Coverage Report

Created: 2026-09-14 18:56

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
/root/doris/be/src/util/s3_util.cpp
Line
Count
Source
1
// Licensed to the Apache Software Foundation (ASF) under one
2
// or more contributor license agreements.  See the NOTICE file
3
// distributed with this work for additional information
4
// regarding copyright ownership.  The ASF licenses this file
5
// to you under the Apache License, Version 2.0 (the
6
// "License"); you may not use this file except in compliance
7
// with the License.  You may obtain a copy of the License at
8
//
9
//   http://www.apache.org/licenses/LICENSE-2.0
10
//
11
// Unless required by applicable law or agreed to in writing,
12
// software distributed under the License is distributed on an
13
// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
14
// KIND, either express or implied.  See the License for the
15
// specific language governing permissions and limitations
16
// under the License.
17
18
#include "util/s3_util.h"
19
20
#include <aws/core/auth/AWSAuthSigner.h>
21
#include <aws/core/auth/AWSCredentials.h>
22
#include <aws/core/auth/AWSCredentialsProviderChain.h>
23
#include <aws/core/auth/STSCredentialsProvider.h>
24
#include <aws/core/client/DefaultRetryStrategy.h>
25
#include <aws/core/utils/logging/LogLevel.h>
26
#include <aws/core/utils/logging/LogSystemInterface.h>
27
#include <aws/core/utils/memory/stl/AWSStringStream.h>
28
#include <aws/identity-management/auth/STSAssumeRoleCredentialsProvider.h>
29
#include <aws/s3/S3Client.h>
30
#include <aws/sts/STSClient.h>
31
#include <bvar/reducer.h>
32
#include <cpp/token_bucket_rate_limiter.h>
33
34
#include <atomic>
35
36
#include "util/string_util.h"
37
38
#ifdef USE_AZURE
39
#include <azure/core/diagnostics/logger.hpp>
40
#include <azure/core/http/curl_transport.hpp>
41
#include <azure/storage/blobs/blob_container_client.hpp>
42
#endif
43
#include <cstdlib>
44
#include <filesystem>
45
#include <fstream>
46
#include <functional>
47
#include <memory>
48
#include <ostream>
49
#include <utility>
50
51
#include "common/config.h"
52
#include "common/logging.h"
53
#include "common/status.h"
54
#include "cpp/aws_logger.h"
55
#include "cpp/custom_aws_credentials_provider_chain.h"
56
#include "cpp/obj_retry_strategy.h"
57
#include "cpp/sync_point.h"
58
#include "cpp/util.h"
59
#ifdef USE_AZURE
60
#include "io/fs/azure_obj_storage_client.h"
61
#endif
62
#include "cloud/config.h"
63
#include "exec/scan/scanner_scheduler.h"
64
#include "io/fs/obj_storage_client.h"
65
#include "io/fs/rate_limited_obj_storage_client.h"
66
#include "io/fs/s3_obj_storage_client.h"
67
#include "runtime/exec_env.h"
68
#include "util/s3_uri.h"
69
70
namespace doris {
71
namespace s3_bvar {
72
bvar::LatencyRecorder s3_get_latency("s3_get");
73
bvar::LatencyRecorder s3_put_latency("s3_put");
74
bvar::LatencyRecorder s3_delete_object_latency("s3_delete_object");
75
bvar::LatencyRecorder s3_delete_objects_latency("s3_delete_objects");
76
bvar::LatencyRecorder s3_head_latency("s3_head");
77
bvar::LatencyRecorder s3_multi_part_upload_latency("s3_multi_part_upload");
78
bvar::LatencyRecorder s3_list_latency("s3_list");
79
bvar::LatencyRecorder s3_list_object_versions_latency("s3_list_object_versions");
80
bvar::LatencyRecorder s3_get_bucket_version_latency("s3_get_bucket_version");
81
bvar::LatencyRecorder s3_copy_object_latency("s3_copy_object");
82
}; // namespace s3_bvar
83
84
namespace {
85
86
35
doris::Status is_s3_conf_valid(const S3ClientConf& conf) {
87
35
    if (conf.endpoint.empty()) {
88
0
        return Status::InvalidArgument<false>("Invalid s3 conf, empty endpoint");
89
0
    }
90
35
    if (conf.region.empty()) {
91
0
        return Status::InvalidArgument<false>("Invalid s3 conf, empty region");
92
0
    }
93
94
35
    if (conf.role_arn.empty()) {
95
        // Allow anonymous access when both ak and sk are empty
96
29
        bool hasAk = !conf.ak.empty();
97
29
        bool hasSk = !conf.sk.empty();
98
99
