//// STRING TRANSFORMATION
/*jslint esversion: 6*/
"use strict";
// Note: 48 is the ASCII code for "0"
var str_transform = {
longestCommonSubstring: function (string1, string2) {
// https://en.wikibooks.org/wiki/Algorithm_Implementation/Strings/Longest_common_substring
// untested
// init max value
let longestCommonSubstring = 0;
// init 2D array with 0
let table = [],
len1 = string1.length,
len2 = string2.length,
row, col;
for (row = 0; row <= len1; row++) {
table[row] = [];
for (col = 0; col <= len2; col++) {
table[row][col] = 0;
}
}
// fill table
let i, j;
for (i = 0; i < len1; i++) {
for (j = 0; j < len2; j++) {
if (string1[i] === string2[j]) {
if (table[i][j] === 0){
table[i+1][j+1] = 1;
} else {
table[i+1][j+1] = table[i][j] + 1;
}
if (table[i+1][j+1] > longestCommonSubstring) {
longestCommonSubstring = table[i+1][j+1];
}
} else {
table[i+1][j+1] = 0;
}
}
}
return longestCommonSubstring;
},
distanceDamerauLevenshtein2: function (s1, s2) {
// distance of Damerau-Levenshtein between <s1> and <s2>
// https://fr.wikipedia.org/wiki/Distance_de_Damerau-Levenshtein
try {
let nLen1 = s1.length;
let nLen2 = s2.length;
let matrix = [];
for (let i = 0; i <= nLen1; i++) {
matrix[i] = new Array(nLen2 + 1);
}
for (let i = 0; i <= nLen1; i++) {
matrix[i][0] = i;
}
for (let j = 0; j <= nLen2; j++) {
matrix[0][j] = j;
}
for (let i = 1; i <= nLen1; i++) {
for (let j = 1; j <= nLen2; j++) {
let nCost = (s1[i] === s2[j]) ? 0 : 1;
matrix[i][j] = Math.min(
matrix[i-1][j] + 1, // Deletion
matrix[i][j-1] + 1, // Insertion
matrix[i-1][j-1] + nCost // Substitution
);
if (i > 1 && j > 1 && s1[i] == s2[j-1] && s1[i-1] == s2[j]) {
matrix[i][j] = Math.min(matrix[i][j], matrix[i-2][j-2] + nCost); // Transposition
}
}
}
return matrix[nLen1][nLen2];
}
catch (e) {
helpers.logerror(e);
}
},
distanceDamerauLevenshtein: function (s1, s2) {
// distance of Damerau-Levenshtein between <s1> and <s2>
// https://fr.wikipedia.org/wiki/Distance_de_Damerau-Levenshtein
try {
let nLen1 = s1.length;
let nLen2 = s2.length;
let INF = nLen1 + nLen2;
let matrix = [];
let sd = {};
for (let i = 0; i < nLen1+2; i++) {
matrix[i] = new Array(nLen2+2);
}
matrix[0][0] = INF;
for (let i = 0; i <= nLen1; i++) {
matrix[i+1][1] = i;
matrix[i+1][0] = INF;
sd[s1[i]] = 0;
}
for (let j = 0; j <= nLen2; j++) {
matrix[1][j+1] = j;
matrix[0][j+1] = INF;
sd[s2[j]] = 0;
}
for (let i = 1; i <= nLen1; i++) {
let DB = 0;
for (let j = 1; j <= nLen2; j++) {
let i1 = sd[s2[j-1]];
let j1 = DB;
if (s1[i-1] === s2[j-1]) {
matrix[i+1][j+1] = matrix[i][j];
DB = j;
}
else {
matrix[i+1][j+1] = Math.min(matrix[i][j], Math.min(matrix[i+1][j], matrix[i][j+1])) + 1;
}
matrix[i+1][j+1] = Math.min(matrix[i+1][j+1], matrix[i1] ? matrix[i1][j1] + (i-i1-1) + 1 + (j-j1-1) : Infinity);
}
sd[s1[i-1]] = i;
}
return matrix[nLen1+1][nLen2+1];
}
catch (e) {
helpers.logerror(e);
}
},
showDistance (s1, s2) {
console.log(`Distance: ${s1} / ${s2} = ${this.distanceDamerauLevenshtein(s1, s2)})`);
},
// Suffix only
defineSuffixCode: function (sFlex, sStem) {
/*
Returns a string defining how to get stem from flexion
"n(sfx)"
with n: a char with numeric meaning, "0" = 0, "1" = 1, ... ":" = 10, etc. (See ASCII table.) Says how many letters to strip from flexion.
