feat: return query that identifies all highlights

gajus · Nov 10, 2021 · 5ae5fa9 · 5ae5fa9
1 parent c2a7e3d
commit 5ae5fa9
Show file tree

Hide file tree

Showing 5 changed files with 125 additions and 24 deletions.
diff --git a/README.md b/README.md
@@ -17,6 +17,8 @@ Lightweight and performant Lucene-like parser and search engine.
   * [Wildcard matching](#wildcard-matching)
   * [Logical Operators](#logical-operators)
 * [Compatibility with Lucene](#compatibility-with-lucene)
+* [Recipes](#recipes)
+  * [Highlighting matches](#highlighting-matches)
 * [Development](#development)
 
 ## Usage
@@ -76,8 +78,8 @@ Highlight matching fields and substrings:
 test(highlight('name:john'), persons[0]);
 // [
 //   {
-//     keyword: 'John',
 //     path: 'name',
+//     query: /(John)/,
 //   }
 // ]
 test(highlight('height:180'), persons[0]);
@@ -242,6 +244,12 @@ The following Lucene abilities are not supported:
 * [Proximity Searches](https://lucene.apache.org/core/2_9_4/queryparsersyntax.html#Proximity%20Searches)
 * [Boosting a Term](https://lucene.apache.org/core/2_9_4/queryparsersyntax.html#Boosting%20a%20Term)
 
+## Recipes
+
+### Highlighting matches
+
+Consider using [`highlight-words`](https://github.com/tricinel/highlight-words) package to highlight Liqe matches.
+
 ## Development
 
 ### Compiling Parser

diff --git a/src/highlight.ts b/src/highlight.ts
@@ -1,16 +1,25 @@
+import {
+  escapeRegexString,
+} from './escapeRegexString';
 import {
   internalFilter,
 } from './internalFilter';
 import type {
   Ast,
   Highlight,
+  InternalHighlight,
 } from './types';
 
+type AggregatedHighlight = {
+  keywords: string[],
+  path: string,
+};
+
 export const highlight = <T extends Object>(
   ast: Ast,
   data: T,
 ): Highlight[] => {
-  const highlights = [];
+  const highlights: InternalHighlight[] = [];
 
   internalFilter(
     ast,
@@ -20,5 +29,39 @@ export const highlight = <T extends Object>(
     highlights,
   );
 
-  return highlights;
+  const aggregatedHighlights: AggregatedHighlight[] = [];
+
+  for (const highlightNode of highlights) {
+    let aggregatedHighlight = aggregatedHighlights.find((maybeTarget) => {
+      return maybeTarget.path === highlightNode.path;
+    });
+
+    if (!aggregatedHighlight) {
+      aggregatedHighlight = {
+        keywords: [],
+        path: highlightNode.path,
+      };
+
+      aggregatedHighlights.push(aggregatedHighlight);
+    }
+
+    if (highlightNode.keyword) {
+      aggregatedHighlight.keywords.push(highlightNode.keyword);
+    }
+  }
+
+  return aggregatedHighlights.map((aggregatedHighlight) => {
+    if (aggregatedHighlight.keywords.length > 0) {
+      return {
+        path: aggregatedHighlight.path,
+        query: new RegExp('(' + aggregatedHighlight.keywords.map((keyword) => {
+          return escapeRegexString(keyword.trim());
+        }).join('|') + ')'),
+      };
+    }
+
+    return {
+      path: aggregatedHighlight.path,
+    };
+  });
 };
diff --git a/src/internalFilter.ts b/src/internalFilter.ts
@@ -9,7 +9,7 @@ import {
 } from './parseRegex';
 import type {
   Ast,
-  Highlight,
+  InternalHighlight,
   InternalTest,
   Range,
   RelationalOperator,
@@ -121,7 +121,7 @@ const testValue = (
   value: unknown,
   resultFast: boolean,
   path: string[],
-  highlights: Highlight[],
+  highlights: InternalHighlight[],
 ) => {
   if (Array.isArray(value)) {
     let foundMatch = false;
@@ -181,7 +181,7 @@ const testField = <T extends Object>(
   ast: Ast,
   resultFast: boolean,
   path: string[],
-  highlights: Highlight[],
+  highlights: InternalHighlight[],
 ): boolean => {
   if (!ast.test) {
     ast.test = createValueTest(ast);
@@ -251,7 +251,7 @@ export const internalFilter = <T extends Object>(
   rows: readonly T[],
   resultFast: boolean = true,
   path: string[] = [],
-  highlights: Highlight[] = [],
+  highlights: InternalHighlight[] = [],
 ): readonly T[] => {
   if (ast.field) {
     return rows.filter((row) => {

diff --git a/src/types.ts b/src/types.ts
@@ -21,9 +21,14 @@ export type Ast = {
   test?: InternalTest,
 };
 
