FEATURE: permalink normalization

Optionally allow admins to apply regex based normalization
to permalinks prior to matching.

This allows us to drop query string, or cleanly ignore slugs, etc.

app/models/permalink.rb

@@ -5,11 +5,72 @@ class Permalink < ActiveRecord::Base
 
   before_validation :normalize_url
 
-  def normalize_url
-    if self.url
-      self.url = self.url.strip
-      self.url = self.url[1..-1] if url[0,1] == '/'
+  class Normalizer
+    attr_reader :source
+
+    def initialize(source)
+      @source = source
+      if source.present?
+        @rules = source.split("|").map do |rule|
+          parse_rule(rule)
+        end.compact
+      end
     end
+
+    def parse_rule(rule)
+      return unless rule =~ /\/.*\//
+
+      escaping = false
+      regex = ""
+      sub = ""
+      c = 0
+
+      rule.chars.each do |l|
+        c += 1 if !escaping && l == "/"
+        escaping = l == "\\"
+
+        if c > 1
+          sub << l
+        else
+          regex << l
+        end
+      end
+
+      if regex.length > 1
+        [Regexp.new(regex[1..-1]), sub[1..-1] || ""] rescue nil
+      end
+
+    end
+
+    def normalize(url)
+      return url unless @rules
+      @rules.each do |(regex,sub)|
+        url = url.sub(regex,sub)
+      end
+
+      url
+    end
+
+  end
+
+  def self.normalize_url(url)
+    if url
+      url = url.strip
+      url = url[1..-1] if url[0,1] == '/'
+    end
+
+    normalizations =  SiteSetting.permalink_normalizations
+
+    @normalizer = Normalizer.new(normalizations) unless @normalizer && @normalizer.source == normalizations
+    @normalizer.normalize(url)
+  end
+
+  def self.find_by_url(url)
+    find_by(url: normalize_url(url))
+  end
+
+  def normalize_url
+    self.url = Permalink.normalize_url(url) if url
   end
 
   def target_url

config/locales/server.en.yml

@@ -1139,6 +1139,8 @@ en:
 
     suppress_uncategorized_badge: "Don't show the badge for uncategorized topics in topic lists."
 
+    permalink_normalizations: "Apply the following regex before matching permalinks, for example: /(\\/topic.*)\\?.*/\\1 will strip query strings from topic routes. Format is regex+string use \\1 etc. to access captures"
+
     global_notice: "Display an URGENT, EMERGENCY global banner notice to all visitors, change to blank to hide it (HTML allowed)."
 
     disable_edit_notifications: "Disables edit notifications by the system user when 'download_remote_images_to_local' is active."

config/site_settings.yml

@@ -806,6 +806,10 @@ uncategorized:
     default: 'ascii'
     enum: 'SlugSetting'
 
+  permalink_normalizations:
+    default: ''
+    type: list
+
   # Search
   min_search_term_length:
     client: true

spec/controllers/permalinks_controller_spec.rb

@@ -10,6 +10,17 @@ describe PermalinksController do
       expect(response.status).to eq(301)
     end
 
+    it "should apply normalizations" do
+      SiteSetting.permalink_normalizations = "/(.*)\\?.*/\\1"
+
+      permalink = Fabricate(:permalink, url: '/topic/bla', external_url: '/topic/100')
+
+      get :show, url: permalink.url, test: "hello"
+
+      expect(response).to redirect_to('/topic/100')
+      expect(response.status).to eq(301)
+    end
+
     it 'return 404 if permalink record does not exist' do
       get :show, url: '/not/a/valid/url'
       expect(response.status).to eq(404)

spec/models/permalink_spec.rb

@@ -2,6 +2,16 @@ require "spec_helper"
 
 describe Permalink do
 
+  describe "normalization" do
+    it "correctly normalizes" do
+      normalizer = Permalink::Normalizer.new("/(\\/hello.*)\\?.*/\\1|/(\\/bye.*)\\?.*/\\1")
+
+      expect(normalizer.normalize("/hello?a=1")).to eq("/hello")
+      expect(normalizer.normalize("/bye?test=1")).to eq("/bye")
+      expect(normalizer.normalize("/bla?a=1")).to eq("/bla?a=1")
+    end
+  end
+
   describe "new record" do
     it "strips blanks" do
       permalink = described_class.create(url: " my/old/url  ")