feat: adds search hits, elapsed time, and count

.eslintignore

@@ -4,4 +4,6 @@ jest.config.js
 .eslintrc.cjs
 *.yaml
 *.md
-*.snap
+*.snap
+*.csv
+*.sh

package.json

@@ -41,5 +41,12 @@
     "ts-jest": "^28.0.2",
     "ts-node": "^10.7.0",
     "typescript": "^4.6.4"
+  },
+  "pnpm": {
+    "peerDependencyRules": {
+      "ignoreMissing": [
+        "typescript"
+      ]
+    }
   }
 }

packages/lyra/package.json

@@ -10,6 +10,7 @@
   "bugs": {
     "url": "https://github.com/nearform/lyra"
   },
+  "main": "./src/lyra.ts",
   "dependencies": {
     "fastq": "^1.13.0",
     "nanoid": "3.3.4"

packages/lyra/src/lyra.ts

@@ -4,6 +4,7 @@ import * as fastq from "fastq";
 import { Trie } from "./prefix-tree/trie";
 import * as ERRORS from "./errors";
 import { tokenize } from "./tokenizer";
+import { formatNanoseconds } from "./utils";
 
 export type PropertyType = "string" | "number" | "boolean";
 
@@ -31,6 +32,12 @@ type QueueDocParams = {
   doc: object;
 };
 
+type SearchResult = Promise<{
+  count: number;
+  hits: object[];
+  elapsed: string;
+}>;
+
 export class Lyra {
   private schema: PropertiesSchema;
   private docs: LyraDocs = new Map();
@@ -60,10 +67,12 @@ export class Lyra {
     }
   }
 
-  async search(params: SearchParams) {
+  async search(params: SearchParams): SearchResult {
     const tokens = tokenize(params.term).values();
     const indices = this.getIndices(params.properties);
-    const results = [];
+    const results: object[] = [];
+
+    const timeStart = process.hrtime.bigint();
 
     for (const token of tokens) {
       for (const index of indices) {
@@ -81,7 +90,11 @@ export class Lyra {
       }
     }
 
-    return results;
+    return {
+      elapsed: formatNanoseconds(process.hrtime.bigint() - timeStart),
+      hits: results.slice(params.offset ?? 0, params.limit ?? 10), // @todo avoid getting all results and slicing them
+      count: results.length,
+    };
   }
 
   private getIndices(indices: SearchParams["properties"]): string[] {
@@ -108,15 +121,14 @@ export class Lyra {
     return indices as string[];
   }
 
-  private async _search(params: SearchParams & { index: string }) {
+  private async _search(
+    params: SearchParams & { index: string }
+  ): Promise<object[]> {
     const idx = this.index.get(params.index);
     const searchResult = idx?.find(params.term);
-    const results = [];
-    let count = 0;
+    const results: object[] = [];
 
     for (const key in searchResult) {
-      if (params.limit && count > params.limit) break;
-
       const docs: string[] = [];
 
       for (const id of (searchResult as any)[key]) {
@@ -126,7 +138,6 @@ export class Lyra {
         docs.push({ id, ...fullDoc });
       }
 
-      count++;
       results.push(docs);
     }
 

packages/repl/commands.ts

@@ -0,0 +1,22 @@
+import yargs from "yargs";
+
+export function commands(input: string) {
+  return yargs(input as unknown as string[])
+    .command("search <text...>", "searches a text")
+    .option("limit", {
+      alias: "l",
+      type: "number",
+      default: 10,
+    })
+    .option("offset", {
+      alias: "o",
+      type: "number",
+      default: 0,
+    })
+    .option("properties", {
+      alias: "p",
+      type: "string",
+      default: "*",
+    })
+    .parse();
+}

packages/repl/get-datasets.sh

@@ -0,0 +1,4 @@
+wget https://datasets.imdbws.com/title.episode.tsv.gz -O ./datasets/title.basics.tsv.gz
+gzip -d ./datasets/title.basics.tsv.gz
+head -n1000000 ./datasets/title.basics.tsv >> ./datasets/title.basics.short.tsv
+rm ./datasets/title.basics.tsv

packages/repl/package.json

@@ -0,0 +1,24 @@
+{
+  "name": "@nearform/lyra-repl",
+  "description": "Repl for the Lyra search engine",
+  "version": "0.0.1",
+  "author": {
+    "name": "Michele Riva",
+    "email": "ciao@micheleriva.it",
+    "url": "https://github.com/MicheleRiva"
+  },
+  "bugs": {
+    "url": "https://github.com/nearform/lyra"
+  },
+  "dependencies": {
+    "@nearform/lyra": "workspace:^",
+    "csv": "^6.0.5",
+    "yargs": "^17.5.1"
+  },
+  "devDependencies": {
+    "@types/jest": "^27.5.0",
+    "@types/yargs": "^17.0.10",
+    "jest": "^28.1.0",
+    "ts-jest": "^28.0.2"
+  }
+}

packages/repl/repl.ts

@@ -0,0 +1,68 @@
+import readline from "readline";
+import { Lyra } from "@nearform/lyra";
+import dataset from "./datasets/reviews.json";
+import { commands } from "./commands";
+
+type Dataset = {
+  "": string;
+  Rating: string;
+  "Review Text": string;
+  "Division Name": string;
+  Title: string;
+  "Recommended IND": string;
+  Age: string;
+  "Department Name": string;
+  "Class Name": string;
+  "Positive Feedback Count": string;
+  "Clothing ID": string;
+};
+
+const rl = readline.createInterface({
+  input: process.stdin,
+  output: process.stdout,
+  terminal: false,
+});
+
+const db = new Lyra({
+  schema: {
+    rating: "string",
+    review: "string",
+    title: "string",
+  },
+});
+
+async function load() {
+  for (const data of dataset as Dataset[]) {
+    await db.insert({
+      rating: data.Rating,
+      review: data["Review Text"],
+      title: data.Title,
+    });
+  }
+}
+
+async function parseLine(input: string) {
+  const cmd = await commands(input);
+  const tokens = (cmd as any).text.join(", ");
+
+  const properties = cmd.properties === "*" ? "*" : cmd.properties.split(",");
+
+  const result = await db.search({
+    term: tokens,
+    limit: cmd.limit,
+    offset: cmd.offset,
+    properties,
+  });
+
+  console.log(result);
+}
+
+async function start() {
+  console.log("loading dataset...");
+  await load();
+  console.log(`${(dataset as any[]).length} reviews loaded`);
+
+  rl.on("line", parseLine);
+}
+
+start();