        // Either both credentials are provided or both are empty (anonymous access)
100
29
        if (hasAk && conf.sk.empty()) {
101
1
            return Status::InvalidArgument<false>("Invalid s3 conf, empty sk");
102
1
        }
103
28
        if (hasSk && conf.ak.empty()) {
104
1
            return Status::InvalidArgument<false>("Invalid s3 conf, empty ak");
105
1
        }
106
28
    }
107
33
    return Status::OK();
108
35
}
109
110
// Return true is convert `str` to int successfully
111
0
bool to_int(std::string_view str, int& res) {
112
0
    auto [_, ec] = std::from_chars(str.data(), str.data() + str.size(), res);
113
0
    return ec == std::errc {};
114
0
}
115
116
#ifdef USE_AZURE
117
0
std::string env_or_empty(const char* env_name) {
118
0
    if (const char* value = std::getenv(env_name); value != nullptr) {
119
0
        return value;
120
0
    }
121
0
    return "";
122
0
}
123
124
0
std::string build_azure_tls_debug_context(const std::string& selected_ca_file) {
125
0
    bool selected_ca_exists = false;
126
0
    bool selected_ca_readable = false;
127
0
    if (!selected_ca_file.empty()) {
128
0
        std::error_code ec;
129
0
        selected_ca_exists = std::filesystem::exists(selected_ca_file, ec) && !ec;
130
0
        std::ifstream input(selected_ca_file);
131
0
        selected_ca_readable = input.good();
132
0
    }
133
134
0
    return fmt::format(
135
0
            "tls_debug(ca_cert_file_paths='{}', selected_ca_file='{}', selected_ca_exists={}, "
136
0
            "selected_ca_readable={}, SSL_CERT_FILE='{}', CURL_CA_BUNDLE='{}', SSL_CERT_DIR='{}')",
137
0
            config::ca_cert_file_paths, selected_ca_file, selected_ca_exists, selected_ca_readable,
138
0
            env_or_empty("SSL_CERT_FILE"), env_or_empty("CURL_CA_BUNDLE"),
139
0
            env_or_empty("SSL_CERT_DIR"));
140
0
}
141
#endif
142
143
constexpr char USE_PATH_STYLE[] = "use_path_style";
144
145
constexpr char AZURE_PROVIDER_STRING[] = "AZURE";
146
constexpr char S3_PROVIDER[] = "provider";
147
constexpr char S3_AK[] = "AWS_ACCESS_KEY";
148
constexpr char S3_SK[] = "AWS_SECRET_KEY";
149
constexpr char S3_ENDPOINT[] = "AWS_ENDPOINT";
150
constexpr char S3_REGION[] = "AWS_REGION";
151
constexpr char S3_TOKEN[] = "AWS_TOKEN";
152
constexpr char S3_MAX_CONN_SIZE[] = "AWS_MAX_CONNECTIONS";
153
constexpr char S3_REQUEST_TIMEOUT_MS[] = "AWS_REQUEST_TIMEOUT_MS";
154
constexpr char S3_CONN_TIMEOUT_MS[] = "AWS_CONNECTION_TIMEOUT_MS";
155
constexpr char S3_NEED_OVERRIDE_ENDPOINT[] = "AWS_NEED_OVERRIDE_ENDPOINT";
156
157
constexpr char S3_ROLE_ARN[] = "AWS_ROLE_ARN";
158
constexpr char S3_EXTERNAL_ID[] = "AWS_EXTERNAL_ID";
159
constexpr char S3_CREDENTIALS_PROVIDER_TYPE[] = "AWS_CREDENTIALS_PROVIDER_TYPE";
160
} // namespace
161
162
1
S3ClientFactory::S3ClientFactory() {
163
1
    _aws_options = Aws::SDKOptions {};
164
1
    auto logLevel = static_cast<Aws::Utils::Logging::LogLevel>(config::aws_log_level);
165
1
    _aws_options.loggingOptions.logLevel = logLevel;
166
1
    _aws_options.loggingOptions.logger_create_fn = [logLevel] {
167
1
        return std::make_shared<DorisAWSLogger>(logLevel);
168
1
    };
169
1
    Aws::InitAPI(_aws_options);
170
1
    _ca_cert_file_path = get_valid_ca_cert_path(doris::split(config::ca_cert_file_paths, ";"));
171
172
1
#ifdef USE_AZURE
173
1
    auto azureLogLevel =
174
1
            static_cast<Azure::Core::Diagnostics::Logger::Level>(config::azure_log_level);
175
1
    Azure::Core::Diagnostics::Logger::SetLevel(azureLogLevel);
176
1
    Azure::Core::Diagnostics::Logger::SetListener(
177
1
            [&](Azure::Core::Diagnostics::Logger::Level level, const std::string& message) {
178
0
                switch (level) {
179
0
                case Azure::Core::Diagnostics::Logger::Level::Verbose:
180
0
                    LOG(INFO) << message;
181
0
                    break;
182
0
                case Azure::Core::Diagnostics::Logger::Level::Informational:
183
0
                    LOG(INFO) << message;
184
0
