Rewrite incomplete time span spellings to extend parsable and splittable time names

This commit is contained in:
Joshua Ramon Enslin 2023-11-20 03:18:02 +01:00
parent 78d5137b96
commit 4496a35f5c
Signed by: jrenslin
GPG Key ID: 46016F84501B70AE
5 changed files with 458 additions and 61 deletions

36
phpstan-baseline.neon Normal file
View File

@ -0,0 +1,36 @@
parameters:
ignoreErrors:
-
message: "#^Constant DATABASENAME_NODA not found\\.$#"
count: 3
path: src/NodaBlacklistedTerms.php
-
message: "#^Constant DATABASENAME_NODA not found\\.$#"
count: 2
path: src/NodaMailChecker.php
-
message: "#^Variable \\$timeInfoToCopy in empty\\(\\) always exists and is not falsy\\.$#"
count: 1
path: src/NodaTimeAutotranslater.php
-
message: "#^Call to an undefined method DOMNode\\:\\:getAttribute\\(\\)\\.$#"
count: 1
path: src/NodaWikidataFetcher.php
-
message: "#^Function printHTMLEnd not found\\.$#"
count: 1
path: src/NodaWikidataFetcher.php
-
message: "#^Function write_get_vars not found\\.$#"
count: 9
path: src/NodaWikidataFetcher.php
-
message: "#^Match expression does not handle remaining value\\: string$#"
count: 1
path: src/enums/NodaTimeAutotranslaterLocales.php

View File

@ -8,3 +8,5 @@ parameters:
- ../ - ../
ignoreErrors: ignoreErrors:
excludePaths: excludePaths:
includes:
- phpstan-baseline.neon

View File

@ -475,7 +475,9 @@ final class NodaTimeAutotranslater {
$output = []; $output = [];
$start = NodaTimeSplitter::attempt_splitting($timespanDates['start_name']); if (empty($start = NodaTimeSplitter::attempt_splitting($timespanDates['start_name']))) {
return [];
}
$startTimeInfo = [ $startTimeInfo = [
"zeit_name" => $timespanDates['start_name'], "zeit_name" => $timespanDates['start_name'],
"zeit_beginn" => $start[0], "zeit_beginn" => $start[0],
@ -486,7 +488,9 @@ final class NodaTimeAutotranslater {
"zeit_zaehlzeit_vorzeichen" => $start[4], "zeit_zaehlzeit_vorzeichen" => $start[4],
]; ];
$end = NodaTimeSplitter::attempt_splitting($timespanDates['end_name']); if (empty($end = NodaTimeSplitter::attempt_splitting($timespanDates['end_name']))) {
return [];
}
$endTimeInfo = [ $endTimeInfo = [
"zeit_name" => $timespanDates['end_name'], "zeit_name" => $timespanDates['end_name'],
"zeit_beginn" => $end[0], "zeit_beginn" => $end[0],

View File

@ -48,7 +48,7 @@ final class NodaTimeSplitter {
"04" => ['április', 'apr.', 'ápr.'], "04" => ['április', 'apr.', 'ápr.'],
"05" => ['május', 'maj.', 'máj.'], "05" => ['május', 'maj.', 'máj.'],
"06" => ['június', 'jun.', 'jún'], "06" => ['június', 'jun.', 'jún'],
"07" => ['július', 'jul.', 'júl.'], "07" => ['július', 'julius', 'jul.', 'júl.'],
"08" => ['augusztus', 'aug.'], "08" => ['augusztus', 'aug.'],
"09" => ['szeptember', 'szp.'], "09" => ['szeptember', 'szp.'],
"10" => ['október', 'okt.'], "10" => ['október', 'okt.'],
@ -132,6 +132,8 @@ final class NodaTimeSplitter {
"decemberig", "decemberig",
]; ];
private const REGEX_CENTURIES = '(\ |)(Jh\.|Jhd(|\.)|Jhdt(|\.)|Jahrhundert|sz|század|th century|ст|ст\.)';
/** /**
* Cleans input strings by trimming obsolete stuff. * Cleans input strings by trimming obsolete stuff.
