1130 lines
37 KiB
PHP
1130 lines
37 KiB
PHP
<?PHP
|
|
/**
|
|
* Splits nodac times.
|
|
*
|
|
* @author Joshua Ramon Enslin <joshua@museum-digital.de>
|
|
*/
|
|
declare(strict_types = 1);
|
|
|
|
/**
|
|
* Class for splitting times.
|
|
*/
|
|
final class NodaTimeSplitter {
|
|
|
|
const MONTH_NAMES_GERMAN = [
|
|
"01" => ['Januar', 'Jan.'],
|
|
"02" => ['Februar', 'Feb'],
|
|
"03" => ['März', 'Mrz.'],
|
|
"04" => ['April', 'Apr.'],
|
|
"05" => ['Mai'],
|
|
"06" => ['Juni', 'Jun.'],
|
|
"07" => ['Juli', 'Jul.'],
|
|
"08" => ['August', 'Aug.'],
|
|
"09" => ['September', 'Sep.', 'Sept.'],
|
|
"10" => ['Oktober', 'Okt.'],
|
|
"11" => ['November', 'Nov.'],
|
|
"12" => ['Dezember', 'Dez.'],
|
|
];
|
|
|
|
const MONTH_NAMES_ENGLISH = [
|
|
"01" => ['January', 'Jan.'],
|
|
"02" => ['February', 'Feb'],
|
|
"03" => ['March', 'Mar.'],
|
|
"04" => ['April', 'Apr.'],
|
|
"05" => ['May'],
|
|
"06" => ['June', 'Jun.'],
|
|
"07" => ['July', 'Jul.'],
|
|
"08" => ['August', 'Aug.'],
|
|
"09" => ['September', 'Sep.', 'Sept.'],
|
|
"10" => ['October', 'Oct.'],
|
|
"11" => ['November', 'Nov.'],
|
|
"12" => ['December', 'Dec.'],
|
|
];
|
|
|
|
const MONTH_NAMES_HUNGARIAN = [
|
|
"01" => ['január', 'jan'],
|
|
"02" => ['február', 'feb'],
|
|
"03" => ['március', 'mar.', 'már.'],
|
|
"04" => ['április', 'apr.', 'ápr.'],
|
|
"05" => ['május', 'maj.', 'máj.'],
|
|
"06" => ['június', 'jun.', 'jún'],
|
|
"07" => ['július', 'jul.', 'júl.'],
|
|
"08" => ['augusztus', 'aug.'],
|
|
"09" => ['szeptember', 'szp.'],
|
|
"10" => ['október', 'okt.'],
|
|
"11" => ['november', 'nov.'],
|
|
"12" => ['december', 'dec.'],
|
|
];
|
|
|
|
const STRINGS_TO_CLEAN = [
|
|
"között" => "",
|
|
"nach Christus" => "",
|
|
"n. Christus" => "",
|
|
"nach Chr." => "",
|
|
"n. Chr." => "",
|
|
"n.Chr." => "",
|
|
// To clean
|
|
"v.Chr." => "v. Chr.",
|
|
"v. Chr" => "v. Chr.",
|
|
"vor Christus" => "v. Chr.",
|
|
];
|
|
|
|
const STRINGS_TO_CLEAN_START = [
|
|
"V. " => "5. ",
|
|
"IV. " => "4. ",
|
|
"III. " => "3. ",
|
|
"II. " => "2. ",
|
|
"I. " => "1. ",
|
|
];
|
|
|
|
const STOP_STRINGS_GERMAN = [
|
|
"-",
|
|
",",
|
|
";",
|
|
":",
|
|
"/",
|
|
"(", ")",
|
|
"[", "]",
|
|
", ",
|
|
" und ",
|
|
"nach ",
|
|
"um ",
|
|
"ca.",
|
|
"ab ",
|
|
"seit ",
|
|
"bis ",
|
|
"vor ",
|
|
"anfang ",
|
|
"ende ",
|
|
];
|
|
|
|
const STOP_STRINGS_HUNGARIAN = [
|
|
"-",
|
|
",",
|
|
";",
|
|
":",
|
|
"/",
|
|
"(", ")",
|
|
"[", "]",
|
|
"ca.",
|
|
", ",
|
|
"-ig",
|
|
"és",
|
|
"eleje",
|
|
"között",
|
|
"töl",
|
|
"tól",
|
|
"januárig",
|
|
"februárig",
|
|
"márciusig",
|
|
"vége",
|
|
"végén",
|
|
"áprilisig",
|
|
"májusig",
|
|
"júniusig",
|
|
"júliusig",
|
|
"augusztusig",
|
|
"szeptemberig",
|
|
"októberig",
|
|
"novemberig",
|
|
"decemberig",
|
|
];
|
|
|
|
/**
|
|
* Cleans input strings by trimming obsolete stuff.
|
|
*
|
|
* @param string $input Input date name.
|
|
*
|
|
* @return string
|
|
*/
|
|
private static function clean_input(string $input):string {
|
|
|
|
while (\strpos($input, " -") !== false) $input = \str_replace(" -", "-", $input);
|
|
while (\strpos($input, "- ") !== false) $input = \str_replace("- ", "-", $input);
|
|
$input = \strtr($input, self::STRINGS_TO_CLEAN);
|
|
foreach (self::STRINGS_TO_CLEAN_START as $toCleanFrom => $toCleanTo) {
|
|
if (strpos($input, $toCleanFrom) === 0) {
|
|
$input = \str_replace($toCleanFrom, $toCleanTo, $input);
|
|
}
|
|
}
|
|
while (strpos($input, "..") !== false) $input = \str_replace("..", ".", $input);
|
|
|
|
return trim($input, ", [](){}");
|
|
|
|
}
|
|
|
|
/**
|
|
* Checks if a string is really numeric, not numeric + space, dot.
|
|
*
|
|
* @param string $input Input string.
|
|
*
|
|
* @return boolean
|
|
*/
|
|
private static function is_numeric(string $input):bool {
|
|
|
|
if (\is_numeric($input)
|
|
and \strpos($input, " ") === false
|
|
and \strpos($input, ".") === false
|
|
) {
|
|
return true;
|
|
}
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
/**
|
|
* Validates a time substr.
