From 46d3325144fbc6fd78a030abd9274c6fb4372aaa Mon Sep 17 00:00:00 2001
From: Matchu <matchu1993@gmail.com>
Date: Fri, 10 May 2024 17:39:10 -0700
Subject: [PATCH] Load *all* NC Mall pages in `nc_mall:sync`

Ta da! Now I can run this and pull 481 records into our database, and
then turn around and run it again and have them all correctly say
"skipped"!
---
 app/services/nc_mall.rb                       | 30 ++++++++++-
 ...que_index_on_item_id_to_nc_mall_records.rb | 11 ++++
 db/schema.rb                                  |  4 +-
 lib/tasks/nc_mall.rake                        | 50 +++++++++++++++++--
 4 files changed, 87 insertions(+), 8 deletions(-)
 create mode 100644 db/migrate/20240511003019_add_unique_index_on_item_id_to_nc_mall_records.rb

diff --git a/app/services/nc_mall.rb b/app/services/nc_mall.rb
index b4168a1d..a0a77537 100644
--- a/app/services/nc_mall.rb
+++ b/app/services/nc_mall.rb
@@ -6,7 +6,7 @@ module NCMall
 	# each request. (This library does that automatically!)
 	INTERNET = Async::HTTP::Internet.instance
 
-	# Load the NC home page, and return its useful data.
+	# Load the NC Mall home page content area, and return its useful data.
 	HOME_PAGE_URL = "https://ncmall.neopets.com/mall/ajax/home_page.phtml"
 	def self.load_home_page
 		load_page_by_url HOME_PAGE_URL
@@ -20,6 +20,30 @@ module NCMall
 		load_page_by_url CATEGORY_PAGE_URL_TEMPLATE.expand(type:, cat:)
 	end
 
+	# Load the NC Mall root document HTML, and extract the list of links to
+	# other pages ("New", "Popular", etc.)
+	ROOT_DOCUMENT_URL = "https://ncmall.neopets.com/mall/shop.phtml"
+	PAGE_LINK_PATTERN = /load_items_pane\(['"](.+?)['"], ([0-9]+)\).+?>(.+?)</
+	def self.load_page_links
+		Sync do
+			response = INTERNET.get(ROOT_DOCUMENT_URL, [
+				["User-Agent", Rails.configuration.user_agent_for_neopets],
+			])
+
+			if response.status != 200
+				raise ResponseNotOK.new(response.status),
+					"expected status 200 but got #{response.status} (#{url})"
+			end
+
+			# Extract `load_items_pane` calls from the root document's HTML. (We use
+			# a very simplified regex, rather than actually parsing the full HTML!)
+			html = response.read
+			html.scan(PAGE_LINK_PATTERN).
+				map { |type, cat, label| {type:, cat:, label:} }.
+				uniq
+		end
+	end
+
 	private
 
 	def self.load_page_by_url(url)
@@ -51,6 +75,10 @@ module NCMall
 			raise UnexpectedResponseFormat, "missing field object_data in NC page"
 		end
 
+		# NOTE: When there's no object data, it will be an empty array instead of
+		# an empty hash. Weird API thing to work around!
+		nc_page["object_data"] = {} if nc_page["object_data"] == []
+
 		items = nc_page["object_data"].values.map do |item_info|
 			{
 				id: item_info["id"],
diff --git a/db/migrate/20240511003019_add_unique_index_on_item_id_to_nc_mall_records.rb b/db/migrate/20240511003019_add_unique_index_on_item_id_to_nc_mall_records.rb
new file mode 100644
index 00000000..9afd67eb
--- /dev/null
+++ b/db/migrate/20240511003019_add_unique_index_on_item_id_to_nc_mall_records.rb
@@ -0,0 +1,11 @@
+class AddUniqueIndexOnItemIdToNCMallRecords < ActiveRecord::Migration[7.1]
+  def change
+    # NOTE: We need to temporarily remove the foreign key, then add it back
+    # once the index is in.
+    remove_foreign_key :nc_mall_records, :items
+    remove_index :nc_mall_records, :item_id
+
+    add_index :nc_mall_records, :item_id, unique: true
+    add_foreign_key :nc_mall_records, :items
+  end
+end
diff --git a/db/schema.rb b/db/schema.rb
index 5fbefe21..3afd9d56 100644
--- a/db/schema.rb
+++ b/db/schema.rb
@@ -10,7 +10,7 @@
 #
 # It's strongly recommended that you check this file into your version control system.
 
