// Package icalfix normalizes non-standard timezone identifiers in raw
// iCalendar data before it's handed to go-ical's decoder.
//
// go-ical resolves a DTSTART/DTEND's TZID parameter (and, for recurring
// events, the RRULE expansion's start time) via a direct
// time.LoadLocation(tzid) call — see
// github.com/emersion/go-ical@.../ical.go's Prop.DateTime and
// components.go's Component.RecurrenceSet. That only understands IANA
// zone names ("Europe/Berlin"), but Windows/Outlook (and anything built
// on the Windows timezone database, e.g. Exchange, Thunderbird's Lightning
// on some setups) commonly emits Windows zone names instead
// ("W. Europe Standard Time"), which time.LoadLocation doesn't recognize.
// A recurring event stored with such a TZID decodes fine (go-ical doesn't
// eagerly parse RRULE dates), but fails later — as a 500 — the moment a
// CalDAV client issues a time-range REPORT that has to expand its
// recurrence, since that's when RecurrenceSet actually calls
// time.LoadLocation.
package icalfix

import "regexp"

// tzidParamRe matches a "TZID=<name>" (optionally quoted) parameter value
// as it appears attached to any property (e.g. "DTSTART;TZID=W. Europe
// Standard Time:..."). The name runs until the next ':' or ';' delimiter
// or line break, so it captures the parameter's full value even though it
// contains spaces and periods.
var tzidParamRe = regexp.MustCompile(`TZID=("?)([^";:\r\n]+)("?)`)

// tzidLineRe matches a standalone "TZID:<name>" property line, as found
// inside a VTIMEZONE component. The trailing "(\r?)" capture group is
// required, not cosmetic: Go's RE2 engine only matches "$" in multi-line
// mode immediately before a literal "\n", not before "\r\n" — so without
// consuming (and re-emitting) an optional trailing "\r" explicitly, this
// regex silently fails to match on any CRLF-terminated feed (which is
// every Outlook/Exchange-published ICS feed in practice), leaving the
// VTIMEZONE's own TZID line un-normalized even though the "TZID=..."
// parameter form (tzidParamRe, used on DTSTART/DTEND/etc.) still matches
// fine. The practical effect was a VTIMEZONE whose declared TZID
// ("W. Europe Standard Time") never matched the now-normalized
// "TZID=Europe/Berlin" parameters referencing it elsewhere in the same
// object, so any code trying to find "the VTIMEZONE for Europe/Berlin"
// (see internal/icssub.Series.referencedTimezones) would never find one.
var tzidLineRe = regexp.MustCompile(`(?m)^TZID:([^\r\n]+)(\r?)$`)

// NormalizeTimeZones rewrites any recognized Windows timezone identifier
// in data to its IANA equivalent, leaving everything else (including
// unrecognized TZIDs) untouched. It's safe to call on data that has no
// Windows TZIDs at all — such data passes through unchanged.
func NormalizeTimeZones(data []byte) []byte {
	data = tzidParamRe.ReplaceAllFunc(data, func(m []byte) []byte {
		sub := tzidParamRe.FindSubmatch(m)
		name := string(sub[2])
		iana, ok := windowsToIANA[name]
		if !ok {
			return m
		}
		return []byte("TZID=" + string(sub[1]) + iana + string(sub[3]))
	})
	data = tzidLineRe.ReplaceAllFunc(data, func(m []byte) []byte {
		sub := tzidLineRe.FindSubmatch(m)
		name := string(sub[1])
		iana, ok := windowsToIANA[name]
		if !ok {
			return m
		}
		// sub[2] is the optional trailing "\r" the regex consumed as part
		// of the match (see the doc comment above) — it must be re-emitted
		// here, or a CRLF-terminated line loses its "\r" and every
		// subsequent line in the file ends up misaligned relative to the
		// original byte offsets a caller might have recorded.
		return []byte("TZID:" + iana + string(sub[2]))
	})
	return data
}