|
|
*
|
|
* @param string $datum Date.
|
|
* @param integer $start Start of substr.
|
|
* @param integer $end End of substr.
|
|
*
|
|
* @return string
|
|
*/
|
|
private static function validateDateSubstr(string $datum, int $start, int $end = 10000):string {
|
|
|
|
if ($start !== 0
|
|
&& !\in_array(\substr($datum, $start - 1, 1), ["-", " ", "."], true)
|
|
) {
|
|
return "";
|
|
}
|
|
|
|
$output = \substr($datum, $start, $end);
|
|
if (self::is_numeric($output)) return $output;
|
|
return "";
|
|
|
|
}
|
|
|
|
/**
|
|
* Generates new time name based on moda.
|
|
*
|
|
* @param array<string> $moda Date strings.
|
|
*
|
|
* @return string
|
|
*/
|
|
public static function timePartsToTimeName(array $moda):string {
|
|
|
|
if (!empty($moda[5]) and $moda[5] === "Vor" and $moda[0] === "?") {
|
|
if (empty(trim($moda[2], "0 "))) $moda[1] = \strval(\intval($moda[1]) + 1);
|
|
$moda[0] = $moda[1];
|
|
$prefix = "Vor ";
|
|
}
|
|
else if (!empty($moda[5]) and $moda[5] === "Nach" and $moda[1] === "?") {
|
|
if (empty(trim($moda[2], "0 "))) $moda[0] = \strval(\intval($moda[0]) - 1);
|
|
$moda[1] = $moda[0];
|
|
$prefix = "Nach ";
|
|
}
|
|
else if ($moda[0] === "?") {
|
|
$prefix = "Bis ";
|
|
$moda[0] = $moda[1];
|
|
}
|
|
else if ($moda[1] === "?") {
|
|
$prefix = "Seit ";
|
|
$moda[1] = $moda[0];
|
|
}
|
|
else $prefix = "";
|
|
|
|
$moda[0] = \intval($moda[0]);
|
|
$moda[1] = \intval($moda[1]);
|
|
|
|
if ($moda[0] < 0 && $moda[1] < 0) {
|
|
$suffix = " v. Chr.";
|
|
}
|
|
else if ($moda[1] < 1000) {
|
|
$suffix = " n. Chr.";
|
|
}
|
|
else $suffix = "";
|
|
|
|
$moda[0] = \abs($moda[0]);
|
|
$moda[1] = \abs($moda[1]);
|
|
|
|
if ($moda[0] !== $moda[1]) {
|
|
return "{$prefix}{$moda[0]}-{$moda[1]}{$suffix}";
|
|
}
|
|
|
|
// A single day of a given month of a given (single) year
|
|
else if (\intval($moda[2]) !== 0 and \intval($moda[3]) !== 0) {
|
|
return "{$prefix}{$moda[3]}.{$moda[2]}.{$moda[0]}{$suffix}";
|
|
}
|
|
|
|
// A single year
|
|
else if ($moda[0] === $moda[1] && trim((string)$moda[2], " 0") === "" && trim((string)$moda[3], " 0") === "") {
|
|
return "{$prefix}{$moda[0]}{$suffix}";
|
|
}
|
|
|
|
// Single month of a given year
|
|
else if ($moda[0] === $moda[1] && trim((string)$moda[2], " 0") !== "" && trim((string)$moda[3], " 0") === "") {
|
|
|
|
$fmt = new IntlDateFormatter(
|
|
'de-DE',
|
|
IntlDateFormatter::FULL,
|
|
IntlDateFormatter::FULL,
|
|
null,
|
|
IntlDateFormatter::GREGORIAN,
|
|
'MMMM Y'
|
|
);
|
|
|
|
try {
|
|
return $prefix . $fmt->format(MD_STD::strtotime("{$moda[0]}-{$moda[2]}-10 01:01:01")) . $suffix;
|
|
}
|
|
catch (MDInvalidInputDate $e) {
|
|
return "";
|
|
}
|
|
}
|
|
|
|
return "";
|
|
|
|
}
|
|
|
|
/**
|
|
* Generates counting year - the middle between start and end year.
|
|
*
|
|
* @param array<string> $moda Date strings.
|
|
*
|
|
* @return integer
|
|
*/
|
|
public static function timePartsToCountingYear(array $moda):int {
|
|
|
|
if ($moda[0] === "?") {
|
|
if (!empty($moda[5]) and $moda[5] === "Vor" and empty(trim($moda[2], " 0"))) {
|
|
return \abs(\intval($moda[1])) + 1;
|
|
}
|
|
return \abs(\intval($moda[1]));
|
|
}
|
|
|
|
if ($moda[1] === "?") {
|
|
if (!empty($moda[5]) and $moda[5] === "Nach" and empty(trim($moda[2], " 0"))) {
|
|
return \abs(\intval($moda[0])) - 1;
|
|
}
|
|
return \abs(\intval($moda[0]));
|
|
}
|
|
|
|
return \abs((int)\ceil(\intval($moda[1]) - ((\intval($moda[1]) - \intval($moda[0])) / 2)));
|
|
|
|
}
|
|
|
|
/**
|
|
* Generates HTML for linking disassembly of times for a single day.
|
|
*
|
|
* @param integer $znum Time ID.
|
|
* @param array<string> $moda Date strings.
|
|
* @param MDTlLoader $tlLoader Translation loader.
