bridgy-fed/common.py

"""Misc common utilities."""
import base64
from datetime import timedelta
import functools
import logging
import os
from pathlib import Path
import re
import threading
import urllib.parse
from urllib.parse import urljoin, urlparse

from Crypto.Util import number
from flask import abort, g, has_request_context, make_response, request
from google.cloud.error_reporting.util import build_flask_context
from google.cloud import ndb
from google.cloud.ndb.key import Key
from google.protobuf.timestamp_pb2 import Timestamp
from granary import as2
from oauth_dropins.webutil import util, webmention
from oauth_dropins.webutil.appengine_config import error_reporting_client, tasks_client
from oauth_dropins.webutil import appengine_info
from oauth_dropins.webutil.appengine_info import DEBUG
from oauth_dropins.webutil import flask_util
from oauth_dropins.webutil.util import json_dumps
from negotiator import ContentNegotiator, AcceptParameters, ContentType

import memcache

logger = logging.getLogger(__name__)

# allow hostname chars (a-z, 0-9, -), allow arbitrary unicode (eg ☃.net), don't
# allow specific chars that we'll often see in webfinger, AP handles, etc. (@, :)
# https://stackoverflow.com/questions/10306690/what-is-a-regular-expression-which-will-match-a-valid-domain-name-without-a-subd
#
# TODO: preprocess with domain2idna, then narrow this to just [a-z0-9-]
DOMAIN_RE = r'^([^/:;@?!\'.]+\.)+[^/:@_?!\'.]+$'

CONTENT_TYPE_HTML = 'text/html; charset=utf-8'

PRIMARY_DOMAIN = 'fed.brid.gy'
# protocol-specific subdomains are under this "super"domain
SUPERDOMAIN = '.brid.gy'
# TODO: add a Flask route decorator version of util.canonicalize_domain, then
# use it to canonicalize most UI routes from these to fed.brid.gy.
# TODO: unify with models.PROTOCOLS
PROTOCOL_DOMAINS = (
    'ap.brid.gy',
    'atproto.brid.gy',
    'bsky.brid.gy',
    'web.brid.gy',
    'efake.brid.gy',
    'fa.brid.gy',
    'other.brid.gy',
)
OTHER_DOMAINS = (
    'bridgy-federated.appspot.com',
    'bridgy-federated.uc.r.appspot.com',
)
LOCAL_DOMAINS = (
  'localhost',
  'localhost:8080',
  'my.dev.com:8080',
)
DOMAINS = (PRIMARY_DOMAIN,) + PROTOCOL_DOMAINS + OTHER_DOMAINS + LOCAL_DOMAINS
# TODO: unify with manual_opt_out
# TODO: unify with Bridgy's
DOMAIN_BLOCKLIST = (
    'bsky.social',
    'facebook.com',
    'fb.com',
    'instagram.com',
    'reddit.com',
    'rumble.com',  # serves infinite HTTP 307 redirects to GCP
    't.co',
    'tiktok.com',
    'twitter.com',
    'x.com',
    'youtu.be',
    'youtube.com',
)

SMTP_HOST = 'smtp.gmail.com'
SMTP_PORT = 587

# populated in models.reset_protocol_properties
SUBDOMAIN_BASE_URL_RE = None
ID_FIELDS = ('id', 'object', 'actor', 'author', 'inReplyTo', 'url')

CACHE_CONTROL = {'Cache-Control': 'public, max-age=3600'}  # 1 hour

NDB_MEMCACHE_TIMEOUT = timedelta(hours=2)

USER_AGENT = 'Bridgy Fed (https://fed.brid.gy/)'
util.set_user_agent(USER_AGENT)

# https://cloud.google.com/appengine/docs/locations
TASKS_LOCATION = 'us-central1'
RUN_TASKS_INLINE = False  # overridden by unit tests

# for Protocol.REQUIRES_OLD_ACCOUNT, how old is old enough
OLD_ACCOUNT_AGE = timedelta(days=7)

# populated later in this file
NDB_CONTEXT_KWARGS = None

_negotiator = ContentNegotiator(acceptable=[
    AcceptParameters(ContentType(CONTENT_TYPE_HTML)),
    AcceptParameters(ContentType(as2.CONTENT_TYPE)),
    AcceptParameters(ContentType(as2.CONTENT_TYPE_LD)),
])