                    break;
185
0
                case Azure::Core::Diagnostics::Logger::Level::Warning:
186
0
                    LOG(WARNING) << message;
187
0
                    break;
188
0
                case Azure::Core::Diagnostics::Logger::Level::Error:
189
0
                    LOG(ERROR) << message;
190
0
                    break;
191
0
                default:
192
0
                    LOG(WARNING) << "Unknown level: " << static_cast<int>(level)
193
0
                                 << ", message: " << message;
194
0
                    break;
195
0
                }
196
0
            });
197
1
#endif
198
1
}
199
200
1
S3ClientFactory::~S3ClientFactory() {
201
1
    Aws::ShutdownAPI(_aws_options);
202
1
}
203
204
50
S3ClientFactory& S3ClientFactory::instance() {
205
50
    static S3ClientFactory ret;
206
50
    return ret;
207
50
}
208
209
20
std::shared_ptr<io::ObjStorageClient> S3ClientFactory::create(const S3ClientConf& s3_conf) {
210
20
    if (!is_s3_conf_valid(s3_conf).ok()) {
211
0
        return nullptr;
212
0
    }
213
214
20
#ifdef BE_TEST
215
20
    {
216
20
        std::lock_guard l(_lock);
217
20
        if (_test_client_creator) {
218
4
            return _test_client_creator(s3_conf);
219
4
        }
220
20
    }
221
16
#endif
222
223
16
    {
224
16
        std::lock_guard l(_lock);
225
16
        auto it = _cache.find(s3_conf);
226
16
        if (it != _cache.end()) {
227
9
            return it->second;
228
9
        }
229
16
    }
230
231
7
    auto obj_client = (s3_conf.provider == io::ObjStorageType::AZURE)
232
7
                              ? _create_azure_client(s3_conf)
233
7
                              : _create_s3_client(s3_conf);
234
235
    // Rate limiting lives in one decorator, decided here at construction time:
236
    // in cloud mode only internal storage-vault buckets are limited; external buckets
237
    // (S3 load, TVF, external catalogs) get the bare client. In non-cloud mode every
238
    // client is wrapped, preserving the legacy behavior.
239
7
    if (obj_client != nullptr && (!config::is_cloud_mode() || s3_conf.is_internal_bucket)) {
240
6
        obj_client = std::make_shared<io::RateLimitedObjStorageClient>(std::move(obj_client));
241
6
    }
242
243
7
    {
244
7
        std::lock_guard l(_lock);
245
7
        auto [it, _] = _cache.emplace(s3_conf, std::move(obj_client));
246
7
        return it->second;
247
16
    }
248
16
}
249
250
#ifdef BE_TEST
251
void S3ClientFactory::set_client_creator_for_test(
252
3
        std::function<std::shared_ptr<io::ObjStorageClient>(const S3ClientConf&)> creator) {
253
3
    std::lock_guard l(_lock);
254
3
    _test_client_creator = std::move(creator);
255
3
}
256
257
18
void S3ClientFactory::clear_client_creator_for_test() {
258
18
    std::lock_guard l(_lock);
259
18
    _test_client_creator = nullptr;
260
18
}
261
#endif
262
263
std::shared_ptr<io::ObjStorageClient> S3ClientFactory::_create_azure_client(
264
0
        const S3ClientConf& s3_conf) {
265
0
#ifdef USE_AZURE
266
0
    auto cred =
267
0
            std::make_shared<Azure::Storage::StorageSharedKeyCredential>(s3_conf.ak, s3_conf.sk);
268
269
0
    const std::string container_name = s3_conf.bucket;
270
0
    std::string uri = fmt::format("{}/{}", s3_conf.endpoint, container_name);
271
0
    if (s3_conf.endpoint.find("://") == std::string::npos) {
272
0
        uri = "https://" + uri;
273
0
    }
274
275
0
    Azure::Storage::Blobs::BlobClientOptions options;
276
0
    options.Retry.StatusCodes.insert(Azure::Core::Http::HttpStatusCode::TooManyRequests);
277
0
    options.Retry.MaxRetries = config::max_s3_client_retry;
278
0
    options.PerRetryPolicies.emplace_back(std::make_unique<AzureRetryRecordPolicy>());
279
0
    if (_ca_cert_file_path.empty()) {
280
0
        _ca_cert_file_path = get_valid_ca_cert_path(doris::split(config::ca_cert_file_paths, ";"));
281
0
    }
282
0
    if (!_ca_cert_file_path.empty()) {
283
0
        Azure::Core::Http::CurlTransportOptions curl_options;
