add better handling of word patterns

keywords.yml.example

@@ -1,4 +1,19 @@
 keywords:
+    # will match just word 'meow' (requires word border on both sides)
     - meow
+
+    # will match plain regex
     - name: woof
       pattern: 'w[oa]+f'
+
+    # will match regex wraped with word borders
+    # will match 'hi, woof :3', 'woof!', 'i heard a woof' but not 'i like subwoofers'
+    - name: woof
+      pattern: 'w[oa]+f'
+      word: true
+
+    # will match any word starting with 'aqua' (aquarium, aquatic, aquaculture, etc...)
+    # requires word border on both sides too
+    - name: aqua
+      pattern: 'aqua.*?'
+      word: true

package.json

@@ -1,7 +1,7 @@
 {
     "name": "mtproto_exporter",
     "type": "module",
-    "version": "1.1.0",
+    "version": "1.2.0",
     "packageManager": "pnpm@10.6.5",
     "license": "MIT",
     "scripts": {

src/config.ts

@@ -1,5 +1,5 @@
 import type { OptionDefinition } from "command-line-args";
-import type { KeywordLike } from "./keywords.js";
+import type { RawKeywordLike } from "./keywords.js";
 import { readFile } from "node:fs/promises";
 import cmdline from "command-line-args";
 import yaml from "js-yaml";
@@ -12,7 +12,7 @@ export interface Configuration {
     watchFile: boolean;
     includePeers?: number[];
     excludePeers?: number[];
-    keywords?: KeywordLike[];
+    keywords?: RawKeywordLike[];
 }
 
 const optionDefinitions: OptionDefinition[] = [
@@ -54,19 +54,22 @@ if (cli["exclude-peers"]) {
     }
 }
 
-export async function readKeywords(filePath: string): Promise<KeywordLike[]> {
+export async function readKeywords(filePath: string): Promise<RawKeywordLike[]> {
     const doc = yaml.load(await readFile(filePath, "utf8")) as { keywords?: any[] };
 
     if (doc.keywords && doc.keywords.constructor.name === "Array") {
-        const keywords: KeywordLike[] = [];
+        const keywords: RawKeywordLike[] = [];
         for (const item of doc.keywords) {
             if (typeof item === "string") {
                 keywords.push(item);
-            } else if (typeof item === "object" && item.name && item.pattern) {
-                keywords.push({
-                    name: item.name,
-                    pattern: new RegExp(item.pattern, "gi"),
-                });
+            } else if (typeof item === "object" && typeof item.name === "string") {
+                if (typeof item.pattern === "string") {
+                    keywords.push({
+                        name: item.name,
+                        pattern: item.pattern,
+                        word: Boolean(item.word ?? false),
+                    });
+                }
             }
         }
         return keywords;

src/keywords.ts

@@ -3,37 +3,55 @@ import { PropagationAction } from "@mtcute/dispatcher";
 import { Counter } from "prom-client";
 import { config } from "./config.js";
 import { peersConfigFilter } from "./filters.js";
+import { escapeRegex } from "./utils.js";
 
-interface KeywordPattern {
+export interface RawKeywordPattern {
+    name: string;
+    pattern: string;
+    word: boolean;
+}
+
+export type RawKeywordLike = string | RawKeywordPattern;
+
+export interface KeywordPattern {
     name: string;
     pattern: RegExp;
 }
 
-export type KeywordLike = string | KeywordPattern;
+export function rawToPatterns(raw: RawKeywordLike[]): KeywordPattern[] {
+    const patterns: KeywordPattern[] = [];
+    for (const keyword of raw) {
+        let pattern;
+        let name;
+        let addBorders = false;
 
-export function newWordsCounter(dp: Dispatcher) {
-    const counter = new Counter({
-        name: "messenger_dialog_words_count",
-        help: "Number of words in messages since exporter startup",
-        labelNames: ["peerId", "word"],
-    });
-    dp.onNewMessage(peersConfigFilter(config), async (msg) => {
-        const words = msg.text.toLowerCase().split(" ");
-        for (const w of words) {
-            counter.inc({
-                peerId: msg.chat.id,
-                word: w,
-            });
+        if (typeof keyword === "string") {
+            pattern = escapeRegex(keyword);
+            name = keyword;
+            addBorders = true;
+        } else {
+            pattern = keyword.pattern;
+            name = keyword.name;
+            addBorders = keyword.word;
         }
-        return PropagationAction.Continue;
-    });
-    return counter;
+
+        const wordBorder = escapeRegex("!\"#$%&'()*+,-./:;<=>?@[\\]^_`{|}~");
+        const borderStart = addBorders ? `(?:[${wordBorder}\\s]|^)` : "";
+        const borderEnd = addBorders ? `(?:[${wordBorder}\\s]|$)` : "";
+
+        patterns.push({
+            name,
+            pattern: new RegExp(borderStart + pattern + borderEnd),
+        });
+    }
+
+    return patterns;
 }
 
