Merge pull request #14 from hadynz/aho-corasick

Use Aho-Corasick algorithm for searching

manifest.json

@@ -2,7 +2,7 @@
   "id": "obsidian-sidekick",
   "name": "Sidekick",
   "description": "A companion to identify hidden connections that match your tags and pages",
-  "version": "1.1.1",
+  "version": "1.2.0",
   "minAppVersion": "0.13.8",
   "author": "Hady Osman",
   "authorUrl": "https://hady.geek.nz",

package.json

@@ -1,6 +1,6 @@
 {
   "name": "obsidian-sidekick",
-  "version": "1.1.1",
+  "version": "1.2.0",
   "description": "A companion to identify hidden connections that match your tags and pages",
   "main": "src/index.ts",
   "repository": {
@@ -35,7 +35,6 @@
     "@types/faker": "^5.5.8",
     "@types/jest": "^26.0.22",
     "@types/lodash": "^4.14.178",
-    "@types/lunr": "^2.3.4",
     "@types/webpack": "^5.28.0",
     "@typescript-eslint/eslint-plugin": "^4.22.0",
     "@typescript-eslint/parser": "^4.22.0",
@@ -69,8 +68,8 @@
     "*.{js,css,md}": "prettier --write"
   },
   "dependencies": {
+    "@tanishiking/aho-corasick": "^0.0.1",
     "lodash": "^4.17.21",
-    "lunr": "^2.3.9",
     "tippy.js": "^6.3.7"
   }
 }

src/indexing/indexModels.ts

@@ -1,45 +1,36 @@
 import { TFile } from 'obsidian';
 
-import { stemmer } from '../utils/stemmer';
-
 export interface SearchIndex {
   replaceText: string;
   originalText: string;
-  stem: string;
 }
 
 export class TagIndex implements SearchIndex {
   public readonly originalText: string;
   public readonly replaceText: string;
-  public readonly stem: string;
 
   constructor(tag: string) {
     this.originalText = tag.replace(/#/, '');
     this.replaceText = tag;
-    this.stem = stemmer(this.originalText);
   }
 }
 
 export class AliasIndex implements SearchIndex {
   public readonly originalText: string;
   public readonly replaceText: string;
-  public readonly stem: string;
 
   constructor(file: TFile, word: string) {
     this.originalText = word;
     this.replaceText = `[[${file.basename}|${word}]]`;
-    this.stem = stemmer(this.originalText);
   }
 }
 
 export class PageIndex implements SearchIndex {
   public readonly originalText: string;
   public readonly replaceText: string;
-  public readonly stem: string;
 
   constructor(file: TFile) {
     this.originalText = file.basename;
     this.replaceText = `[[${file.basename}]]`;
-    this.stem = stemmer(this.originalText);
   }
 }

src/indexing/indexer.ts

@@ -17,7 +17,7 @@ export class Indexer {
     return this.indexAllTags(allFiles)
       .concat(allFiles.map((file) => this.indexFile(file)).flat())
       .reduce((acc: Index, index) => {
-        return { ...acc, [index.stem]: index };
+        return { ...acc, [index.originalText.toLowerCase()]: index };
       }, {});
   }
 

src/search/index.ts

@@ -1,4 +1,4 @@
-import lunr from 'lunr';
+import { Trie, Emit } from '@tanishiking/aho-corasick';
 
 import { Indexer, Index } from '../indexing/indexer';
 
@@ -8,58 +8,34 @@ type SearchResult = {
   replaceText: string;
 };
 
-// Any arbitrary key to use for the search index
-const DocumentKey = 'text';
-
 export default class Search {
   constructor(private indexer: Indexer) {}
 
   public find(text: string): SearchResult[] {
-    // Redact text that we don't want to be searched
-    const redactedText = this.redactText(text);
+    const indices = this.indexer.getIndices();
 
-    const idx = lunr(function () {
-      this.metadataWhitelist = ['position'];
-      this.ref(DocumentKey);
-      this.field(DocumentKey);
-      this.add({ [DocumentKey]: redactedText });
+    const trie = new Trie(Object.keys(indices), {
+      allowOverlaps: false,
+      onlyWholeWords: true,
+      caseInsensitive: true,
     });
 
-    const indices = this.indexer.getIndices();
+    // Redact text that we don't want to be searched
+    const redactedText = this.redactText(text);
 
-    const results = idx.query(function () {
-      Object.keys(indices).map((index) => {
-        this.term(index, {});
-      });
-    });
+    const results = trie.parseText(redactedText);
 
     return this.toSearchResults(results, indices);
   }
 
-  private toSearchResults(results: lunr.Index.Result[], indices: Index): SearchResult[] {
-    if (results.length === 0) {
-      return [];
-    }
-
-    // We will always ever only have one result as we only index one document
-    const indexHits = results[0].matchData.metadata;
-
-    return Object.keys(indexHits)
-      .filter((indexHit) => this.existsInIndex(indexHit, indices))
-      .reduce((acc: SearchResult[], indexHit) => {
-        const positions: number[][] = indexHits[indexHit][DocumentKey].position;
-
-        const searchResults = positions.map(
-          (position): SearchResult => ({
-            start: position[0],
-            end: position[0] + position[1],
-            replaceText: indices[indexHit].replaceText,
-          })
-        );
-
-        acc.push(...searchResults);
-        return acc;
-      }, [])
+  private toSearchResults(results: Emit[], indices: Index): SearchResult[] {
+    return results
+      .filter((result) => this.existsInIndex(result.keyword, indices))
+      .map((result) => ({
+        start: result.start,
+        end: result.end + 1,
+        replaceText: indices[result.keyword].replaceText,
+      }))
       .sort((a, b) => a.start - b.start); // Must sort by start position to prepare for highlighting
   }