Refactor to remove route generator

benchmarks/lobsters/benchmark.rb

@@ -7,58 +7,14 @@
 use_gemfile extra_setup_cmd: "bin/rails db:drop db:create && sqlite3 db/production.sqlite3 < db/faked_bench_data.sql"
 
 require_relative 'config/environment'
+require_relative "route_generator"
 
 app = Rails.application
+generator = RouteGenerator.new(app)
+generator.do_login # We want to be logged in for all generated requests
+generator.routes # Make sure routes have been pregenerated
 
-# TODO: a wider variety of routes might better show megamorphism in call sites
-
-# Do we need to distinguish between e.g. banned and non-banned users?
-
-ROUTE_GROUPS = [
-  { num: 50, method: :GET, routes: ["/u"] }, # Users tree, showing order of invitation - lots of view logic
-  { num: 50, method: :GET, routes: ["/active", "/newest", "/recent", "/hottest"] }, # Views of the stories by attributes
-  { num: 25, method: :GET, routes: ["/rss", "/privacy", "/about", "/settings"] }, # Less-common and less-interesting routes for variation
-  { num: 25, method: :GET, routes: ["/top?length=1d", "/top?length=1w", "/top?length=1y"] }, # Top stories by time
-  { num: 50, method: :GET, routes: ["/hidden", "/saved", "/upvoted/stories"] }, # These all required being logged in
-
-  # Turned off flag_warning check for /threads -- to hard to port to SQLite
-  { num: 50, routes: ["/comments", "/upvoted/comments", "/threads", "/comments/:comment_id/reply"] },
-  { num: 25, routes: ["/threads/:username", "/u/:username"] },
-
-  { num: 50, routes: ["/replies", "/replies/comments", "/replies/stories", "/replies/unread"] },  # replies#stories, replies#unread, replies#comments
-
-  #{ num: 25, routes: ["/s/:story_id"] },  # need to check more into how stories work - e.g. story#show doesn't seem side-effect-free
-
-  # /categories: - admin-only
-  #{ num: 50, routes: ["/stats"] },  # Stats gets a 500, needs more MySQL->SQLite porting
-  # Shouldn't add /404, because that returns status 404, not 200
-  # No messages added during fake-data task, so skip messages controller
-  # The moderators controller isn't high-traffic, plus it has various MySQL time code that needs porting - skip it?
-
-]
-
-rng = Random.new(0x1be52551fc152997)
-
-db_ids = {
-  # Find appropriate model files
-  comment_id: Comment.all.pluck(:short_id),
-  #story_id: Story.all.pluck(:short_id),
-  username: User.all.pluck(:username),
-}
-
-visiting_envs = []
-ROUTE_GROUPS.each do |group|
-  group[:num].times do
-    route = group[:routes].sample(random: rng)
-    if route.include?(":")
-      route = route.gsub(/:(\w+)/) do |match|
-        db_ids[$1.to_sym].sample(random: rng)
-      end
-    end
-    visiting_envs << Rack::MockRequest::env_for("https://localhost#{route}", method: group[:method])
-  end
-end
-
+# Track ActiveRecord time
 if ENV['TRACK_AR_TIME']
   ar_total_duration = 0.0
   process_start_t = Time.now
@@ -76,36 +32,13 @@
   end
 end
 
-# Let's be able to log in as one specific user...
-# With the srand seed given in lib/tasks/fake_data.rake, we use the fake data for one of the users
-
-# Lobsters doesn't love setting a logged-in cookie if you previously had no cookie. So first GET /login
-# and set the cookie from there.
-login_get_env = Rack::MockRequest::env_for("https://localhost/login")
-login_get_resp = app.call(login_get_env)
-auth_token_line = login_get_resp[2].join.lines.detect { |line| line.include?("authenticity_token") && line.include?("value") }
-auth_token = auth_token_line.scan(/value="([^"]+)"/)[0][0]
-resp_cookie_header = login_get_resp[1]["Set-Cookie"] #+ "; tag_filters=NOCACHE" # turn off the file cache
-
-# Let's log in as one specific user...
-# With the srand seed given in lib/tasks/fake_data.rake, we use the fake data for one of the users
-login_post_env = Rack::MockRequest::env_for("https://localhost/login", method: "POST", params: { email: "[email protected]", password: "ji3W36xR", authenticity_token: auth_token })
-login_post_env["HTTP_COOKIE"] = resp_cookie_header
-login_post_resp = app.call(login_post_env)
-raise("Can't log in as fake user wiegand.michell: #{login_post_resp.inspect}") unless login_post_resp[0] == 302
-resp_cookie_header = login_post_resp[1]["Set-Cookie"] #+ "; tag_filters=NOCACHE" # turn off the file cache
-
 run_benchmark(10) do
-  visiting_envs.each_with_index do |env, idx|
+  generator.routes.each_with_index do |env, idx|
     path = env["PATH_INFO"] # app.call mutates the path
-    env["HTTP_COOKIE"] = resp_cookie_header # TODO: disable file cache w/ cookie?
-    response_array = app.call(env)
+    response_array = generator.visit(env) # Track HTTP cookies as we go along
     unless response_array.first == 200
       puts response_array.inspect
-      raise "HTTP status is #{response_array.first} instead of 200 for req #{idx}/#{visiting_envs.size}, #{path.inspect}. Is the benchmark app properly set up? See README.md."
-    end
-    if response_array[1]["Set-Cookie"]
-      resp_cookie_header = response_array[1]["Set-Cookie"]
+      raise "HTTP status is #{response_array.first} instead of 200 for req #{idx}/#{generator.routes.size}, #{path.inspect}. Is the benchmark app properly set up? See README.md."
     end
   end
 end

