Merge pull request #31 from quentingodeau/feature/proxy

Add proxy configuration support

.github/workflows/Linux.yml

@@ -26,6 +26,7 @@ jobs:
       GEN: Ninja
       VCPKG_TOOLCHAIN_PATH: ${{ github.workspace }}/vcpkg/scripts/buildsystems/vcpkg.cmake
       AZURE_STORAGE_CONNECTION_STRING: 'DefaultEndpointsProtocol=http;AccountName=devstoreaccount1;AccountKey=Eby8vdM02xNOcqFlqUwJPLlmEtlCDXJ1OUzFT50uSRZ6IFsuFq2UVErCz4I6tq/K1SZFPTOtr/KBHBeksoGMGw==;BlobEndpoint=http://127.0.0.1:10000/devstoreaccount1;QueueEndpoint=http://127.0.0.1:10001/devstoreaccount1;TableEndpoint=http://127.0.0.1:10002/devstoreaccount1;'
+      HTTP_PROXY_RUNNING: '1'
 
     steps:
     - name: Install required ubuntu packages
@@ -56,7 +57,7 @@ jobs:
     - name: install Azure test service
       if: ${{ matrix.arch == 'linux_amd64_gcc4' }}
       run: |
-        yum install -y nodejs npm
+        yum install -y nodejs npm squid
         npm install -g azurite
         echo -e "[azure-cli]\nname=Azure CLI\nbaseurl=https://packages.microsoft.com/yumrepos/azure-cli\nenabled=1\ngpgcheck=1\ngpgkey=https://packages.microsoft.com/keys/microsoft.asc" | tee /etc/yum.repos.d/azure-cli.repo
         yum install -y azure-cli
@@ -65,16 +66,19 @@ jobs:
       if: ${{ matrix.arch == 'linux_amd64' }}
       run: |
         curl -fsSL https://deb.nodesource.com/setup_16.x | bash
-        apt-get install -y -qq nodejs
+        apt-get install -y -qq nodejs squid
         node -v 
         npm -v
+        squid --version
         npm install -g azurite
         curl -sL https://aka.ms/InstallAzureCLIDeb | bash
 
     - name: Launch & populate Azure test service
       if: ${{ matrix.arch == 'linux_amd64' || matrix.arch == 'linux_amd64_gcc4' }}
       run: |
         azurite > azurite_log.txt 2>&1 &
+        ./scripts/run_squid.sh --port 3128 --log_dir squid_logs &
+        ./scripts/run_squid.sh --port 3129 --log_dir squid_auth_logs --auth &
         sleep 10
         ./scripts/upload_test_files_to_azurite.sh
 
@@ -121,4 +125,11 @@ jobs:
       if: always() && matrix.arch == 'linux_amd64_gcc4'
       shell: bash
       run: |
-        cat azurite_log.txt
+        echo "## azurite"
+        cat azurite_log.txt
+
+        echo "## squid"
+        cat squid_logs/*
+
+        echo "## squid auth"
+        cat squid_auth_logs/*

.github/workflows/MacOS.yml

@@ -25,6 +25,7 @@ jobs:
       GEN: Ninja
       VCPKG_TOOLCHAIN_PATH: ${{ github.workspace }}/vcpkg/scripts/buildsystems/vcpkg.cmake
       AZURE_STORAGE_CONNECTION_STRING: 'DefaultEndpointsProtocol=http;AccountName=devstoreaccount1;AccountKey=Eby8vdM02xNOcqFlqUwJPLlmEtlCDXJ1OUzFT50uSRZ6IFsuFq2UVErCz4I6tq/K1SZFPTOtr/KBHBeksoGMGw==;BlobEndpoint=http://127.0.0.1:10000/devstoreaccount1;QueueEndpoint=http://127.0.0.1:10001/devstoreaccount1;TableEndpoint=http://127.0.0.1:10002/devstoreaccount1;'
+      HTTP_PROXY_RUNNING: '1'
 
     steps:
       - uses: actions/checkout@v3
@@ -59,7 +60,10 @@ jobs:
       - name: Launch & populate Azure test service
         if: ${{ matrix.osx_build_arch == 'x86_64'}}
         run: |
+          brew install squid
           npm install -g azurite
+          ./scripts/run_squid.sh --port 3128 --log_dir squid_logs &
+          ./scripts/run_squid.sh --port 3129 --log_dir squid_auth_logs --auth &
           azurite > azurite_log.txt 2>&1 &
           sleep 10
           ./scripts/upload_test_files_to_azurite.sh
@@ -74,4 +78,11 @@ jobs:
         if: always() && matrix.osx_build_arch == 'x86_64'
         shell: bash
         run: |
-          cat azurite_log.txt
+          echo "## azurite"
+          cat azurite_log.txt
+          
+          echo "## squid"
+          cat squid_logs/*
+
+          echo "## squid with auth"
+          cat squid_auth_logs/*

