Encode ARNs in RFC compliant mode

aws · Oct 17, 2023 · 0545f80 · 0545f80
1 parent 06c16a2
commit 0545f80
Show file tree

Hide file tree

Showing 9 changed files with 802 additions and 137 deletions.
diff --git a/src/aws-cpp-sdk-core/include/aws/core/endpoint/AWSEndpoint.h b/src/aws-cpp-sdk-core/include/aws/core/endpoint/AWSEndpoint.h
@@ -44,6 +44,11 @@ namespace Aws
                 m_uri.AddPathSegments(std::forward<T>(pathSegments));
             }
 
+            inline void SetRfc3986Encoded(bool rfcEncoded)
+            {
+                m_uri.SetRfc3986Encoded(rfcEncoded);
+            }
+
             using OptionalError = Crt::Optional<Aws::Client::AWSError<Aws::Client::CoreErrors>>;
             OptionalError AddPrefixIfMissing(const Aws::String& prefix);
 

diff --git a/src/aws-cpp-sdk-core/include/aws/core/http/URI.h b/src/aws-cpp-sdk-core/include/aws/core/http/URI.h
@@ -180,6 +180,16 @@ namespace Aws
             */
             Aws::String GetURIString(bool includeQueryString = true) const;
 
+            /**
+            * Returns true if this URI is going to be encoded in Rfc3986 compliant mode
+            */
+            inline bool IsRfc3986Encoded() const { return m_useRfcEncoding; }
+
+            /**
+            * Sets Rfc3986 compliant encoding mode. False (i.e. use legacy encoding with some chars unescaped) is the default.
+            */
+            inline void SetRfc3986Encoded(const bool value) { m_useRfcEncoding = value; }
+
             /**
              * URLEncodes the path portions of path (doesn't encode the "/" portion)
              * Keeps the first and the last "/".
@@ -189,7 +199,7 @@ namespace Aws
             /**
              * URLEncodes the path portion of the URI according to RFC3986
              */
-            static Aws::String URLEncodePathRFC3986(const Aws::String& path);
+            static Aws::String URLEncodePathRFC3986(const Aws::String& path, bool  = false);
 
         private:
             void ParseURIParts(const Aws::String& uri);
@@ -205,6 +215,7 @@ namespace Aws
             uint16_t m_port = HTTP_DEFAULT_PORT;
             Aws::Vector<Aws::String> m_pathSegments;
             bool m_pathHasTrailingSlash = false;
+            bool m_useRfcEncoding = false;
             Aws::String m_queryString;
         };
 

diff --git a/src/aws-cpp-sdk-core/include/aws/core/utils/StringUtils.h b/src/aws-cpp-sdk-core/include/aws/core/utils/StringUtils.h
@@ -49,6 +49,11 @@ namespace Aws
             */
             static Aws::String URLEncode(const char* unsafe);
 
+            static inline Aws::String URLEncode(const Aws::String& unsafe)
+            {
+              return URLEncode(unsafe.c_str());
+            }
+
             /**
             * Http Clients tend to escape some characters but not all. Escaping all of them causes problems, because the client
             * will also try to escape them.

diff --git a/src/aws-cpp-sdk-core/source/http/URI.cpp b/src/aws-cpp-sdk-core/source/http/URI.cpp
@@ -27,10 +27,10 @@ const char* SEPARATOR = "://";
 bool s_compliantRfc3986Encoding = false;
 void SetCompliantRfc3986Encoding(bool compliant) { s_compliantRfc3986Encoding = compliant; }
 
-Aws::String urlEncodeSegment(const Aws::String& segment)
+Aws::String urlEncodeSegment(const Aws::String& segment, bool rfcEncoded = false)
 {
     // consolidates legacy escaping logic into one local method
-    if (s_compliantRfc3986Encoding)
+    if (rfcEncoded || s_compliantRfc3986Encoding)
     {
         return StringUtils::URLEncode(segment.c_str());
     }
@@ -141,7 +141,7 @@ void URI::SetScheme(Scheme value)
     }
 }
 
-Aws::String URI::URLEncodePathRFC3986(const Aws::String& path)
+Aws::String URI::URLEncodePathRFC3986(const Aws::String& path, bool rfcEncoded)
 {
     if (path.empty())
     {
@@ -155,7 +155,7 @@ Aws::String URI::URLEncodePathRFC3986(const Aws::String& path)
     // escape characters appearing in a URL path according to RFC 3986
     for (const auto& segment : pathParts)
     {
-        ss << '/' << urlEncodeSegment(segment);
+        ss << '/' << urlEncodeSegment(segment, rfcEncoded);
     }
 
     // if the last character was also a slash, then add that back here.
@@ -237,7 +237,7 @@ Aws::String URI::GetURLEncodedPathRFC3986() const
     // (mostly; there is some non-standards legacy support that can be disabled)
     for (const auto& segment : m_pathSegments)
     {
-        ss << '/' << urlEncodeSegment(segment);
+        ss << '/' << urlEncodeSegment(segment, m_useRfcEncoding);
     }
 
     if (m_pathSegments.empty() || m_pathHasTrailingSlash)

diff --git a/tests/aws-cpp-sdk-core-tests/http/URITest.cpp b/tests/aws-cpp-sdk-core-tests/http/URITest.cpp
@@ -286,6 +286,15 @@ TEST_F(URITest, TestParseWithColon)
     EXPECT_EQ(80, complexUri.GetPort());
     EXPECT_STREQ("/awsnativesdkputobjectstestbucket20150702T200059Z/TestObject:1234/awsnativesdkputobjectstestbucket20150702T200059Z/TestObject:Key", complexUri.GetPath().c_str());
     EXPECT_STREQ(strComplexUri, complexUri.GetURIString().c_str());
+
+    URI complexUriCompliant(strComplexUri);
+    complexUriCompliant.SetRfc3986Encoded(true);
+    EXPECT_STREQ("s3.us-east-1.amazonaws.com", complexUriCompliant.GetAuthority().c_str());
+    EXPECT_EQ(80, complexUriCompliant.GetPort());
+    EXPECT_STREQ("/awsnativesdkputobjectstestbucket20150702T200059Z/TestObject:1234/awsnativesdkputobjectstestbucket20150702T200059Z/TestObject:Key",
+                 complexUri.GetPath().c_str());
+    EXPECT_STREQ("http://s3.us-east-1.amazonaws.com/awsnativesdkputobjectstestbucket20150702T200059Z/TestObject%3A1234/awsnativesdkputobjectstestbucket20150702T200059Z/TestObject%3AKey",
+                 complexUriCompliant.GetURIString().c_str());
 }
 
 TEST_F(URITest, TestParseWithColonCompliant)