284
0
        curl_options.CAInfo = _ca_cert_file_path;
285
0
        options.Transport.Transport =
286
0
                std::make_shared<Azure::Core::Http::CurlTransport>(std::move(curl_options));
287
0
    }
288
289
0
    std::string normalized_uri = normalize_http_uri(uri);
290
0
    VLOG_DEBUG << "uri:" << uri << ", normalized_uri:" << normalized_uri;
291
0
    std::string tls_debug_context = build_azure_tls_debug_context(_ca_cert_file_path);
292
293
0
    auto containerClient = std::make_shared<Azure::Storage::Blobs::BlobContainerClient>(
294
0
            uri, cred, std::move(options));
295
0
    LOG_INFO("create one azure client with {}", s3_conf.to_string());
296
0
    return std::make_shared<io::AzureObjStorageClient>(std::move(containerClient),
297
0
                                                       std::move(tls_debug_context));
298
#else
299
    LOG_FATAL("BE is not compiled with azure support, export BUILD_AZURE=ON before building");
300
    return nullptr;
301
#endif
302
0
}
303
304
std::shared_ptr<Aws::Auth::AWSCredentialsProvider>
305
6
S3ClientFactory::_get_aws_credentials_provider_v1(const S3ClientConf& s3_conf) {
306
6
    if (!s3_conf.ak.empty() && !s3_conf.sk.empty()) {
307
2
        Aws::Auth::AWSCredentials aws_cred(s3_conf.ak, s3_conf.sk);
308
2
        DCHECK(!aws_cred.IsExpiredOrEmpty());
309
2
        if (!s3_conf.token.empty()) {
310
0
            aws_cred.SetSessionToken(s3_conf.token);
311
0
        }
312
2
        return std::make_shared<Aws::Auth::SimpleAWSCredentialsProvider>(std::move(aws_cred));
313
2
    }
314
315
4
    if (s3_conf.cred_provider_type == CredProviderType::InstanceProfile) {
316
2
        if (s3_conf.role_arn.empty()) {
317
1
            return std::make_shared<Aws::Auth::InstanceProfileCredentialsProvider>();
318
1
        }
319
320
1
        Aws::Client::ClientConfiguration clientConfiguration =
321
1
                S3ClientFactory::getClientConfiguration();
322
323
1
        if (_ca_cert_file_path.empty()) {
324
0
            _ca_cert_file_path =
325
0
                    get_valid_ca_cert_path(doris::split(config::ca_cert_file_paths, ";"));
326
0
        }
327
1
        if (!_ca_cert_file_path.empty()) {
328
1
            clientConfiguration.caFile = _ca_cert_file_path;
329
1
        }
330
331
1
        auto stsClient = std::make_shared<Aws::STS::STSClient>(
332
1
                std::make_shared<Aws::Auth::InstanceProfileCredentialsProvider>(),
333
1
                clientConfiguration);
334
335
1
        return std::make_shared<Aws::Auth::STSAssumeRoleCredentialsProvider>(
336
1
                s3_conf.role_arn, Aws::String(), s3_conf.external_id,
337
1
                Aws::Auth::DEFAULT_CREDS_LOAD_FREQ_SECONDS, stsClient);
338
2
    }
339
340
    // Support anonymous access for public datasets when no credentials are provided
341
2
    if (s3_conf.ak.empty() && s3_conf.sk.empty()) {
342
2
        return std::make_shared<Aws::Auth::AnonymousAWSCredentialsProvider>();
343
2
    }
344
345
0
    return std::make_shared<Aws::Auth::DefaultAWSCredentialsProviderChain>();
346
2
}
347
348
std::shared_ptr<Aws::Auth::AWSCredentialsProvider> S3ClientFactory::_create_credentials_provider(
349
26
        CredProviderType type) {
350
26
    switch (type) {
351
2
    case CredProviderType::Env:
352
2
        return std::make_shared<Aws::Auth::EnvironmentAWSCredentialsProvider>();
353
2
    case CredProviderType::SystemProperties:
354
2
        return std::make_shared<Aws::Auth::ProfileConfigFileAWSCredentialsProvider>();
355
3
    case CredProviderType::WebIdentity:
356
3
        return std::make_shared<Aws::Auth::STSAssumeRoleWebIdentityCredentialsProvider>();
357
6
    case CredProviderType::Container:
358
6
        return create_container_credentials_provider();
359
4
    case CredProviderType::InstanceProfile:
360
4
        return std::make_shared<Aws::Auth::InstanceProfileCredentialsProvider>();
361
6
    case CredProviderType::Anonymous:
362
6
