From f6902e370dd0c156d66fac95160f0b9db0daf5a7 Mon Sep 17 00:00:00 2001
From: =?utf8?q?Roland=20H=C3=A4der?= <roland@mxchange.org>
Date: Sat, 9 Dec 2023 03:46:25 +0100
Subject: [PATCH] Continued: - check all URLs against validator

---
 fba/helpers/domain.py     | 2 ++
 fba/helpers/processing.py | 2 ++
 fba/http/federation.py    | 2 ++
 fba/http/network.py       | 2 ++
 fba/models/instances.py   | 2 ++
 fba/utils.py              | 3 +++
 6 files changed, 13 insertions(+)

diff --git a/fba/helpers/domain.py b/fba/helpers/domain.py
index 4141258..2df92e8 100644
--- a/fba/helpers/domain.py
+++ b/fba/helpers/domain.py
@@ -76,6 +76,8 @@ def is_in_url(domain: str, url: str) -> bool:
         raise ValueError(f"Parameter url[]='{type(url)}' is not of type 'str'")
     elif url == "":
         raise ValueError("Parameter 'url' is empty")
+    elif not validators.url(url):
+        raise ValueError(f"Parameter url='{url}' is not a valid URL")
     elif domain + url in _cache["is_in_url"]:
         logger.debug("Returning cached is_in_url='%s' - EXIT!", _cache["is_in_url"][domain + url])
         return _cache["is_in_url"][domain + url]
diff --git a/fba/helpers/processing.py b/fba/helpers/processing.py
index d8f3d12..37a4706 100644
--- a/fba/helpers/processing.py
+++ b/fba/helpers/processing.py
@@ -120,6 +120,8 @@ def csv_block(blocker: str, url: str, command: str):
         raise ValueError(f"url[]='{url}' is not of type 'str'")
     elif url == "":
         raise ValueError("Parameter 'url' is empty")
+    elif not validators.url(url):
+        raise ValueError(f"Parameter url='{url}' is not a valid URL")
     elif not isinstance(command, str):
         raise ValueError(f"command[]='{command}' is not of type 'str'")
     elif command == "":
diff --git a/fba/http/federation.py b/fba/http/federation.py
index e0bd1ae..8b1508d 100644
--- a/fba/http/federation.py
+++ b/fba/http/federation.py
@@ -477,6 +477,7 @@ def determine_software(domain: str, path: str = None) -> str:
 
 def find_domains(tag: bs4.element.Tag) -> list:
     logger.debug("tag[]='%s' - CALLED!", type(tag))
+
     if not isinstance(tag, bs4.element.Tag):
         raise ValueError(f"Parameter tag[]='{type(tag)}' is not type of bs4.element.Tag")
     elif len(tag.select("tr")) == 0:
@@ -527,6 +528,7 @@ def find_domains(tag: bs4.element.Tag) -> list:
 
 def add_peers(rows: dict) -> list:
     logger.debug("rows[]='%s' - CALLED!", type(rows))
+
     if not isinstance(rows, dict):
         raise ValueError(f"Parameter rows[]='{type(rows)}' is not of type 'dict'")
 
diff --git a/fba/http/network.py b/fba/http/network.py
index a5ad140..ec548d8 100644
--- a/fba/http/network.py
+++ b/fba/http/network.py
@@ -125,6 +125,8 @@ def fetch_api_url(url: str, timeout: tuple) -> dict:
         raise ValueError(f"Parameter url[]='{type(url)}' is not of type 'str'")
     elif url == "":
         raise ValueError("Parameter 'url' is empty")
+    elif not validators.url(url):
+        raise ValueError(f"Parameter url='{url}' is not a valid URL")
     elif not isinstance(timeout, tuple):
         raise ValueError(f"timeout[]='{type(timeout)}' is not of type 'tuple'")
 
diff --git a/fba/models/instances.py b/fba/models/instances.py
index 619d05b..f01dd36 100644
--- a/fba/models/instances.py
+++ b/fba/models/instances.py
@@ -483,6 +483,8 @@ def set_nodeinfo_url(domain: str, url: str):
         raise ValueError(f"Parameter url[]='{type(url)}' is not of type 'str'")
     elif url == "":
         raise ValueError("Parameter 'url' is empty")
+    elif not validators.url(url):
+        raise ValueError(f"Parameter url='{url}' is not a valid URL")
 
     # Set timestamp
     _set_data("nodeinfo_url", domain, url)
diff --git a/fba/utils.py b/fba/utils.py
index 562abde..f698194 100644
--- a/fba/utils.py
+++ b/fba/utils.py
@@ -20,6 +20,7 @@ from urllib.parse import urlparse
 
 import bs4
 import requests
+import validators
 
 from fba.helpers import blacklist
 from fba.helpers import config
@@ -52,6 +53,8 @@ def fetch_url(url: str, headers: dict, timeout: tuple) -> requests.models.Respon
         raise ValueError(f"Parameter url[]='{type(url)}' is not of type 'str'")
     elif url == "":
         raise ValueError("Parameter 'url' is empty")
+    elif not validators.url(url):
+        raise ValueError(f"Parameter url='{url}' is not a valid URL")
     elif not isinstance(headers, dict):
         raise ValueError(f"Parameter headers[]='{type(headers)}' is not of type 'dict'")
     elif not isinstance(timeout, tuple):
-- 
2.39.5