|
|
*
|
|
* @return string
|
|
*/
|
|
public static function generateDisassemblyForDay(int $znum, array $moda, MDTlLoader $tlLoader):string {
|
|
|
|
$zaehlzeit_jahr = self::pad_to_four((string)self::timePartsToCountingYear($moda));
|
|
|
|
// Wenn Datum in Form von tt.mm.jjjj, dann biete zerlegen an
|
|
$output = '<hr>';
|
|
$output .= '<table>';
|
|
$output .= '<tr><td width="250px">' . $tlLoader->tl("tempi", "tempi", "time_tool") . '</td>';
|
|
$output .= '<td><a href="tempi_md/zeit_cha.php?znum=' . $znum . '&kontrolle=todo';
|
|
if (($newTimeName = self::timePartsToTimeName($moda)) !== "") {
|
|
$output .= "&zeit_name_neu={$newTimeName}";
|
|
}
|
|
$output .= '&zeit_beginn_neu=' . $moda[0] . '&zeit_ende_neu=' . $moda[1] . '&zeit_zaehlzeit_vorzeichen_neu=' . urlencode($moda[4]) . '&zeit_zaehlzeit_jahr_neu=' . $zaehlzeit_jahr . '&zeit_zaehlzeit_monat_neu=' . $moda[2] . '&zeit_zaehlzeit_tag_neu=' . $moda[3] . '&zeit_status_neu=%2B" class="icons iconsBell buttonLike" id="splitTimeLink">+';
|
|
|
|
if (!empty($newTimeName)) {
|
|
$output .= $newTimeName;
|
|
}
|
|
else {
|
|
if (!empty(trim($moda[3], " 0")) and !empty(trim($moda[2], " 0"))) $output .= $moda[3] . '.' . $moda[2] . '.' . $moda[0];
|
|
else if ($moda[0] !== $moda[1]) $output .= $moda[0] . "-" . $moda[1];
|
|
else if (!empty(trim($moda[2], " 0"))) $output .= "{$moda[2]}.{$moda[0]}";
|
|
else $output .= $moda[0];
|
|
}
|
|
$output .= ' - ' . $tlLoader->tl("tempi", "tempi", "time_disassemble") . '</a></td>';
|
|
$output .= '</tr>';
|
|
$output .= '</table>';
|
|
|
|
return $output;
|
|
|
|
}
|
|
|
|
/**
|
|
* Checks if any string of a list occurs in the haystack input string.
|
|
*
|
|
* @param string $haystack Haystack.
|
|
* @param array<string> $needles Needles.
|
|
*
|
|
* @return boolean
|
|
*/
|
|
private static function stri_occurs(string $haystack, array $needles):bool {
|
|
|
|
foreach ($needles as $needle) {
|
|
if (stripos($haystack, $needle) !== false) return true;
|
|
}
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
/**
|
|
* Pads to four digits. E.g. 20 > 0020.
|
|
*
|
|
* @param string $input Input string.
|
|
*
|
|
* @return string
|
|
*/
|
|
public static function pad_to_four(string $input):string {
|
|
|
|
return \substr("0000" . $input, -4);
|
|
|
|
}
|
|
|
|
/**
|
|
* Pads to four digits. E.g. 2 > 02.
|
|
*
|
|
* @param string $input Input string.
|
|
*
|
|
* @return string
|
|
*/
|
|
public static function pad_to_two(string $input):string {
|
|
|
|
return \substr("00" . $input, -2);
|
|
|
|
}
|
|
|
|
/**
|
|
* Translate German month to two digits number.
|
|
*
|
|
* @param string $datum Date.
|
|
*
|
|
* @return array<string>
|
|
*/
|
|
public static function is_valid_date(string $datum):array {
|
|
|
|
$datum = self::clean_input($datum);
|
|
|
|
if (\preg_match("/\ v\.\ Chr\.$/", $datum)) {
|
|
if ($output = self::attempt_splitting(\substr($datum, 0, -8))) {
|
|
$start = \strval(-1 * \intval($output[1]));
|
|
$end = \strval(-1 * \intval($output[0]));
|
|
if (\intval($start) > \intval($end)) {
|
|
$startToSet = $end;
|
|
$end = $start;
|
|
$start = $startToSet;
|
|
}
|
|
return [$start, $end, $output[2], $output[3], '-', ""];
|
|
}
|
|
}
|
|
|
|
if (\preg_match("/^[0-9][0-9][0-9][0-9]\ bis [0-9][0-9][0-9][0-9]$/", $datum)) {
|
|
$start = \substr($datum, 0, 4);
|
|
$end = \substr($datum, -4);
|
|
return [$start, $end, "00", "00", "+", ""];
|
|
}
|
|
|
|
$datum = \str_replace(". ", ".", $datum);
|
|
|
|
if (self::stri_occurs($datum, self::STOP_STRINGS_GERMAN)) {
|
|
return [];
|
|
}
|
|
|
|
if (strlen($datum) <= 6) return [];
|
|
if (strlen($datum) <= 9) $use_day = false;
|
|
else $use_day = true;
|
|
|
|
if (self::is_numeric((string)\substr($datum, -4))) $year = \substr($datum, -4);
|
|
// Further code requires a year to be present, skip if none is set
|
|
if (empty($year)) return [];
|
|
|
|
foreach (self::MONTH_NAMES_ENGLISH as $monthVal => $monthValidNames) {
|
|
if (self::stri_occurs($datum, $monthValidNames)) {
|
|
if (!empty($monat)) break;
|
|
$monat = (string)$monthVal;
|
|
}
|
|
}
|
|
|
|
foreach (self::MONTH_NAMES_GERMAN as $monthVal => $monthValidNames) {
|
|
if (self::stri_occurs($datum, $monthValidNames)) {
|
|
if (!empty($monat)) break;
|
|
$monat = (string)$monthVal;
|
|
}
|
|
}
|
|
|
|
if (empty($monat) and self::is_numeric((string)\substr($datum, 3, 2))) $monat = \substr($datum, 3, 2);
|
|
|
|
if (self::is_numeric((string)\substr($datum, 0, 2))) {
|
|
$day = \substr($datum, 0, 2);
|
|
}
|
|
else if (\in_array(\substr($datum, 1, 1), [".", " "]) && self::is_numeric((string)\substr($datum, 0, 1))) {
|
|
$day = "0" . \substr($datum, 0, 1);
|
|
}
|
|
|
|
if (!empty($year) and !empty($monat) and !empty($day) and $use_day) {
|
|
return [$year, $year, $monat, $day, '+', ""];
|
|
}
|
|
else if (!empty($year) and !empty($monat)) {
|
|
return [$year, $year, $monat, "00", '+', ""];
|
|
}
|
|
return [];
|
|
|
|
}
|
|
|
|
/**
|
|
* Translate Hungarian month to two digits number.
|
|
*
|
|
* @param string $datum Date.
