museum-digital/MDNodaHelpers
36
0
Fork 0
Code Issues 6 Pull Requests Releases Wiki Activity

Extend NodaTimeSplitter

Browse Source
This commit is contained in:
Joshua Ramon Enslin
2023-12-09 23:54:45 +01:00
parent 4a49c7a4e7
commit f3831965a3
2 changed files with 54 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

27
src/NodaTimeSplitter.php
View File

@@ -58,6 +58,8 @@ final class NodaTimeSplitter {
const STRINGS_TO_CLEAN = [
"között" => "",
" рр." => "",
" рр" => "",
"nach Christus" => "",
"n. Christus" => "",
"nach Chr." => "",
@@ -69,6 +71,7 @@ final class NodaTimeSplitter {
"BCE" => "v. Chr.",
"CE" => "",
"vor Christus" => "v. Chr.",
" до н. е." => "v. Chr.",
];
const STRINGS_TO_CLEAN_START = [
@@ -133,6 +136,7 @@ final class NodaTimeSplitter {
];
private const REGEX_CENTURIES = '(\ |)(Jh\.|Jhd(|\.)|Jhdt(|\.)|Jahrhundert|sz|század|th century|ст|ст\.)';
private const REGEX_DECADES = '(s|er|er\ Jahre|(\-|\ )es\ évek|(\-|\ )as\ \évek|\ évek|\-es\ években|\-ті)';
/**
* Cleans input strings by trimming obsolete stuff.
@@ -555,10 +559,6 @@ final class NodaTimeSplitter {
}
}
if ($datum === '1978. július 7 elött') {
throw new Exception(var_export($monat, true));
}
if (!empty($monat) && empty($day) && preg_match('~[0-9]+~', substr($datum, -3))) {
return [];
}
@@ -864,14 +864,17 @@ final class NodaTimeSplitter {
}
}
if (\preg_match("/\ (\(vor|\(Vor|vor)$/", $datum)) {
if (($spacePos = \strpos($datum, " ")) === false) {
if (\preg_match("/\ (\(vor|\(Vor|vor|előtt)$/", $datum)) {
if (($spacePos = \strrpos($datum, " ")) === false) {
return [];
}
if ($output = self::attempt_splitting(\substr($datum, 0, $spacePos))) {
$output[0] = "?";
if (empty(trim($output[2], "0 .,"))) $output[1] = \strval(\intval($output[1]) - 1);
// If month and day are unknown, lower end year by one
if (empty(trim($output[2], "0 .,"))) {
$output[1] = \strval(\intval($output[1]) - 1);
}
$output[5] = "Vor";
return $output;
}
@@ -951,6 +954,12 @@ final class NodaTimeSplitter {
return $output;
}
}
if (str_ends_with($datum, '-as évekig') || str_ends_with($datum, '-es évekig')) {
if ($output = self::attempt_splitting(\substr($datum, 0, -2))) {
$output[0] = "?";
return $output;
}
}
return [];
@@ -1054,14 +1063,14 @@ final class NodaTimeSplitter {
$bcBceIndicator = '+';
// 20er Jahre
if (\preg_match("/^[0-9]0(er|er\ Jahre|\-es\ évek|\-as\ \évek)$/", $datum)) {
if (\preg_match("/^[0-9]0" . self::REGEX_DECADES . "$/", $datum)) {
$start = "19" . \substr($datum, 0, 2);
$ende = (string)(\intval($start) + 9);
return [$start, $ende, "00", "00", $bcBceIndicator, ""];
}
// 1920er Jahre
if (\preg_match("/^[0-9]{3}0(s|er|er\ Jahre|(\-|\ )es\ évek|(\-|\ )as\ \évek)$/", $datum)) {
if (\preg_match("/^[0-9]{3}0" . self::REGEX_DECADES . "$/", $datum)) {
$start = \substr($datum, 0, 4);
$ende = (string)(\intval($start) + 9);
return [$start, $ende, "00", "00", $bcBceIndicator, ""];