From 4ab98ec12bff5d04c8e9647e4b65350b107cf0e2 Mon Sep 17 00:00:00 2001
From: db0 <mail@dbzer0.com>
Date: Sun, 15 Oct 2023 14:31:06 +0200
Subject: [PATCH] fix: handle bot blocking siteinfo

---
 fediseer/fediverse.py | 28 ++++++++++++++++++++++++----
 fediseer/register.py  |  2 +-
 2 files changed, 25 insertions(+), 5 deletions(-)

diff --git a/fediseer/fediverse.py b/fediseer/fediverse.py
index 92ae6ea..2549660 100644
--- a/fediseer/fediverse.py
+++ b/fediseer/fediverse.py
@@ -176,7 +176,12 @@ class InstanceInfo():
 
     def get_mastodon_info(self):
         site = requests.get(f"https://{self.domain}/api/v1/instance",timeout=self._req_timeout)
-        self.instance_info = site.json()
+        try:
+            self.instance_info = site.json()
+        except Exception as err:
+            if "challenge-error-text" in site.text:
+                raise Exception("Instance is preventing scripted retrieval of their site info.")
+            raise err
         self.approval_required = self.instance_info["approval_required"]
         if self.node_info is None:
             raise Exception("Error retrieving nodeinfo")
@@ -186,7 +191,12 @@ class InstanceInfo():
 
     def get_pleroma_info(self):
         site = requests.get(f"https://{self.domain}/api/v1/instance",timeout=self._req_timeout)
-        self.instance_info = site.json()
+        try:
+            self.instance_info = site.json()
+        except Exception as err:
+            if "challenge-error-text" in site.text:
+                raise Exception("Instance is preventing scripted retrieval of their site info.")
+            raise err
         self.approval_required = self.instance_info["approval_required"]
         if self.node_info is None:
             raise Exception("Error retrieving nodeinfo")
@@ -196,7 +206,12 @@ class InstanceInfo():
 
     def get_firefish_info(self):
         site = requests.get(f"https://{self.domain}/api/v1/instance",timeout=self._req_timeout)
-        self.instance_info = site.json()
+        try:
+            self.instance_info = site.json()
+        except Exception as err:
+            if "challenge-error-text" in site.text:
+                raise Exception("Instance is preventing scripted retrieval of their site info.")
+            raise err
         self.approval_required = self.instance_info["approval_required"]
         if self.node_info is None:
             raise Exception("Error retrieving nodeinfo")
@@ -213,7 +228,12 @@ class InstanceInfo():
         site = requests.get(f"https://{self.domain}/api/v1/instance",timeout=self._req_timeout,allow_redirects=False)
         if site.status_code != 200:
             raise Exception(f"Unexpected status code retrieved when discovering instance info: {site.status_code}")
-        self.instance_info = site.json()
+        try:
+            self.instance_info = site.json()
+        except Exception as err:
+            if "challenge-error-text" in site.text:
+                raise Exception("Instance is preventing scripted retrieval of their site info.")
+            raise err
         self.approval_required = self.instance_info.get("approval_required")
         if self.node_info is None:
             raise Exception("Error retrieving nodeinfo")
diff --git a/fediseer/register.py b/fediseer/register.py
index 43ee970..13a0904 100644
--- a/fediseer/register.py
+++ b/fediseer/register.py
@@ -26,7 +26,7 @@ def ensure_instance_registered(domain, allow_unreachable=False, record_unreachab
                     instance.poll_failures += 60
                 db.session.commit()
         if not allow_unreachable:
-            raise e.BadRequest(f"Error encountered while polling domain {domain}. Please check it's running correctly")
+            raise e.BadRequest(str(err))
     if instance:
         if (
             instance.software != instance_info.software or