|
|
*
|
|
* @return array<string>
|
|
*/
|
|
public static function is_valid_date_hungarian(string $datum):array {
|
|
|
|
$datum = self::clean_input($datum);
|
|
|
|
if (\preg_match("/^Kr\.\ e\.\ /", $datum)) {
|
|
if ($output = self::attempt_splitting(\substr($datum, 7))) {
|
|
$start = \strval(-1 * \intval($output[1]));
|
|
$end = \strval(-1 * \intval($output[0]));
|
|
if (\intval($start) > \intval($end)) {
|
|
$startToSet = $end;
|
|
$end = $start;
|
|
$start = $startToSet;
|
|
}
|
|
return [$start, $end, $output[2], $output[3], '-', ""];
|
|
}
|
|
}
|
|
|
|
if (\preg_match("/^[0-9][0-9][0-9][0-9]\-t(ő|ó)l(\ |\-)[0-9][0-9][0-9][0-9]\-ig$/", $datum)) {
|
|
$start = \substr($datum, 0, 4);
|
|
$end = \substr($datum, -7, 4);
|
|
return [$start, $end, "00", "00", "+", ""];
|
|
}
|
|
|
|
if (self::stri_occurs($datum, self::STOP_STRINGS_HUNGARIAN)) {
|
|
return [];
|
|
}
|
|
|
|
//
|
|
// Rest: Only those entries, where there are spelled out months
|
|
//
|
|
if (strlen($datum) <= 9) return [];
|
|
|
|
// The year is only parse-able if it is a four digit year at the start
|
|
if (self::is_numeric((string)\substr($datum, 0, 4))) {
|
|
$year = \substr($datum, 0, 4);
|
|
}
|
|
|
|
// Further code requires a year to be present, skip if none is set
|
|
if (empty($year)) return [];
|
|
|
|
foreach (self::MONTH_NAMES_HUNGARIAN as $monthVal => $monthValidNames) {
|
|
if (self::stri_occurs($datum, $monthValidNames)) {
|
|
if (!empty($monat)) return [];
|
|
$monat = (string)$monthVal;
|
|
}
|
|
}
|
|
|
|
if (empty($monat) and self::is_numeric((string)\substr($datum, 5, 2))) $monat = \substr($datum, 5, 2);
|
|
else if (empty($monat) and self::is_numeric((string)\substr($datum, 6, 2))) $monat = \substr($datum, 6, 2);
|
|
|
|
$day = self::validateDateSubstr($datum, -2);
|
|
if (empty($day)) $day = self::validateDateSubstr($datum, -3, 2);
|
|
if (empty($day)) $day = self::validateDateSubstr($datum, -4, 2);
|
|
if (empty($day)) $day = self::validateDateSubstr($datum, -5, 2);
|
|
if (empty($day)) $day = self::validateDateSubstr($datum, -6, 2);
|
|
|
|
if (empty($day)) {
|
|
if (\substr($datum, -2, 1) === " " and self::is_numeric((string)\substr($datum, -1, 1))) {
|
|
$day = "0" . \substr($datum, -1, 1);
|
|
}
|
|
else if (\substr($datum, -3, 1) === " " and self::is_numeric((string)\substr($datum, -2, 1))) {
|
|
$day = "0" . \substr($datum, -2, 1);
|
|
}
|
|
}
|
|
|
|
if (!empty($year) and !empty($monat) and !empty($day)) {
|
|
return [$year, $year, $monat, $day, '+', ""];
|
|
}
|
|
else if (!empty($year) and !empty($monat)) {
|
|
return [$year, $year, $monat, "00", '+', ""];
|
|
}
|
|
return [];
|
|
|
|
}
|
|
|
|
/**
|
|
* Translate German month to two digits number.
|
|
*
|
|
* @param string $datum Date.
|
|
*
|
|
* @return array<string>
|
|
*/
|
|
public static function is_valid_date_by_php(string $datum):array {
|
|
|
|
$datum = self::clean_input($datum);
|
|
|
|
if (!($timeInt = \strtotime($datum))) {
|
|
return [];
|
|
}
|
|
|
|
return [\date("Y", $timeInt), \date("m", $timeInt), \date("d", $timeInt), '+', ""];
|
|
|
|
}
|
|
|
|
/**
|
|
* Checks if an input date is a timespan.
|
|
*
|
|
* @param string $datum Input date.
|
|
*
|
|
* @return array<string>
|
|
*/
|
|
public static function is_timespan(string $datum):array {
|
|
|
|
$datum = self::clean_input($datum);
|
|
|
|
// 10000-20000
|
|
if (!empty(\preg_match("/^[0-9]{5}(\-|\/)[0-9]{5}$/", $datum))) {
|
|
$start = \substr($datum, 0, 5);
|
|
$end = \substr($datum, 6, 5);
|
|
return [$start, $end, "00", "00", "+", ""];
|
|
}
|
|
|
|
// 0000-0000
|
|
if (\preg_match("/^[0-9]{4}(\-|\/)[0-9]{4}(\.|)$/", $datum)) {
|
|
$start = \substr($datum, 0, 4);
|
|
$end = \substr($datum, 5, 4);
|
|
return [$start, $end, "00", "00", "+", ""];
|
|
}
|
|
|
|
if (\preg_match("/^[0-9]\.[0-9][0-9][0-9](\-|\/)[0-9]\.[0-9][0-9][0-9]$/", $datum)) {
|
|
$datum = \str_replace(".", "", $datum);
|
|
$start = \substr($datum, 0, 4);
|
|
$end = \substr($datum, 5, 4);
|
|
return [$start, $end, "00", "00", "+", ""];
|
|
}
|
|
|
|
// German TT.MM.JJJJ / TT.MM.JJJ / TT.MM.JJ / TT.MM.J