sfx [optional]: string to add on flexion
Examples:
"0": strips nothing, adds nothing
"1er": strips 1 letter, adds "er"
"2": strips 2 letters, adds nothing
*/
if (sFlex == sStem) {
return "0";
}
let jSfx = 0;
for (let i = 0; i < Math.min(sFlex.length, sStem.length); i++) {
if (sFlex[i] !== sStem[i]) {
break;
}
jSfx += 1;
}
return String.fromCharCode(sFlex.length-jSfx+48) + sStem.slice(jSfx);
},
changeWordWithSuffixCode: function (sWord, sSfxCode) {
if (sSfxCode == "0") {
return sWord;
}
return sSfxCode[0] == '0' ? sWord + sSfxCode.slice(1) : sWord.slice(0, -(sSfxCode.charCodeAt(0)-48)) + sSfxCode.slice(1);
},
// Prefix and suffix
defineAffixCode: function (sFlex, sStem) {
/*
UNTESTED!
Returns a string defining how to get stem from flexion. Examples:
"0" if stem = flexion
"stem" if no common substring
"n(pfx)/m(sfx)"
with n and m: chars with numeric meaning, "0" = 0, "1" = 1, ... ":" = 10, etc. (See ASCII table.) Says how many letters to strip from flexion.
pfx [optional]: string to add before the flexion
sfx [optional]: string to add after the flexion
*/
if (sFlex == sStem) {
return "0";
}
// is stem a substring of flexion?
let n = sFlex.indexOf(sStem);
if (n >= 0) {
return String.fromCharCode(n+48) + "/" + String.fromCharCode(sFlex.length-(sStem.length+n)+48);
}
// no, so we are looking for common substring
let sSubs = this.longestCommonSubstring(sFlex, sStem);
if (sSubs.length > 1) {
let iPos = sStem.indexOf(sSubs);
let sPfx = sStem.slice(0, iPos);
let sSfx = sStem.slice(iPos+sSubs.length);
let n = sFlex.indexOf(sSubs);
let m = sFlex.length - (sSubs.length+n);
return String.fromCharCode(n+48) + sPfx + "/" + String.fromCharCode(m+48) + sSfx;
}
return sStem;
},
changeWordWithAffixCode: function (sWord, sAffCode) {
if (sAffCode == "0") {
return sWord;
}
if (!sAffCode.includes("/")) {
return sAffCode;
}
let [sPfxCode, sSfxCode] = sAffCode.split('/');
sWord = sPfxCode.slice(1) + sWord.slice(sPfxCode.charCodeAt(0)-48);
return sSfxCode[0] == '0' ? sWord + sSfxCode.slice(1) : sWord.slice(0, -(sSfxCode.charCodeAt(0)-48)) + sSfxCode.slice(1);
}
};
if (typeof(exports) !== 'undefined') {
exports.longestCommonSubstring = str_transform.longestCommonSubstring;
exports.distanceDamerauLevenshtein = str_transform.distanceDamerauLevenshtein;
exports.distanceDamerauLevenshtein2 = str_transform.distanceDamerauLevenshtein2;
exports.showDistance = str_transform.showDistance;
exports.changeWordWithSuffixCode = str_transform.changeWordWithSuffixCode;
exports.changeWordWithAffixCode = str_transform.changeWordWithAffixCode;
exports.defineAffixCode = str_transform.defineAffixCode;
exports.defineSuffixCode = str_transform.defineSuffixCode;
}