From 100697127189b8e45c91f5906c1d157d6dc130fc Mon Sep 17 00:00:00 2001
From: rubenwardy <rw@rubenwardy.com>
Date: Thu, 4 Jul 2024 21:52:24 +0100
Subject: [PATCH] Add admin action to check for broken links

Fixes #546
---
 app/blueprints/admin/actions.py |  8 +++++++-
 app/tasks/pkgtasks.py           | 36 ++++++++++++++++++++++++++++-----
 2 files changed, 38 insertions(+), 6 deletions(-)

diff --git a/app/blueprints/admin/actions.py b/app/blueprints/admin/actions.py
index 18fa6280..0f5c3f72 100644
--- a/app/blueprints/admin/actions.py
+++ b/app/blueprints/admin/actions.py
@@ -29,7 +29,7 @@ from app.tasks.forumtasks import import_topic_list, check_all_forum_accounts
 from app.tasks.importtasks import import_repo_screenshot, check_zip_release, check_for_updates, update_all_game_support, \
 	import_languages
 from app.tasks.usertasks import import_github_user_ids
-from app.tasks.pkgtasks import notify_about_git_forum_links, clear_removed_packages
+from app.tasks.pkgtasks import notify_about_git_forum_links, clear_removed_packages, check_package_for_broken_links
 from app.utils import add_notification, get_system_user
 
 actions = {}
@@ -404,3 +404,9 @@ def delete_empty_threads():
 	flash(f"Deleted {count} threads", "success")
 
 	return redirect(url_for("admin.admin_page"))
+
+
+@action("DANGER: Check for broken links in all packages")
+def check_for_broken_links():
+	for package in Package.query.filter_by(state=PackageState.APPROVED).order_by(db.desc(Package.score)).limit(100).all():
+		check_package_for_broken_links.delay(package.id)
diff --git a/app/tasks/pkgtasks.py b/app/tasks/pkgtasks.py
index 1aadc9a3..1cf6b04b 100644
--- a/app/tasks/pkgtasks.py
+++ b/app/tasks/pkgtasks.py
@@ -16,6 +16,7 @@
 
 import datetime
 import re
+import sys
 from typing import Optional
 
 import requests
@@ -114,12 +115,13 @@ def _url_exists(url: str) -> str:
 			response.raise_for_status()
 			return ""
 	except requests.exceptions.HTTPError as e:
+		print(f"   - [{e.response.status_code}] {url}", file=sys.stderr)
 		return str(e.response.status_code)
 	except requests.exceptions.ConnectionError:
 		return "ConnectionError"
 
 
-def check_for_dead_links(package: Package) -> dict[str, str]:
+def _check_for_dead_links(package: Package) -> dict[str, str]:
 	links: list[Optional[str]] = [
 		package.repo,
 		package.website,
@@ -133,6 +135,8 @@ def check_for_dead_links(package: Package) -> dict[str, str]:
 	if package.desc:
 		links.extend(get_links(render_markdown(package.desc), package.get_url("packages.view", absolute=True)))
 
+	print(f"Checking {package.title} ({len(links)} links) for broken links", file=sys.stderr)
+
 	bad_urls = {}
 
 	for link in links:
@@ -146,20 +150,42 @@ def check_for_dead_links(package: Package) -> dict[str, str]:
 	return bad_urls
 
 
+def _check_package(package: Package) -> Optional[str]:
+	bad_urls = _check_for_dead_links(package)
+	if len(bad_urls) > 0:
+		return ("The following broken links were found on your package:\n\n" +
+				"\n".join([f"- {link} [{res}]" for link, res in bad_urls.items()]))
+
+	return None
+
+
 @celery.task()
 def check_package_on_submit(package_id: int):
 	package = Package.query.get(package_id)
 	if package is None:
 		raise TaskError("No such package")
 
-	bad_urls = check_for_dead_links(package)
-	if len(bad_urls) > 0:
+	if package.state != PackageState.READY_FOR_REVIEW:
+		return
+
+	msg = _check_package(package)
+	if msg != "":
 		marked = f"Marked {package.title} as Changed Needed"
-		msg = ("The following broken links were found on your package:\n\n" +
-				"\n".join([f"- {link} [{res}]" for link, res in bad_urls.items()]))
 
 		system_user = get_system_user()
 		post_to_approval_thread(package, system_user, marked, is_status_update=True, create_thread=True)
 		post_to_approval_thread(package, system_user, msg, is_status_update=False, create_thread=True)
 		package.state = PackageState.CHANGES_NEEDED
 		db.session.commit()
+
+
+@celery.task(rate_limit="5/m")
+def check_package_for_broken_links(package_id: int):
+	package = Package.query.get(package_id)
+	if package is None:
+		raise TaskError("No such package")
+
+	msg = _check_package(package)
+	if msg:
+		post_bot_message(package, "Broken links", msg)
+		db.session.commit()