fetch: Send User-Agent header

[paperdoorknob] / fetch.py

# paperdoorknob: Print glowfic
#
# This program is free software: you can redistribute it and/or modify it
# under the terms of the GNU General Public License as published by the
# Free Software Foundation, version 3.


from abc import ABC, abstractmethod
from contextlib import contextmanager
from sys import stderr
from typing import IO, Iterator

import requests
import requests_cache


_headers = {
    'User-Agent': 'paperdoorknob/0.0.1 (https://git.scottworley.com/paperdoorknob/)'}


class Fetcher(ABC):
    @abstractmethod
    def fetch(self, url: str) -> bytes:
        raise NotImplementedError()


class _SessionFetcher(Fetcher):

    def __init__(self, session: requests.Session, timeout: int) -> None:
        self._session = session
        self._timeout = timeout

    def fetch(self, url: str) -> bytes:
        with self._session.get(url, timeout=self._timeout, headers=_headers) as r:
            r.raise_for_status()
            return r.content


class _CachingFetcher(Fetcher):

    def __init__(
            self,
            session: requests_cache.CachedSession,
            timeout: int) -> None:
        self._session = session
        self._timeout = timeout
        self._request_count = 0
        self._cache_hit_count = 0

    def fetch(self, url: str) -> bytes:
        with self._session.get(url, timeout=self._timeout, headers=_headers) as r:
            r.raise_for_status()
            self._request_count += 1
            self._cache_hit_count += int(r.from_cache)
            return r.content

    @property
    def request_count(self) -> int:
        return self._request_count

    @property
    def cache_hit_count(self) -> int:
        return self._cache_hit_count


@contextmanager
def DirectFetcher(timeout: int) -> Iterator[_SessionFetcher]:
    with requests.session() as session:
        yield _SessionFetcher(session, timeout)


@contextmanager
def CachingFetcher(
        cache_path: str,
        timeout: int,
        report_stream: IO[str] = stderr) -> Iterator[_CachingFetcher]:
    with requests_cache.CachedSession(cache_path, cache_control=True) as session:
        fetcher = _CachingFetcher(session, timeout)
        yield fetcher
        if fetcher.request_count > 0:
            percent = 100.0 * fetcher.cache_hit_count / fetcher.request_count
            print(
                f"Fetch cache hits: {fetcher.cache_hit_count} ({percent:.1f}%)",
                file=report_stream)


class FakeFetcher(Fetcher):

    def __init__(self, resources: dict[str, bytes]) -> None:
        self._resources = resources
        self._fetch_count = 0

    def fetch(self, url: str) -> bytes:
        self._fetch_count += 1
        if url not in self._resources:
            raise requests.HTTPError("URL not found", url)
        return self._resources[url]

    def request_count(self) -> int:
        return self._fetch_count