prompts.py

# Prompt builder utilities for the Open Subtitle Translator

STYLE_PREFIX = """You are a professional subtitle translator.

HARD CONSTRAINTS:
- Output SRT only between <<<SRT>>> and <<<END>>>.
- Do NOT change block numbers or timecodes.
- Do NOT add/remove lines within a block; preserve exact line breaks.
- Keep any tags/speaker labels (e.g., <i>, ♪) exactly as-is.
- No commentary or explanations outside the SRT.
"""

RTL_LANGS = {"he", "ar", "fa", "ur"}

def build_prompt(source_lang: str,
                 target_lang: str,
                 batch_srt: str,
                 glossary_text: str | None,
                 extra_instructions: str | None,
                 prev_source: str | None,
                 prev_target: str | None) -> str:
    """
    Compose a cache-friendly prompt. Keep the prefix byte-identical across calls
    to leverage provider-side prompt caching where available.
    """
    prefix = STYLE_PREFIX + f"""

TASK:
- Translate from {source_lang} to {target_lang}.
- Input format is SRT blocks.

STYLE & GLOSSARY (project-provided; must follow if applicable):
"""
    if glossary_text and glossary_text.strip():
        prefix += glossary_text.strip() + "\n"
    if extra_instructions and extra_instructions.strip():
        prefix += "\nEXTRA INSTRUCTIONS (apply carefully):\n" + extra_instructions.strip() + "\n"

    context = "\nCONTEXT (do not modify):\n"
    if prev_source:
        context += "[Previous batch — Source]\n" + prev_source.strip() + "\n"
    if prev_target:
        context += "[Previous batch — Target]\n" + prev_target.strip() + "\n"

    task = (
        "\nCURRENT BATCH TO TRANSLATE:\n"
        + batch_srt
        + "\nReturn only:\n<<<SRT>>>\n[Translated SRT blocks]\n<<<END>>>"
    )
    return prefix + context + task