transcribe/pipelines/pipe_translate.py

from .base import MetaItem, BasePipe, Segment
from llama_cpp import Llama
from ..helpers.translator import QwenTranslator
from config import LLM_MODEL_PATH, LLM_SYS_PROMPT_EN, LLM_SYS_PROMPT_ZH, LLM_LARGE_MODEL_PATH, ALL_MARKERS, LLM_SYS_7B_PROMPT_EN, LLM_SYS_7B_PROMPT_ZH


class TranslatePipe(BasePipe):
    translator = None

    @classmethod
    def init(cls):
        if cls.translator is None:
            cls.translator = QwenTranslator(LLM_MODEL_PATH, LLM_SYS_PROMPT_EN, LLM_SYS_PROMPT_ZH)


    def process(self, in_data: MetaItem) -> MetaItem:
        context = in_data.transcribe_content
        if not context.strip():
            result = ""
        elif all([ch in ALL_MARKERS for ch in context.strip()]):
            result = ""
        else:
            result = self.translator.translate(
                context, src_lang=in_data.source_language, dst_lang=in_data.destination_language)
        in_data.translate_content = result
        return in_data


class Translate7BPipe(TranslatePipe):
    translator = None

    @classmethod
    def init(cls):
        if cls.translator is None:
            cls.translator = QwenTranslator(LLM_LARGE_MODEL_PATH, LLM_SYS_7B_PROMPT_EN, LLM_SYS_7B_PROMPT_ZH)