        return std::make_shared<Aws::Auth::AnonymousAWSCredentialsProvider>();
363
3
    case CredProviderType::Default:
364
3
    default:
365
3
        return std::make_shared<CustomAwsCredentialsProviderChain>();
366
26
    }
367
26
}
368
369
std::shared_ptr<Aws::Auth::AWSCredentialsProvider>
370
30
S3ClientFactory::_get_aws_credentials_provider_v2(const S3ClientConf& s3_conf) {
371
30
    if (!s3_conf.ak.empty() && !s3_conf.sk.empty()) {
372
4
        Aws::Auth::AWSCredentials aws_cred(s3_conf.ak, s3_conf.sk);
373
4
        DCHECK(!aws_cred.IsExpiredOrEmpty());
374
4
        if (!s3_conf.token.empty()) {
375
0
            aws_cred.SetSessionToken(s3_conf.token);
376
0
        }
377
4
        return std::make_shared<Aws::Auth::SimpleAWSCredentialsProvider>(std::move(aws_cred));
378
4
    }
379
380
    // Handle role_arn for assume role scenario
381
26
    if (!s3_conf.role_arn.empty()) {
382
8
        Aws::Client::ClientConfiguration clientConfiguration =
383
8
                S3ClientFactory::getClientConfiguration();
384
385
8
        if (_ca_cert_file_path.empty()) {
386
0
            _ca_cert_file_path =
387
0
                    get_valid_ca_cert_path(doris::split(config::ca_cert_file_paths, ";"));
388
0
        }
389
8
        if (!_ca_cert_file_path.empty()) {
390
8
            clientConfiguration.caFile = _ca_cert_file_path;
391
8
        }
392
393
8
        auto baseProvider = _create_credentials_provider(s3_conf.cred_provider_type);
394
8
        auto stsClient = std::make_shared<Aws::STS::STSClient>(baseProvider, clientConfiguration);
395
396
8
        return std::make_shared<Aws::Auth::STSAssumeRoleCredentialsProvider>(
397
8
                s3_conf.role_arn, Aws::String(), s3_conf.external_id,
398
8
                Aws::Auth::DEFAULT_CREDS_LOAD_FREQ_SECONDS, stsClient);
399
8
    }
400
401
    // Return provider based on cred_provider_type
402
18
    return _create_credentials_provider(s3_conf.cred_provider_type);
403
26
}
404
405
std::shared_ptr<Aws::Auth::AWSCredentialsProvider> S3ClientFactory::get_aws_credentials_provider(
406
36
        const S3ClientConf& s3_conf) {
407
36
    if (config::aws_credentials_provider_version == "v2") {
408
30
        return _get_aws_credentials_provider_v2(s3_conf);
409
30
    }
410
6
    return _get_aws_credentials_provider_v1(s3_conf);
411
36
}
412
413
std::shared_ptr<io::ObjStorageClient> S3ClientFactory::_create_s3_client(
414
7
        const S3ClientConf& s3_conf) {
415
7
    TEST_SYNC_POINT_RETURN_WITH_VALUE(
416
6
            "s3_client_factory::create",
417
6
            std::make_shared<io::S3ObjStorageClient>(std::make_shared<Aws::S3::S3Client>()));
418
6
    Aws::Client::ClientConfiguration aws_config = S3ClientFactory::getClientConfiguration();
419
6
    if (s3_conf.need_override_endpoint) {
420
6
        aws_config.endpointOverride = s3_conf.endpoint;
421
6
    }
422
6
    aws_config.region = s3_conf.region;
423
424
6
    if (_ca_cert_file_path.empty()) {
425
0
        _ca_cert_file_path = get_valid_ca_cert_path(doris::split(config::ca_cert_file_paths, ";"));
426
0
    }
427
428
6
    if (!_ca_cert_file_path.empty()) {
429
6
        aws_config.caFile = _ca_cert_file_path;
430
6
    }
431
432
6
    if (s3_conf.max_connections > 0) {
433
0
        aws_config.maxConnections = s3_conf.max_connections;
434
6
    } else {
435
6
        aws_config.maxConnections = 102400;
436
6
    }
437
438
6
    aws_config.requestTimeoutMs = 30000;
439
6
    if (s3_conf.request_timeout_ms > 0) {
440
0
        aws_config.requestTimeoutMs = s3_conf.request_timeout_ms;
441
0
    }
442
443
6
    if (s3_conf.connect_timeout_ms > 0) {
444
0
        aws_config.connectTimeoutMs = s3_conf.connect_timeout_ms;
445
0
    }
446
447
6
    if (config::s3_client_http_scheme == "http") {
448
6
        aws_config.scheme = Aws::Http::Scheme::HTTP;
449
6
    }
450
451
6
    aws_config.retryStrategy = std::make_shared<S3CustomRetryStrategy>(
452
6
            config::max_s3_client_retry /*scaleFactor = 25*/, /*retry_slow_down=*/true);
