Skip to main content

​
Overview

Extracts all translation marker IDs from normalized text, preserving their order of appearance. This is useful for validating response structure and detecting missing or duplicate IDs.

​
Function Signature

extractTranslationIds(text: string): string[]

​
Parameters

​
text
string
required
Translation text containing markers in the format “ID - Translation”. Should be normalized using normalizeTranslationText first for best results.

​
Returns

​
ids
string[]
Array of extracted IDs in the order they appear in the text.

​
Usage

​
Basic Example

import { extractTranslationIds } from 'wobble-bibble';

const response = `P1 - First translation
P2 - Second translation
P3 - Third translation`;

const ids = extractTranslationIds(response);
console.log(ids); // ['P1', 'P2', 'P3']

​
Detecting Duplicates

const response = `P1 - First
P2 - Second
P1 - Duplicate`;

const ids = extractTranslationIds(response);
console.log(ids); // ['P1', 'P2', 'P1']

// Check for duplicates
const hasDuplicates = ids.length !== new Set(ids).size;
console.log(hasDuplicates); // true

​
Validating Expected IDs

const expected = ['P1', 'P2', 'P3'];
const response = `P1 - Text
P3 - More text`;

const actual = extractTranslationIds(response);
const missing = expected.filter(id => !actual.includes(id));

console.log(missing); // ['P2']

​
Handling Complex ID Formats

// Works with various ID formats
const response = `P1 - Text
P2b - More text
P123 - Even more`;

const ids = extractTranslationIds(response);
console.log(ids); // ['P1', 'P2b', 'P123']

​
ID Format Requirements

The function recognizes IDs that match the pattern defined by MARKER_ID_PATTERN. Valid IDs typically:
  • Start with a letter (often ‘P’ for paragraph)
  • May contain additional letters or numbers
  • Are followed by optional space and dashes (- or - or - or -)
Examples of valid markers:
  • P1 - text
  • P2b- text
  • P123 -text
  • H5 - text (for Hadith segments)

​
When to Use

Use extractTranslationIds when:
  • Validating that all expected segments were translated
  • Checking for duplicate ID entries in responses
  • Verifying correct ID ordering in sequential translations
  • Building validation pipelines for LLM outputs
  • Debugging translation response structure issues
This function is a key component of the validation system used to detect LLM hallucinations and output errors.

​
Best Practices

  1. Always normalize first: Use normalizeTranslationText before calling this function to ensure reliable ID extraction 
    const normalized = normalizeTranslationText(rawResponse);
const ids = extractTranslationIds(normalized);
  2. Compare with expected: Always validate extracted IDs against your input segments 
    const expectedIds = segments.map(s => s.id);
const actualIds = extractTranslationIds(response);
const allPresent = expectedIds.every(id => actualIds.includes(id));
  3. Check for order: For sequential translations, verify IDs appear in the correct order 
    const isOrdered = JSON.stringify(actualIds) === JSON.stringify(expectedIds);

Build docs developers (and LLMs) love

Get started for freeTalk to us