From 62244ffeae129207cd2b677f8310aab5e2c1a1e9 Mon Sep 17 00:00:00 2001
From: Claiyc <58397976+Claiyc@users.noreply.github.com>
Date: Tue, 13 Dec 2022 20:19:24 +0100
Subject: [PATCH] fix(regex): remove substring caching & add genMatches caching
 (#68)

---
 src/proc/regex/Regex.ts        | 13 ++++++++++++-
 src/proc/regex/RegexHandler.ts | 32 ++++++++++++++------------------
 2 files changed, 26 insertions(+), 19 deletions(-)

diff --git a/src/proc/regex/Regex.ts b/src/proc/regex/Regex.ts
index 1648afe..97f6df2 100644
--- a/src/proc/regex/Regex.ts
+++ b/src/proc/regex/Regex.ts
@@ -13,6 +13,9 @@ export abstract class Regex {
         rating: number;
         match: { index: number; element: string };
     };
+    protected lastRegex: RegExp; // used for checking whether regex has changed -> regenerating matches
+    protected lastMatchesNum: number; // used for checking whether matchesNum has changed -> regenerating matches
+    protected generatedMatches: string[]; // generated matches
 
     protected constructor() {
         this.id = 0;
@@ -23,6 +26,9 @@ export abstract class Regex {
         this.matchesNum = 10000;
         this.substrings = [];
         this.lastBestMatch = { rating: -1, match: { index: -1, element: '' } };
+        this.lastRegex = new RegExp('');
+        this.lastMatchesNum = this.matchesNum;
+        this.generatedMatches = [];
     }
 
     public getId(): number {
@@ -195,9 +201,14 @@ export abstract class Regex {
                 });
                 break;
             case Matching.APPROX:
+                if (this.generatedMatches.length === 0 || this.lastRegex.toString() !== this.regex.toString() || this.matchesNum !== this.lastMatchesNum) {
+                    this.generatedMatches = this.genMatches();
+                    this.lastRegex = this.regex;
+                    this.lastMatchesNum = this.matchesNum;
+                }
                 bestMatch = RegexHandler.approxMatching(
                     this.substrings,
-                    this.genMatches()
+                    this.generatedMatches
                 );
                 break;
         }
diff --git a/src/proc/regex/RegexHandler.ts b/src/proc/regex/RegexHandler.ts
index 23be3e6..b07c096 100644
--- a/src/proc/regex/RegexHandler.ts
+++ b/src/proc/regex/RegexHandler.ts
@@ -44,14 +44,12 @@ export class RegexHandler {
 
         // set required substrings for constraint regex & apply similarity conversion
         for (let i = 0; i < constraintRegex.length; i++) {
-            if (constraintRegex[i].getSubstrings().length === 0) {  // if substrings are not already set
-                if (constraintRegex[i].getSlicing() === Slicing.SUBSTR && allSubstrings.length !== 0) {
-                    constraintRegex[i].setSubstrings(allSubstrings.slice());   // clone array
-                } else if (constraintRegex[i].getSlicing() === Slicing.SPACES && spacesSubstrings.length !== 0) {
-                    constraintRegex[i].setSubstrings(spacesSubstrings.slice());   // clone array
-                } else {
-                    constraintRegex[i].genSubstrings(data);
-                }
+            if (constraintRegex[i].getSlicing() === Slicing.SUBSTR && allSubstrings.length !== 0) {
+                constraintRegex[i].setSubstrings(allSubstrings.slice());   // clone array
+            } else if (constraintRegex[i].getSlicing() === Slicing.SPACES && spacesSubstrings.length !== 0) {
+                constraintRegex[i].setSubstrings(spacesSubstrings.slice());   // clone array
+            } else {
+                constraintRegex[i].genSubstrings(data);
             }
             if (constraintRegex[i].getSlicing() === Slicing.SUBSTR) {
                 allSubstrings = constraintRegex[i].getSubstrings(); // cache substrings
@@ -89,18 +87,16 @@ export class RegexHandler {
         }
 
         // set required substrings for value regex & apply similarity conversion
-        if (valueRegex.getSubstrings().length === 0) {  // if substrings are not already set
-            if (highestLowIndex === 0 && lowestHighIndex === data.length) { // if no constraint regex
-                if (valueRegex.getSlicing() === Slicing.SUBSTR && allSubstrings.length !== 0) {
-                    valueRegex.setSubstrings(allSubstrings.slice());   // clone array
-                } else if (valueRegex.getSlicing() === Slicing.SPACES && spacesSubstrings.length !== 0) {
-                    valueRegex.setSubstrings(spacesSubstrings.slice());   // clone array
-                } else {
-                    valueRegex.genSubstrings(data);
-                }
+        if (highestLowIndex === 0 && lowestHighIndex === data.length) { // if no constraint regex
+            if (valueRegex.getSlicing() === Slicing.SUBSTR && allSubstrings.length !== 0) {
+                valueRegex.setSubstrings(allSubstrings.slice());   // clone array
+            } else if (valueRegex.getSlicing() === Slicing.SPACES && spacesSubstrings.length !== 0) {
+                valueRegex.setSubstrings(spacesSubstrings.slice());   // clone array
             } else {
-                valueRegex.genSubstrings(data.slice(highestLowIndex, lowestHighIndex));
+                valueRegex.genSubstrings(data);
             }
+        } else {
+            valueRegex.genSubstrings(data.slice(highestLowIndex, lowestHighIndex));
         }
 
         valueRegex.applySimilarity();