From df35ce16cce423732e1dbc99139a11ab21d7eda4 Mon Sep 17 00:00:00 2001
From: Ryan Barrett <git@ryanb.org>
Date: Wed, 24 May 2023 15:18:31 -0700
Subject: [PATCH] AP users: add convert.py and /convert/... endpoint

#512
---
 app.py                |  2 +-
 convert.py            | 49 ++++++++++++++++++++++++++++++++++++++++++
 tests/test_convert.py | 50 +++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 100 insertions(+), 1 deletion(-)
 create mode 100644 convert.py
 create mode 100644 tests/test_convert.py

diff --git a/app.py b/app.py
index 43d2a41..3cf62cf 100644
--- a/app.py
+++ b/app.py
@@ -6,4 +6,4 @@ registered.
 from flask_app import app
 
 # import all modules to register their Flask handlers
-import activitypub, follow, pages, redirect, render, superfeedr, webfinger, webmention, xrpc_actor, xrpc_feed, xrpc_graph
+import activitypub, convert, follow, pages, redirect, render, superfeedr, webfinger, webmention, xrpc_actor, xrpc_feed, xrpc_graph
diff --git a/convert.py b/convert.py
new file mode 100644
index 0000000..e8c63af
--- /dev/null
+++ b/convert.py
@@ -0,0 +1,49 @@
+"""Serves /convert/... URLs to convert data from one protocol to another.
+
+URL pattern is /convert/SOURCE/DEST , where SOURCE and DEST are the LABEL
+constants from the :class:`Protocol` subclasses.
+
+Currently only supports /convert/activitypub/webmention/...
+"""
+import logging
+import re
+import urllib.parse
+
+from flask import request
+from oauth_dropins.webutil import flask_util, util
+from oauth_dropins.webutil.flask_util import error
+
+from activitypub import ActivityPub
+from common import CACHE_TIME
+from flask_app import app, cache
+from protocol import protocols
+from webmention import Webmention
+
+logger = logging.getLogger(__name__)
+
+SOURCES = frozenset((
+    ActivityPub.LABEL,
+))
+DESTS = frozenset((
+    Webmention.LABEL,
+))
+
+
+@app.get(f'/convert/<any({",".join(SOURCES)}):src>/<any({",".join(DESTS)}):dest>/<path:url>')
+@flask_util.cached(cache, CACHE_TIME, headers=['Accept'])
+def convert(src, dest, url):
+    """Converts data from one protocol to another and serves it.
+
+    Fetches the source data if it's not already stored.
+    """
+    if request.args:
+        url += '?' + urllib.parse.urlencode(request.args)
+    # some browsers collapse repeated /s in the path down to a single slash.
+    # if that happened to this URL, expand it back to two /s.
+    url = re.sub(r'^(https?:/)([^/])', r'\1/\2', url)
+
+    if not util.is_web(url):
+        error(f'Expected fully qualified URL; got {url}')
+
+    obj = protocols[src].load(url)
+    return protocols[dest].serve(obj)
diff --git a/tests/test_convert.py b/tests/test_convert.py
new file mode 100644
index 0000000..6cffd6f
--- /dev/null
+++ b/tests/test_convert.py
@@ -0,0 +1,50 @@
+"""Unit tests for convert.py.
+"""
+from unittest.mock import patch
+
+from oauth_dropins.webutil.testutil import requests_response
+import requests
+
+from common import CONTENT_TYPE_HTML
+
+from .test_redirect import (
+    REPOST_AS2,
+    REPOST_HTML,
+)
+from . import testutil
+
+
+@patch('requests.get')
+class ConvertTest(testutil.TestCase):
+
+    def test_unknown_source(self, _):
+        got = self.client.get('/convert/nope/webmention/http://foo')
+        self.assertEqual(404, got.status_code)
+
+    def test_unknown_dest(self, _):
+        got = self.client.get('/convert/activitypub/nope/http://foo')
+        self.assertEqual(404, got.status_code)
+
+    def test_missing_url(self, _):
+        got = self.client.get('/convert/activitypub/webmention/')
+        self.assertEqual(404, got.status_code)
+
+    def test_url_not_web(self, _):
+        got = self.client.get('/convert/activitypub/webmention/git+ssh://foo/bar')
+        self.assertEqual(400, got.status_code)
+
+    def test_activitypub_to_web(self, mock_get):
+        mock_get.return_value = self.as2_resp(REPOST_AS2)
+
+        got = self.client.get('/convert/activitypub/webmention/https://user.com/bar?baz=baj&biff')
+        self.assertEqual(200, got.status_code)
+        self.assertEqual(CONTENT_TYPE_HTML, got.content_type)
+
+        mock_get.assert_has_calls((self.as2_req('https://user.com/bar?baz=baj&biff='),))
+
+    def test_activitypub_to_web_fetch_fails(self, mock_get):
+        mock_get.side_effect = [requests_response('', status=405)]
+
+        got = self.client.get('/convert/activitypub/webmention/http://foo')
+        self.assertEqual(502, got.status_code)
+        mock_get.assert_has_calls((self.as2_req('http://foo'),))