Downcase encoded slug by default and more specs

lib/slug.rb

@@ -3,8 +3,9 @@
 module Slug
 
   CHAR_FILTER_REGEXP = /[:\/\?#\[\]@!\$&'\(\)\*\+,;=_\.~%\\`^\s|\{\}"<>]+/ # :/?#[]@!$&'()*+,;=_.~%\`^|{}"<>
+  MAX_LENGTH = 255
 
-  def self.for(string, default = 'topic')
+  def self.for(string, default = 'topic', max_length = MAX_LENGTH)
     slug =
       case (SiteSetting.slug_generation_method || :ascii).to_sym
       when :ascii then self.ascii_generator(string)
@@ -13,30 +14,38 @@ module Slug
       end
     # Reject slugs that only contain numbers, because they would be indistinguishable from id's.
     slug = (slug =~ /[^\d]/ ? slug : '')
+    slug = self.prettify_slug(slug, max_length: max_length)
     slug.blank? ? default : slug
   end
 
-  def self.sanitize(string)
+  def self.sanitize(string, downcase: false, max_length: MAX_LENGTH)
-    self.encoded_generator(string)
+    slug = self.encoded_generator(string, downcase: downcase)
+    self.prettify_slug(slug, max_length: max_length)
   end
 
   private
 
-  def self.ascii_generator(string)
+  def self.prettify_slug(slug, max_length: MAX_LENGTH)
-    string.tr("'", "")
+    slug.tr!("_", "-")
-      .parameterize
+    slug = slug.squeeze('-') # squeeze continuous dashes to prettify slug
-      .tr("_", "-")
+    slug.gsub!(/\A-+|-+\z/, '') # remove possible trailing and preceding dashes
+    slug.truncate(max_length, omission: '')
   end
 
-  def self.encoded_generator(string)
+  def self.ascii_generator(string)
+    string.tr!("'", "")
+    string.parameterize
+  end
+
+  def self.encoded_generator(string, downcase: true)
     # This generator will sanitize almost all special characters,
     # including reserved characters from RFC3986.
     # See also URI::REGEXP::PATTERN.
-    string.strip
+    string.strip!
-      .gsub(/\s+/, '-')
+    string.gsub!(/\s+/, '-')
-      .gsub(CHAR_FILTER_REGEXP, '')
+    string.gsub!(CHAR_FILTER_REGEXP, '')
-      .gsub(/\A-+|-+\z/, '') # remove possible trailing and preceding dashes
+    string.downcase! if downcase
-      .squeeze('-') # squeeze continuous dashes to prettify slug
+    string
   end
 
   def self.none_generator(string)

spec/components/slug_spec.rb

@@ -6,6 +6,24 @@ require 'slug'
 describe Slug do
 
   describe '#for' do
+    let(:default_slug) { 'topic' }
+
+    let(:very_long_string) do
+      '内容似乎不清晰，这是个完整的句子吗？内容似乎不清晰，这是个完整的句子吗？' * 10
+    end
+
+    it 'returns topic by default' do
+      expect(Slug.for('')).to eq default_slug
+    end
+
+    it 'accepts fallback' do
+      expect(Slug.for('', 'king')).to eq 'king'
+    end
+
+    it 'replaces the underscore' do
+      expect(Slug.for("o_o_o")).to eq("o-o-o")
+    end
+
     context 'ascii generator' do
       before { SiteSetting.slug_generation_method = 'ascii' }
 
@@ -14,11 +32,15 @@ describe Slug do
       end
 
       it 'generates default slug when nothing' do
-        expect(Slug.for('')).to eq('topic')
+        expect(Slug.for('')).to eq(default_slug)
       end
 
       it "doesn't generate slugs that are just numbers" do
-        expect(Slug.for('123')).to eq('topic')
+        expect(Slug.for('123')).to eq(default_slug)
+      end
+
+      it "fallbacks to empty string if it's too long" do
+        expect(Slug.for(very_long_string)).to eq(default_slug)
       end
     end
 
