From 923505f146cc01ecb4960906a1d8ec38934e0342 Mon Sep 17 00:00:00 2001
From: Joshua Ramon Enslin <joshua@museum-digital.de>
Date: Mon, 21 Sep 2020 01:24:07 +0200
Subject: [PATCH] Add NodaUncertaintyHelper for guessing uncertainty of noda
 entries

---
 src/NodaUncertaintyHelper.php | 100 ++++++++++++++++++++++++++++++++++
 1 file changed, 100 insertions(+)
 create mode 100644 src/NodaUncertaintyHelper.php

diff --git a/src/NodaUncertaintyHelper.php b/src/NodaUncertaintyHelper.php
new file mode 100644
index 0000000..b9ee265
--- /dev/null
+++ b/src/NodaUncertaintyHelper.php
@@ -0,0 +1,100 @@
+<?PHP
+/**
+ * Contains class NodaUncertaintyHelper.
+ *
+ * @author Joshua Ramon Enslin <joshua@museum-digital.de>
+ */
+declare(strict_types = 1);
+
+/**
+ * Contains static functions for getting IDs for noda entries by various means.
+ */
+final class NodaUncertaintyHelper {
+
+    const TIME_UNCERTAINTY_PREFIXES = [
+        "um ",
+        "wohl um ",
+        "circa ",
+        "ca. ",
+        "ca "
+    ];
+
+    const TIME_UNCERTAINTY_SUFFIXES = [
+        "(?)",
+        "?",
+    ];
+
+    /**
+     * Substrings used to express uncertainty about the validity of a place name.
+     */
+    const PLACE_UNCERTAINTY_PREFIXES = [
+        "vlt. ",
+        "circa ",
+        "ca. ",
+        "ca ",
+    ];
+
+    const PLACE_UNCERTAINTY_SUFFIXES = [
+        "(?)",
+        "?",
+    ];
+
+    /**
+     * Attempts guessing whether time is uncertain.
+     *
+     * @param string $zeit_name Time name.
+     *
+     * @return boolean
+     */
+    public static function guessTimeCertainty(string $zeit_name):bool {
+
+        $zeit_name = \strtolower($zeit_name);
+
+        // Attempt to guess uncertainty based on prefixes.
+        foreach (self::TIME_UNCERTAINTY_PREFIXES as $prefix) {
+            if (\substr($zeit_name, 0, \strlen($prefix)) === $prefix) {
+                return false; // Uncertainty found
+            }
+        }
+
+        // Attempt to guess uncertainty based on prefixes.
+        foreach (self::TIME_UNCERTAINTY_SUFFIXES as $prefix) {
+            if (\substr($zeit_name, -1 * \strlen($prefix)) === $prefix) {
+                return false; // Uncertainty found
+            }
+        }
+
+        return true; // No uncertainty found
+
+    }
+
+    /**
+     * Attempts guessing whether place is uncertain.
+     *
+     * @param string $ort_name Place name.
+     *
+     * @return boolean
+     */
+    public static function guessPlaceCertainty(string $ort_name):bool {
+
+        $ort_name = \strtolower($ort_name);
+
+        // Attempt to guess uncertainty based on prefixes.
+        foreach (NodaUncertaintyHelper::PLACE_UNCERTAINTY_PREFIXES as $prefix) {
+            if (\substr($ort_name, 0, \strlen($prefix)) === $prefix) {
+                return false; // Uncertain
+            }
+        }
+
+        // Attempt to guess uncertainty based on prefixes.
+        foreach (NodaUncertaintyHelper::PLACE_UNCERTAINTY_SUFFIXES as $prefix) {
+            if (\substr($ort_name, -1 * \strlen($prefix)) === $prefix) {
+                return false; // Uncertain
+            }
+        }
+
+        return true; // Certain / no uncertainty found
+
+    }
+
+}