Improve URI#normalize (#7635)

Adds normalization for scheme, host and path.

spec/std/uri_spec.cr

@@ -159,31 +159,19 @@ describe "URI" do
     it { URI.parse("/foo").relative?.should be_true }
   end
 
-  describe "normalize" do
-    it "removes dot notation from path" do
-      cases = {
-        "../bar"      => "bar",
-        "./bar"       => "bar",
-        ".././bar"    => "bar",
-        "/foo/./bar"  => "/foo/bar",
-        "/bar/./"     => "/bar/",
-        "/."          => "/",
-        "/bar/."      => "/bar/",
-        "/foo/../bar" => "/bar",
-        "/bar/../"    => "/",
-        "/.."         => "/",
-        "/bar/.."     => "/",
-        "/foo/bar/.." => "/foo/",
-        "."           => "",
-        ".."          => "",
-      }
-
-      cases.each do |input, expected|
-        uri = URI.parse(input)
-        uri = uri.normalize
-
-        uri.path.should eq(expected), "failed to remove dot notation from #{input}"
-      end
+  describe "#normalize" do
+    it "doesn't modify instance" do
+      uri = URI.parse("HTTP://example.COM:80/./foo/../bar/")
+      uri.normalize.should eq URI.parse("http://example.com/bar/")
+      uri.should eq URI.parse("HTTP://example.COM:80/./foo/../bar/")
+    end
+
+    it "normalizes scheme" do
+      URI.parse("HtTp://").normalize.should eq URI.parse("http://")
+    end
+
+    it "normalizes host" do
+      URI.parse("http://FoO.cOm/").normalize.should eq URI.parse("http://foo.com/")
     end
 
     it "removes default port" do
@@ -194,6 +182,34 @@ describe "URI" do
       URI.new("ldap", "www.example.com", 389).normalize.to_s.should eq("ldap://www.example.com")
       URI.new("ldaps", "www.example.com", 636).normalize.to_s.should eq("ldaps://www.example.com")
     end
+
+    it "removes dot notation from path" do
+      URI.new(path: "../bar").normalize.path.should eq "bar"
+      URI.new(path: "./bar").normalize.path.should eq "bar"
+      URI.new(path: ".././bar").normalize.path.should eq "bar"
+      URI.new(path: "/foo/./bar").normalize.path.should eq "/foo/bar"
+      URI.new(path: "/bar/./").normalize.path.should eq "/bar/"
+      URI.new(path: "/.").normalize.path.should eq "/"
+      URI.new(path: "/bar/.").normalize.path.should eq "/bar/"
+      URI.new(path: "/foo/../bar").normalize.path.should eq "/bar"
+      URI.new(path: "/bar/../").normalize.path.should eq "/"
+      URI.new(path: "/..").normalize.path.should eq "/"
+      URI.new(path: "/bar/..").normalize.path.should eq "/"
+      URI.new(path: "/foo/bar/..").normalize.path.should eq "/foo/"
+      URI.new(path: ".").normalize.path.should eq ""
+      URI.new(path: "..").normalize.path.should eq ""
+    end
+
+    it "prefixes relative path with colon with `./`" do
+      URI.parse("./a:b").normalize.should eq URI.parse("./a:b")
+      URI.parse("http:a:b").normalize.should eq URI.parse("http:./a:b")
+    end
+  end
+
+  it "#normalize!" do
+    uri = URI.parse("HTTP://example.COM:80/./foo/../bar/")
+    uri.normalize!
+    uri.should eq URI.parse("http://example.com/bar/")
   end
 
   it "implements ==" do

src/uri.cr

@@ -212,17 +212,35 @@ class URI
     end
   end
 
-  # Returns normalized URI.
-  def normalize : self
-    uri = dup
-    uri.normalize!
-    uri
+  # Returns a normalized copy of this URI.
+  #
+  # See `#normalize!` for details.
+  def normalize : URI
+    dup.normalize!
   end
 
-  # Destructive normalize.
-  def normalize! : Nil
-    @path = remove_dot_segments(path)
+  # Normalizes this URI instance.
+  #
+  # The following normalizations are applied to the individual components (if available):
+  #
+  # * `scheme` is lowercased.
+  # * `host` is lowercased.
+  # * `port` is removed if it is the `.default_port?` of the scheme.
+  # * `path` is resolved to a minimal, semantical equivalent representation removing
+  #    dot segments `/.` and `/..`.
+  #
+  # ```
+  # uri = URI.parse("HTTP://example.COM:80/./foo/../bar/")
+  # uri.normalize!
+  # uri # => http://example.com/bar/
+  # ```
+  def normalize! : URI
+    @scheme = @scheme.try &.downcase
+    @host = @host.try &.downcase
     @port = nil if default_port?
+    @path = remove_dot_segments(path)
+
+    self
   end
 
   # Parses `raw_url` into an URI. The `raw_url` may be relative or absolute.
@@ -468,6 +486,10 @@ class URI
         path = path[segment_end_idx..-1]
       end
     end
+    first = result.first?
+    if first && !first.starts_with?('/') && first.includes?(':')
+      result.unshift "./"
+    end
 
     result.join
   end