* *
@ -395,7 +397,7 @@ final class NodaTimeSplitter {
* *
* @param string $datum Date. * @param string $datum Date.
* *
* @return array<string> * @return array{0: string, 1: string, 2: string, 3: string, 4: '+'|'-'|'', 5: string}|array{}
*/ */
public static function is_valid_date(string $datum):array { public static function is_valid_date(string $datum):array {
@ -472,7 +474,7 @@ final class NodaTimeSplitter {
* *
* @param string $datum Date. * @param string $datum Date.
* *
* @return array<string> * @return array{0: string, 1: string, 2: string, 3: string, 4: '+'|'-'|'', 5: string}|array{}
*/ */
public static function is_valid_date_hungarian(string $datum):array { public static function is_valid_date_hungarian(string $datum):array {
@ -491,6 +493,8 @@ final class NodaTimeSplitter {
} }
} }
// Example: 2009-tol 2010-ig
// From 2009 to 2010
if (\preg_match("/^[0-9][0-9][0-9][0-9]\-t(ő|ó)l(\ |\-)[0-9][0-9][0-9][0-9]\-ig$/", $datum)) { if (\preg_match("/^[0-9][0-9][0-9][0-9]\-t(ő|ó)l(\ |\-)[0-9][0-9][0-9][0-9]\-ig$/", $datum)) {
$start = \substr($datum, 0, 4); $start = \substr($datum, 0, 4);
$end = \substr($datum, -7, 4); $end = \substr($datum, -7, 4);
@ -576,7 +580,7 @@ final class NodaTimeSplitter {
* *
* @param string $datum Input date. * @param string $datum Input date.
* *
* @return array<string> * @return array{0: string, 1: string, 2: string, 3: string, 4: '+'|'-'|'', 5: string}|array{}
*/ */
public static function is_timespan(string $datum):array { public static function is_timespan(string $datum):array {
@ -670,13 +674,19 @@ final class NodaTimeSplitter {
$month = "0" . \substr($datum, 0, 1); $month = "0" . \substr($datum, 0, 1);
return [$start, $start, $month, "00", "+", ""]; return [$start, $start, $month, "00", "+", ""];
} }
if (\preg_match("/^[0-9]{4}\.[0-9]{2}\.[0-9]{1,2}(\.|)$/", $datum)) { // Hungarian Y-m-d if (\preg_match("/^[0-9]{4}\.[0-3][0-9]\.[0-9]{1,2}(\.|)$/", $datum)) { // Hungarian Y-m-d
$start = \substr($datum, 0, 4); $start = \substr($datum, 0, 4);
$month = \substr($datum, 5, 2); $month = \substr($datum, 5, 2);
$day = self::pad_to_two(\substr($datum, 8, 2)); $day = self::pad_to_two(\rtrim(\substr($datum, 8, 2), '.'));
if (\intval($month) < 13) return [$start, $start, $month, $day, "+", ""]; if (\intval($month) < 13) return [$start, $start, $month, $day, "+", ""];
} }
if (\preg_match("/^[0-9]{4}\.[0-9]{2}(\.|)$/", $datum)) { // Hungarian Y-m if (\preg_match("/^[0-9]{4}\.[0-9]\.[0-9]{1,2}\.$/", $datum)) { // Hungarian Y-m-d > 2005.1.1.
$start = \substr($datum, 0, 4);
$month = self::pad_to_two(\substr($datum, 5, 1));
$day = self::pad_to_two(\rtrim(\substr($datum, 7, 2), '.'));
if (\intval($month) < 13) return [$start, $start, $month, $day, "+", ""];
}
if (\preg_match("/^[0-9]{4}\.[0-3][0-9](\.|)$/", $datum)) { // Hungarian Y-m
$start = \substr($datum, 0, 4); $start = \substr($datum, 0, 4);
$month = \substr($datum, 5, 2); $month = \substr($datum, 5, 2);
if (\intval($month) < 13) return [$start, $start, $month, "00", "+", ""]; if (\intval($month) < 13) return [$start, $start, $month, "00", "+", ""];
@ -750,7 +760,7 @@ final class NodaTimeSplitter {
* *
* @param string $datum Input date. * @param string $datum Input date.