// windowsToIANA maps common Windows timezone identifiers (as found in the
// CLDR windowsZones.xml "001"/default territory mapping) to an IANA zone
// name recognized by Go's tzdata. It's not exhaustive, but covers the
// zones seen in practice from Outlook/Exchange-generated ICS data.
var windowsToIANA = map[string]string{
	"Dateline Standard Time":          "Etc/GMT+12",
	"UTC-11":                          "Etc/GMT+11",
	"Aleutian Standard Time":          "America/Adak",
	"Hawaiian Standard Time":          "Pacific/Honolulu",
	"Marquesas Standard Time":         "Pacific/Marquesas",
	"Alaskan Standard Time":           "America/Anchorage",
	"UTC-09":                          "Etc/GMT+9",
	"Pacific Standard Time (Mexico)":  "America/Tijuana",
	"UTC-08":                          "Etc/GMT+8",
	"Pacific Standard Time":           "America/Los_Angeles",
	"US Mountain Standard Time":       "America/Phoenix",
	"Mountain Standard Time (Mexico)": "America/Chihuahua",
	"Mountain Standard Time":          "America/Denver",
	"Central America Standard Time":   "America/Guatemala",
	"Central Standard Time":           "America/Chicago",
	"Central Standard Time (Mexico)":  "America/Mexico_City",
	"Canada Central Standard Time":    "America/Regina",
	"SA Pacific Standard Time":        "America/Bogota",
	"Eastern Standard Time (Mexico)":  "America/Cancun",
	"Eastern Standard Time":           "America/New_York",
	"Haiti Standard Time":             "America/Port-au-Prince",
	"Cuba Standard Time":              "America/Havana",
	"US Eastern Standard Time":        "America/Indianapolis",
	"Turks And Caicos Standard Time":  "America/Grand_Turk",
	"Paraguay Standard Time":          "America/Asuncion",
	"Atlantic Standard Time":          "America/Halifax",
	"Venezuela Standard Time":         "America/Caracas",
	"Central Brazilian Standard Time": "America/Cuiaba",
	"SA Western Standard Time":        "America/La_Paz",
	"Pacific SA Standard Time":        "America/Santiago",
	"Newfoundland Standard Time":      "America/St_Johns",
	"Tocantins Standard Time":         "America/Araguaina",
	"E. South America Standard Time":  "America/Sao_Paulo",
	"SA Eastern Standard Time":        "America/Cayenne",
	"Argentina Standard Time":         "America/Buenos_Aires",
	"Greenland Standard Time":         "America/Godthab",
	"Montevideo Standard Time":        "America/Montevideo",
	"Magallanes Standard Time":        "America/Punta_Arenas",
	"Saint Pierre Standard Time":      "America/Miquelon",
	"Bahia Standard Time":             "America/Bahia",
	"UTC-02":                          "Etc/GMT+2",
	"Azores Standard Time":            "Atlantic/Azores",
	"Cape Verde Standard Time":        "Atlantic/Cape_Verde",
	"UTC":                             "Etc/UTC",
	"GMT Standard Time":               "Europe/London",
	"Greenwich Standard Time":         "Atlantic/Reykjavik",
	"Sao Tome Standard Time":          "Africa/Sao_Tome",
	"Morocco Standard Time":           "Africa/Casablanca",
	"W. Europe Standard Time":         "Europe/Berlin",
	"Central Europe Standard Time":    "Europe/Budapest",
	"Romance Standard Time":           "Europe/Paris",
	"Central European Standard Time":  "Europe/Warsaw",
	"W. Central Africa Standard Time": "Africa/Lagos",
	"Jordan Standard Time":            "Asia/Amman",
	"GTB Standard Time":               "Europe/Bucharest",
	"Middle East Standard Time":       "Asia/Beirut",
	"Egypt Standard Time":             "Africa/Cairo",
	"E. Europe Standard Time":         "Europe/Chisinau",
	"Syria Standard Time":             "Asia/Damascus",
	"West Bank Standard Time":         "Asia/Hebron",
	"South Africa Standard Time":      "Africa/Johannesburg",
	"FLE Standard Time":               "Europe/Kiev",
	"Israel Standard Time":            "Asia/Jerusalem",
	"Kaliningrad Standard Time":       "Europe/Kaliningrad",
	"Sudan Standard Time":             "Africa/Khartoum",
