From e53eec84e6ba8cbb3a4c00d52afe1f745f085e6e Mon Sep 17 00:00:00 2001
From: Joshua Ramon Enslin <joshua@museum-digital.de>
Date: Mon, 21 Sep 2020 01:57:21 +0200
Subject: [PATCH] Add functions cleaning of uncertainty indicators to
 NodaUncertaintyHelper

---
 src/NodaUncertaintyHelper.php | 120 ++++++++++++++++++++++++++++++++++
 1 file changed, 120 insertions(+)

diff --git a/src/NodaUncertaintyHelper.php b/src/NodaUncertaintyHelper.php
index b9ee265..aa0d2c8 100644
--- a/src/NodaUncertaintyHelper.php
+++ b/src/NodaUncertaintyHelper.php
@@ -11,6 +11,18 @@ declare(strict_types = 1);
  */
 final class NodaUncertaintyHelper {
 
+    const PERSINST_UNCERTAINTY_PREFIXES = [
+        "wohl ",
+        "wahrscheinlich ",
+        "Wohl ",
+        "Wahrscheinlich ",
+    ];
+
+    const PERSINST_UNCERTAINTY_SUFFIXES = [
+        "(?)",
+        "?",
+    ];
+
     const TIME_UNCERTAINTY_PREFIXES = [
         "um ",
         "wohl um ",
@@ -39,6 +51,33 @@ final class NodaUncertaintyHelper {
         "?",
     ];
 
+    /**
+     * Removes uncertainty indicators from an time name.
+     *
+     * @param string $name Input string.
+     *
+     * @return string
+     */
+    public static function cleanUncertaintyIndicatorsTime(string $name):string {
+
+        // Remove uncertainty prefixes
+        foreach (NodaUncertaintyHelper::TIME_UNCERTAINTY_PREFIXES as $prefix) {
+            if (\substr($name, 0, \strlen($prefix)) === "$prefix") {
+                $name = trim(substr($name, \strlen($prefix)));
+            }
+        }
+
+        // Remove uncertainty sufixes
+        foreach (NodaUncertaintyHelper::TIME_UNCERTAINTY_SUFFIXES as $suffix) {
+            if (\substr($name, \strlen($suffix) * -1) === "$suffix") {
+                $name = \trim(\substr($name, 0, \strlen($suffix) * -1));
+            }
+        }
+
+        return \trim($name);
+
+    }
+
     /**
      * Attempts guessing whether time is uncertain.
      *
@@ -68,6 +107,33 @@ final class NodaUncertaintyHelper {
 
     }
 
+    /**
+     * Removes uncertainty indicators from an place name.
+     *
+     * @param string $ort_name Input string.
+     *
+     * @return string
+     */
+    public static function cleanUncertaintyIndicatorsPlace(string $ort_name):string {
+
+        // Remove uncertainty prefixes
+        foreach (NodaUncertaintyHelper::PLACE_UNCERTAINTY_PREFIXES as $prefix) {
+            if (\substr($ort_name, 0, \strlen($prefix)) === "$prefix") {
+                $ort_name = trim(substr($ort_name, \strlen($prefix)));
+            }
+        }
+
+        // Remove uncertainty sufixes
+        foreach (NodaUncertaintyHelper::PLACE_UNCERTAINTY_SUFFIXES as $suffix) {
+            if (\substr($ort_name, \strlen($suffix) * -1) === "$suffix") {
+                $ort_name = \trim(\substr($ort_name, 0, \strlen($suffix) * -1));
+            }
+        }
+
+        return \trim($ort_name);
+
+    }
+
     /**
      * Attempts guessing whether place is uncertain.
      *
@@ -97,4 +163,58 @@ final class NodaUncertaintyHelper {
 
     }
 
+    /**
+     * Removes uncertainty indicators from an actor name.
+     *
+     * @param string $value Input string.
+     *
+     * @return string
+     */
+    public static function cleanUncertaintyIndicatorsPersinst(string $value):string {
+
+        foreach (self::PERSINST_UNCERTAINTY_PREFIXES as $toRemove) {
+            if (\mb_substr($value, 0, \mb_strlen($toRemove)) === $toRemove) {
+                $value = substr($value, \mb_strlen($toRemove));
+            }
+        }
+
+        foreach (self::PLACE_UNCERTAINTY_SUFFIXES as $suffix) {
+            if (\mb_substr($value, \mb_strlen($suffix) * -1) === "$suffix") {
+                $value = \mb_substr($value, 0, \mb_strlen($suffix) * -1);
+            }
+        }
+
+        return \trim($value);
+
+    }
+
+    /**
+     * Attempts guessing whether persinst is uncertain.
+     *
+     * @param string $name Persinst name.
+     *
+     * @return boolean
+     */
+    public static function guessPersinstCertainty(string $name):bool {
+
+        $name = \strtolower($name);
+
+        // Attempt to guess uncertainty based on prefixes.
+        foreach (NodaUncertaintyHelper::PERSINST_UNCERTAINTY_PREFIXES as $prefix) {
+            if (\substr($name, 0, \strlen($prefix)) === $prefix) {
+                return false; // Uncertain
+            }
+        }
+
+        // Attempt to guess uncertainty based on prefixes.
+        foreach (NodaUncertaintyHelper::PERSINST_UNCERTAINTY_SUFFIXES as $prefix) {
+            if (\substr($name, -1 * \strlen($prefix)) === $prefix) {
+                return false; // Uncertain
+            }
+        }
+
+        return true; // Certain / no uncertainty found
+
+    }
+
 }