* *
* @return array<string> * @return array{0: string, 1: string, 2: string, 3: string, 4: '+'|'-'|'', 5: string}|array{}
*/ */
public static function is_incomplete_date(string $datum):array { public static function is_incomplete_date(string $datum):array {
@ -860,7 +870,7 @@ final class NodaTimeSplitter {
} }
} }
// Endings beginning with a space // Endings beginning with a dash
if (\preg_match("/(\-től|\-tól)$/", $datum)) { if (\preg_match("/(\-től|\-tól)$/", $datum)) {
if (($spacePos = strrpos($datum, "-")) === false) { if (($spacePos = strrpos($datum, "-")) === false) {
return []; return [];
@ -880,8 +890,25 @@ final class NodaTimeSplitter {
return $output; return $output;
} }
} }
// Endings that are extensions of an existing word
if (\preg_match("/évektől$/", $datum)) {
if ($output = self::attempt_splitting(\substr($datum, 0, -4))) {
$output[1] = "?";
return $output;
}
}
// Endings beginning with a space // Endings beginning with a space (after)
if (\preg_match("/ (utantól|utántól)$/", $datum)) {
if (($spacePos = strrpos($datum, " ")) === false) {
return [];
}
if ($output = self::attempt_splitting(\substr($datum, 0, $spacePos))) {
$output[1] = "?";
return $output;
}
}
// Endings beginning with a space (until)
if (\preg_match("/ (\(bis)$/", $datum)) { if (\preg_match("/ (\(bis)$/", $datum)) {
if (($spacePos = strrpos($datum, " ")) === false) { if (($spacePos = strrpos($datum, " ")) === false) {
return []; return [];
@ -891,6 +918,7 @@ final class NodaTimeSplitter {
return $output; return $output;
} }
} }
// Ends beginning with a hyphen
if (\preg_match("/\-ig(\.|)$/", $datum)) { if (\preg_match("/\-ig(\.|)$/", $datum)) {
if (($spacePos = strrpos($datum, "-")) === false) { if (($spacePos = strrpos($datum, "-")) === false) {
return []; return [];
@ -931,23 +959,18 @@ final class NodaTimeSplitter {
* *
* @param string $datum Input date. * @param string $datum Input date.
* *
* @return array<string> * @return array{0: string, 1: string, 2: string, 3: string, 4: '+'|'-'|'', 5: string}|array{}
*/ */
public static function is_century(string $datum):array { public static function is_century(string $datum):array {
$datum = self::clean_input($datum); $datum = self::clean_input($datum);
$bcBceIndicator = '+'; $bcBceIndicator = '+';
// 17. Jahrhundert
if (\preg_match("/^[0-9]{2}\.\ (Jh\.|Jhd(|\.)|Jhdt(|\.)|Jahrhundert|sz|század)$/", $datum)) {
if ($centuryNo = \intval(\substr($datum, 0, 2))) {
$centuryNo--;
return [(string)$centuryNo . "01", \strval($centuryNo + 1) . "00", "00", "00", $bcBceIndicator, ""];
}
}
// 17th century // TODO: Check if this is duplicate
if (\preg_match("/^[0-9]{2}th century$/", $datum)) {
// 17. Jahrhundert
if (\preg_match("/^[0-9]{2}(\.|)" . self::REGEX_CENTURIES ."$/", $datum)) {
if ($centuryNo = \intval(\substr($datum, 0, 2))) { if ($centuryNo = \intval(\substr($datum, 0, 2))) {
$centuryNo--; $centuryNo--;
return [(string)$centuryNo . "01", \strval($centuryNo + 1) . "00", "00", "00", $bcBceIndicator, ""]; return [(string)$centuryNo . "01", \strval($centuryNo + 1) . "00", "00", "00", $bcBceIndicator, ""];
@ -1000,20 +1023,22 @@ final class NodaTimeSplitter {
* *
* @param string $datum Input date. * @param string $datum Input date.
