@spec build_translategemma_prompt(String.t(), String.t(), String.t()) :: String.t()

Builds a TranslateGemma-formatted prompt for translation.

TranslateGemma expects a single user message with a specific structure: professional translator persona, followed by two blank lines before the text.

Parameters

• message - The text to translate
• source_code - POSIX code of the source language (e.g., "en")
• target_code - POSIX code of the target language (e.g., "es")

Examples

iex> GettextTranslator.Processor.LLM.build_translategemma_prompt("Hello", "en", "es")
"You are a professional English (en) to Spanish (es) translator. " <>
"Your goal is to accurately convey the meaning and nuances of the original English text " <>
"while adhering to Spanish grammar, vocabulary, and cultural sensitivities.\n" <>
"Produce only the Spanish translation, without any additional explanations or commentary. " <>
"Please translate the following English text into Spanish:\n\n\nHello"

@spec translate(
  GettextTranslator.Processor.Translator.provider(),
  GettextTranslator.Processor.Translator.opts()
) :: {:ok, String.t()} | {:error, any()}

Callback implementation for GettextTranslator.Processor.Translator.translate/2.

@spec translate_single(
  GettextTranslator.Processor.Translator.provider(),
  map(),
  String.t() | nil
) ::
  {:ok, map()} | {:error, any()}

Translates a single message using the configured LLM provider, with optional additional instructions to guide the translation.

Parameters

• provider - The provider config map from Parser.parse_provider/0
• opts - Map with :language_code, :message, :type, and optionally :plural_message
• additional_instructions - Optional string with extra instructions for the LLM

Returns

• {:ok, %{translation: String.t(), plural_translation: String.t() | nil}}

• {:error, reason}

@spec translategemma?(String.t()) :: boolean()

Checks if the given model name is a TranslateGemma model.

Examples

iex> GettextTranslator.Processor.LLM.translategemma?("translategemma-27b")
true

iex> GettextTranslator.Processor.LLM.translategemma?("gpt-4")
false