|
|
if (\preg_match("/^[0-9][0-9]\.[0-9][0-9]\.([0-9][0-9][0-9][0-9]|[0-9][0-9][0-9]|[0-9][0-9]|[0-9])$/", $datum)) { // German T.MM.JJJJ
|
|
$start = \substr($datum, 6, 4);
|
|
$month = \substr($datum, 3, 2);
|
|
$day = \substr($datum, 0, 2);
|
|
return [$start, $start, $month, $day, "+", ""];
|
|
}
|
|
|
|
// German TT.M.JJJJ / TT.M.JJJ / TT.M.JJ / TT.M.J
|
|
if (\preg_match("/^[0-9][0-9]\.[0-9]\.([0-9][0-9][0-9][0-9]|[0-9][0-9][0-9]|[0-9][0-9]|[0-9])$/", $datum)) { // German T.MM.JJJJ
|
|
$start = \substr($datum, 5, 4);
|
|
$month = "0" . \substr($datum, 3, 1);
|
|
$day = \substr($datum, 0, 2);
|
|
return [$start, $start, $month, $day, "+", ""];
|
|
}
|
|
|
|
// German T.MM.JJJJ / T.MM.JJJ / T.MM.JJ / T.MM.J
|
|
if (\preg_match("/^[0-9]\.[0-9][0-9]\.([0-9][0-9][0-9][0-9]|[0-9][0-9][0-9]|[0-9][0-9]|[0-9])$/", $datum)) {
|
|
$start = \substr($datum, 5, 4);
|
|
$month = \substr($datum, 2, 2);
|
|
$day = "0" . \substr($datum, 0, 1);
|
|
return [$start, $start, $month, $day, "+", ""];
|
|
}
|
|
|
|
// German T.M.JJJJ / T.M.JJJ / T.M.JJ / T.M.J
|
|
if (\preg_match("/^[0-9]\.[0-9]\.([0-9][0-9][0-9][0-9]|[0-9][0-9][0-9]|[0-9][0-9]|[0-9])$/", $datum)) {
|
|
$start = \substr($datum, 4, 4);
|
|
$month = "0" . \substr($datum, 2, 1);
|
|
$day = "0" . \substr($datum, 0, 1);
|
|
return [$start, $start, $month, $day, "+", ""];
|
|
}
|
|
|
|
// Intl': 2020-12-20
|
|
if (\preg_match("/^[0-9]{4}\-[0-9]{2}\-[0-9]{2}$/", $datum)) { // German Y-m
|
|
$start = \substr($datum, 0, 4);
|
|
$month = \substr($datum, 5, 2);
|
|
if (\intval($month) < 13) {
|
|
$day = \substr($datum, 8, 2);
|
|
return [$start, $start, $month, $day, "+", ""];
|
|
}
|
|
else {
|
|
unset($start, $month);
|
|
}
|
|
}
|
|
// Intl': 2020-12
|
|
if (\preg_match("/^[0-9]{4}\-[0-9]{2}$/", $datum)) { // German Y-m
|
|
$start = \substr($datum, 0, 4);
|
|
$month = \substr($datum, 5, 2);
|
|
if (\intval($month) < 13) {
|
|
return [$start, $start, $month, "00", "+", ""];
|
|
}
|
|
else {
|
|
unset($start, $month);
|
|
}
|
|
}
|
|
|
|
// German MM.JJJJ
|
|
if (\preg_match("/^[0-9]{2}\.[0-9]{4}$/", $datum)) { // German Y-m
|
|
$start = \substr($datum, 3, 4);
|
|
$month = \substr($datum, 0, 2);
|
|
return [$start, $start, $month, "00", "+", ""];
|
|
}
|
|
if (\preg_match("/^[0-9]\.[0-9]{4}$/", $datum)) { // German Y-m
|
|
$start = \substr($datum, 2, 4);
|
|
$month = "0" . \substr($datum, 0, 1);
|
|
return [$start, $start, $month, "00", "+", ""];
|
|
}
|
|
if (\preg_match("/^[0-9]{4}\.[0-9]{2}\.[0-9]{1,2}(\.|)$/", $datum)) { // Hungarian Y-m-d
|
|
$start = \substr($datum, 0, 4);
|
|
$month = \substr($datum, 5, 2);
|
|
$day = self::pad_to_two(\substr($datum, 8, 2));
|
|
if (\intval($month) < 13) return [$start, $start, $month, $day, "+", ""];
|
|
}
|
|
if (\preg_match("/^[0-9]{4}\.[0-9]{2}(\.|)$/", $datum)) { // Hungarian Y-m
|
|
$start = \substr($datum, 0, 4);
|
|
$month = \substr($datum, 5, 2);
|
|
if (\intval($month) < 13) return [$start, $start, $month, "00", "+", ""];
|
|
}
|
|
if (\preg_match("/^[0-9]{4}\-[0-9]{2}$/", $datum)) { // Time spans: 1945-46
|
|
$start = \substr($datum, 0, 4);
|
|
$endDigits = \substr($datum, 5, 2);
|
|
if (\intval($endDigits) > 12) return [$start, \substr($datum, 0, 2) . $endDigits, "00", "00", "+", ""];
|
|
}
|
|
|
|
// 01.01.1920-31.12.1930
|
|
if (\preg_match("/^01\.01\.[0-9]{4}\-31\.12\.[0-9]{4}$/", $datum)) { // Hungarian Y-m
|
|
$start = \substr($datum, 6, 4);
|
|
$end = \substr($datum, -4);
|
|
return [$start, $end, "00", "00", "+", ""];
|
|
}
|
|
|
|
// 303-305 (n. Chr.)
|
|
if (\preg_match("/^[0-9]{3}\-[0-9]{3}$/", $datum)) { // Hungarian Y-m
|
|
$start = \substr($datum, 0, 3);
|
|
$end = \substr($datum, -3);
|
|
return ["0" . $start, "0" . $end, "00", "00", "+", ""];
|
|
}
|
|
|
|
// 1720-120
|
|
if (\preg_match("/^[0-9]{4}\-[0-9]{3}$/", $datum)) { // Hungarian Y-m
|
|
$start = \substr($datum, 0, 4);
|
|
$end = \substr($datum, -3);
|
|
return ["0" . $start, "0" . $end, "00", "00", "+", ""];
|
|
}
|
|
|
|
// 20-30 (n. Chr.)
|
|
if (\preg_match("/^[0-9]{2}\-[0-9]{2}$/", $datum)) { // 20-40 (n. Chr.)