 export class KeywordsCounter extends Counter {
     private _dp: Dispatcher;
-    private _keywords: KeywordLike[];
-    constructor(dp: Dispatcher, keywords: KeywordLike[] = []) {
+    private _keywords: KeywordPattern[];
+    constructor(dp: Dispatcher, keywords: KeywordPattern[] = []) {
         super({
             name: "messenger_dialog_keywords_count",
             help: "Number of keywords found in messages since exporter startup",
@@ -42,24 +60,18 @@ export class KeywordsCounter extends Counter {
         this._dp = dp;
         this._keywords = keywords;
 
-        dp.onNewMessage(peersConfigFilter(config), async (msg) => {
+        this._dp.onNewMessage(peersConfigFilter(config), async (msg) => {
             for (const kw of this._keywords) {
-                let count;
-                let kwname;
-                if (typeof kw === "string") {
-                    const words = msg.text.toLowerCase().split(" ");
-                    count = words.filter(w => w === kw).length;
-                    kwname = kw;
-                } else {
-                    count = (msg.text.match(kw.pattern) || []).length;
-                    kwname = kw.name;
-                }
+                const count = (msg.text.match(kw.pattern) ?? []).length;
+
+                // this will prevent from flooding metrics with keywords that had never been triggered yet
                 if (count === 0) {
                     continue;
                 }
+
                 this.inc({
                     peerId: msg.chat.id,
-                    keyword: kwname,
+                    keyword: kw.name,
                 }, count);
             }
             return PropagationAction.Continue;
@@ -70,7 +82,7 @@ export class KeywordsCounter extends Counter {
         return this._keywords;
     }
 
-    public setKeywords(keywords: KeywordLike[]) {
+    public setKeywords(keywords: KeywordPattern[]) {
         this._keywords = keywords;
     }
 }

src/main.ts

@@ -5,6 +5,7 @@ import { collectDefaultMetrics, Registry } from "prom-client";
 
 import { config, readKeywords } from "./config.js";
 import * as env from "./env.js";
+import { rawToPatterns } from "./keywords.js";
 import * as metrics from "./metrics.js";
 import MetricsServer from "./server.js";
 
@@ -36,7 +37,7 @@ registry.registerMetric(metrics.newUnreadCountGauge(tg));
 registry.registerMetric(metrics.newMessagesCounter(dp));
 
 if (config.keywords) {
-    const counter = new metrics.KeywordsCounter(dp, config.keywords);
+    const counter = new metrics.KeywordsCounter(dp, rawToPatterns(config.keywords));
     registry.registerMetric(counter);
 
     if (config.watchFile) {
@@ -47,7 +48,7 @@ if (config.keywords) {
             console.log("[watch-file] Keywords file was updated. Re-reading keywords configuration...");
             try {
                 config.keywords = await readKeywords(config.keywordsFile);
-                counter.setKeywords(config.keywords);
+                counter.setKeywords(rawToPatterns(config.keywords));
             } catch (e) {
                 console.error("Failed to read keywords file", config.keywordsFile, e);
             }

src/metrics.ts

@@ -5,7 +5,7 @@ import { Counter, Gauge } from "prom-client";
 
 import { config } from "./config.js";
 import { peersConfigBoolFilter, peersConfigFilter } from "./filters.js";
-import { KeywordsCounter, newWordsCounter } from "./keywords.js";
+import { KeywordsCounter } from "./keywords.js";
 
 function newMessagesCounter(dp: Dispatcher) {
     const counter = new Counter({
@@ -67,6 +67,25 @@ function newUnreadCountGauge(tg: TelegramClient) {
     return gauge;
 }
 
+function newWordsCounter(dp: Dispatcher) {
+    const counter = new Counter({
+        name: "messenger_dialog_words_count",
+        help: "Number of words in messages since exporter startup",
+        labelNames: ["peerId", "word"],
+    });
+    dp.onNewMessage(peersConfigFilter(config), async (msg) => {
+        const words = msg.text.toLowerCase().split(" ");
+        for (const w of words) {
+            counter.inc({
+                peerId: msg.chat.id,
+                word: w,
+            });
+        }
+        return PropagationAction.Continue;
+    });
+    return counter;
+}
+
 export {
     KeywordsCounter,
     newMessagesCounter,

src/utils.ts

@@ -0,0 +1,3 @@
+export function escapeRegex(text: string) {
+    return text.replace(/[/\-\\^$*+?.()|[\]{}]/g, "\\$&");
+}