Language Translator

Purpose & User Intent: The translator tool solves fast, accurate text translation between major languages for writers, developers, educators, travelers. Users provide input text, select source and target languages, or allow auto-detection for the source. It is used when understanding foreign text, producing draft translations, or validating multilingual content. Example: translate a product description from English to Spanish for market localization.
Core Logic & Features: Domain capabilities include language detection, two way translation, preservation of punctuation and numbers, and handling of paragraphs and line breaks. It supports common language pairs via ISO codes, and produces clean, plain text output. Required features: input text, source and target languages, translation output, and optional detection results. Advanced options may include glossary term mapping, tone or formality hints, and batch translation. The tool relies on neural translation with subword modeling plus dictionary lookups for glossaries, and applies post processing to fix spacing and punctuation.
Inputs & Outputs: Inputs are text (string, required, sample Hello, world!), source_language (string, optional ISO code e.g. en), target_language (string, required ISO code e.g. es). Outputs include translated_text (string), detected_source_language (string, ISO code if provided or auto detected), and confidence (numeric 0-1). Validation enforces characters up to a practical limit and ensures language codes are valid ISO 639-1 in the given system.
Algorithms & Calculations: The core uses neural encoder-decoder translation with sentence and phrase level alignment. Language detection uses lightweight classifier on the input; glossary terms are applied before final surface realignment; punctuation normalization and numeric preservation are performed. Results may be refined by post processing and small heuristic corrections.
Error & Edge Cases: Handles unsupported languages with a clear failure signal for the content side; empty text returns an empty translation; extremely long inputs may be chunked; network errors raise retry suggestions; mixed language input may translate phrases separately; glossary conflicts fall back to standard translation.
Industry/Region & Localization: Built on ISO language codes and UTF-8 text. Outputs respect locale specific punctuation where possible; no currency or date formatting unless provided in text; privacy and ephemeral processing unless storage is requested. Assumptions & Exclusions: Assumes input text is legible and represents the intended source language; translations are not guaranteed to capture idiomatic nuance or domain specific terms without a glossary; excludes real-time speech or audio input.