|
|
$start = \substr($datum, 0, 2);
|
|
$end = \substr($datum, -2);
|
|
return ["00" . $start, "00" . $end, "00", "00", "+", ""];
|
|
}
|
|
|
|
// 1920
|
|
if (\preg_match("/^[0-9]{4}(\.|)$/", $datum)) {
|
|
$start = \substr($datum, 0, 4);
|
|
return [$start, $start, "00", "00", "+", ""];
|
|
}
|
|
|
|
// 1920
|
|
if (\preg_match("/^[0-9]{3}$/", $datum)) {
|
|
$start = "0" . \substr($datum, 0, 3);
|
|
return [$start, $start, "00", "00", "+", ""];
|
|
}
|
|
|
|
if (\preg_match("/^[0-9]{2}$/", $datum)) {
|
|
$start = "00" . \substr($datum, 0, 2);
|
|
return [$start, $start, "00", "00", "+", ""];
|
|
}
|
|
|
|
// Special case for SMB: YYYY, MM. DD and YYYY, MM.
|
|
|
|
if (\preg_match("/^[0-9]{4}\,\ [0-9]{2}\.(|\ [0-9]{2})$/", $datum)) {
|
|
$start = \substr($datum, 0, 4);
|
|
$month = \substr($datum, 6, 2);
|
|
$day = self::pad_to_two(\substr($datum, 10, 2));
|
|
return [$start, $start, $month, $day, "+", ""];
|
|
}
|
|
|
|
return [];
|
|
|
|
}
|
|
|
|
/**
|
|
* Checks if an input date is an incomplete date: Before 1920, after 1930.
|
|
*
|
|
* @param string $datum Input date.
|
|
*
|
|
* @return array<string>
|
|
*/
|
|
public static function is_incomplete_date(string $datum):array {
|
|
|
|
$datum = self::clean_input($datum);
|
|
|
|
if (\preg_match("/^[0-9]{4}\.[0-9]{2}\.[0-9]{2}(\.|)\-$/", $datum)) { // Hungarian Y-m
|
|
$start = \substr($datum, 0, 4);
|
|
$month = \substr($datum, 5, 2);
|
|
$day = \substr($datum, 8, 2);
|
|
return [$start, "?", $month, $day, "+", ""];
|
|
}
|
|
if (\preg_match("/^[0-9]{4}\.[0-9]{2}(\.|)\-$/", $datum)) { // Hungarian Y-m
|
|
$start = \substr($datum, 0, 4);
|
|
$month = \substr($datum, 5, 2);
|
|
return [$start, "?", $month, "00", "+", ""];
|
|
}
|
|
if (\preg_match("/^[0-9]{4}\-$/", $datum)) { // Hungarian Y-
|
|
$start = \substr($datum, 0, 4);
|
|
return [$start, "?", "00", "00", "+", ""];
|
|
}
|
|
|
|
if (\preg_match("/^\-[0-9]{4}\.[0-9]{2}\.[0-9]{2}$/", $datum)) { // Hungarian Y-m
|
|
$start = \substr($datum, 1, 4);
|
|
$month = \substr($datum, 6, 2);
|
|
$day = \substr($datum, 9, 2);
|
|
return ["?", $start, $month, $day, "+", ""];
|
|
}
|
|
if (\preg_match("/^\-[0-9]{4}\.[0-9]{2}$/", $datum)) { // Hungarian Y-m
|
|
$start = \substr($datum, 1, 4);
|
|
$month = \substr($datum, 6, 2);
|
|
return ["?", $start, $month, "00", "+", ""];
|
|
}
|
|
if (\preg_match("/^\-[0-9]{4}$/", $datum)) { // Hungarian -Y
|
|
$start = \substr($datum, 1, 4);
|
|
return ["?", $start, "00", "00", "+", ""];
|
|
}
|
|
|
|
if (\preg_match("/^(Nach|nach)\ /", $datum)) {
|
|
if (($spacePos = \strpos($datum, " ")) === false) {
|
|
return [];
|
|
}
|
|
if ($output = self::attempt_splitting(\substr($datum, $spacePos))) {
|
|
|
|
// Handle cases like "Nach 60er Jahre" with divergent start and end year
|
|
if ($output[1] !== $output[0] and \intval($output[1]) > \intval($output[0])) {
|
|
$output[0] = $output[1];
|
|
}
|
|
|
|
$output[1] = "?";
|
|
if (empty(trim($output[2], "0 .,"))) $output[0] = \strval(\intval($output[0]) + 1);
|
|
$output[5] = "Nach";
|
|
return $output;
|
|
}
|
|
}
|
|
|
|
if (\preg_match("/\ (\(nach|nach)$/", $datum)) {
|
|
if (($spacePos = \strpos($datum, " ")) === false) {
|
|
return [];
|
|
}
|
|
if ($output = self::attempt_splitting(\substr($datum, 0, $spacePos))) {
|
|
|
|
// Handle cases like "Nach 60er Jahre" with divergent start and end year
|
|
if ($output[1] !== $output[0] and \intval($output[1]) > \intval($output[0])) {
|
|
$output[0] = $output[1];
|
|
}
|
|
|
|
$output[1] = "?";
|
|
if (empty(trim($output[2], "0 .,"))) $output[0] = \strval(\intval($output[0]) + 1);
|
|
$output[5] = "Nach";
|
|
return $output;
|
|
}
|
|
}
|
|
|
|
if (\preg_match("/^(Vor|vor)\ /", $datum)) {
|
|
if (($spacePos = \strpos($datum, " ")) === false) {
|
|
return [];
|
|
}
|
|
if ($output = self::attempt_splitting(\substr($datum, $spacePos))) {
|
|
|
|
$output[0] = "?";
|
|
if (empty(trim($output[2], "0 .,"))) $output[1] = \strval(\intval($output[1]) - 1);
|
|
$output[5] = "Vor";
|
|
return $output;
|
|
}
|
|
}
|
|
|
|
if (\preg_match("/\ (\(vor|\(Vor|vor)$/", $datum)) {
|
|
if (($spacePos = \strpos($datum, " ")) === false) {
|
|
return [];
|
|
}
|
|
if ($output = self::attempt_splitting(\substr($datum, 0, $spacePos))) {
|
|
|
|
$output[0] = "?";
|
|
if (empty(trim($output[2], "0 .,"))) $output[1] = \strval(\intval($output[1]) - 1);
|
|
$output[5] = "Vor";
|
|
return $output;
|
|
}
|
|
}
|
|
|
|
if (\preg_match("/^(Ab|Seit|seit)\ /", $datum)) {
|
|
if (($spacePos = \strpos($datum, " ")) === false) {
