From b122279ece06a9381e805c6087f130b41616fe3d Mon Sep 17 00:00:00 2001
From: Geo Halkiadakis <gchalkiadakis@sklavenitis.co.gr>
Date: Fri, 12 Apr 2024 18:31:00 +0300
Subject: added utils folder; added several suplamentary modules

---
 utils/kb-util.js | 88 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 88 insertions(+)
 create mode 100644 utils/kb-util.js

(limited to 'utils/kb-util.js')

diff --git a/utils/kb-util.js b/utils/kb-util.js
new file mode 100644
index 0000000..aac88f9
--- /dev/null
+++ b/utils/kb-util.js
@@ -0,0 +1,88 @@
+/**
+ * fast string manipulation utilities
+ * for bi-lingual (EL/EN) words/phrases
+ * based on the keyboard layout
+ */
+
+// suplamentary arrays (mostly for cache)
+// --- -- -- - - -
+
+var ORiGiNal = 'ςερτυθιοπασδφγηξκλζχψωβνμΕΡΤΥΘΙΟΠΑΣΔΦΓΗΞΚΛΖΧΨΩΒΝΜάέήίόύώϊΐϋΆΈΉΊΌΎΏΪΫQWERTYUIOPASDFGHJKLZXCVBNMqwertyuiopasdfghjklzxcvbnm0123456789- '.split('');
+
+var kbKeyZed = 'sertyuiopasdfghjklzxcvbnmertyuiopasdfghjklzxcvbnmaehioyviiyaehioyviyqwertyuiopasdfghjklzxcvbnmqwertyuiopasdfghjklzxcvbnm0123456789- '.split('');
+
+var map = new Map();
+for (var i=0; i<ORiGiNal.length; i++) map.set(ORiGiNal[i], kbKeyZed[i]);
+
+
+// cache (=create a global array)
+// of accended to non-accended vowels mapping
+// --- -- -- - - -
+accented_vowels = [];   
+[
+    'ά α', 'έ ε', 'ή η', 'ί ι', 'ϊ ι', 'ΐ ι', 'ό ο', 'ύ υ', 'ϋ υ', 'ώ ω',
+    'Ά Α', 'Έ Ε', 'Ή Η', 'Ί Ι', 'Ϊ Ι', 'Ό Ο', 'Ύ Υ', 'Ϋ Υ', 'Ώ Ω'
+].forEach( pair => {
+    ap = pair.split(' ');
+    accented_vowels.push({
+        a: ap[0],   // accented
+        p: ap[1]    // pure = non accended
+    });
+});
+
+
+// translates string to keyboard-latin keys
+// (the ones that used whan typing each letter of the word)
+const keyboardize = (str) => {
+    str = str.replace('\'','');
+    var out = '';
+    // [map]'s implementation is 40x faster than [for]'s
+    for (var i=0 ; i< str.length; i++)  out += map.get(str[i]);
+    return out;
+}
+
+// keyboardize an array of strings
+const keyb_array = (arr) => {
+    kb_arr = [];
+    arr.forEach( w => {
+        kb_arr.push(keyboardize(w));
+    });
+    return kb_arr;
+}
+
+
+// transforms to lowercase; handles sigma-teliko
+const sanitizeGR = (str) => {
+    str = str.toLowerCase();
+
+    // replace accended vowels with pure ones
+    accented_vowels.forEach( v => {
+        str = str.replaceAll(v.a, v.p);
+    });
+
+    // replace sigma on the end of words
+    str = str + ' ';
+    str = str.replaceAll('σ-', 'ς-');
+    str = str.replaceAll('σ ', 'ς ');
+    
+    return str;
+}
+
+// removes non keyword characters [+ . , !] and internal multiple-spaces
+// @param txt (string): product description
+const clean = (txt) => {
+    return txt.replace('+',' ').replace('.',' ').replace(',',' ')   // change to space
+        .replace('!','').replace('\"', '')                  // remove character
+        .replace('   ',' ').replace('  ',' ');          // remove multiple spaces
+}
+
+
+// exports
+// --- -- -- - - -
+
+module.exports = {
+    keyboardize,
+    keyb_array,
+    sanitizeGR,
+    clean
+};
\ No newline at end of file
-- 
cgit v1.2.3