From 35d6fff69e86704081270333fbcb5248202a246a Mon Sep 17 00:00:00 2001
From: Vinoth Kannan <vinothkannan@vinkas.com>
Date: Thu, 13 Jun 2019 22:03:58 +0530
Subject: [PATCH] PERF: use url instead of file key in temporary inventory
 table.

---
 lib/s3_inventory.rb                  | 14 ++++++++------
 spec/components/s3_inventory_spec.rb |  4 ++--
 2 files changed, 10 insertions(+), 8 deletions(-)

diff --git a/lib/s3_inventory.rb b/lib/s3_inventory.rb
index cba0f9cfa76..4afc545167e 100644
--- a/lib/s3_inventory.rb
+++ b/lib/s3_inventory.rb
@@ -37,13 +37,14 @@ class S3Inventory
       multisite_prefix = "uploads/#{RailsMultisite::ConnectionManagement.current_db}/"
       ActiveRecord::Base.transaction do
         begin
-          connection.exec("CREATE TEMP TABLE #{table_name}(key text UNIQUE, etag text, PRIMARY KEY(etag, key))")
+          connection.exec("CREATE TEMP TABLE #{table_name}(url text UNIQUE, etag text, PRIMARY KEY(etag, url))")
           connection.copy_data("COPY #{table_name} FROM STDIN CSV") do
             files.each do |file|
               CSV.foreach(file[:filename][0...-3], headers: false) do |row|
                 key = row[CSV_KEY_INDEX]
                 next if Rails.configuration.multisite && key.exclude?(multisite_prefix)
-                connection.put_copy_data("#{key},#{row[CSV_ETAG_INDEX]}\n")
+                url = File.join(Discourse.store.absolute_base_url, key)
+                connection.put_copy_data("#{url},#{row[CSV_ETAG_INDEX]}\n")
               end
             end
           end
@@ -53,7 +54,7 @@ class S3Inventory
             SET etag = #{table_name}.etag
             FROM #{table_name}
             WHERE #{model.table_name}.etag IS NULL
-              AND url ILIKE '%' || #{table_name}.key")
+              AND #{model.table_name}.url = #{table_name}.url")
 
           list_missing_post_uploads if type == "original"
 
@@ -85,11 +86,12 @@ class S3Inventory
       next if sha1.blank?
 
       upload_id = nil
-      result = connection.exec("SELECT * FROM #{table_name} WHERE key LIKE '%original/%/#{sha1}%'")
+      result = connection.exec("SELECT * FROM #{table_name} WHERE url LIKE '%original/%/#{sha1}%'")
 
       if result.count >= 1
         begin
-          key = result[0]["key"]
+          url = result[0]["url"]
+          key = url.sub(/^#{Discourse.store.absolute_base_url}\//, "")
           data = @s3_helper.object(key).data
           filename = (data.content_disposition&.match(/filename=\"(.*)\"/) || [])[1]
 
@@ -97,7 +99,7 @@ class S3Inventory
             user_id: Discourse.system_user.id,
             original_filename: filename || File.basename(key),
             filesize: data.content_length,
-            url: File.join(Discourse.store.absolute_base_url, key),
+            url: url,
             sha1: sha1,
             etag: result[0]["etag"]
           )
diff --git a/spec/components/s3_inventory_spec.rb b/spec/components/s3_inventory_spec.rb
index 44ae55c4ff4..0b8c66d45e8 100644
--- a/spec/components/s3_inventory_spec.rb
+++ b/spec/components/s3_inventory_spec.rb
@@ -82,8 +82,8 @@ describe "S3Inventory" do
 
   it "should backfill etags to uploads table correctly" do
     files = [
-      ["//bucket.amazonaws.com/original/0184537a4f419224404d013414e913a4f56018f2.jpg", "defcaac0b4aca535c284e95f30d608d0"],
-      ["//bucket.amazonaws.com/original/0789fbf5490babc68326b9cec90eeb0d6590db05.png", "25c02eaceef4cb779fc17030d33f7f06"]
+      ["#{Discourse.store.absolute_base_url}/original/0184537a4f419224404d013414e913a4f56018f2.jpg", "defcaac0b4aca535c284e95f30d608d0"],
+      ["#{Discourse.store.absolute_base_url}/original/0789fbf5490babc68326b9cec90eeb0d6590db05.png", "25c02eaceef4cb779fc17030d33f7f06"]
     ]
     files.each { |file| Fabricate(:upload, url: file[0]) }