May 2010 21:03:38 -00:00")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
d = RSDateWithString("Fri, 28 May 2010 21:03:38 -0000")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
d = RSDateWithString("Fri, 28 May 2010 21:03:38 GMT")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
d = RSDateWithString("2010-05-28T21:03:38+00:00")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
d = RSDateWithString("2010-05-28T21:03:38+0000")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
d = RSDateWithString("2010-05-28T21:03:38-0000")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
d = RSDateWithString("2010-05-28T21:03:38-00:00")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
d = RSDateWithString("2010-05-28T21:03:38Z")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
expectedDateResult = Self.dateWithValues(2010, 7, 13, 17, 6, 40)
|
||||
d = RSDateWithString("2010-07-13T17:06:40+00:00")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
expectedDateResult = Self.dateWithValues(2010, 4, 30, 12, 0, 0)
|
||||
d = RSDateWithString("30 Apr 2010 5:00 PDT")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
expectedDateResult = Self.dateWithValues(2010, 5, 21, 21, 22, 53)
|
||||
d = RSDateWithString("21 May 2010 21:22:53 GMT")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
expectedDateResult = Self.dateWithValues(2010, 6, 9, 5, 0, 0)
|
||||
d = RSDateWithString("Wed, 09 Jun 2010 00:00 EST")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
expectedDateResult = Self.dateWithValues(2010, 6, 23, 3, 43, 50)
|
||||
d = RSDateWithString("Wed, 23 Jun 2010 03:43:50 Z")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
expectedDateResult = Self.dateWithValues(2010, 6, 22, 3, 57, 49)
|
||||
d = RSDateWithString("2010-06-22T03:57:49+00:00")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
|
||||
expectedDateResult = Self.dateWithValues(2010, 11, 17, 13, 40, 07)
|
||||
d = RSDateWithString("2010-11-17T08:40:07-05:00")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
}
|
||||
|
||||
func testAtomDateWithMissingTCharacter() {
|
||||
let expectedDateResult = Self.dateWithValues(2010, 11, 17, 13, 40, 07)
|
||||
let d = RSDateWithString("2010-11-17 08:40:07-05:00")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
}
|
||||
|
||||
func testFeedbinDate() {
|
||||
let expectedDateResult = Self.dateWithValues(2019, 9, 27, 21, 01, 48)
|
||||
let d = RSDateWithString("2019-09-27T21:01:48.000000Z")
|
||||
XCTAssertEqual(d, expectedDateResult)
|
||||
}
|
||||
|
||||
// func testHighMillisecondDate() {
|
||||
// let expectedDateResult = Self.dateWithValues(2021, 03, 29, 10, 46, 56)
|
||||
// let d = RSDateWithString("2021-03-29T10:46:56.516941+00:00")
|
||||
// XCTAssertEqual(d, expectedDateResult)
|
||||
// }
|
||||
}
|
||||
Loading…
x
Reference in New Issue
Block a user