|
|
return [];
|
|
}
|
|
if ($output = self::attempt_splitting(\substr($datum, $spacePos))) {
|
|
$output[1] = "?";
|
|
return $output;
|
|
}
|
|
}
|
|
|
|
// Endings beginning with a space
|
|
if (\preg_match("/(\-től|\-tól)$/", $datum)) {
|
|
if (($spacePos = strrpos($datum, "-")) === false) {
|
|
return [];
|
|
}
|
|
if ($output = self::attempt_splitting(\substr($datum, 0, $spacePos))) {
|
|
$output[1] = "?";
|
|
return $output;
|
|
}
|
|
}
|
|
|
|
if (\preg_match("/^(Bis|bis)\ /", $datum)) {
|
|
if (($spacePos = \strpos($datum, " ")) === false) {
|
|
return [];
|
|
}
|
|
if ($output = self::attempt_splitting(\substr($datum, $spacePos))) {
|
|
$output[0] = "?";
|
|
return $output;
|
|
}
|
|
}
|
|
|
|
// Endings beginning with a space
|
|
if (\preg_match("/ (\(bis)$/", $datum)) {
|
|
if (($spacePos = strrpos($datum, " ")) === false) {
|
|
return [];
|
|
}
|
|
if ($output = self::attempt_splitting(\substr($datum, 0, $spacePos))) {
|
|
$output[0] = "?";
|
|
return $output;
|
|
}
|
|
}
|
|
if (\preg_match("/\-ig(\.|)$/", $datum)) {
|
|
if (($spacePos = strrpos($datum, "-")) === false) {
|
|
return [];
|
|
}
|
|
if ($output = self::attempt_splitting(\substr($datum, 0, $spacePos))) {
|
|
$output[0] = "?";
|
|
return $output;
|
|
}
|
|
}
|
|
|
|
return [];
|
|
|
|
}
|
|
|
|
/**
|
|
* Negotiates century spans before times.
|
|
*
|
|
* @param string $start Begin time.
|
|
* @param string $end End time.
|
|
*
|
|
* @return array<string>
|
|
*/
|
|
public static function negotiate_century_span_bce_ce(string $start, string $end):array {
|
|
|
|
$start = \intval($start);
|
|
$end = \intval($end);
|
|
|
|
if ($start < $end) {
|
|
return [(string)($start - 1) . "01", $end . "00", "00", "00", "+", ""];
|
|
}
|
|
|
|
return [(string)($start) . "00", ($end - 1) . "01", "00", "00", "+", ""];
|
|
|
|
}
|
|
|
|
/**
|
|
* Checks if an input date is a century.
|
|
*
|
|
* @param string $datum Input date.
|
|
*
|
|
* @return array<string>
|
|
*/
|
|
public static function is_century(string $datum):array {
|
|
|
|
$datum = self::clean_input($datum);
|
|
$bcBceIndicator = '+';
|
|
|
|
// 17. Jahrhundert
|
|
if (\preg_match("/^[0-9]{2}\.\ (Jh\.|Jhd(|\.)|Jhdt(|\.)|Jahrhundert|sz|század)$/", $datum)) {
|
|
if ($centuryNo = \intval(\substr($datum, 0, 2))) {
|
|
$centuryNo--;
|
|
return [(string)$centuryNo . "01", \strval($centuryNo + 1) . "00", "00", "00", $bcBceIndicator, ""];
|
|
}
|
|
}
|
|
|
|
// 17th century
|
|
if (\preg_match("/^[0-9]{2}th century$/", $datum)) {
|
|
if ($centuryNo = \intval(\substr($datum, 0, 2))) {
|
|
$centuryNo--;
|
|
return [(string)$centuryNo . "01", \strval($centuryNo + 1) . "00", "00", "00", $bcBceIndicator, ""];
|
|
}
|
|
}
|
|
|
|
// 1. Jahrhundert
|
|
if (\preg_match("/^[0-9]\.\ (Jh\.|Jahrhundert|sz|század)$/", $datum)) {
|
|
if ($centuryNo = \intval(\substr($datum, 0, 1))) {
|
|
$centuryNo--;
|
|
return [(string)$centuryNo . "01", \strval($centuryNo + 1) . "00", "00", "00", $bcBceIndicator, ""];
|
|
}
|
|
}
|
|
|
|
// 17.-18. Jahrhundert
|
|
if (\preg_match("/^[0-9]{2}(\.|)(|\ Jh\.||\ Jahrhundert||\ sz||\ század)(\-|\/)[0-9]{2}\.\ (Jh\.|Jahrhundert|sz|század)$/", $datum)) {
|
|
if (\strpos($datum, '/') !== false) {
|
|
$datum = str_replace('/', '-', $datum);
|
|
}
|
|
if (($dashPos = \strpos($datum, "-")) !== false) {
|
|
return self::negotiate_century_span_bce_ce(\substr($datum, 0, 2), \substr($datum, $dashPos + 1, 2));
|
|
}
|
|
}
|
|
|
|
// 1.-12. Jahrhundert
|
|
if (\preg_match("/^[0-9](\.|)(|\ Jh\.||\ Jahrhundert||\ sz||\ század)\-[0-9]{2}\.\ (Jh\.|Jahrhundert|sz|század)$/", $datum)) {
|
|
if (($dashPos = \strpos($datum, "-")) !== false) {
|
|
return self::negotiate_century_span_bce_ce(\substr($datum, 0, 1), \substr($datum, $dashPos + 1, 2));
|
|
}
|
|
}
|
|
// 1.-2. Jahrhundert
|
|
if (\preg_match("/^[0-9](\.|)(|\ Jh\.||\ Jahrhundert||\ sz||\ század)\-[0-9]\.\ (Jh\.|Jahrhundert|sz|század)$/", $datum)) {
|
|
if (($dashPos = \strpos($datum, "-")) !== false) {
|
|
return self::negotiate_century_span_bce_ce(\substr($datum, 0, 1), \substr($datum, $dashPos + 1, 1));
|
|
}
|
|
}
|
|
// 1-2. Jahrhundert
|
|
if (\preg_match("/^[0-9](\.|)(|\ Jh\.||\ Jahrhundert||\ sz||\ század)\-[0-9]\.\ (Jh\.|Jahrhundert|sz|század)$/", $datum)) {
|
|
if (($dashPos = \strpos($datum, "-")) !== false) {
|
|
return self::negotiate_century_span_bce_ce(\substr($datum, 0, 1), \substr($datum, $dashPos + 1, 1));
|
|
}
|
|
}
|
|
|
|
return [];
|
|
|
|
}
|
|
|
|
/**
|
|
* Checks if an input date is a decade.
