museum-digital/MDNodaHelpers
33
0
Fork 0
Code Issues 3 Pull Requests Releases Wiki Activity

Identify uncertainty before brackets ("Berlin ? (Germany)" > "Berlin

Browse Source 
(Germany)" + Uncertain)
This commit is contained in:
Joshua Ramon Enslin 2024-11-09 18:42:18 +01:00
parent 7cfe752c94
commit 48355a6a36
Signed by: jrenslin
GPG Key ID: 46016F84501B70AE
2 changed files with 79 additions and 61 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
src/NodaUncertaintyHelper.php
View File

@ -328,6 +328,14 @@ final class NodaUncertaintyHelper {
} }
} }
// If brackets are included in the name, try removing prefixes and suffixes
// from the beginning.
if (($bracketPos = strpos($ort_name, "(")) !== false) {
$start = substr($ort_name, 0, $bracketPos);
$end = substr($ort_name, $bracketPos);
$ort_name = self::cleanUncertaintyIndicatorsPlace($start) . ' ' . $end;
}
return self::trim($ort_name); return self::trim($ort_name);
} }
@ -358,6 +366,13 @@ final class NodaUncertaintyHelper {
} }
} }
// If brackets are included in the name, try the same for everything up to the
// first brackets.
if (($bracketPos = strpos($ort_name, "(")) !== false) {
$name = substr($ort_name, 0, $bracketPos);
return self::guessPlaceCertainty($name);
}
return true; // Certain / no uncertainty found return true; // Certain / no uncertainty found
} }