-ActiveRecord::Schema[7.1].define(version: 2024_05_07_235742) do
+ActiveRecord::Schema[7.1].define(version: 2024_05_11_003019) do
   create_table "alt_styles", charset: "utf8mb4", collation: "utf8mb4_unicode_520_ci", force: :cascade do |t|
     t.integer "species_id", null: false
     t.integer "color_id", null: false
@@ -162,7 +162,7 @@ ActiveRecord::Schema[7.1].define(version: 2024_05_07_235742) do
     t.datetime "discount_ends_at"
     t.datetime "created_at", null: false
     t.datetime "updated_at", null: false
-    t.index ["item_id"], name: "index_nc_mall_records_on_item_id"
+    t.index ["item_id"], name: "index_nc_mall_records_on_item_id", unique: true
   end
 
   create_table "neopets_connections", id: :integer, charset: "utf8mb4", collation: "utf8mb4_unicode_520_ci", force: :cascade do |t|
diff --git a/lib/tasks/nc_mall.rake b/lib/tasks/nc_mall.rake
index 85a3e8bc..837e12fd 100644
--- a/lib/tasks/nc_mall.rake
+++ b/lib/tasks/nc_mall.rake
@@ -4,9 +4,12 @@ namespace :nc_mall do
 		# Log to STDOUT.
 		Rails.logger = Logger.new(STDOUT)
 
-		# First, load all records of what's being sold in the live NC Mall.
-		# TODO: Load from other pages, too!
-		live_item_records = NCMall.load_home_page[:items]
+		# First, load all records of what's being sold in the live NC Mall. We load
+		# the homepage and all pages linked from the main document, and extract the
+		# items from each. (We also de-duplicate the items, which is important
+		# because the algorithm expects to only process each item once!)
+		pages = load_all_nc_mall_pages
+		live_item_records = pages.map { |p| p[:items] }.flatten.uniq
 
 		# Then, get the existing NC Mall records in our database. (We include the
 		# items, to be able to output the item name during logging.)
@@ -17,7 +20,8 @@ namespace :nc_mall do
 		# we've seen before. (We'll skip records for items we don't know.)
 		live_item_ids = live_item_records.map { |r| r[:id] }
 		recognized_item_ids = Item.where(id: live_item_ids).pluck(:id).to_set
-		Rails.logger.debug "We recognize #{recognized_item_ids.size} of these items"
+		Rails.logger.debug "We found #{live_item_records.size} items, and we " +
+			"recognize #{recognized_item_ids.size} of them."
 
 		# For each record in the live NC Mall, check if there's an existing record.
 		# If so, update it, and remove it from the existing records hash. If not,
@@ -33,8 +37,19 @@ namespace :nc_mall do
 			record.discount_price = record_data.dig(:discount, :price)
 			record.discount_begins_at = record_data.dig(:discount, :begins_at)
 			record.discount_ends_at = record_data.dig(:discount, :ends_at)
+
+			if !record.changed?
+				Rails.logger.info "Skipping record for item #{record_data[:name]} " +
+					"(unchanged)"
+				next
+			end
+
 			if record.save
-				Rails.logger.info "Saved record for item #{record_data[:name]}"
+				if record.previously_new_record?
+					Rails.logger.info "Created record for item #{record_data[:name]}"
+				else
+					Rails.logger.info "Updated record for item #{record_data[:name]}"
+				end
 			else
 				Rails.logger.error "Failed to save record for item " +
 					"#{record_data[:name]}: " +
@@ -58,3 +73,28 @@ namespace :nc_mall do
 		end
 	end
 end
+
+def load_all_nc_mall_pages
+	Sync do
+		# First, start loading the homepage.
+		homepage_task = Async { NCMall.load_home_page }
+
+		# Next, load the page links for different categories etc.
+		links = NCMall.load_page_links
+
+		# Next, load the linked pages, 10 at a time.
+		barrier = Async::Barrier.new
+		semaphore = Async::Semaphore.new(10, parent: barrier)
+		begin
+			linked_page_tasks = links.map do |link|
+				semaphore.async { NCMall.load_page link[:type], link[:cat] }
+			end
+			barrier.wait # Load all the pages.
+		ensure
+			barrier.stop # If any pages failed, cancel the rest.
+		end
+
+		# Finally, return all the pages: the homepage, and the linked pages.
+		[homepage_task.wait] + linked_page_tasks.map(&:wait)
+	end
+end