language-learning-app/api/app/domain/services/summarise_service.py

import asyncio
import random
import re
import uuid
from typing import Any, Callable, Coroutine

import anthropic
from sqlalchemy.ext.asyncio import AsyncSession

from ...outbound.postgres.repositories import summarise_job_repository
from ...outbound.postgres.repositories.translated_article_repository import TranslatedArticleRepository
from ...outbound.anthropic.anthropic_client import AnthropicClient
from ...outbound.deepgram.deepgram_client import LocalDeepgramClient
from ...outbound.deepl.deepl_client import DeepLClient
from ...outbound.gemini.gemini_client import GeminiClient
from ...outbound.spacy.spacy_client import SpacyClient
from ...storage import upload_audio
from ...languages import SUPPORTED_LANGUAGES



_ANTHROPIC_RETRYABLE = (
    anthropic.RateLimitError,
    anthropic.InternalServerError,
    anthropic.APITimeoutError,
    anthropic.APIConnectionError,
)
_MAX_RETRIES = 4
_BASE_DELAY = 1.0
_MAX_DELAY = 60.0


async def _anthropic_with_backoff(
    coro_fn: Callable[..., Coroutine[Any, Any, Any]],
    *args: Any,
    **kwargs: Any,
) -> Any:
    for attempt in range(_MAX_RETRIES + 1):
        try:
            return await coro_fn(*args, **kwargs)
        except _ANTHROPIC_RETRYABLE as exc:
            if attempt == _MAX_RETRIES:
                raise
            retry_after: float | None = None
            if isinstance(exc, anthropic.RateLimitError):
                raw = exc.response.headers.get("retry-after")
                if raw is not None:
                    retry_after = float(raw)
            if retry_after is None:
                retry_after = min(_BASE_DELAY * (2 ** attempt), _MAX_DELAY)
            jittered = retry_after * (0.8 + random.random() * 0.4)
            await asyncio.sleep(jittered)


class SummariseService:
    def __init__(
        self,
        anthropic_client: AnthropicClient,
        deepgram_client: LocalDeepgramClient,
        deepl_client: DeepLClient,
        gemini_client: GeminiClient,
        spacy_client: SpacyClient,
    ) -> None:
        self.anthropic_client = anthropic_client
        self.deepgram_client = deepgram_client
        self.deepl_client = deepl_client
        self.gemini_client = gemini_client
        self.spacy_client = spacy_client

    def _first_heading(self, md: str) -> str | None:
        m = re.search(r'^#{1,2}\s+(.+)', md, re.MULTILINE)
        return m.group(1).strip() if m else None
    
    def _split_title_and_body(self, text: str) -> tuple[str, str]:
        """Splits the text into a title (first heading) and body (the rest)."""
        title = self._first_heading(text) or ""
        body = text[len(title):].lstrip() if title else text
        if title == "":
            title = "Untitled Article"

        return title, body

    async def run(
        self,
        db: AsyncSession,
        job_id: uuid.UUID,
        article_id: uuid.UUID,
        source_language: str,
        target_language: str,
        complexity_level: str,
        input_texts: list[str],
    ) -> None:
        article_repo = TranslatedArticleRepository(db)
        job = await summarise_job_repository.get_by_id(db, job_id)
        await summarise_job_repository.mark_processing(db, job)

        try:
            language_name = SUPPORTED_LANGUAGES[target_language]
            source_material = "\n\n".join(input_texts[:3])

            generated_text = await _anthropic_with_backoff(
                self.anthropic_client.generate_summary_text,
                content_to_summarise=source_material,
                complexity_level=complexity_level,
                from_language=language_name,
                to_language=language_name,
                length_preference="200-400 words",
            )
            
            generated_title, generated_text_without_title = self._split_title_and_body(generated_text)

            await article_repo.update_content(
                article_id,
                target_title=generated_title,
                target_body=generated_text_without_title,
                source_title="",
                source_body="",
            )
            
            translated_text = await self.deepl_client.translate(generated_text, source_language)           

            translated_title, translated_text_without_title = self._split_title_and_body(translated_text)

            await article_repo.update_content(
                article_id,
                target_title=generated_title,
                target_body=generated_text_without_title,
                source_title=translated_title,
                source_body=translated_text_without_title,
            )

            target_pos_data = self.spacy_client.get_parts_of_speech(generated_text_without_title, target_language)
            source_pos_data = self.spacy_client.get_parts_of_speech(translated_text_without_title, source_language)

            await article_repo.update_pos(article_id, target_pos_data, source_pos_data)

            voice = self.gemini_client.get_voice_by_language(target_language)
            wav_bytes = await self.gemini_client.generate_audio(generated_text, voice)
            audio_key = f"audio/{job_id}.wav"
            upload_audio(audio_key, wav_bytes)

            transcript = await self.deepgram_client.transcribe_bytes(wav_bytes, target_language)

            await article_repo.update_audio(article_id, audio_key, transcript)

            await summarise_job_repository.mark_succeeded(db, job)

        except Exception as exc:
            await summarise_job_repository.mark_failed(db, job, str(exc))