@@ -28,14 +50,25 @@ describe Slug do
 
       it 'generates the slug' do
         expect(Slug.for("熱帶風暴畫眉")).to eq('熱帶風暴畫眉')
+        expect(Slug.for("Jeff hate's !~-_|,=#this")).to eq("jeff-hates-this")
       end
 
       it 'generates default slug when nothing' do
-        expect(Slug.for('')).to eq('topic')
+        expect(Slug.for('')).to eq(default_slug)
       end
 
       it "doesn't generate slugs that are just numbers" do
-        expect(Slug.for('123')).to eq('topic')
+        expect(Slug.for('123')).to eq(default_slug)
+      end
+
+      it "handles the special characters" do
+        expect(Slug.for(
+          " - English and Chinese title with special characters / 中文标题 !@:?\\:'`#^& $%&*()` -- "
+        )).to eq("english-and-chinese-title-with-special-characters-中文标题")
+      end
+
+      it "kills the trailing dash" do
+        expect(Slug.for("2- -this!~-_|,we-#-=^-")).to eq('2-this-we')
       end
     end
 
@@ -45,9 +78,9 @@ describe Slug do
 
       it 'generates the slug' do
         expect(Slug.for("hello world", 'category')).to eq('category')
-        expect(Slug.for("hello world")).to eq('topic')
+        expect(Slug.for("hello world")).to eq(default_slug)
-        expect(Slug.for('')).to eq('topic')
+        expect(Slug.for('')).to eq(default_slug)
-        expect(Slug.for('123')).to eq('topic')
+        expect(Slug.for('123')).to eq(default_slug)
       end
     end
   end
@@ -89,10 +122,6 @@ describe Slug do
       expect(Slug.ascii_generator(from)).to eq(to)
     end
 
-    it 'replaces underscores' do
-      expect(Slug.ascii_generator("o_o_o")).to eq("o-o-o")
-    end
-
     it "doesn't keep single quotes within word" do
       expect(Slug.ascii_generator("Jeff hate's this")).to eq("jeff-hates-this")
     end
@@ -111,15 +140,13 @@ describe Slug do
     after { SiteSetting.slug_generation_method = 'ascii' }
 
     it 'generates precentage encoded string' do
-      expect(Slug.encoded_generator("Jeff hate's !~-_|,=#this")).to eq("Jeff-hates-this")
       expect(Slug.encoded_generator("뉴스피드")).to eq("뉴스피드")
       expect(Slug.encoded_generator("آموزش اضافه کردن لینک اختیاری به هدر")).to eq("آموزش-اضافه-کردن-لینک-اختیاری-به-هدر")
       expect(Slug.encoded_generator("熱帶風暴畫眉")).to eq("熱帶風暴畫眉")
     end
 
     it 'reject RFC 3986 reserved character and blank' do
-      expect(Slug.encoded_generator(":/?#[]@!$ &'()*+,;=% -_`~.")).to eq("")
+      expect(Slug.encoded_generator(":/?#[]@!$ &'()*+,;=% -_`~.")).to eq("---") # will be clear by #for
-      expect(Slug.encoded_generator(" - English and Chinese title with special characters / 中文标题 !@:?\\:'`#^& $%&*()` -- ")).to eq("English-and-Chinese-title-with-special-characters-中文标题")
     end
 
     it 'generates null when nothing' do
@@ -129,6 +156,10 @@ describe Slug do
     it "keeps number unchanged" do
       expect(Slug.encoded_generator('123')).to eq('123')
     end
+
+    it 'downcase the string' do
+      expect(Slug.encoded_generator("LoWer")).to eq('lower')
+    end
   end
 
   describe '#none_generator' do

spec/controllers/categories_controller_spec.rb

@@ -338,7 +338,7 @@ describe CategoriesController do
         expect(@category.reload.slug).to eq('valid-slug')
       end
 
-      it 'accepts and sanitize custom slug when the slug generation method is not english' do
+      it 'accepts and sanitize custom slug when the slug generation method is not ascii' do
         SiteSetting.slug_generation_method = 'none'
         put :update_slug,
           params: { category_id: @category.id, slug: ' another !_ slug @' },