feat(s3): Use multipart upload for chunked uploading

This allows to stream file chunks directly to S3 during upload.

Signed-off-by: Julius Härtl <jus@bitgrid.net>

3 files changed, 200 insertions, 6 deletions

diff --git a/build/integration/features/bootstrap/BasicStructure.php b/build/integration/features/bootstrap/BasicStructure.php
index 9060c85c756..e12a40ac6b4 100644
--- a/build/integration/features/bootstrap/BasicStructure.php
+++ b/build/integration/features/bootstrap/BasicStructure.php
@@ -179,7 +179,7 @@ trait BasicStructure {
 			$options['auth'] = [$this->currentUser, $this->regularUser];
 		}
 		$options['headers'] = [
-			'OCS_APIREQUEST' => 'true'
+			'OCS-APIRequest' => 'true'
 		];
 		if ($body instanceof TableNode) {
 			$fd = $body->getRowsHash();
@@ -306,7 +306,7 @@ trait BasicStructure {
 	 * @param string $user
 	 */
 	public function loggingInUsingWebAs($user) {
-		$loginUrl = substr($this->baseUrl, 0, -5) . '/login';
+		$loginUrl = substr($this->baseUrl, 0, -5) . '/index.php/login';
 		// Request a new session and extract CSRF token
 		$client = new Client();
 		$response = $client->get(
diff --git a/build/integration/features/bootstrap/WebDav.php b/build/integration/features/bootstrap/WebDav.php
index 680db01a260..00ba5c28862 100644
--- a/build/integration/features/bootstrap/WebDav.php
+++ b/build/integration/features/bootstrap/WebDav.php
@@ -54,6 +54,9 @@ trait WebDav {
 	/** @var int */
 	private $storedFileID = null;
 
+	private string $s3MultipartDestination;
+	private string $uploadId;
+
 	/**
 	 * @Given /^using dav path "([^"]*)"$/
 	 */
@@ -751,6 +754,7 @@ trait WebDav {
 	 * @Given user :user creates a new chunking upload with id :id
 	 */
 	public function userCreatesANewChunkingUploadWithId($user, $id) {
+		$this->parts = [];
 		$destination = '/uploads/' . $user . '/' . $id;
 		$this->makeDavRequest($user, 'MKCOL', $destination, [], null, "uploads");
 	}
@@ -792,6 +796,60 @@ trait WebDav {
 		}
 	}
 
