utils/cache_layer.py

from typing import List

from diskcache import Cache

from utils.data_proto import Page
from utils.summary_utils import summarize

cache = Cache(directory='/.cache/tmp/summary', size_limit=int(1e9))  # 1GB


def get_summarize_from_cache(pages: List[Page]):
    pages_summaries = []
    uncached_pages = []
    for page in pages:
        id = page['id']
        summary = cache.get(id)
        if summary is not None:
            pages_summaries.append(summary)
        else:
            uncached_pages.append(page)
    return pages_summaries, uncached_pages


def summarize_un_cache_page(pages: List[Page]):
    for page in pages:
        id = page['id']
        summary = summarize(id, page['text'])
        cache.set(id, summary)
        print(f'processed page {id}')