	"Libya Standard Time":             "Africa/Tripoli",
	"Namibia Standard Time":           "Africa/Windhoek",
	"Arabic Standard Time":            "Asia/Baghdad",
	"Turkey Standard Time":            "Europe/Istanbul",
	"Arab Standard Time":              "Asia/Riyadh",
	"Belarus Standard Time":           "Europe/Minsk",
	"Russian Standard Time":           "Europe/Moscow",
	"E. Africa Standard Time":         "Africa/Nairobi",
	"Volgograd Standard Time":         "Europe/Volgograd",
	"Iran Standard Time":              "Asia/Tehran",
	"Arabian Standard Time":           "Asia/Dubai",
	"Astrakhan Standard Time":         "Europe/Astrakhan",
	"Azerbaijan Standard Time":        "Asia/Baku",
	"Russia Time Zone 3":              "Europe/Samara",
	"Mauritius Standard Time":         "Indian/Mauritius",
	"Saratov Standard Time":           "Europe/Saratov",
	"Georgian Standard Time":          "Asia/Tbilisi",
	"Caucasus Standard Time":          "Asia/Yerevan",
	"Afghanistan Standard Time":       "Asia/Kabul",
	"West Asia Standard Time":         "Asia/Tashkent",
	"Ekaterinburg Standard Time":      "Asia/Yekaterinburg",
	"Pakistan Standard Time":          "Asia/Karachi",
	"Qyzylorda Standard Time":         "Asia/Qyzylorda",
	"India Standard Time":             "Asia/Calcutta",
	"Sri Lanka Standard Time":         "Asia/Colombo",
	"Nepal Standard Time":             "Asia/Katmandu",
	"Central Asia Standard Time":      "Asia/Almaty",
	"Bangladesh Standard Time":        "Asia/Dhaka",
	"Omsk Standard Time":              "Asia/Omsk",
	"Myanmar Standard Time":           "Asia/Rangoon",
	"SE Asia Standard Time":           "Asia/Bangkok",
	"Altai Standard Time":             "Asia/Barnaul",
	"W. Mongolia Standard Time":       "Asia/Hovd",
	"North Asia Standard Time":        "Asia/Krasnoyarsk",
	"N. Central Asia Standard Time":   "Asia/Novosibirsk",
	"Tomsk Standard Time":             "Asia/Tomsk",
	"China Standard Time":             "Asia/Shanghai",
	"North Asia East Standard Time":   "Asia/Irkutsk",
	"Singapore Standard Time":         "Asia/Singapore",
	"W. Australia Standard Time":      "Australia/Perth",
	"Taipei Standard Time":            "Asia/Taipei",
	"Ulaanbaatar Standard Time":       "Asia/Ulaanbaatar",
	"Aus Central W. Standard Time":    "Australia/Eucla",
	"Transbaikal Standard Time":       "Asia/Chita",
	"Tokyo Standard Time":             "Asia/Tokyo",
	"North Korea Standard Time":       "Asia/Pyongyang",
	"Korea Standard Time":             "Asia/Seoul",
	"Yakutsk Standard Time":           "Asia/Yakutsk",
	"Cen. Australia Standard Time":    "Australia/Adelaide",
	"AUS Central Standard Time":       "Australia/Darwin",
	"E. Australia Standard Time":      "Australia/Brisbane",
	"AUS Eastern Standard Time":       "Australia/Sydney",
	"West Pacific Standard Time":      "Pacific/Port_Moresby",
	"Tasmania Standard Time":          "Australia/Hobart",
	"Vladivostok Standard Time":       "Asia/Vladivostok",
	"Lord Howe Standard Time":         "Australia/Lord_Howe",
	"Bougainville Standard Time":      "Pacific/Bougainville",
	"Russia Time Zone 10":             "Asia/Srednekolymsk",
	"Magadan Standard Time":           "Asia/Magadan",
	"Norfolk Standard Time":           "Pacific/Norfolk",
	"Sakhalin Standard Time":          "Asia/Sakhalin",
	"Central Pacific Standard Time":   "Pacific/Guadalcanal",
	"Russia Time Zone 11":             "Asia/Kamchatka",
	"New Zealand Standard Time":       "Pacific/Auckland",
	"UTC+12":                          "Etc/GMT-12",
	"Fiji Standard Time":              "Pacific/Fiji",
	"Kamchatka Standard Time":         "Asia/Kamchatka",
	"Chatham Islands Standard Time":   "Pacific/Chatham",
	"UTC+13":                          "Etc/GMT-13",
	"Tonga Standard Time":             "Pacific/Tongatapu",
	"Samoa Standard Time":             "Pacific/Apia",
	"Line Islands Standard Time":      "Pacific/Kiritimati",
}