| import { normalizeText } from './text_preprocessing.js'; |
|
|
| let generator = null; |
| let loadingPromise = null; |
| let precomputedMap = null; |
|
|
| const RULES = [ |
| [/\b[Ss]he go\b/g, (match) => `${match[0]}es`], |
| [/\b[Hh]e go\b/g, (match) => `${match[0]}es`], |
| [/\b[Ee]very (\w+) have\b/g, 'Every $1 has'], |
| [/\b([Tt]he quality results) is\b/g, '$1 are'], |
| [/\bdiscussed about\b/gi, 'discussed'], |
| [/\bequipments\b/gi, 'equipment'], |
| [/\bpeople is\b/gi, 'people are'], |
| [/\bthere is many\b/gi, 'there are many'], |
| [/\bwe was\b/gi, 'we were'], |
| [/\bcan sings\b/gi, 'can sing'], |
| [/\bdidn['’]?t went\b/gi, "didn't go"], |
| [/\bdidn['’]?t knew\b/gi, "didn't know"], |
| [/\bmore better\b/gi, 'better'], |
| [/\bmarried with\b/gi, 'married to'], |
| [/\blook forward to meet\b/gi, 'look forward to meeting'], |
| [/\bI has\b/g, 'I have'], |
| [/\bi has\b/g, 'i have'], |
| [/\bworks to do\b/gi, 'work to do'], |
| ]; |
|
|
| async function loadPrecomputed() { |
| if (precomputedMap) return precomputedMap; |
| const response = await fetch(new URL('./sample_outputs/precomputed.json', import.meta.url)); |
| const samples = await response.json(); |
| precomputedMap = new Map(samples.map((item) => [normalizeText(item.original).toLowerCase(), item])); |
| return precomputedMap; |
| } |
|
|
| function buildPrompt(text) { |
| return `Task:\nCorrect the grammar, spelling, punctuation, and wording of the text while preserving the original meaning.\n\nRules:\n- Preserve the original meaning.\n- Do not add unsupported information.\n- Do not remove important details.\n- Keep named entities, numbers, technical terms, and product names unchanged unless clearly incorrect.\n- Return only the corrected text.\n\nInput:\n${text}\n\nCorrected Text:`; |
| } |
|
|
| export async function loadLiveModel(metadata, onProgress = () => {}) { |
| if (!metadata.enable_live_model || metadata.browser_model_id.startsWith('YOUR_')) { |
| throw new Error('Live browser model is not configured. The app will use the disclosed fallback mode.'); |
| } |
| if (generator) return generator; |
| if (!loadingPromise) { |
| loadingPromise = (async () => { |
| const { pipeline } = await import('https://cdn.jsdelivr.net/npm/@huggingface/transformers@3.8.1/+esm'); |
| const options = { |
| dtype: metadata.browser_dtype || 'q8', |
| device: metadata.preferred_device || 'wasm', |
| progress_callback: onProgress, |
| }; |
| generator = await pipeline(metadata.browser_task || 'text2text-generation', metadata.browser_model_id, options); |
| return generator; |
| })(); |
| } |
| return loadingPromise; |
| } |
|
|
| function deterministicFallback(text) { |
| let corrected = text; |
| RULES.forEach(([pattern, replacement]) => { corrected = corrected.replace(pattern, replacement); }); |
| corrected = corrected.replace(/\bi\b/g, 'I').replace(/\s+([.,!?;:])/g, '$1'); |
| if (corrected && !/[.!?]$/.test(corrected)) corrected += '.'; |
| return corrected; |
| } |
|
|
| export async function correctText(rawText, metadata, onProgress = () => {}) { |
| const original = normalizeText(rawText); |
| if (!original) throw new Error('Enter text before running correction.'); |
| const start = performance.now(); |
|
|
| if (metadata.enable_live_model && !metadata.browser_model_id.startsWith('YOUR_')) { |
| const pipe = await loadLiveModel(metadata, onProgress); |
| const output = await pipe(buildPrompt(original), { max_new_tokens: 256, num_beams: 4, do_sample: false }); |
| const corrected = output?.[0]?.generated_text?.trim() ?? ''; |
| return { original, corrected, latencyMs: performance.now() - start, mode: 'live-browser-model', confidenceSource: 'generation output proxy' }; |
| } |
|
|
| const examples = await loadPrecomputed(); |
| const sample = examples.get(original.toLowerCase()); |
| const corrected = sample?.corrected ?? deterministicFallback(original); |
| return { |
| original, |
| corrected, |
| latencyMs: performance.now() - start, |
| mode: sample ? 'precomputed-fallback-sample' : 'rule-based-fallback-demo', |
| confidenceSource: 'heuristic only', |
| disclosure: metadata.fallback_disclosure, |
| }; |
| } |
|
|