453
454
6
    std::shared_ptr<Aws::S3::S3Client> new_client = std::make_shared<Aws::S3::S3Client>(
455
6
            get_aws_credentials_provider(s3_conf), std::move(aws_config),
456
6
            Aws::Client::AWSAuthV4Signer::PayloadSigningPolicy::Never,
457
6
            s3_conf.use_virtual_addressing);
458
459
6
    auto obj_client = std::make_shared<io::S3ObjStorageClient>(std::move(new_client));
460
6
    LOG_INFO("create one s3 client with {}", s3_conf.to_string());
461
6
    return obj_client;
462
7
}
463
464
Status S3ClientFactory::convert_properties_to_s3_conf(
465
15
        const std::map<std::string, std::string>& prop, const S3URI& s3_uri, S3Conf* s3_conf) {
466
15
    StringCaseMap<std::string> properties(prop.begin(), prop.end());
467
15
    if (auto it = properties.find(S3_AK); it != properties.end()) {
468
2
        s3_conf->client_conf.ak = it->second;
469
2
    }
470
15
    if (auto it = properties.find(S3_SK); it != properties.end()) {
471
2
        s3_conf->client_conf.sk = it->second;
472
2
    }
473
15
    if (auto it = properties.find(S3_TOKEN); it != properties.end()) {
474
0
        s3_conf->client_conf.token = it->second;
475
0
    }
476
15
    if (auto it = properties.find(S3_ENDPOINT); it != properties.end()) {
477
15
        s3_conf->client_conf.endpoint = it->second;
478
15
    }
479
15
    if (auto it = properties.find(S3_NEED_OVERRIDE_ENDPOINT); it != properties.end()) {
480
0
        s3_conf->client_conf.need_override_endpoint = (it->second == "true");
481
0
    }
482
15
    if (auto it = properties.find(S3_REGION); it != properties.end()) {
483
15
        s3_conf->client_conf.region = it->second;
484
15
    }
485
15
    if (auto it = properties.find(S3_MAX_CONN_SIZE); it != properties.end()) {
486
0
        if (!to_int(it->second, s3_conf->client_conf.max_connections)) {
487
0
            return Status::InvalidArgument("invalid {} value \"{}\"", S3_MAX_CONN_SIZE, it->second);
488
0
        }
489
0
    }
490
15
    if (auto it = properties.find(S3_REQUEST_TIMEOUT_MS); it != properties.end()) {
491
0
        if (!to_int(it->second, s3_conf->client_conf.request_timeout_ms)) {
492
0
            return Status::InvalidArgument("invalid {} value \"{}\"", S3_REQUEST_TIMEOUT_MS,
493
0
                                           it->second);
494
0
        }
495
0
    }
496
15
    if (auto it = properties.find(S3_CONN_TIMEOUT_MS); it != properties.end()) {
497
0
        if (!to_int(it->second, s3_conf->client_conf.connect_timeout_ms)) {
498
0
            return Status::InvalidArgument("invalid {} value \"{}\"", S3_CONN_TIMEOUT_MS,
499
0
                                           it->second);
500
0
        }
501
0
    }
502
15
    if (auto it = properties.find(S3_PROVIDER); it != properties.end()) {
503
        // S3 Provider properties should be case insensitive.
504
0
        if (0 == strcasecmp(it->second.c_str(), AZURE_PROVIDER_STRING)) {
505
0
            s3_conf->client_conf.provider = io::ObjStorageType::AZURE;
506
0
        }
507
0
    }
508
509
15
    if (s3_uri.get_bucket().empty()) {
510
0
        return Status::InvalidArgument("Invalid S3 URI {}, bucket is not specified",
511
0
                                       s3_uri.to_string());
512
0
    }
513
15
    s3_conf->bucket = s3_uri.get_bucket();
514
    // For azure's compatibility
515
15
    s3_conf->client_conf.bucket = s3_uri.get_bucket();
516
15
    s3_conf->prefix = "";
517
518
    // See https://sdk.amazonaws.com/cpp/api/LATEST/class_aws_1_1_s3_1_1_s3_client.html
519
15
    s3_conf->client_conf.use_virtual_addressing = true;
520
15
    if (auto it = properties.find(USE_PATH_STYLE); it != properties.end()) {
521
0
        s3_conf->client_conf.use_virtual_addressing = it->second != "true";
522
0
    }
523
524
15
    if (auto it = properties.find(S3_ROLE_ARN); it != properties.end()) {
525
        // Keep provider type as Default unless explicitly configured by
526
        // AWS_CREDENTIALS_PROVIDER_TYPE, consistent with FE behavior.