* *
* @return array<string> * @return array{0: string, 1: string, 2: string, 3: string, 4: '+'|'-'|'', 5: string}|array{}
*/ */
public static function is_decade(string $datum):array { public static function is_decade(string $datum):array {
$datum = self::clean_input($datum); $datum = self::clean_input($datum);
$bcBceIndicator = '+'; $bcBceIndicator = '+';
// 20er Jahre
if (\preg_match("/^[0-9]0(er|er\ Jahre|\-es\ évek|\-as\ \évek)$/", $datum)) { if (\preg_match("/^[0-9]0(er|er\ Jahre|\-es\ évek|\-as\ \évek)$/", $datum)) {
$start = "19" . \substr($datum, 0, 2); $start = "19" . \substr($datum, 0, 2);
$ende = (string)(\intval($start) + 9); $ende = (string)(\intval($start) + 9);
return [$start, $ende, "00", "00", $bcBceIndicator, ""]; return [$start, $ende, "00", "00", $bcBceIndicator, ""];
} }
if (\preg_match("/^[0-9]{3}0(s|er|er\ Jahre|\-es\ évek|\-as\ \évek)$/", $datum)) { // 1920er Jahre
if (\preg_match("/^[0-9]{3}0(s|er|er\ Jahre|(\-|\ )es\ évek|(\-|\ )as\ \évek)$/", $datum)) {
$start = \substr($datum, 0, 4); $start = \substr($datum, 0, 4);
$ende = (string)(\intval($start) + 9); $ende = (string)(\intval($start) + 9);
return [$start, $ende, "00", "00", $bcBceIndicator, ""]; return [$start, $ende, "00", "00", $bcBceIndicator, ""];
@ -1032,10 +1057,19 @@ final class NodaTimeSplitter {
*/ */
public static function check_is_timespan_from_till(string $datum):array { public static function check_is_timespan_from_till(string $datum):array {
if (substr_count($datum, '-') !== 1) return []; if (substr_count($datum, '-') !== 1) {
return [];
}
list($start_str, $end_str) = explode('-', $datum); list($start_str, $end_str) = explode('-', $datum);
if (strlen($end_str) < 4 && strlen($end_str) < strlen($start_str)) {
return [];
}
if (strlen($start_str) < 4 && strlen($start_str) < strlen($end_str)) {
return [];
}
if (empty($start = self::attempt_splitting($start_str))) { if (empty($start = self::attempt_splitting($start_str))) {
return []; return [];
} }
@ -1048,6 +1082,107 @@ final class NodaTimeSplitter {
} }
/**
* Contains special rules for incorrectly or incompletely spelled out timespan names.
* To be called by self::attempt_splitting_from_till().
*
* @param string $datum Date.
*
* @return string
*/
public static function _attempt_rewriting_special_cases_from_till(string $datum):string {
if (empty($datum)) return '';
$inputLength = strlen($datum);
// Hungarian year and month until month
// 2005.01.-02. => 2005.01.-2005.02.
if ($inputLength === 12 && \preg_match("/^[0-9]{4}\.[0-1][0-9]\.\-[0-1][0-9]\.$/", $datum)) {
$reconstituted = substr($datum, 0, 8) . '-';
$reconstituted .= substr($datum, 0, 4) . '.' . substr($datum, -3);
return $reconstituted;
}
// Hungarian year and month until month without a dot after the first YYYY-MM
// 2005.01-02. => 2005.01.-2005.02.
if (in_array($inputLength, [10, 11], true) && \preg_match("/^[0-9]{4}\.[0-1][0-9]\-[0-1][0-9](\.|)$/", $datum)) {
$reconstituted = substr($datum, 0, 7) . '.-';
$reconstituted .= substr($datum, 0, 4) . '.' . substr(rtrim($datum, '.'), -2) . '.';
return $reconstituted;
}
// Hungarian year and month until month
// 2005.01.01.-02.02. => 2005.01.01-2005.02.02.