@functools.cache
def protocol_user_copy_ids():
    """Returns all copy ids for protocol bot users."""
    ids = []

    from web import Web
    for user in ndb.get_multi(Web(id=domain).key for domain in PROTOCOL_DOMAINS):
        if user:
            ids.extend(copy.uri for copy in user.copies)

    return tuple(ids)


def base64_to_long(x):
    """Converts from URL safe base64 encoding to long integer.

    Originally from ``django_salmon.magicsigs``. Used in :meth:`User.public_pem`
    and :meth:`User.private_pem`.
    """
    return number.bytes_to_long(base64.urlsafe_b64decode(x))


def long_to_base64(x):
    """Converts from long integer to base64 URL safe encoding.

    Originally from ``django_salmon.magicsigs``. Used in :meth:`User.get_or_create`.
    """
    return base64.urlsafe_b64encode(number.long_to_bytes(x))


def host_url(path_query=None):
    base = request.host_url
    if (util.domain_or_parent_in(request.host, OTHER_DOMAINS)
            # when running locally against prod datastore
            or (not DEBUG and request.host in LOCAL_DOMAINS)):
        base = f'https://{PRIMARY_DOMAIN}'

    assert base
    return urljoin(base, path_query)


def error(err, status=400, exc_info=None, **kwargs):
    """Like :func:`oauth_dropins.webutil.flask_util.error`, but wraps body in JSON."""
    msg = str(err)
    logger.info(f'Returning {status}: {msg}', exc_info=exc_info)
    abort(status, response=make_response({'error': msg}, status), **kwargs)


def pretty_link(url, text=None, user=None, **kwargs):
    """Wrapper around :func:`oauth_dropins.webutil.util.pretty_link` that converts Mastodon user URLs to @-@ handles.

    Eg for URLs like https://mastodon.social/@foo and
    https://mastodon.social/users/foo, defaults text to ``@foo@mastodon.social``
    if it's not provided.

    Args:
      url (str)
      text (str)
      user (models.User): current user
      kwargs: passed through to :func:`oauth_dropins.webutil.util.pretty_link`
    """
    if user and user.is_web_url(url):
        return user.user_link(handle=False, pictures=True)

    if text is None:
        match = re.match(r'https?://([^/]+)/(@|users/)([^/]+)$', url)
        if match:
            text = match.expand(r'@\3@\1')

    return util.pretty_link(url, text=text, **kwargs)


def content_type(resp):
    """Returns a :class:`requests.Response`'s Content-Type, without charset suffix."""
    type = resp.headers.get('Content-Type')
    if type:
        return type.split(';')[0]


def redirect_wrap(url, domain=None):
    """Returns a URL on our domain that redirects to this URL.

    ...to satisfy Mastodon's non-standard domain matching requirement. :(

    Args:
      url (str)
      domain (str): optional Bridgy Fed domain to use. Must be in :attr:`DOMAINS`

    * https://github.com/snarfed/bridgy-fed/issues/16#issuecomment-424799599
    * https://github.com/tootsuite/mastodon/pull/6219#issuecomment-429142747

    Returns:
      str: redirect url
    """
    if not url or util.domain_from_link(url) in DOMAINS:
        return url

    path = '/r/' + url

    if domain:
        assert domain in DOMAINS, (domain, url)
        return urljoin(f'https://{domain}/', path)

    return host_url(path)


def subdomain_wrap(proto, path=None):
    """Returns the URL for a given path on this protocol's subdomain.

    Eg for the path ``foo/bar`` on ActivityPub, returns
    ``https://ap.brid.gy/foo/bar``.

    Args:
      proto (subclass of :class:`protocol.Protocol`)

    Returns:
      str: URL
    """
    subdomain = proto.ABBREV if proto and proto.ABBREV else 'fed'
    return urljoin(f'https://{subdomain}{SUPERDOMAIN}/', path)


def unwrap(val, field=None):
    """Removes our subdomain/redirect wrapping from a URL, if it's there.