527
5
        s3_conf->client_conf.role_arn = it->second;
528
5
    }
529
530
15
    if (auto it = properties.find(S3_EXTERNAL_ID); it != properties.end()) {
531
0
        s3_conf->client_conf.external_id = it->second;
532
0
    }
533
534
15
    if (auto it = properties.find(S3_CREDENTIALS_PROVIDER_TYPE); it != properties.end()) {
535
8
        s3_conf->client_conf.cred_provider_type = cred_provider_type_from_string(it->second);
536
8
    }
537
538
15
    if (auto st = is_s3_conf_valid(s3_conf->client_conf); !st.ok()) {
539
2
        return st;
540
2
    }
541
13
    return Status::OK();
542
15
}
543
544
0
static CredProviderType cred_provider_type_from_thrift(TCredProviderType::type cred_provider_type) {
545
0
    switch (cred_provider_type) {
546
0
    case TCredProviderType::DEFAULT:
547
0
        return CredProviderType::Default;
548
0
    case TCredProviderType::SIMPLE:
549
0
        return CredProviderType::Simple;
550
0
    case TCredProviderType::INSTANCE_PROFILE:
551
0
        return CredProviderType::InstanceProfile;
552
0
    case TCredProviderType::ENV:
553
0
        return CredProviderType::Env;
554
0
    case TCredProviderType::SYSTEM_PROPERTIES:
555
0
        return CredProviderType::SystemProperties;
556
0
    case TCredProviderType::WEB_IDENTITY:
557
0
        return CredProviderType::WebIdentity;
558
0
    case TCredProviderType::CONTAINER:
559
0
        return CredProviderType::Container;
560
0
    case TCredProviderType::ANONYMOUS:
561
0
        return CredProviderType::Anonymous;
562
0
    default:
563
0
        __builtin_unreachable();
564
0
        LOG(WARNING) << "Invalid TCredProviderType value: " << cred_provider_type
565
0
                     << ", use default instead.";
566
0
        return CredProviderType::Default;
567
0
    }
568
0
}
569
570
0
S3Conf S3Conf::get_s3_conf(const cloud::ObjectStoreInfoPB& info) {
571
0
    S3Conf ret {
572
0
            .bucket = info.bucket(),
573
0
            .prefix = info.prefix(),
574
0
            .client_conf {
575
0
                    .endpoint = info.endpoint(),
576
0
                    .region = info.region(),
577
0
                    .ak = info.ak(),
578
0
                    .sk = info.sk(),
579
0
                    .token {},
580
0
                    .bucket = info.bucket(),
581
0
                    .provider = io::ObjStorageType::AWS,
582
0
                    .use_virtual_addressing =
583
0
                            info.has_use_path_style() ? !info.use_path_style() : true,
584
585
0
                    .role_arn = info.role_arn(),
586
0
                    .external_id = info.external_id(),
587
                    // ObjectStoreInfoPB always describes a storage vault, i.e. a Doris
588
                    // internal bucket in cloud mode.