// 2005.01.01-02.02 => 2005.01.01-2005.02.02.
if ($inputLength >= 16 && $inputLength <= 18 && \preg_match("/^[0-9]{4}\.[0-1][0-9]\.[0-3][0-9](\.|)\-[0-1][0-9]\.[0-3][0-9](\.|)$/", $datum)) {
$parts = explode('-', $datum);
if (count($parts) !== 2) return '';
$reconstituted = substr($datum, 0, 10) . '.-';
$reconstituted .= substr($datum, 0, 4) . '.' . rtrim($parts[1], '.') . '.';
return $reconstituted;
}
// Hungarian; without trailing dots: YYYY.MM.DD-DD
if ($inputLength >= 13 && $inputLength <= 15 && \preg_match("/^[0-9]{4}\.[0-1][0-9]\.[0-3][0-9](\.|)\-[0-3][0-9](\.|)$/", $datum)) {
$parts = explode('-', $datum);
if (count($parts) !== 2) return '';
$reconstituted = substr($datum, 0, 10) . '.-';
$reconstituted .= substr($datum, 0, 7) . '.' . substr(rtrim($parts[1], '.'), -2);
return $reconstituted;
}
// 17-19. Jahrhundert
if (\preg_match("/^[0-9]{2}(\.|)\-[0-9]{2}(\.|)" . self::REGEX_CENTURIES . "$/", $datum)) {
$parts = explode('-', $datum);
$reconstituted = ((int)substr($parts[0] ?? "", 0, 2) - 1) . '01-';
$reconstituted .= substr($parts[1] ?? "", 0, 2) . '. Jahrhundert';
return $reconstituted;
}
// 1950-60-as évek
if (\preg_match("/^[0-9]{4}\-[0-9]{2} (a|e)s évek$/", $datum)) {
$reconstituted = substr($datum, 0, 4) . '-';
$reconstituted .= substr($datum, 5, 2) . 'er Jahre';
return $reconstituted;
}
// If es évek / as évek is contained in the string (e.g. 1880-1990-es évek), there
// will be more than one hyphens
if (MD_STD::stri_contains_any($datum, ['-as évek', '-es évek'])) {
return strtr($datum, ['-as évek' => ' as évek', '-es évek' => ' es évek']);
}
// 1981. július-augusztus > 1981.07-08
if (is_numeric(substr($datum, 0, 4)) && substr($datum, 4, 2) === '. ') {
$monthNames = [];
foreach (self::MONTH_NAMES_ENGLISH as $month => $names) {
foreach ($names as $name) $monthNames[$name] = $month;
}
foreach (self::MONTH_NAMES_GERMAN as $month => $names) {
foreach ($names as $name) $monthNames[$name] = $month;
}
foreach (self::MONTH_NAMES_HUNGARIAN as $month => $names) {
foreach ($names as $name) $monthNames[$name] = $month;
}
$rewrite = strtr($datum, $monthNames);
if ($rewrite !== $datum) {
return str_replace('..', '.', str_replace(" ", ".", $rewrite));
}
}
if (str_contains($datum, ',')) {
return str_replace(',', '-', $datum);
}
return '';
}
/** /**
* Checks if the string is a time span with given start and end dates. * Checks if the string is a time span with given start and end dates.
* *
@ -1062,6 +1197,9 @@ final class NodaTimeSplitter {
if (strlen($datum) === 9 and substr($datum, 4, 1) !== '-') return []; if (strlen($datum) === 9 and substr($datum, 4, 1) !== '-') return [];
if (empty($startEnd = self::check_is_timespan_from_till($datum))) { if (empty($startEnd = self::check_is_timespan_from_till($datum))) {
if ($rewritten = self::_attempt_rewriting_special_cases_from_till($datum)) {
return self::attempt_splitting_from_till($rewritten);
}
return []; return [];
} }
list($start, $end) = $startEnd; list($start, $end) = $startEnd;
@ -1105,38 +1243,76 @@ final class NodaTimeSplitter {
} }
/** /**
* Wrapper to check if any splitting command works. * Cleans invalid outputs from splitting.
* *
* @param string $datum Input date. * @param array{0: string, 1: string, 2: string, 3: string, 4: '+'|'-'|'', 5: string}|array{} $moda Split time to check.
