fix word borders in regexps; add regexp flags to config file

keywords.yml.example

@@ -5,6 +5,11 @@ keywords:
     # will match plain regex
     - name: woof
       pattern: 'w[oa]+f'
+      # you can also specify flags for your regex
+      flags:
+        global: true
+        multi_line: false
+        insensitive: true
 
     # will match regex wraped with word borders
     # will match 'hi, woof :3', 'woof!', 'i heard a woof' but not 'i like subwoofers'

package.json

@@ -1,7 +1,7 @@
 {
     "name": "mtproto_exporter",
     "type": "module",
-    "version": "1.4.2",
+    "version": "1.5.0",
     "packageManager": "pnpm@10.6.5",
     "license": "MIT",
     "scripts": {

src/config.ts

@@ -94,11 +94,32 @@ export async function readKeywords(filePath: string): Promise<RawKeywordLike[]>
                 keywords.push(item);
             } else if (typeof item === "object" && typeof item.name === "string") {
                 if (typeof item.pattern === "string") {
-                    keywords.push({
+                    let result = {
                         name: item.name,
                         pattern: item.pattern,
                         word: Boolean(item.word ?? false),
-                    });
+                        flags: {
+                            global: true,
+                            multi_line: false,
+                            insensitive: true,
+                        }
+                    };
+
+                    if (typeof item.flags === "object") {
+                        if (typeof item.flags.global === "boolean") {
+                            result.flags.global = item.flags.global;
+                        }
+
+                        if (typeof item.flags.multi_line === "boolean") {
+                            result.flags.multi_line = item.flags.multi_line;
+                        }
+
+                        if (typeof item.flags.insensitive === "boolean") {
+                            result.flags.insensitive = item.flags.insensitive;
+                        }
+                    }
+
+                    keywords.push(result);
                 }
             }
         }

src/metrics/keywords.ts

@@ -11,6 +11,11 @@ export interface RawKeywordPattern {
     name: string;
     pattern: string;
     word: boolean;
+    flags: {
+        global: boolean;
+        multi_line: boolean;
+        insensitive: boolean;
+    }
 }
 
 export type RawKeywordLike = string | RawKeywordPattern;
@@ -26,6 +31,7 @@ export function rawToPatterns(raw: RawKeywordLike[]): KeywordPattern[] {
         let pattern;
         let name;
         let addBorders = false;
+        let flags = "giu";
 
         if (typeof keyword === "string") {
             pattern = escapeRegex(keyword);
@@ -35,15 +41,28 @@ export function rawToPatterns(raw: RawKeywordLike[]): KeywordPattern[] {
             pattern = keyword.pattern;
             name = keyword.name;
             addBorders = keyword.word;
+
+            flags = "u";
+            if (keyword.flags.global) {
+                flags += "g";
+            }
+
+            if (keyword.flags.insensitive) {
+                flags += "i";
+            }
+
+            if (keyword.flags.multi_line) {
+                flags += "m";
+            }
         }
 
         const wordBorder = escapeRegex("!\"#$%&'()*+,-./:;<=>?@[\\]^_`{|}~");
-        const borderStart = addBorders ? `(?:[${wordBorder}\\s]|^)` : "";
+        const borderStart = addBorders ? `(?<=[${wordBorder}\\s]|^)` : "";
-        const borderEnd = addBorders ? `(?:[${wordBorder}\\s]|$)` : "";
+        const borderEnd = addBorders ? `(?=[${wordBorder}\\s]|$)` : "";
 
         patterns.push({
             name,
-            pattern: new RegExp(borderStart + pattern + borderEnd),
+            pattern: new RegExp(`${borderStart}(?:${pattern})${borderEnd}`, flags),
         });
     }