From ad87596e37eb6cac52ee0ee3474e4e843420ff1c Mon Sep 17 00:00:00 2001
From: Ben Kurtovic <ben.kurtovic@gmail.com>
Date: Thu, 4 Feb 2021 21:21:00 -0500
Subject: [PATCH] Better error handling for revision IDs

---
 copyvios/api.py      |  1 +
 copyvios/checker.py  | 10 ++++++++--
 templates/index.mako |  4 +++-
 3 files changed, 12 insertions(+), 3 deletions(-)

diff --git a/copyvios/api.py b/copyvios/api.py
index ffec124..4da06fd 100644
--- a/copyvios/api.py
+++ b/copyvios/api.py
@@ -11,6 +11,7 @@ __all__ = ["format_api_error", "handle_api_request"]
 
 _CHECK_ERRORS = {
     "no search method": "Either 'use_engine' or 'use_links' must be true",
+    "bad oldid": "The revision ID is invalid",
     "no URL": "The parameter 'url' is required for URL comparisons",
     "bad URI": "The given URI scheme is unsupported",
     "no data": "No text could be found in the given URL (note that only HTML "
diff --git a/copyvios/checker.py b/copyvios/checker.py
index 4e25eeb..6bd8aad 100644
--- a/copyvios/checker.py
+++ b/copyvios/checker.py
@@ -3,6 +3,7 @@
 from datetime import datetime, timedelta
 from hashlib import sha256
 from logging import getLogger
+import re
 from urlparse import urlparse
 
 from earwigbot import exceptions
@@ -28,11 +29,16 @@ def do_check(query=None):
     if not query:
         query = Query()
     if query.lang:
-        query.lang = query.orig_lang = query.lang.lower()
+        query.lang = query.orig_lang = query.lang.strip().lower()
         if "::" in query.lang:
             query.lang, query.name = query.lang.split("::", 1)
     if query.project:
-        query.project = query.project.lower()
+        query.project = query.project.strip().lower()
+    if query.oldid:
+        query.oldid = query.oldid.strip().lstrip("0")
+        if not re.match(r"^\d+$", query.oldid):
+            query.error = "bad oldid"
+            return
 
     query.submitted = query.project and query.lang and (query.title or query.oldid)
     if query.submitted:
diff --git a/templates/index.mako b/templates/index.mako
index fa5bc28..44586c8 100644
--- a/templates/index.mako
+++ b/templates/index.mako
@@ -19,6 +19,8 @@
                 Unknown action: <b><span class="mono">${query.action | h}</span></b>.
             % elif query.error == "no search method":
                 No copyvio search methods were selected. A check can only be made using the search engine, links present in the page, Turnitin, or some combination of these.
+            % elif query.error == "bad oldid":
+                The revision ID <code>${query.oldid | h}</code> is invalid. It should be an integer.
             % elif query.error == "no URL":
                 URL comparison mode requires a URL to be entered. Enter one in the text box below, or choose copyvio search mode to look for content similar to the article elsewhere on the web.
             % elif query.error == "bad URI":
@@ -215,7 +217,7 @@
     <% attrib = get_attribution_info(query.site, query.page) %>
     % if attrib:
         <div id="attribution-warning" class="yellow-box">
-            This article contains an attribution template: <tt>{{<a href="${attrib[1]}">${attrib[0] | h}</a>}}</tt>. Please verify that any potential copyvios are not from properly attributed sources.
+            This article contains an attribution template: <code>{{<a href="${attrib[1]}">${attrib[0] | h}</a>}}</code>. Please verify that any potential copyvios are not from properly attributed sources.
         </div>
     % endif