From 327ab4dfbf9670bc3144e348fba837c3f2019e1d Mon Sep 17 00:00:00 2001
From: sfc-gh-ext-simba-hx <harry.xi@insightsoftware.com>
Date: Wed, 25 Oct 2023 14:21:33 -0700
Subject: [PATCH] put with GCP to skip existing file

---
 cpp/SnowflakeGCSClient.cpp      | 31 +++++++++++++++++----
 tests/test_unit_put_get_gcs.cpp | 49 +++++++++++++++++++++++++++++++--
 2 files changed, 72 insertions(+), 8 deletions(-)

diff --git a/cpp/SnowflakeGCSClient.cpp b/cpp/SnowflakeGCSClient.cpp
index 7fd5d41a1c..209c13a2ff 100644
--- a/cpp/SnowflakeGCSClient.cpp
+++ b/cpp/SnowflakeGCSClient.cpp
@@ -66,6 +66,18 @@ RemoteStorageRequestOutcome SnowflakeGCSClient::upload(FileMetadata *fileMetadat
   if (!m_gcsAccessToken.empty())
   {
     std::string filePathFull = m_stageInfo->location + fileMetadata->destFileName;
+
+    //check if file exists if overwrite is not specified.
+    if (!fileMetadata->overWrite)
+    {
+      RemoteStorageRequestOutcome outcome = GetRemoteFileMetadata(&filePathFull, fileMetadata);
+      if (RemoteStorageRequestOutcome::SUCCESS == outcome)
+      {
+        CXX_LOG_DEBUG("File already exists skipping the file upload %s",
+                      fileMetadata->srcFileToUpload.c_str());
+        return RemoteStorageRequestOutcome::SKIP_UPLOAD_FILE;
+      }
+    }
     buildGcsRequest(filePathFull, url, reqHeaders);
   }
   else
@@ -134,12 +146,21 @@ RemoteStorageRequestOutcome SnowflakeGCSClient::GetRemoteFileMetadata(
     url = fileMetadata->presignedUrl;
   }
 
-  if (!m_statement->http_get(url,
-                             reqHeaders,
-                             NULL,
-                             headerString,
-                             true))
+  try
+  {
+    if (!m_statement->http_get(url,
+                               reqHeaders,
+                               NULL,
+                               headerString,
+                               true))
+    {
+      return RemoteStorageRequestOutcome::FAILED;
+    }
+  }
+  catch (...)
   {
+    // It's expected to fail when file doesn't exist, while http_get()
+    // could throw excpetion on that. Catch that and return FAILED
     return RemoteStorageRequestOutcome::FAILED;
   }
 
diff --git a/tests/test_unit_put_get_gcs.cpp b/tests/test_unit_put_get_gcs.cpp
index a2a6699c24..53872e9ff2 100644
--- a/tests/test_unit_put_get_gcs.cpp
+++ b/tests/test_unit_put_get_gcs.cpp
@@ -32,7 +32,10 @@ class MockedGCSStatement : public Snowflake::Client::IStatementPutGet
 {
 public:
   MockedGCSStatement(bool useGcsToken)
-    : m_useGcsToken(useGcsToken), Snowflake::Client::IStatementPutGet()
+    : m_useGcsToken(useGcsToken),
+      m_firstTime(true),
+      m_isPut(false),
+      Snowflake::Client::IStatementPutGet()
   {
     m_stageInfo.stageType = Snowflake::Client::StageType::GCS;
     if (m_useGcsToken)
@@ -63,9 +66,18 @@ class MockedGCSStatement : public Snowflake::Client::IStatementPutGet
     }
     else
     {
+      m_isPut = true;
       putGetParseResponse->command = CommandType::UPLOAD;
       putGetParseResponse->srcLocations = m_srcLocationsPut;
       putGetParseResponse->threshold = DEFAULT_UPLOAD_DATA_SIZE_THRESHOLD;
+      if (sql->find("OVERWRITE=true") != std::string::npos)
+      {
+        putGetParseResponse->overwrite = true;
+      }
+      else
+      {
+        putGetParseResponse->overwrite = false;
+      }
     }
     if (!m_useGcsToken)
     {
@@ -106,8 +118,17 @@ class MockedGCSStatement : public Snowflake::Client::IStatementPutGet
                         std::string& responseHeaders,
                         bool headerOnly) override
   {
-    if (headerOnly)
+    if (headerOnly) // try getting metadata
     {
+      if (m_isPut)
+      {
+        if ((strcasecmp(url.c_str(), m_expectedUrl.c_str()) == 0) && m_firstTime)
+        {
+          // first time retry false to mock the case of file doesn't exist
+          m_firstTime = false;
+          return false;
+        }
+      }
       responseHeaders = "HTTP/1.1 200 OK\n"
                         "x-goog-meta-encryptiondata: {\"WrappedContentKey\":{\"KeyId\":\"symmKey1\",\"EncryptedKey\":\"MyZBZNLcndKTeR+xC8Msle5IcSYKsx/nYNn93OONSqs=\",\"Algorithm\":\"AES_CBC_256\"},\"ContentEncryptionIV\":\"30fmhKrf1aKyWidrv06NNA==\"}\n"
                         "Content-Type: application/octet-stream\n"
@@ -143,12 +164,16 @@ class MockedGCSStatement : public Snowflake::Client::IStatementPutGet
   bool m_useGcsToken;
 
   std::string m_expectedUrl;
+
+  bool m_firstTime;
+
+  bool m_isPut;
 };
 
 // test helper for put
 void put_gcs_test_core(bool useGcsToken)
 {
-  std::string cmd = std::string("put file://small_file.csv.gz @odbctestStage AUTO_COMPRESS=false OVERWRITE=true");
+  std::string cmd = std::string("put file://small_file.csv.gz @odbctestStage AUTO_COMPRESS=false");
 
   MockedGCSStatement mockedStatementPut(useGcsToken);
 
@@ -162,6 +187,24 @@ void put_gcs_test_core(bool useGcsToken)
     result->getColumnAsString(6, put_status);
     assert_string_equal("UPLOADED", put_status.c_str());
   }
+
+  if (useGcsToken)
+  {
+    result = agent.execute(&cmd);
+    while (result->next())
+    {
+      result->getColumnAsString(6, put_status);
+      assert_string_equal("SKIPPED", put_status.c_str());
+    }
+  }
+
+  cmd += " OVERWRITE=true";
+  result = agent.execute(&cmd);
+  while (result->next())
+  {
+    result->getColumnAsString(6, put_status);
+    assert_string_equal("UPLOADED", put_status.c_str());
+  }
 }
 
 // test helper for get