benchmarks/lobsters/route_generator.rb

@@ -0,0 +1,121 @@
+# Generate a set of routes for Lobsters
+
+class RouteGenerator
+  # Take a variety of routes and randomise order, distribution and specific data items (comments, users.)
+  ROUTE_GROUPS = [
+    { num: 50, method: :GET, routes: ["/u"] }, # Users tree, showing order of invitation - lots of view logic
+    { num: 50, method: :GET, routes: ["/active", "/newest", "/recent", "/hottest"] }, # Views of the stories by attributes
+    { num: 25, method: :GET, routes: ["/rss", "/privacy", "/about", "/settings"] }, # Less-common and less-interesting routes for variation
+    { num: 25, method: :GET, routes: ["/top?length=1d", "/top?length=1w", "/top?length=1y"] }, # Top stories by time
+    { num: 50, method: :GET, routes: ["/hidden", "/saved", "/upvoted/stories"] }, # These all required being logged in
+
+    # Turned off flag_warning check for /threads -- to hard to port to SQLite
+    { num: 50, routes: ["/comments", "/upvoted/comments", "/threads", "/comments/:comment_id/reply"] },
+    { num: 25, routes: ["/threads/:username", "/u/:username"] },
+
+    { num: 50, routes: ["/replies", "/replies/comments", "/replies/stories", "/replies/unread"] },  # replies#stories, replies#unread, replies#comments
+
+    #{ num: 25, routes: ["/s/:story_id"] },  # need to check more into how stories work - e.g. story#show doesn't seem side-effect-free
+
+    # /categories: - admin-only
+    #{ num: 50, routes: ["/stats"] },  # Stats gets a 500, needs more MySQL->SQLite porting
+    # Shouldn't add /404, because that returns status 404, not 200
+    # No messages added during fake-data task, so skip messages controller
+    # The moderators controller isn't high-traffic, plus it has various MySQL time code that needs porting - skip it?
+
+    # POSTs are harder here. Comments seem to exist mostly in the context of stories, which changes their behaviour.
+    # We'd need to do roughly what the Faker does, where we create a story and do various interaction in the context
+    # of it. For now, skip it.
+    #{ num: 10, method: :POST, routes: ["/comments/:comment_id/upvote"], post_opts: {} },
+  ]
+
+  def initialize(app, rng: nil)
+    @app = app
+
+    @auth_token = nil
+    @resp_cookie_header = nil
+    @logged_in = false
+
+    @rng = rng || Random.new(0x1be52551fc152997)
+  end
+
+  def routes
+    @routes ||= generate_routes
+  end
+
+  def visit(route)
+    route["HTTP_COOKIE"] = @resp_cookie_header
+    response_array = @app.call(route)
+    if response_array[1]["Set-Cookie"]
+      @resp_cookie_header = response_array[1]["Set-Cookie"]
+    end
+    response_array
+  end
+
+  ### Helpers to Query Rails Data
+
+  def auth_token
+    return @auth_token if @auth_token
+
+    # We need to log in to get a CSRF token. We'll use the same token for all requests.
+    # We also need to get the CSRF token before generating the env hashes for later requests.
+
+    # First GET /login and set the cookie from there. CSRF token from a single session should work throughout that session.
+    login_get_env = Rack::MockRequest::env_for("https://localhost/login")
+    login_get_resp = @app.call(login_get_env)
+    auth_token_line = login_get_resp[2].join.lines.detect { |line| line.include?("authenticity_token") && line.include?("value") }
+    @auth_token = auth_token_line.scan(/value="([^"]+)"/)[0][0]
+    @resp_cookie_header = login_get_resp[1]["Set-Cookie"] #+ "; tag_filters=NOCACHE" # turn off the file cache
+
+    @auth_token
+  end
+
+  def do_login
+    return if @logged_in
+
+    auth_token # make sure we have the auth token
+
+    # Let's log in as one specific user...
+    # With the srand seed given in lib/tasks/fake_data.rake, we use the fake data for one of the users
+    login_post_env = Rack::MockRequest::env_for("https://localhost/login", method: "POST", params: { email: "[email protected]", password: "ji3W36xR", authenticity_token: @auth_token })
+    login_post_env["HTTP_COOKIE"] = @resp_cookie_header
+    login_post_resp = @app.call(login_post_env)
+    raise("Can't log in as fake user wiegand.michell: #{login_post_resp.inspect}") unless login_post_resp[0] == 302
+    @resp_cookie_header = login_post_resp[1]["Set-Cookie"] #+ "; tag_filters=NOCACHE" # turn off the file cache
+    @logged_in = true
+  end
+
+  private
+
+  def generate_routes
+    db_ids = {
+      comment_id: Comment.all.pluck(:short_id),
+      username: User.all.pluck(:username),
+    }
+
+    # We want to randomise the order, but we need to make sure a user, comment, etc. exists when it's referenced.
+    # So we start by creating a set of references to "this group is at this point in the order" and then
+    # fill them in, keeping track of what data items exist as we go along.
+
+    group_list = ROUTE_GROUPS.flat_map { |group| (1..group[:num]).map { group } } # group[:num] references to each group
+    group_list.shuffle!(random: @rng)
+    route_group_envs = []
+    group_list.each do |group|
+      route = group[:routes].sample(random: @rng)
+      if route.include?(":")
+        route = route.gsub(/:(\w+)/) do |match|
+          db_ids[$1.to_sym].sample(random: @rng)
+        end
+      end
+      route_group_envs << Rack::MockRequest::env_for("https://localhost#{route}", method: group[:method])
+
+      # Do we need to mess with our list of data items?
+      # If we figure out comment upvote/flag/delete etc. we'll need some of this.
+      # For now, ignore.
+      #if group[:method] != :GET && group[:post_opts]
+      #end
+    end
+
+    route_group_envs
+  end
+end