From fabe469a53e41d0490b4500c567f5e49af261e3e Mon Sep 17 00:00:00 2001
From: Thomas Sileo <t@a4.io>
Date: Thu, 4 Aug 2022 19:13:32 +0200
Subject: [PATCH] Add missing file

---
 app/utils/privacy_replace.py | 37 ++++++++++++++++++++++++++++++++++++
 1 file changed, 37 insertions(+)
 create mode 100644 app/utils/privacy_replace.py

diff --git a/app/utils/privacy_replace.py b/app/utils/privacy_replace.py
new file mode 100644
index 0000000..c3795b6
--- /dev/null
+++ b/app/utils/privacy_replace.py
@@ -0,0 +1,37 @@
+from urllib.parse import urlparse
+
+from bs4 import BeautifulSoup  # type: ignore
+from loguru import logger
+
+from app.config import PRIVACY_REPLACE
+
+
+def replace_content(content: str) -> str:
+    if not PRIVACY_REPLACE:
+        return content
+
+    soup = BeautifulSoup(content, "html5lib")
+    links = list(soup.find_all("a", href=True))
+    if not links:
+        return content
+
+    for link in links:
+        link.attrs["href"] = replace_url(link.attrs["href"])
+
+    return soup.find("body").decode_contents()
+
+
+def replace_url(u: str) -> str:
+    if not PRIVACY_REPLACE:
+        return u
+
+    try:
+        parsed_href = urlparse(u)
+    except Exception:
+        logger.warning(f"Failed to parse url={u}")
+        return u
+
+    if new_netloc := PRIVACY_REPLACE.get(parsed_href.netloc.removeprefix("www.")):
+        return parsed_href._replace(netloc=new_netloc).geturl()
+
+    return u