* *
* @return array<string> * @return array{0: string, 1: string, 2: string, 3: string, 4: '+'|'-'|'', 5: string}|array{}
*/ */
public static function attempt_splitting(string $datum):array { private static function validate_split_time(array $moda):array {
$moda = NodaTimeSplitter::is_timespan($datum); if (empty($moda)) return [];
if (!$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)$moda[3] > 31) { if ((int)$moda[2] > 12 || (int)$moda[3] > 31) {
return []; return [];
} }
$month_no_zero = strtr($moda[2], ["0" => "", "1" => "", "2" => "", "3" => "", "4" => "", "5" => "", "6" => "", "7" => "", "8" => "", "9" => ""]);
$day_no_zero = strtr($moda[3], ["0" => "", "1" => "", "2" => "", "3" => "", "4" => "", "5" => "", "6" => "", "7" => "", "8" => "", "9" => ""]);
if (!empty($month_no_zero)) {
throw new Exception("Invalid split month: " . var_export($moda, true));
}
if (!empty($day_no_zero)) {
throw new Exception("Invalid split day: " . var_export($moda, true));
} }
return $moda; return $moda;
} }
/**
* Wrapper to check if any splitting command works.
*
* @param string $datum Input date.
*
* @return array{0: string, 1: string, 2: string, 3: string, 4: '+'|'-'|'', 5: string}|array{}
*/
public static function attempt_splitting(string $datum):array {
if (!empty($moda = self::is_timespan($datum))) {
return self::validate_split_time($moda);
}
if (!empty($moda = self::is_incomplete_date($datum))) {
return self::validate_split_time($moda);
}
if (!empty($moda = self::is_valid_date($datum))) {
return self::validate_split_time($moda);
}
if (!empty($moda = self::is_valid_date_hungarian($datum))) {
return self::validate_split_time($moda);
}
if (!empty($moda = self::is_century($datum))) {
return self::validate_split_time($moda);
}
if (!empty($moda = self::is_decade($datum))) {
return self::validate_split_time($moda);
}
// 2015. 05.
if (str_contains($datum, ' ')) {
$rewrite = str_replace(' ', '', $datum);
if (is_numeric(str_replace('.', '', $datum))) {
return self::attempt_splitting($rewrite);
}
}
return [];
}
} }

View File

@ -132,6 +132,18 @@ final class NodaTimeSplitterTest extends TestCase {
self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "1920-1929"); self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "1920-1929");
self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 1925); self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 1925);
$output = NodaTimeSplitter::attempt_splitting("1920-1929");
self::assertEquals($output, [
0 => "1920",
1 => "1929",
2 => "00",
3 => "00",
4 => "+",
5 => "",
]);
self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "1920-1929");
self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 1925);
$output = NodaTimeSplitter::attempt_splitting("1920er Jahre"); $output = NodaTimeSplitter::attempt_splitting("1920er Jahre");
self::assertEquals($output, [ self::assertEquals($output, [
0 => "1920", 0 => "1920",
@ -488,6 +500,30 @@ final class NodaTimeSplitterTest extends TestCase {
self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "02.01.2020"); self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "02.01.2020");
self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 2020); self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 2020);
$output = NodaTimeSplitter::attempt_splitting("2020.01.2.");
self::assertEquals($output, [
0 => "2020",
1 => "2020",
2 => "01",
3 => "02",
4 => "+",
5 => "",
]);
self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "02.01.2020");
self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 2020);
$output = NodaTimeSplitter::attempt_splitting("2020.1.2.");
self::assertEquals($output, [
0 => "2020",
1 => "2020",
2 => "01",
3 => "02",
4 => "+",
5 => "",
]);
self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "02.01.2020");
self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 2020);
$output = NodaTimeSplitter::attempt_splitting("2020. Januar 2."); $output = NodaTimeSplitter::attempt_splitting("2020. Januar 2.");
self::assertEquals($output, [ self::assertEquals($output, [
0 => "2020", 0 => "2020",
@ -548,6 +584,19 @@ final class NodaTimeSplitterTest extends TestCase {
self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "1920-1929"); self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "1920-1929");
self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 1925); self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 1925);