-export type Highlight = {
+export type InternalHighlight = {
   keyword?: string,
   path: string,
 };
 
+export type Highlight = {
+  path: string,
+  query?: RegExp,
+};
+
 export type InternalTest = (value: unknown) => boolean | string;
diff --git a/test/liqe/highlight.ts b/test/liqe/highlight.ts
@@ -23,11 +23,11 @@ test.skip(
   },
   [
     {
-      keyword: 'foo@bar.com',
       path: 'email',
+      query: /(foo@bar.com)/,
     },
     {
-      keyword: 'foo bar',
+      keyword: /(foo bar)/,
       path: 'name',
     },
   ],
@@ -43,12 +43,12 @@ test(
   },
   [
     {
-      keyword: 'foo',
       path: 'email',
+      query: /(foo)/,
     },
     {
-      keyword: 'foo',
       path: 'name',
+      query: /(foo)/,
     },
   ],
 );
@@ -62,8 +62,8 @@ test(
   },
   [
     {
-      keyword: 'foo',
       path: 'name',
+      query: /(foo)/,
     },
   ],
 );
@@ -77,8 +77,8 @@ test(
   },
   [
     {
-      keyword: 'Foo',
       path: 'name',
+      query: /(Foo)/,
     },
   ],
 );
@@ -93,8 +93,8 @@ test(
   },
   [
     {
-      keyword: 'bar',
       path: 'name',
+      query: /(bar)/,
     },
     {
       path: 'height',
@@ -111,8 +111,8 @@ test(
   },
   [
     {
-      keyword: 'foo',
       path: 'name',
+      query: /(foo)/,
     },
   ],
 );
@@ -126,8 +126,8 @@ test(
   },
   [
     {
-      keyword: 'foo',
       path: 'name',
+      query: /(foo)/,
     },
   ],
 );
@@ -141,8 +141,8 @@ test(
   },
   [
     {
-      keyword: 'foo',
       path: 'name',
+      query: /(foo)/,
     },
   ],
 );
@@ -156,11 +156,11 @@ test.skip(
   },
   [
     {
-      keyword: 'foo',
       path: 'name',
+      query: /(foo)/,
     },
     {
-      keyword: 'bar',
+      keyword: /(bar)/,
       path: 'name',
     },
   ],
@@ -221,8 +221,8 @@ test(
   },
   [
     {
-      keyword: 'bar',
       path: 'tags.1',
+      query: /(bar)/,
     },
   ],
 );
@@ -240,12 +240,12 @@ test(
   },
   [
     {
-      keyword: 'ba',
       path: 'tags.1',
+      query: /(ba)/,
     },
     {
-      keyword: 'ba',
       path: 'tags.2',
+      query: /(ba)/,
     },
   ],
 );
@@ -279,8 +279,53 @@ test(
   },
   [
     {
-      keyword: 'foo',
       path: 'name',
+      query: /(foo)/,
+    },
+  ],
+);
+
+test(
+  'aggregates multiple highlights',
+  testQuery,
+  'foo AND bar AND baz',
+  {
+    name: 'foo bar baz',
+  },
+  [
+    {
+      path: 'name',
+      query: /(foo|bar|baz)/,
+    },
+  ],
+);
+
+test(
+  'aggregates multiple highlights (phrases)',
+  testQuery,
+  '"foo bar" AND baz',
+  {
+    name: 'foo bar baz',
+  },
+  [
+    {
+      path: 'name',
+      query: /(foo bar|baz)/,
+    },
+  ],
+);
+
+test(
+  'aggregates multiple highlights (escaping)',
+  testQuery,
+  '"(foo bar)" AND baz',
+  {
+    name: '(foo bar) baz',
+  },
+  [
+    {
+      path: 'name',
+      query: /(\(foo bar\)|baz)/,
     },
   ],
 );