fix(cli): split block extraction into focused functions and fix regex to match exact keywords

Author: AndreyHirsa

packages/cli/src/cli/loaders/vtt.ts

@@ -2,44 +2,36 @@ import webvtt from "node-webvtt";
 import { ILoader } from "./_types";
 import { createLoader } from "./_utils";
 
-// node-webvtt doesn't handle STYLE/REGION blocks — strip them before parsing
-function extractUnsupportedBlocks(input: string): {
-  cleaned: string;
-  blocks: string[];
-} {
-  const parts = input.replace(/\r\n/g, "\n").split("\n\n");
-  const blocks: string[] = [];
-  const kept: string[] = [];
-
-  const unsupportedRegex = /^(?:STYLE|REGION)/;
+const UNSUPPORTED_BLOCK_REGEX = /^(?:STYLE|REGION)\s*$/;
 
-  for (const part of parts) {
-    const firstLine = part.trimStart().split("\n", 1)[0];
+function isUnsupportedBlock(block: string): boolean {
+  const firstLine = block.trimStart().split("\n", 1)[0];
+  return UNSUPPORTED_BLOCK_REGEX.test(firstLine);
+}
 
-    if (unsupportedRegex.test(firstLine)) {
-      blocks.push(part);
-    } else {
-      kept.push(part);
-    }
-  }
+// node-webvtt doesn't handle STYLE/REGION blocks — strip them before parsing
+function stripUnsupportedBlocks(input: string): string {
+  return input
+    .replace(/\r\n/g, "\n")
+    .split("\n\n")
+    .filter((part) => !isUnsupportedBlock(part))
+    .join("\n\n");
+}
 
-  return { cleaned: kept.join("\n\n"), blocks };
+function getUnsupportedBlocks(input: string): string[] {
+  return input.replace(/\r\n/g, "\n").split("\n\n").filter(isUnsupportedBlock);
 }
 
 export default function createVttLoader(): ILoader<
   string,
   Record<string, any>
 > {
-  let savedBlocks: string[] = [];
-
   return createLoader({
     async pull(locale, input) {
       if (!input) {
         return ""; // if VTT file does not exist yet we can not parse it - return empty string
       }
-      const { cleaned, blocks } = extractUnsupportedBlocks(input);
-      savedBlocks = blocks;
-      const vtt = webvtt.parse(cleaned)?.cues;
+      const vtt = webvtt.parse(stripUnsupportedBlocks(input))?.cues;
       if (Object.keys(vtt).length === 0) {
         return {};
       } else {
@@ -50,7 +42,7 @@ export default function createVttLoader(): ILoader<
         }, {});
       }
     },
-    async push(locale, payload) {
+    async push(locale, payload, originalInput) {
       const output = Object.entries(payload).map(([key, text]) => {
         const [id, timeRange, identifier] = key.split("#");
         const [startTime, endTime] = timeRange.split("-");
@@ -71,11 +63,13 @@ export default function createVttLoader(): ILoader<
       };
 
       const compiled = webvtt.compile(input);
-      if (savedBlocks.length === 0) return compiled;
 
       // Re-insert STYLE/REGION blocks after the WEBVTT header
+      const blocks = getUnsupportedBlocks(originalInput ?? "");
+      if (blocks.length === 0) return compiled;
+
       const [header, ...rest] = compiled.split("\n\n");
-      return [header, ...savedBlocks, ...rest].join("\n\n");
+      return [header, ...blocks, ...rest].join("\n\n");
     },
   });
 }