// From 1920 onwards
$output = NodaTimeSplitter::attempt_splitting("1920 utántól");
self::assertEquals($output, [
0 => "1920",
1 => "?",
2 => "00",
3 => "00",
4 => "+",
5 => "",
]);
self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "Seit 1920");
self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 1920);
$output = NodaTimeSplitter::attempt_splitting("1920-es évek"); $output = NodaTimeSplitter::attempt_splitting("1920-es évek");
self::assertEquals($output, [ self::assertEquals($output, [
0 => "1920", 0 => "1920",
@ -608,6 +657,18 @@ final class NodaTimeSplitterTest extends TestCase {
self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "201-300 n. Chr."); self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "201-300 n. Chr.");
self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 251); self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 251);
$output = NodaTimeSplitter::attempt_splitting("20th century");
self::assertEquals($output, [
0 => "1901",
1 => "2000",
2 => "00",
3 => "00",
4 => "+",
5 => "",
]);
self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "1901-2000");
self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 1951);
$output = NodaTimeSplitter::attempt_splitting("20. század"); $output = NodaTimeSplitter::attempt_splitting("20. század");
self::assertEquals($output, [ self::assertEquals($output, [
0 => "1901", 0 => "1901",
@ -620,6 +681,18 @@ final class NodaTimeSplitterTest extends TestCase {
self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "1901-2000"); self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "1901-2000");
self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 1951); self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 1951);
$output = NodaTimeSplitter::attempt_splitting("20.század");
self::assertEquals($output, [
0 => "1901",
1 => "2000",
2 => "00",
3 => "00",
4 => "+",
5 => "",
]);
self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "1901-2000");
self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 1951);
$output = NodaTimeSplitter::attempt_splitting("Kr. e. 20. század"); $output = NodaTimeSplitter::attempt_splitting("Kr. e. 20. század");
self::assertEquals($output, [ self::assertEquals($output, [
0 => "-2000", 0 => "-2000",
@ -658,6 +731,109 @@ final class NodaTimeSplitterTest extends TestCase {
} }
/**
* Test to check special cases can be parsed.
*
* @author Joshua Ramon Enslin <joshua@museum-digital.de>
* @group ValidOutput
* @small
*
* @return void
*/
public function testSplittingFromTill():void {
// Regular
$output = NodaTimeSplitter::attempt_splitting_from_till("2004.01.-2004.02.");
self::assertNotEmpty($output);
self::assertEquals($output, [
'start_name' => "Januar 2004",
'end_name' => "Februar 2004",
"start_year" => '2004',
"end_year" => '2004',
"counting_time_year" => "2004",
"counting_time_month" => "01",
"counting_time_day" => "16",
"counting_time_bcce" => "+",
]);
// Rewritten / Hungarian YYYY.MM.-MM.
$output = NodaTimeSplitter::attempt_splitting_from_till("2004.01.-02.");
self::assertNotEmpty($output);
self::assertEquals($output, [
'start_name' => "Januar 2004",
'end_name' => "Februar 2004",
"start_year" => '2004',
"end_year" => '2004',
"counting_time_year" => "2004",
"counting_time_month" => "01",
"counting_time_day" => "16",
"counting_time_bcce" => "+",
]);
// Rewritten / Hungarian YYYY.MM.-MM.
$output = NodaTimeSplitter::attempt_splitting_from_till("2003.04-05.");
self::assertNotEmpty($output);
// Rewritten / Hungarian YYYY.MM.-MM.
$output = NodaTimeSplitter::attempt_splitting_from_till("1981. július-augusztus");
self::assertNotEmpty($output);
$output = NodaTimeSplitter::attempt_splitting_from_till("2019.03.14.,04.15.");
self::assertNotEmpty($output);
self::assertEquals($output, [
'start_name' => "14.03.2019",
'end_name' => "15.04.2019",
"start_year" => '2019',
"end_year" => '2019',
"counting_time_year" => "2019",
"counting_time_month" => "03",
"counting_time_day" => "30",
"counting_time_bcce" => "+",
]);
$output = NodaTimeSplitter::attempt_splitting_from_till("2019.03.14-15");
self::assertNotEmpty($output);
self::assertEquals($output, [
'start_name' => "14.03.2019",
'end_name' => "15.03.2019",
"start_year" => '2019',
"end_year" => '2019',
"counting_time_year" => "2019",
"counting_time_month" => "03",
"counting_time_day" => "15",
"counting_time_bcce" => "+",
]);
// Rewritten / Hungarian YYYY.MM.-MM.
