Handle URLs with invalid characters

lib/datadog/tracing/contrib/utils/quantization/http.rb

@@ -19,7 +19,9 @@ module HTTP
             def url(url, options = {})
               url!(url, options)
             rescue StandardError
-              options[:placeholder] || PLACEHOLDER
+              placeholder = options[:placeholder] || PLACEHOLDER
+
+              options[:base] == :exclude ? placeholder : "#{base_url(url)}/#{placeholder}"
             end
 
             def base_url(url, options = {})
@@ -28,6 +30,12 @@ def base_url(url, options = {})
                 uri.query = nil
                 uri.fragment = nil
               end.to_s
+            rescue StandardError
+              if (m = %r{([a-z]+://[^/]+)(?:/|$)}.match(url))
+                m[1]
+              else
+                ''
+              end
             end
 
             def url!(url, options = {})

spec/datadog/tracing/contrib/utils/quantization/http_spec.rb

@@ -69,7 +69,26 @@
         # URLs do not permit unencoded non-ASCII characters in the URL.
         let(:url) { 'http://example.com/path?繋がってて' }
 
-        it { is_expected.to eq(described_class::PLACEHOLDER) }
+        it { is_expected.to eq(format('http://example.com/%s', described_class::PLACEHOLDER)) }
+
+        context 'and base: :exclude' do
+          let(:options) { { base: :exclude } }
+
+          it { is_expected.to eq(described_class::PLACEHOLDER) }
+        end
+      end
+
+      context 'with unencoded ASCII characters' do
+        # URLs do not permit all ASCII characters to be unencoded in the URL.
+        let(:url) { 'http://example.com/|' }
+
+        it { is_expected.to eq(format('http://example.com/%s', described_class::PLACEHOLDER)) }
+
+        context 'and base: :exclude' do
+          let(:options) { { base: :exclude } }
+
+          it { is_expected.to eq(described_class::PLACEHOLDER) }
+        end
       end
 
       context 'with internal obfuscation and the default replacement' do
@@ -95,6 +114,46 @@
     end
   end
 
+  describe '#base_url' do
+    subject(:result) { described_class.base_url(url, options) }
+
+    let(:options) { {} }
+
+    context 'given a URL' do
+      let(:url) { 'http://example.com/path?category_id=1&sort_by=asc#featured' }
+
+      context 'default behavior' do
+        it { is_expected.to eq('http://example.com') }
+      end
+
+      context 'with Unicode characters' do
+        # URLs do not permit unencoded non-ASCII characters in the URL.
+        let(:url) { 'http://example.com/path?繋がってて' }
+
+        it { is_expected.to eq('http://example.com') }
+      end
+
+      context 'with unencoded ASCII characters' do
+        # URLs do not permit all ASCII characters to be unencoded in the URL.
+        let(:url) { 'http://example.com/|' }
+
+        it { is_expected.to eq('http://example.com') }
+      end
+
+      context 'without a base' do
+        let(:url) { '/foo' }
+
+        it { is_expected.to eq('') }
+      end
+
+      context 'that is entirely invalid' do
+        let(:url) { "\x00" }
+
+        it { is_expected.to eq('') }
+      end
+    end
+  end
+
   describe '#query' do
     subject(:result) { described_class.query(query, options) }