    ``val`` may be a string, dict, or list. dicts and lists are unwrapped
    recursively.

    Strings that aren't wrapped URLs are left unchanged.

    Args:
      val (str or dict or list)
      field (str): optional field name for this value

    Returns:
      str: unwrapped url
    """

    if isinstance(val, dict):
        # TODO: clean up. https://github.com/snarfed/bridgy-fed/issues/967
        id = val.get('id')
        if (isinstance(id, str)
                and urlparse(id).path.strip('/') in DOMAINS + ('',)
                and util.domain_from_link(id) in DOMAINS):
            # protocol bot user, don't touch its URLs
            return {**val, 'id': unwrap(id)}

        return {f: unwrap(v, field=f) for f, v in val.items()}

    elif isinstance(val, list):
        return [unwrap(v) for v in val]

    elif isinstance(val, str):
        if match := SUBDOMAIN_BASE_URL_RE.match(val):
            unwrapped = match.group('path')
            if field in ID_FIELDS and re.fullmatch(DOMAIN_RE, unwrapped):
                return f'https://{unwrapped}/'
            return unwrapped

    return val


def create_task(queue, delay=None, **params):
    """Adds a Cloud Tasks task.

    If running in a local server, runs the task handler inline instead of
    creating a task.

    Args:
      queue (str): queue name
      delay (:class:`datetime.timedelta`): optional, used as task ETA (from now)
      params: form-encoded and included in the task request body

    Returns:
      flask.Response or (str, int): response from either running the task
      inline, if running in a local server, or the response from creating the
      task.
    """
    assert queue
    path = f'/queue/{queue}'

    # removed from "Added X task ..." log messae below to cut logging costs
    # https://github.com/snarfed/bridgy-fed/issues/1149#issuecomment-2265861956
    # loggable = {k: '{...}' if isinstance(v, dict) else v for k, v in params.items()}
    params = {
        k: json_dumps(v, sort_keys=True) if isinstance(v, dict) else v
        for k, v in params.items()
        if v is not None
    }

    if RUN_TASKS_INLINE or appengine_info.LOCAL_SERVER:
        logger.info(f'Running task inline: {queue} {params}')
        from router import app
        return app.test_client().post(
            path, data=params, headers={flask_util.CLOUD_TASKS_TASK_HEADER: 'x'})

        # # alternative: run inline in this request context
        # request.form = params
        # endpoint, args = app.url_map.bind(request.server[0])\
        #                             .match(path, method='POST')
        # return app.view_functions[endpoint](**args)

    body = urllib.parse.urlencode(sorted(params.items())).encode()
    try:
        traceparent = request.headers.get('traceparent', '')
    except RuntimeError:  # not currently in a request context
        traceparent = ''
    task = {
        'app_engine_http_request': {
            'http_method': 'POST',
            'relative_uri': path,
            'body': body,
            'headers': {
                'Content-Type': 'application/x-www-form-urlencoded',
                # propagate trace id
                # https://cloud.google.com/trace/docs/trace-context#http-requests
                # https://stackoverflow.com/a/71343735/186123
                'traceparent': traceparent,
            },
        },
    }
    if delay:
        eta_seconds = int(util.to_utc_timestamp(util.now()) + delay.total_seconds())
        task['schedule_time'] = Timestamp(seconds=eta_seconds)

    parent = tasks_client.queue_path(appengine_info.APP_ID, TASKS_LOCATION, queue)
    task = tasks_client.create_task(parent=parent, task=task)
    msg = f'Added {queue} {task.name.split("/")[-1]}'
    if not traceparent:
        logger.info(msg)
    return msg, 202


def report_exception(**kwargs):
    return report_error(msg=None, exception=True, **kwargs)


def report_error(msg, *, exception=False, **kwargs):
    """Reports an error to StackDriver Error Reporting.

    https://cloud.google.com/python/docs/reference/clouderrorreporting/latest/google.cloud.error_reporting.client.Client

    If ``DEBUG`` and ``exception`` are ``True``, re-raises the exception instead.

