From c75d2d208eb9ed6d9e364003456ab40ee884fd3f Mon Sep 17 00:00:00 2001
From: evazion <noizave@gmail.com>
Date: Sun, 5 Oct 2014 12:13:38 -0500
Subject: [PATCH] normalize_for_artist_finder!: Don't crash on bad URLs

If we can't normalize the URL (because of bad IDs, it's malformed, or
the HTML page changed), just return the unnormalized URL.
---
 app/logical/sources/site.rb             | 8 +++++++-
 app/logical/sources/strategies/pixiv.rb | 8 ++++----
 2 files changed, 11 insertions(+), 5 deletions(-)

diff --git a/app/logical/sources/site.rb b/app/logical/sources/site.rb
index 9c13ba32b..933a32050 100644
--- a/app/logical/sources/site.rb
+++ b/app/logical/sources/site.rb
@@ -1,6 +1,8 @@
 # encoding: UTF-8
 
 module Sources
+  class Error < Exception ; end
+
   class Site
     attr_reader :url, :strategy
     delegate :get, :referer_url, :site_name, :artist_name, :profile_url, :image_url, :tags, :artist_record, :unique_id, :page_count, :to => :strategy
@@ -22,7 +24,11 @@ module Sources
 
     def normalize_for_artist_finder!
       if available?
-        return strategy.normalize_for_artist_finder!
+        begin
+          return strategy.normalize_for_artist_finder!
+        rescue Sources::Error
+          return url
+        end
       else
         return url
       end
diff --git a/app/logical/sources/strategies/pixiv.rb b/app/logical/sources/strategies/pixiv.rb
index 7f55679b4..67d107c82 100644
--- a/app/logical/sources/strategies/pixiv.rb
+++ b/app/logical/sources/strategies/pixiv.rb
@@ -149,7 +149,7 @@ module Sources
         if not stacc_link.nil?
           stacc_link.attr("href").sub(%r!^/stacc/!i, '')
         else
-          raise "Couldn't find Pixiv moniker in page: #{normalized_url}"
+          raise Sources::Error.new("Couldn't find Pixiv moniker in page: #{normalized_url}")
         end
       end
 
@@ -162,7 +162,7 @@ module Sources
           thumbnail_url = elements.first.attr("src")
           return rewrite_thumbnails(thumbnail_url, is_manga)
         else
-          raise "Couldn't find image thumbnail URL in page: #{normalized_url}"
+          raise Sources::Error.new("Couldn't find image thumbnail URL in page: #{normalized_url}")
         end
       end
 
@@ -217,7 +217,7 @@ module Sources
           metadata = CSV.parse(response.content.force_encoding("UTF-8")).first
 
           if metadata.nil?
-            raise "Couldn't get Pixiv API metadata from #{spapi_url}."
+            raise Sources::Error.new("Couldn't get Pixiv API metadata from #{spapi_url}.")
           else
             yield metadata
           end
@@ -255,7 +255,7 @@ module Sources
           $1
 
         else
-          raise "Couldn't get illust ID from URL: #{url}"
+          raise Sources::Error.new("Couldn't get illust ID from URL: #{url}")
         end
       end