|
|
*
|
|
* @param string $datum Input date.
|
|
*
|
|
* @return array<string>
|
|
*/
|
|
public static function is_decade(string $datum):array {
|
|
|
|
$datum = self::clean_input($datum);
|
|
$bcBceIndicator = '+';
|
|
|
|
if (\preg_match("/^[0-9]0(er|er\ Jahre|\-es\ évek|\-as\ \évek)$/", $datum)) {
|
|
$start = "19" . \substr($datum, 0, 2);
|
|
$ende = (string)(\intval($start) + 9);
|
|
return [$start, $ende, "00", "00", $bcBceIndicator, ""];
|
|
}
|
|
|
|
if (\preg_match("/^[0-9]{3}0(s|er|er\ Jahre|\-es\ évek|\-as\ \évek)$/", $datum)) {
|
|
$start = \substr($datum, 0, 4);
|
|
$ende = (string)(\intval($start) + 9);
|
|
return [$start, $ende, "00", "00", $bcBceIndicator, ""];
|
|
}
|
|
|
|
return [];
|
|
|
|
}
|
|
|
|
/**
|
|
* Checks if the string is a time span with given start and end dates.
|
|
*
|
|
* @param string $datum Date.
|
|
*
|
|
* @return array<array<string>>
|
|
*/
|
|
public static function check_is_timespan_from_till(string $datum):array {
|
|
|
|
if (substr_count($datum, '-') !== 1) return [];
|
|
|
|
list($start_str, $end_str) = explode('-', $datum);
|
|
|
|
if (empty($start = self::attempt_splitting($start_str))) {
|
|
return [];
|
|
}
|
|
|
|
if (empty($end = self::attempt_splitting($end_str))) {
|
|
return [];
|
|
}
|
|
|
|
return [$start, $end];
|
|
|
|
}
|
|
|
|
/**
|
|
* Checks if the string is a time span with given start and end dates.
|
|
*
|
|
* @param string $datum Date.
|
|
*
|
|
* @return array<string>
|
|
*/
|
|
public static function attempt_splitting_from_till(string $datum):array {
|
|
|
|
// Skip 1200-1300
|
|
if (!empty(self::attempt_splitting($datum))) return [];
|
|
if (strlen($datum) === 9 and substr($datum, 4, 1) !== '-') return [];
|
|
|
|
if (empty($startEnd = self::check_is_timespan_from_till($datum))) {
|
|
return [];
|
|
}
|
|
list($start, $end) = $startEnd;
|
|
|
|
if ($start[4] === '-') return [];
|
|
|
|
try {
|
|
$startDate = new DateTime($start[4] . $start[0] . '-' . $start[2] . '-' . $start[3]);
|
|
$endDate = new DateTime($end[4] . $end[1] . '-' . $end[2] . '-' . $end[3]);
|
|
}
|
|
catch (Exception $e) {
|
|
return [];
|
|
}
|
|
|
|
$interval = $startDate->diff($endDate);
|
|
$days_diff = (int)$interval->format('%a');
|
|
|
|
$middle_substraction = round($days_diff / 2);
|
|
|
|
if (!($startDateTimestamp = strtotime($startDate->format('Y-m-d')))) return [];
|
|
if (!($middleDayTimestamp = strtotime('+' . $middle_substraction . ' days', $startDateTimestamp))) return [];
|
|
$middle_day = date('Y-m-d', $middleDayTimestamp);
|
|
|
|
$start_name = self::timePartsToTimeName($start);
|
|
$end_name = self::timePartsToTimeName($end);
|
|
|
|
if (strlen($start_name) === 9 and substr($start_name, 4, 1) === '-') $start_name = substr($start_name, 0, 4);
|
|
if (strlen($end_name) === 9 and substr($end_name, 4, 1) === '-') $end_name = substr($end_name, 5, 4);
|
|
|
|
$output = [
|
|
"start_name" => $start_name,
|
|
"end_name" => $end_name,
|
|
"start_year" => $start[0],
|
|
"end_year" => $end[1],
|
|
"counting_time_year" => substr($middle_day, 0, 4),
|
|
"counting_time_month" => substr($middle_day, 5, 2),
|
|
"counting_time_day" => substr($middle_day, 8, 2),
|
|
"counting_time_bcce" => "+",
|
|
];
|
|
|
|
return $output;
|
|
|
|
}
|
|
|
|
/**
|
|
* Wrapper to check if any splitting command works.
|
|
*
|
|
* @param string $datum Input date.
|
|
*
|
|
* @return array<string>
|
|
*/
|
|
public static function attempt_splitting(string $datum):array {
|
|
|
|
$moda = NodaTimeSplitter::is_timespan($datum);
|
|
if (!$moda) $moda = NodaTimeSplitter::is_incomplete_date($datum);
|
|
if (!$moda) $moda = NodaTimeSplitter::is_valid_date($datum);
|
|
if (!$moda) $moda = NodaTimeSplitter::is_valid_date_hungarian($datum);
|
|
if (!$moda) $moda = NodaTimeSplitter::is_century($datum);
|
|
if (!$moda) $moda = NodaTimeSplitter::is_decade($datum);
|
|
|
|
if (!empty($moda)) {
|
|
if ((int)$moda[2] > 12 || (int)$moda[3] > 31) {
|
|
return [];
|
|
}
|
|
}
|
|
|
|
return $moda;
|
|
|
|
}
|
|
}
|