    Duplicated in ``bridgy.util``.
    """
    if DEBUG:
        if exception:
            raise
        else:
            logger.error(msg)
            return

    http_context = build_flask_context(request) if has_request_context() else None

    try:
        if exception:
            logger.error('', exc_info=True)
            error_reporting_client.report_exception(
                http_context=http_context, **kwargs)
        else:
            logger.error(msg)
            error_reporting_client.report(
                msg, http_context=http_context, **kwargs)
    except BaseException:
        kwargs['exception'] = exception
        logger.warning(f'Failed to report error! {kwargs}', exc_info=exception)


def cache_policy(key):
    """In memory ndb cache. Cache everything!

    https://github.com/snarfed/bridgy-fed/issues/1149#issuecomment-2261383697

    Keep an eye on this in case we start seeing problems due to this ndb bug
    where unstored in-memory modifications get returned by later gets:
    https://github.com/googleapis/python-ndb/issues/888

    Args:
      key (google.cloud.datastore.key.Key or google.cloud.ndb.key.Key):
        see https://github.com/googleapis/python-ndb/issues/987

    Returns:
      bool: whether to cache this object
    """
    return True

    # if isinstance(key, Key):
    #     # use internal google.cloud.datastore.key.Key
    #     # https://github.com/googleapis/python-ndb/issues/987
    #     key = key._key

    # return key and key.kind == 'Object' and key.name.startswith('did:')


def global_cache_policy(key):
    return True


PROFILE_ID_RE = re.compile(
    fr"""
      /users?/[^/]+$ |
      /app.bsky.actor.profile/self$ |
      ^did:[a-z0-9:.]+$ |
      ^https://{DOMAIN_RE[1:-1]}/?$
    """, re.VERBOSE)

def global_cache_timeout_policy(key):
    """Cache users and profile objects indefinitely, everything else 2h.

    Args:
      key (google.cloud.datastore.key.Key or google.cloud.ndb.key.Key):
        see https://github.com/googleapis/python-ndb/issues/987

    Returns:
      int: cache expiration for this object, in seconds
    """
    if isinstance(key, Key):
        # use internal google.cloud.datastore.key.Key
        # https://github.com/googleapis/python-ndb/issues/987
        key = key._key

    if (key and (key.kind in ('ActivityPub', 'ATProto', 'MagicKey')
                 or key.kind == 'Object' and PROFILE_ID_RE.search(key.name))):
        return None

    return int(NDB_MEMCACHE_TIMEOUT.total_seconds())


NDB_CONTEXT_KWARGS = {
    'cache_policy': cache_policy,
    'global_cache': memcache.global_cache,
    'global_cache_policy': global_cache_policy,
    'global_cache_timeout_policy': global_cache_timeout_policy,
}


def as2_request_type():
    """If this request has conneg (ie the ``Accept`` header) for AS2, returns its type.

    Specifically, returns either
    ``application/ld+json; profile="https://www.w3.org/ns/activitystreams"`` or
    ``application/activity+json``.

    If the current request's conneg isn't asking for AS2, returns None.

    https://www.w3.org/TR/activitypub/#retrieving-objects
    https://snarfed.org/2023-03-24_49619-2
    """
    if accept := request.headers.get('Accept'):
        try:
            negotiated = _negotiator.negotiate(accept)
        except ValueError:
            # work around https://github.com/CottageLabs/negotiator/issues/6
            negotiated = None
        if negotiated:
            accept_type = str(negotiated.content_type)
            if accept_type == as2.CONTENT_TYPE:
                return as2.CONTENT_TYPE
            elif accept_type in (as2.CONTENT_TYPE_LD, as2.CONTENT_TYPE_LD_PROFILE):
                return as2.CONTENT_TYPE_LD_PROFILE
            logger.debug(f'Conneg resolved {accept_type} for Accept: {accept}')


def log_request():
    """Logs GET query params and POST form.

    Limits each value to 1000 chars."""
    logger.info(f'Params:\n' + '\n'.join(
        f'{k} = {v[:1000]}' for k, v in request.values.items()))