From d378e171735e5676f2a9e925be7fc45d8d19fdb4 Mon Sep 17 00:00:00 2001
From: Thomas Sileo <t@a4.io>
Date: Thu, 4 Aug 2022 19:11:14 +0200
Subject: [PATCH] Improve privacy relace

---
 app/ap_object.py         | 18 ++----------------
 app/templates.py         |  2 ++
 app/templates/utils.html |  2 +-
 app/utils/opengraph.py   |  5 +++++
 docs/user_guide.md       |  3 +++
 5 files changed, 13 insertions(+), 17 deletions(-)

diff --git a/app/ap_object.py b/app/ap_object.py
index 72fd871..f96b5c6 100644
--- a/app/ap_object.py
+++ b/app/ap_object.py
@@ -2,7 +2,6 @@ import hashlib
 from datetime import datetime
 from functools import cached_property
 from typing import Any
-from urllib.parse import urlparse
 
 import pydantic
 from bs4 import BeautifulSoup  # type: ignore
@@ -12,8 +11,8 @@ from app import activitypub as ap
 from app.actor import LOCAL_ACTOR
 from app.actor import Actor
 from app.actor import RemoteActor
-from app.config import PRIVACY_REPLACE
 from app.media import proxied_media_url
+from app.utils import privacy_replace
 from app.utils.datetime import now
 from app.utils.datetime import parse_isoformat
 
@@ -179,20 +178,7 @@ class Object:
         if self.ap_object.get("mediaType") == "text/markdown":
             content = markdown(content, extensions=["mdx_linkify"])
 
-        if not PRIVACY_REPLACE:
-            return content
-
-        soup = BeautifulSoup(content, "html5lib")
-        links = soup.find_all("a", href=True)
-
-        for link in links:
-            parsed_href = urlparse(link.attrs["href"])
-            if new_netloc := PRIVACY_REPLACE.get(
-                parsed_href.netloc.removeprefix("www.")
-            ):
-                link.attrs["href"] = parsed_href._replace(netloc=new_netloc).geturl()
-
-        return soup.find("body").decode_contents()
+        return privacy_replace.replace_content(content)
 
     @property
     def summary(self) -> str | None:
diff --git a/app/templates.py b/app/templates.py
index 2c51bbe..30af066 100644
--- a/app/templates.py
+++ b/app/templates.py
@@ -32,6 +32,7 @@ from app.config import generate_csrf_token
 from app.config import session_serializer
 from app.database import AsyncSession
 from app.media import proxied_media_url
+from app.utils import privacy_replace
 from app.utils.datetime import now
 from app.utils.highlight import HIGHLIGHT_CSS
 from app.utils.highlight import highlight
@@ -400,3 +401,4 @@ _templates.env.filters["emojify"] = _emojify
 _templates.env.filters["pluralize"] = _pluralize
 _templates.env.filters["parse_datetime"] = _parse_datetime
 _templates.env.filters["poll_item_pct"] = _poll_item_pct
+_templates.env.filters["privacy_replace_url"] = privacy_replace.replace_url
diff --git a/app/templates/utils.html b/app/templates/utils.html
index bfc6331..a2040a5 100644
--- a/app/templates/utils.html
+++ b/app/templates/utils.html
@@ -317,7 +317,7 @@
     </div>
     {% endif %}
     <div>
-        <a href="{{ og_meta.url }}">{{ og_meta.title }}</a>
+        <a href="{{ og_meta.url | privacy_replace_url }}">{{ og_meta.title }}</a>
         {% if og_meta.site_name %}
             <small style="display:block;">{{ og_meta.site_name }}</small>
         {% endif %}
diff --git a/app/utils/opengraph.py b/app/utils/opengraph.py
index 13b8c35..fccf43b 100644
--- a/app/utils/opengraph.py
+++ b/app/utils/opengraph.py
@@ -15,6 +15,7 @@ from app.database import AsyncSession
 from app.models import InboxObject
 from app.models import OutboxObject
 from app.utils.url import is_url_valid
+from app.utils.url import make_abs
 
 
 class OpenGraphMeta(BaseModel):
@@ -46,6 +47,10 @@ def _scrap_og_meta(url: str, html: str) -> OpenGraphMeta | None:
     if "title" not in raw:
         return None
 
+    for maybe_rel in {"url", "image"}:
+        if u := raw.get(maybe_rel):
+            raw[maybe_rel] = make_abs(u, url)
+
     return OpenGraphMeta.parse_obj(raw)
 
 
diff --git a/docs/user_guide.md b/docs/user_guide.md
index 7109119..b129eda 100644
--- a/docs/user_guide.md
+++ b/docs/user_guide.md
@@ -43,6 +43,9 @@ To do so, just add as these extra config items, this is a sample config that rew
 domain = "youtube.com"
 replace_by  ="yewtu.be"
 [[privacy_replace]]
+domain = "youtu.be"
+replace_by = "yewtu.be"
+[[privacy_replace]]
 domain = "twitter.com"
 replace_by = "nitter.net"
 [[privacy_replace]]