$output = NodaTimeSplitter::attempt_splitting_from_till("17-19.század");
self::assertNotEmpty($output);
self::assertEquals($output, [
'start_name' => "1601",
'end_name' => "1900",
"start_year" => '1601',
"end_year" => '1900',
"counting_time_year" => "1750",
"counting_time_month" => "06",
"counting_time_day" => "01",
"counting_time_bcce" => "+",
]);
// Rewritten / 1950-60-as évek
$output = NodaTimeSplitter::attempt_splitting_from_till("1950-60-as évek");
self::assertNotEmpty($output);
self::assertEquals($output, [
'start_name' => "1950",
'end_name' => "1969",
"start_year" => '1950',
"end_year" => '1969',
"counting_time_year" => "1959",
"counting_time_month" => "06",
"counting_time_day" => "01",
"counting_time_bcce" => "+",
]);
}
/** /**
* Test to check whether the HTML page is correctly generated. * Test to check whether the HTML page is correctly generated.
* *
@ -670,40 +846,43 @@ final class NodaTimeSplitterTest extends TestCase {
public function testSplitDoesNotWorkWhenItShouldNot():void { public function testSplitDoesNotWorkWhenItShouldNot():void {
$output = NodaTimeSplitter::attempt_splitting(""); $output = NodaTimeSplitter::attempt_splitting("");
self::assertEquals($output, []); self::assertEmpty($output);
$output = NodaTimeSplitter::attempt_splitting("1.2.2020-2.2.2020"); $output = NodaTimeSplitter::attempt_splitting("1.2.2020-2.2.2020");
self::assertEquals($output, []); self::assertEmpty($output);
$output = NodaTimeSplitter::attempt_splitting("2020 Januar 2-2020 Februar 2"); $output = NodaTimeSplitter::attempt_splitting("2020 Januar 2-2020 Februar 2");
self::assertEquals($output, []); self::assertEmpty($output);
$output = NodaTimeSplitter::attempt_splitting("2020 Januar-2020 Februar"); $output = NodaTimeSplitter::attempt_splitting("2020 Januar-2020 Februar");
self::assertEquals($output, []); self::assertEmpty($output);
$output = NodaTimeSplitter::attempt_splitting("Januar-Februar"); $output = NodaTimeSplitter::attempt_splitting("Januar-Februar");
self::assertEquals($output, []); self::assertEmpty($output);
$output = NodaTimeSplitter::attempt_splitting("13.13.2022"); $output = NodaTimeSplitter::attempt_splitting("13.13.2022");
self::assertEquals($output, []); self::assertEmpty($output);
$output = NodaTimeSplitter::attempt_splitting("2022-13-13"); $output = NodaTimeSplitter::attempt_splitting("2022-13-13");
self::assertEquals($output, []); self::assertEmpty($output);
$output = NodaTimeSplitter::attempt_splitting("40.10.2022"); $output = NodaTimeSplitter::attempt_splitting("40.10.2022");
self::assertEquals($output, []); self::assertEmpty($output);
$output = NodaTimeSplitter::attempt_splitting("2022-10-40"); $output = NodaTimeSplitter::attempt_splitting("2022-10-40");
self::assertEquals($output, []); self::assertEmpty($output);
$output = NodaTimeSplitter::attempt_splitting("6;November 1978"); $output = NodaTimeSplitter::attempt_splitting("6;November 1978");
self::assertEquals($output, []); self::assertEmpty($output);
$output = NodaTimeSplitter::attempt_splitting("65497028c51eb"); $output = NodaTimeSplitter::attempt_splitting("65497028c51eb");
self::assertEquals($output, []); self::assertEmpty($output);
$output = NodaTimeSplitter::attempt_splitting("6552cf08b0196 test tag"); $output = NodaTimeSplitter::attempt_splitting("6552cf08b0196 test tag");
self::assertEquals($output, []); self::assertEmpty($output);
$output = NodaTimeSplitter::attempt_splitting("Anfang September 1903");
self::assertEmpty($output);
# $output = NodaTimeSplitter::attempt_splitting("Nach 1944-1964"); # $output = NodaTimeSplitter::attempt_splitting("Nach 1944-1964");
# self::assertEquals($output, []); # self::assertEmpty($output);
} }
} }