.gitignore

@@ -1,8 +1,22 @@
 build
 .idea
+.vscode
 cmake-build-debug
 duckdb_unittest_tempdir/
 .DS_Store
 testext
 test/python/__pycache__/
 .Rhistory
+
+# Azurite
+__blobstorage__/
+__queuestorage__/
+__azurite_db_*__.json
+
+# squid
+squid_logs
+squid_users
+squid.pid
+squid.pid
+squid.conf
+squid_auth.conf

scripts/run_squid.sh

@@ -0,0 +1,84 @@
+#!/bin/bash
+
+help() {
+    echo "Usage: ${0} [port] [auth]"
+    echo "  port    Port number for squid to lisen to (by default 3128)"
+    echo "  auth    Optional string ('auth') to force user basic authentification (autherwise no authentification is required)"
+    exit 0
+}
+
+port='3128'
+auth='false'
+log_dir="squid_logs"
+conf_file="squid.conf"
+pid_file='${service_name}.pid'
+
+while [[ $# -gt 0 ]]; do
+  case "${1}" in
+    -h|--help)
+      help
+      ;;
+    -p|--port)
+      port="${2}"
+      shift # past argument
+      shift # past value
+      ;;
+    --auth)
+      auth='true'
+      conf_file="squid_auth.conf"
+      pid_file='${service_name}_auth.pid'
+      shift # past argument
+      ;;
+    --log_dir)
+      log_dir="${2}"
+      shift # past argument
+      shift # past value
+      ;;
+    *)
+      echo "Unknown option ${1}"
+      exit 1
+      ;;
+  esac
+done
+
+mkdir "${log_dir}"
+touch "${log_dir}/daemon.log"
+chmod -R 777 "${log_dir}"
+
+echo "http_port 127.0.0.1:${port}"                  >"${conf_file}"
+echo "pid_filename ${pid_file}"                    >>"${conf_file}"
+
+echo 'logfile_rotate 0'                            >>"${conf_file}"
+echo "logfile_daemon ${log_dir}/daemon.log"        >>"${conf_file}"
+echo "access_log ${log_dir}/access.log"            >>"${conf_file}"
+echo "cache_log ${log_dir}/cache.log"              >>"${conf_file}"
+echo "cache_store_log ${log_dir}/cache_store.log"  >>"${conf_file}"
+
+
+if [[ "${auth}" == "true" ]]; then
+    # User 'john' with password 'doe'
+    echo 'john:$apr1$dalj9e7s$AhqY28Hvl3EcNblNJMiXa0' >squid_users
+
+    squid_version="$(squid -v | head -n1 | grep -o 'Version [^ ]*' | cut -d ' ' -f 2)"
+    if [[ "$(uname)" == "Darwin" ]]; then
+        auth_basic_program="/usr/local/Cellar/squid/${squid_version}/libexec/basic_ncsa_auth"
+    else
+        if [[ -e '/usr/lib64/squid/basic_ncsa_auth' ]]; then
+            auth_basic_program="/usr/lib64/squid/basic_ncsa_auth"
+        else
+            auth_basic_program="/usr/lib/squid/basic_ncsa_auth"
+        fi
+    fi
+
+    echo '# Add authentification options'       >>"${conf_file}"
+    echo "auth_param basic program ${auth_basic_program} squid_users" >>"${conf_file}"
+    echo 'auth_param basic children 3'          >>"${conf_file}"
+    echo 'auth_param basic realm Squid BA'      >>"${conf_file}"
+    echo 'acl auth_users proxy_auth REQUIRED'   >>"${conf_file}"
+    echo 'http_access allow auth_users'         >>"${conf_file}"
+    echo 'http_access deny all'                 >>"${conf_file}"
+else
+    echo 'http_access allow localhost'          >>"${conf_file}"
+fi
+
+exec squid -N -f "${conf_file}"