125
tests/NodaUncertaintyHelperTest.php
View File

@ -8,6 +8,7 @@ declare(strict_types = 1);
use PHPUnit\Framework\TestCase; use PHPUnit\Framework\TestCase;
use PHPUnit\Framework\Attributes\CoversClass; use PHPUnit\Framework\Attributes\CoversClass;
use PHPUnit\Framework\Attributes\Small; use PHPUnit\Framework\Attributes\Small;
use PHPUnit\Framework\Attributes\DataProvider;
/** /**
* This script contains tests for the uncertainty helper. * This script contains tests for the uncertainty helper.
@ -16,111 +17,113 @@ use PHPUnit\Framework\Attributes\Small;
#[CoversClass(\NodaUncertaintyHelper::class)] #[CoversClass(\NodaUncertaintyHelper::class)]
final class NodaUncertaintyHelperTest extends TestCase { final class NodaUncertaintyHelperTest extends TestCase {
/** /**
* Removes uncertainty indicators from an time name. * Returns time names with expected cleaned version and expected parsed certainty.
* *
* @group ValidOutput * @return array<array{0: string, 1: string, 2: boolean}>
* @small
*
* @return void
*/ */
public function testCleanUncertaintyIndicatorsTime():void { public static function uncertainTimesProvider():array {
self::assertEquals("1950", NodaUncertaintyHelper::cleanUncertaintyIndicatorsTime("wohl 1950")); return [
self::assertEquals("1950", NodaUncertaintyHelper::cleanUncertaintyIndicatorsTime("1950?")); 'uncertainty prefix: "wohl 1950"' => ["wohl 1950", "1950", false],
self::assertEquals("1950", NodaUncertaintyHelper::cleanUncertaintyIndicatorsTime("1950?,")); 'uncertainty suffix: "1950?"' => ["1950?", "1950", false],
self::assertEquals("1950", NodaUncertaintyHelper::cleanUncertaintyIndicatorsTime("1950,")); 'uncertainty suffix and superfluous chars: "1950 ?,"' => ["1950 ?,", "1950", false],
self::assertEquals("1950", NodaUncertaintyHelper::cleanUncertaintyIndicatorsTime("1950")); 'certain term with superfluous chars: "1950 ,"' => ["1950 ,", "1950", true],
'certain term: 1950' => ["1950", "1950", true],
];
} }
/** /**
* Attempts guessing whether time is uncertain. * Returns place names with expected cleaned version and expected parsed certainty.
* *
* @group ValidOutput * @return array<array{0: string, 1: string, 2: boolean}>
* @small
*
* @return void
*/ */
public function testGuessTimeCertainty():void { public static function uncertainPlacesProvider():array {
self::assertFalse(NodaUncertaintyHelper::guessTimeCertainty("wohl 1950")); return [
self::assertTrue(NodaUncertaintyHelper::guessTimeCertainty("1950"));
'uncertainty prefix: "wohl Berlin"' => ["wohl Berlin", "Berlin", false],
'uncertainty prefix: "vermutl. Berlin"' => ["vermutl. Berlin", "Berlin", false],
'uncertainty prefix and superfluous chars: "?-Berlin"' => ["?-Berlin", "Berlin", false],
'uncertainty suffix: "Berlin?"' => ["Berlin?", "Berlin", false],
'uncertainty suffix: "Berlin (?)"' => ["Berlin (?)", "Berlin", false],
'uncertainty suffix and superfluous chars: "Berlin ?,"' => ["Berlin ?,", "Berlin", false],
'certain term with superfluous chars: "Berlin ,"' => ["Berlin ,", "Berlin", true],
'certain term: Berlin' => ["Berlin", "Berlin", true],
'Berlin ? (Deutschland)' => ["Berlin ? (Deutschland)", "Berlin (Deutschland)", false],
'Berli?n (Deutschland)' => ["Berl?n (Deutschland)", "Berl?n (Deutschland)", true],
];
} }
/** /**
* Removes uncertainty indicators from an place name. * Returns actor names with expected cleaned version and expected parsed certainty.
* *
* @group ValidOutput * @return array<array{0: string, 1: string, 2: boolean}>
* @small
*
* @return void
*/ */
public static function testCleanUncertaintyIndicatorsPlace():void { public static function uncertainPersinstProvider():array {
self::assertEquals("Berlin", NodaUncertaintyHelper::cleanUncertaintyIndicatorsPlace("wohl Berlin")); return [
self::assertEquals("Berlin", NodaUncertaintyHelper::cleanUncertaintyIndicatorsPlace("Berlin")); 'uncertainty prefix: "wohl Barbarossa"' => ["wohl Barbarossa", "Barbarossa", false],
'uncertainty prefix: "vermutl. Barbarossa"' => ["vermutl. Barbarossa", "Barbarossa", false],
// Real-life examples that previously passed unencumbered 'uncertainty prefix and superfluous chars: "?-Barbarossa"' => ["?-Barbarossa", "Barbarossa", false],
self::assertEquals("Augsburg", NodaUncertaintyHelper::cleanUncertaintyIndicatorsPlace("vermutlich: Augsburg")); 'uncertainty suffix: "Barbarossa?"' => ["Barbarossa?", "Barbarossa", false],
self::assertEquals("Augsburg", NodaUncertaintyHelper::cleanUncertaintyIndicatorsPlace("vermutl. Augsburg")); 'uncertainty suffix and superfluous chars: "Barbarossa ?,"' => ["Barbarossa ?,", "Barbarossa", false],
'certain term with superfluous chars: "Barbarossa ,"' => ["Barbarossa ,", "Barbarossa", true],
'certain term: Barbarossa' => ["Barbarossa", "Barbarossa", true],
];
} }
/** /**
* Attempts guessing whether place is uncertain. * Test to ensure times are correctly cleaned and parsed.
* *
* @group ValidOutput * @param string $term Term to check.
* @small * @param string $target_output Expected output name.
* @param boolean $target_certainty Expected output certainty.
* *
* @return void * @return void
*/ */
public static function testGuessPlaceCertainty():void { #[DataProvider('uncertainTimesProvider')]
public function testParsingUncertaintyFromTimes(string $term, string $target_output, bool $target_certainty):void {
self::assertFalse(NodaUncertaintyHelper::guessPlaceCertainty("vermutlich: Augsburg")); self::assertEquals($target_output, NodaUncertaintyHelper::cleanUncertaintyIndicatorsTime($term));
self::assertFalse(NodaUncertaintyHelper::guessPlaceCertainty("vermutl. Augsburg")); self::assertEquals($target_certainty, NodaUncertaintyHelper::guessTimeCertainty($term));
self::assertFalse(NodaUncertaintyHelper::guessPlaceCertainty("wohl Berlin"));
self::assertFalse(NodaUncertaintyHelper::guessPlaceCertainty("Berlin?"));
self::assertFalse(NodaUncertaintyHelper::guessPlaceCertainty("Berlin?,"));
self::assertTrue(NodaUncertaintyHelper::guessPlaceCertainty("Berlin,"));
self::assertTrue(NodaUncertaintyHelper::guessPlaceCertainty("Berlin"));
self::assertFalse(NodaUncertaintyHelper::guessPlaceCertainty("?-Italien"));
} }
/** /**
* Removes uncertainty indicators from an actor name. * Test to ensure places are correctly cleaned and parsed.
* *
* @group ValidOutput * @param string $term Term to check.
* @small * @param string $target_output Expected output name.
* @param boolean $target_certainty Expected output certainty.
* *
* @return void * @return void
*/ */
public static function testCleanUncertaintyIndicatorsPersinst():void { #[DataProvider('uncertainPlacesProvider')]
public function testParsingUncertaintyFromPlaces(string $term, string $target_output, bool $target_certainty):void {
self::assertEquals("Barbarossa", NodaUncertaintyHelper::cleanUncertaintyIndicatorsPersinst("wohl Barbarossa")); self::assertEquals($target_output, NodaUncertaintyHelper::cleanUncertaintyIndicatorsPlace($term));
self::assertEquals("Barbarossa", NodaUncertaintyHelper::cleanUncertaintyIndicatorsPersinst("Barbarossa")); self::assertEquals($target_certainty, NodaUncertaintyHelper::guessPlaceCertainty($term));
self::assertEquals("Barbarossa", NodaUncertaintyHelper::cleanUncertaintyIndicatorsPersinst("?-Barbarossa"));
} }
/** /**
* Attempts guessing whether persinst is uncertain. * Test to ensure actor names are correctly cleaned and parsed.
* *
* @group ValidOutput * @param string $term Term to check.
* @small * @param string $target_output Expected output name.
* @param boolean $target_certainty Expected output certainty.
* *
* @return void * @return void
*/ */
public static function testGuessPersinstCertainty():void { #[DataProvider('uncertainPersinstProvider')]
public function testParsingUncertaintyFromPersinst(string $term, string $target_output, bool $target_certainty):void {
self::assertFalse(NodaUncertaintyHelper::guessPlaceCertainty("wohl Barbarossa")); self::assertEquals($target_output, NodaUncertaintyHelper::cleanUncertaintyIndicatorsPersinst($term));
self::assertFalse(NodaUncertaintyHelper::guessPlaceCertainty("Barbarossa?")); self::assertEquals($target_certainty, NodaUncertaintyHelper::guessPersinstCertainty($term));
self::assertFalse(NodaUncertaintyHelper::guessPlaceCertainty("Barbarossa?,"));
self::assertTrue(NodaUncertaintyHelper::guessPlaceCertainty("Barbarossa,"));
self::assertTrue(NodaUncertaintyHelper::guessPlaceCertainty("Barbarossa"));
} }
} }