589
0
                    .is_internal_bucket = true,
590
0
            },
591
0
            .sse_enabled = info.sse_enabled(),
592
0
    };
593
594
0
    if (info.has_cred_provider_type()) {
595
0
        ret.client_conf.cred_provider_type = cred_provider_type_from_pb(info.cred_provider_type());
596
0
    }
597
598
0
    io::ObjStorageType type = io::ObjStorageType::AWS;
599
0
    switch (info.provider()) {
600
0
    case cloud::ObjectStoreInfoPB_Provider_OSS:
601
0
        type = io::ObjStorageType::OSS;
602
0
        break;
603
0
    case cloud::ObjectStoreInfoPB_Provider_S3:
604
0
        type = io::ObjStorageType::AWS;
605
0
        break;
606
0
    case cloud::ObjectStoreInfoPB_Provider_COS:
607
0
        type = io::ObjStorageType::COS;
608
0
        break;
609
0
    case cloud::ObjectStoreInfoPB_Provider_OBS:
610
0
        type = io::ObjStorageType::OBS;
611
0
        break;
612
0
    case cloud::ObjectStoreInfoPB_Provider_BOS:
613
0
        type = io::ObjStorageType::BOS;
614
0
        break;
615
0
    case cloud::ObjectStoreInfoPB_Provider_GCP:
616
0
        type = io::ObjStorageType::GCP;
617
0
        break;
618
0
    case cloud::ObjectStoreInfoPB_Provider_AZURE:
619
0
        type = io::ObjStorageType::AZURE;
620
0
        break;
621
0
    case cloud::ObjectStoreInfoPB_Provider_TOS:
622
0
        type = io::ObjStorageType::TOS;
623
0
        break;
624
0
    default:
625
0
        __builtin_unreachable();
626
0
        LOG_FATAL("unknown provider type {}, info {}", info.provider(), ret.to_string());
627
0
    }
628
0
    ret.client_conf.provider = type;
629
0
    return ret;
630
0
}
631
632
0
S3Conf S3Conf::get_s3_conf(const TS3StorageParam& param) {
633
0
    S3Conf ret {
634
0
            .bucket = param.bucket,
635
0
            .prefix = param.root_path,
636
0
            .client_conf = {
637
0
                    .endpoint = param.endpoint,
638
0
                    .region = param.region,
639
0
                    .ak = param.ak,
640
0
                    .sk = param.sk,
641
0
                    .token = param.token,
642
0
                    .bucket = param.bucket,
643
0
                    .provider = io::ObjStorageType::AWS,
644
0
                    .max_connections = param.max_conn,
645
0
                    .request_timeout_ms = param.request_timeout_ms,
646
0
                    .connect_timeout_ms = param.conn_timeout_ms,
647
                    // When using cold heat separation in minio, user might use ip address directly,
648
                    // which needs enable use_virtual_addressing to true
649
0
                    .use_virtual_addressing = !param.use_path_style,
650
0
                    .role_arn = param.role_arn,
651
0
                    .external_id = param.external_id,
652
0
            }};
653
654
0
    if (param.__isset.cred_provider_type) {
655
0
        ret.client_conf.cred_provider_type =
656
0
                cred_provider_type_from_thrift(param.cred_provider_type);
657
0
    }
658
659
0
    io::ObjStorageType type = io::ObjStorageType::AWS;
660
0
    switch (param.provider) {
661
0
    case TObjStorageType::UNKNOWN:
662
0
        LOG_INFO("Receive one legal storage resource, set provider type to aws, param detail {}",
663
0
                 ret.to_string());
664
0
        type = io::ObjStorageType::AWS;
665
0
        break;
666
0
    case TObjStorageType::AWS:
667
0
        type = io::ObjStorageType::AWS;
668
0
        break;
669
0
    case TObjStorageType::AZURE:
670
0
        type = io::ObjStorageType::AZURE;
671
0
        break;
672
0
    case TObjStorageType::BOS:
673
0
        type = io::ObjStorageType::BOS;
674
0
        break;
675
0
    case TObjStorageType::COS:
676
0
        type = io::ObjStorageType::COS;
677
0
        break;
678
0
    case TObjStorageType::OBS:
679
0
        type = io::ObjStorageType::OBS;
680
0
        break;
681
0
    case TObjStorageType::OSS:
682
0
        type = io::ObjStorageType::OSS;
683
0
        break;
684
0
    case TObjStorageType::GCP:
685
0
        type = io::ObjStorageType::GCP;
686
0
        break;
687
0
    case TObjStorageType::TOS:
688
0
        type = io::ObjStorageType::TOS;
689
0
        break;
690
0
    default:
691
0
        LOG_FATAL("unknown provider type {}, info {}", param.provider, ret.to_string());
692
0
        __builtin_unreachable();
693
0
    }
694
0
    ret.client_conf.provider = type;
695
0
    return ret;
696
0
}
697
698
18
std::string hide_access_key(const std::string& ak) {
699
18
    std::string key = ak;
700
18
    size_t key_len = key.length();
701
18
    size_t reserved_count;
702
18
    if (key_len > 7) {
703
3
        reserved_count = 6;
704
15
    } else if (key_len > 2) {
705
6
        reserved_count = key_len - 2;
706
9
    } else {
707
9
        reserved_count = 0;
708
9
    }
709
710
18
    size_t x_count = key_len - reserved_count;
711
18
    size_t left_x_count = (x_count + 1) / 2;
712
713
18
    if (left_x_count > 0) {
714
12
        key.replace(0, left_x_count, left_x_count, 'x');
715
12
    }
716
717
18
    if (x_count - left_x_count > 0) {
718
11
        key.replace(key_len - (x_count - left_x_count), x_count - left_x_count,
719
11
                    x_count - left_x_count, 'x');
720
11
    }
721
18
    return key;
722
18
}
723
724
} // end namespace doris