src/azure_extension.cpp

@@ -25,6 +25,7 @@
 #include <azure/storage/blobs.hpp>
 #include <duckdb/parser/parsed_data/create_scalar_function_info.hpp>
 #include <iostream>
+#include <cstdlib>
 
 namespace duckdb {
 
@@ -56,19 +57,22 @@ static void Log(Logger::Level level, std::string const &message) {
 }
 
 static Azure::Identity::ChainedTokenCredential::Sources
-CreateCredentialChainFromSetting(const string &credential_chain) {
+CreateCredentialChainFromSetting(const string &credential_chain,
+                                 const Azure::Core::Http::Policies::TransportOptions &transport_options) {
 	auto chain_list = StringUtil::Split(credential_chain, ';');
 	Azure::Identity::ChainedTokenCredential::Sources result;
 
+	Azure::Core::Credentials::TokenCredentialOptions options;
+	options.Transport = transport_options;
 	for (const auto &item : chain_list) {
 		if (item == "cli") {
-			result.push_back(std::make_shared<Azure::Identity::AzureCliCredential>());
+			result.push_back(std::make_shared<Azure::Identity::AzureCliCredential>(options));
 		} else if (item == "managed_identity") {
-			result.push_back(std::make_shared<Azure::Identity::ManagedIdentityCredential>());
+			result.push_back(std::make_shared<Azure::Identity::ManagedIdentityCredential>(options));
 		} else if (item == "env") {
-			result.push_back(std::make_shared<Azure::Identity::EnvironmentCredential>());
+			result.push_back(std::make_shared<Azure::Identity::EnvironmentCredential>(options));
 		} else if (item == "default") {
-			result.push_back(std::make_shared<Azure::Identity::DefaultAzureCredential>());
+			result.push_back(std::make_shared<Azure::Identity::DefaultAzureCredential>(options));
 		} else if (item != "none") {
 			throw InvalidInputException("Unknown credential provider found: " + item);
 		}
@@ -114,6 +118,22 @@ static AzureAuthentication ParseAzureAuthSettings(FileOpener *opener, const stri
 		}
 	}
 
+	// Load proxy options
+	Value http_proxy;
+	if (FileOpener::TryGetCurrentSetting(opener, "azure_http_proxy", http_proxy)) {
+		auth.proxy_options.http_proxy = http_proxy.ToString();
+	}
+
+	Value http_proxy_user_name;
+	if (FileOpener::TryGetCurrentSetting(opener, "azure_proxy_user_name", http_proxy_user_name)) {
+		auth.proxy_options.user_name = http_proxy_user_name.ToString();
+	}
+
+	Value http_proxy_password;
+	if (FileOpener::TryGetCurrentSetting(opener, "azure_proxy_password", http_proxy_password)) {
+		auth.proxy_options.password = http_proxy_password.ToString();
+	}
+
 	return auth;
 }
 
@@ -138,20 +158,65 @@ static AzureReadOptions ParseAzureReadOptions(FileOpener *opener) {
 	return options;
 }
 
+static Azure::Core::Http::Policies::TransportOptions GetTransportOptions(AzureAuthentication &auth) {
+	Azure::Core::Http::Policies::TransportOptions options;
+	if (auth.secret) {
+		auto http_proxy = auth.secret->TryGetValue("http_proxy");
+		if (!http_proxy.IsNull()) {
+			options.HttpProxy = http_proxy.ToString();
+		} else {
+			// Keep honoring the env variable if present
+			auto *http_proxy_env = std::getenv("HTTP_PROXY");
+			if (http_proxy_env != nullptr) {
+				options.HttpProxy = http_proxy_env;
+			}
+		}
+
+		auto http_proxy_user_name = auth.secret->TryGetValue("proxy_user_name");
+		if (!http_proxy_user_name.IsNull()) {
+			options.ProxyUserName = http_proxy_user_name.ToString();
+		}
+
+		auto http_proxypassword = auth.secret->TryGetValue("proxy_password");
+		if (!http_proxypassword.IsNull()) {
+			options.ProxyPassword = http_proxypassword.ToString();
+		}
+	} else {
+		const auto &proxy_options = auth.proxy_options;
+		if (!proxy_options.http_proxy.empty()) {
+			options.HttpProxy = proxy_options.http_proxy;
+		}
+
+		if (!proxy_options.user_name.empty()) {
+			options.ProxyUserName = proxy_options.user_name;
+		}
+
+		if (!proxy_options.password.empty()) {
+			options.ProxyPassword = proxy_options.password;
+		}
+	}
+
+	return options;
+}
+
 static Azure::Storage::Blobs::BlobContainerClient GetContainerClient(AzureAuthentication &auth, AzureParsedUrl &url) {
 	string connection_string;
 	bool use_secret = false;
 	string chain;
 	string account_name;
 	string endpoint;
 
+	auto transport_options = GetTransportOptions(auth);
+	Azure::Storage::Blobs::BlobClientOptions options;
+	options.Transport = transport_options;
+
 	// Firstly, try to use the auth from the secret
 	if (auth.secret) {
 		// If connection string, we're done heres
 		auto connection_string_value = auth.secret->TryGetValue("connection_string");
 		if (!connection_string_value.IsNull()) {
 			return Azure::Storage::Blobs::BlobContainerClient::CreateFromConnectionString(
-			    connection_string_value.ToString(), url.container);
+			    connection_string_value.ToString(), url.container, options);
 		}
 
 		// Account_name can be used both for unauthenticated
@@ -181,7 +246,7 @@ static Azure::Storage::Blobs::BlobContainerClient GetContainerClient(AzureAuthen
 
 		if (!auth.connection_string.empty()) {
 			return Azure::Storage::Blobs::BlobContainerClient::CreateFromConnectionString(auth.connection_string,
-			                                                                              url.container);
+			                                                                              url.container, options);
 		}
 	}
 
@@ -192,17 +257,17 @@ static Azure::Storage::Blobs::BlobContainerClient GetContainerClient(AzureAuthen
 	// Build credential chain, from last to first
 	Azure::Identity::ChainedTokenCredential::Sources credential_chain;
 	if (!chain.empty()) {
-		credential_chain = CreateCredentialChainFromSetting(chain);
+		credential_chain = CreateCredentialChainFromSetting(chain, transport_options);
 	}
 
 	auto accountURL = "https://" + account_name + "." + endpoint;
 	if (!credential_chain.empty()) {
 		// A set of credentials providers was passed
 		auto chainedTokenCredential = std::make_shared<Azure::Identity::ChainedTokenCredential>(credential_chain);
-		Azure::Storage::Blobs::BlobServiceClient blob_service_client(accountURL, chainedTokenCredential);
+		Azure::Storage::Blobs::BlobServiceClient blob_service_client(accountURL, chainedTokenCredential, options);
 		return blob_service_client.GetBlobContainerClient(url.container);
 	} else if (!account_name.empty()) {
-		return Azure::Storage::Blobs::BlobContainerClient(accountURL + "/" + url.container);
+		return Azure::Storage::Blobs::BlobContainerClient(accountURL + "/" + url.container, options);
 	} else {
 		throw InvalidInputException("No valid Azure credentials found!");
 	}
@@ -361,6 +426,17 @@ static void LoadInternal(DatabaseInstance &instance) {
 	                          "Size of the read buffer.  It is recommended that this is evenly divisible by "
 	                          "azure_read_transfer_chunk_size.",
 	                          LogicalType::UBIGINT, Value::UBIGINT(default_read_options.buffer_size));
+
+	auto *http_proxy = std::getenv("HTTP_PROXY");
+	Value default_http_value = http_proxy ? Value(http_proxy) : Value(nullptr);
+	config.AddExtensionOption("azure_http_proxy",
+	                          "Proxy to use when login & performing request to azure. "
+	                          "By default it will use the HTTP_PROXY environment variable if set.",
+	                          LogicalType::VARCHAR, default_http_value);
+	config.AddExtensionOption("azure_proxy_user_name", "Http proxy user name if needed.", LogicalType::VARCHAR,
+	                          Value(nullptr));
+	config.AddExtensionOption("azure_proxy_password", "Http proxy password if needed.", LogicalType::VARCHAR,
+	                          Value(nullptr));
 }
 
 int64_t AzureStorageFileSystem::Read(FileHandle &handle, void *buffer, int64_t nr_bytes) {