+
+	/**
+	 * @Given user :user creates a new chunking v2 upload with id :id and destination :targetDestination
+	 */
+	public function userCreatesANewChunkingv2UploadWithIdAndDestination($user, $id, $targetDestination) {
+		$this->s3MultipartDestination = $this->getTargetDestination($user, $targetDestination);
+		$this->newUploadId();
+		$destination = '/uploads/' . $user . '/' . $this->getUploadId($id);
+		$this->response = $this->makeDavRequest($user, 'MKCOL', $destination, [
+			'Destination' => $this->s3MultipartDestination,
+		], null, "uploads");
+	}
+
+	/**
+	 * @Given user :user uploads new chunk v2 file :num to id :id
+	 */
+	public function userUploadsNewChunkv2FileToIdAndDestination($user, $num, $id) {
+		$data = \GuzzleHttp\Psr7\Utils::streamFor(fopen('/tmp/part-upload-' . $num, 'r'));
+		$destination = '/uploads/' . $user . '/' . $this->getUploadId($id) . '/' . $num;
+		$this->response = $this->makeDavRequest($user, 'PUT', $destination, [
+			'Destination' => $this->s3MultipartDestination
+		], $data, "uploads");
+	}
+
+	/**
+	 * @Given user :user moves new chunk v2 file with id :id
+	 */
+	public function userMovesNewChunkv2FileWithIdToMychunkedfileAndDestination($user, $id) {
+		$source = '/uploads/' . $user . '/' . $this->getUploadId($id) . '/.file';
+		try {
+			$this->response = $this->makeDavRequest($user, 'MOVE', $source, [
+				'Destination' => $this->s3MultipartDestination,
+			], null, "uploads");
+		} catch (\GuzzleHttp\Exception\ServerException $e) {
+			// 5xx responses cause a server exception
+			$this->response = $e->getResponse();
+		} catch (\GuzzleHttp\Exception\ClientException $e) {
+			// 4xx responses cause a client exception
+			$this->response = $e->getResponse();
+		}
+	}
+
+	private function getTargetDestination(string $user, string $destination): string {
+		return substr($this->baseUrl, 0, -4) . $this->getDavFilesPath($user) . $destination;
+	}
+
+	private function getUploadId(string $id): string {
+		return $id . '-' . $this->uploadId;
+	}
+
+	private function newUploadId() {
+		$this->uploadId = (string)time();
+	}
+
 	/**
 	 * @Given /^Downloading file "([^"]*)" as "([^"]*)"$/
 	 */
@@ -980,4 +1038,44 @@ trait WebDav {
 		$currentFileID = $this->getFileIdForPath($user, $path);
 		Assert::assertEquals($currentFileID, $this->storedFileID);
 	}
+
+	/**
+	 * @Given /^user "([^"]*)" creates a file locally with "([^"]*)" x 5 MB chunks$/
+	 */
+	public function userCreatesAFileLocallyWithChunks($arg1, $chunks) {
+		$this->parts = [];
+		for ($i = 1;$i <= (int)$chunks;$i++) {
+			$randomletter = substr(str_shuffle("abcdefghijklmnopqrstuvwxyz"), 0, 1);
+			file_put_contents('/tmp/part-upload-' . $i, str_repeat($randomletter, 5 * 1024 * 1024));
+			$this->parts[] = '/tmp/part-upload-' . $i;
+		}
+	}
+
+	/**
+	 * @Given user :user creates the chunk :id with a size of :size MB
+	 */
+	public function userCreatesAChunk($user, $id, $size) {
+		$randomletter = substr(str_shuffle("abcdefghijklmnopqrstuvwxyz"), 0, 1);
+		file_put_contents('/tmp/part-upload-' . $id, str_repeat($randomletter, (int)$size * 1024 * 1024));
+		$this->parts[] = '/tmp/part-upload-' . $id;
+	}
+
+	/**
+	 * @Then /^Downloaded content should be the created file$/
+	 */
+	public function downloadedContentShouldBeTheCreatedFile() {
+		$content = '';
+		sort($this->parts);
+		foreach ($this->parts as $part) {
+			$content .= file_get_contents($part);
+		}
+		Assert::assertEquals($content, (string)$this->response->getBody());
+	}
+
+	/**
+	 * @Then /^the S3 multipart upload was successful with status "([^"]*)"$/
+	 */
+	public function theSmultipartUploadWasSuccessful($status) {
+		Assert::assertEquals((int)$status, $this->response->getStatusCode());
+	}
 }
diff --git a/build/integration/features/webdav-related.feature b/build/integration/features/webdav-related.feature
index 21e195af115..f63ee24527f 100644
--- a/build/integration/features/webdav-related.feature
+++ b/build/integration/features/webdav-related.feature
@@ -191,10 +191,10 @@ Feature: webdav-related
 		And As an "user1"
 		And user "user1" created a folder "/testquota"
 		And as "user1" creating a share with
-		  | path | testquota |
-		  | shareType | 0 |
-		  | permissions | 31 |
-		  | shareWith | user0 |
+			| path | testquota |
+			| shareType | 0 |
+			| permissions | 31 |
+			| shareWith | user0 |
 		And user "user0" accepts last share
 		And As an "user0"
 		When User "user0" uploads file "data/textfile.txt" to "/testquota/asdf.txt"
@@ -630,3 +630,99 @@ Feature: webdav-related
 		And As an "user1"
 		And user "user1" created a folder "/testshare	"
 		Then the HTTP status code should be "400"
+
+	@s3-multipart
+	Scenario: Upload chunked file asc with new chunking v2
+		Given using new dav path
+		And user "user0" exists
+		And user "user0" creates a file locally with "3" x 5 MB chunks
+		And user "user0" creates a new chunking v2 upload with id "chunking-42" and destination "/myChunkedFile1.txt"
+		And user "user0" uploads new chunk v2 file "1" to id "chunking-42"
+		And user "user0" uploads new chunk v2 file "2" to id "chunking-42"
+		And user "user0" uploads new chunk v2 file "3" to id "chunking-42"
+		And user "user0" moves new chunk v2 file with id "chunking-42"
+		Then the S3 multipart upload was successful with status "201"
+		When As an "user0"
+		And Downloading file "/myChunkedFile1.txt"
+		Then Downloaded content should be the created file
+
+	@s3-multipart
+	Scenario: Upload chunked file desc with new chunking v2
+		Given using new dav path
+		And user "user0" exists
+		And user "user0" creates a file locally with "3" x 5 MB chunks
+		And user "user0" creates a new chunking v2 upload with id "chunking-42" and destination "/myChunkedFile.txt"
+		And user "user0" uploads new chunk v2 file "3" to id "chunking-42"
+		And user "user0" uploads new chunk v2 file "2" to id "chunking-42"
+		And user "user0" uploads new chunk v2 file "1" to id "chunking-42"
+		And user "user0" moves new chunk v2 file with id "chunking-42"
+		Then the S3 multipart upload was successful with status "201"
+		When As an "user0"
+		And Downloading file "/myChunkedFile.txt"
+		Then Downloaded content should be the created file
+
+	@s3-multipart
+	Scenario: Upload chunked file with random chunk sizes
+		Given using new dav path
+		And user "user0" exists
+		And user "user0" creates a new chunking v2 upload with id "chunking-random" and destination "/myChunkedFile.txt"
+		And user user0 creates the chunk 1 with a size of 5 MB
+		And user user0 creates the chunk 2 with a size of 7 MB
+		And user user0 creates the chunk 3 with a size of 9 MB
+		And user user0 creates the chunk 4 with a size of 1 MB
+		And user "user0" uploads new chunk v2 file "1" to id "chunking-random"
+		And user "user0" uploads new chunk v2 file "3" to id "chunking-random"
+		And user "user0" uploads new chunk v2 file "2" to id "chunking-random"
+		And user "user0" uploads new chunk v2 file "4" to id "chunking-random"
+		And user "user0" moves new chunk v2 file with id "chunking-random"
+		Then the S3 multipart upload was successful with status "201"
+		When As an "user0"
+		And Downloading file "/myChunkedFile.txt"
+		Then Downloaded content should be the created file
+
+	@s3-multipart
+	Scenario: Upload chunked file with too low chunk sizes
+		Given using new dav path
+		And user "user0" exists
+		And user "user0" creates a new chunking v2 upload with id "chunking-random" and destination "/myChunkedFile.txt"
+		And user user0 creates the chunk 1 with a size of 5 MB
+		And user user0 creates the chunk 2 with a size of 2 MB
+		And user user0 creates the chunk 3 with a size of 5 MB
+		And user user0 creates the chunk 4 with a size of 1 MB
+		And user "user0" uploads new chunk v2 file "1" to id "chunking-random"
+		And user "user0" uploads new chunk v2 file "3" to id "chunking-random"
+		And user "user0" uploads new chunk v2 file "2" to id "chunking-random"
+		And user "user0" uploads new chunk v2 file "4" to id "chunking-random"
+		And user "user0" moves new chunk v2 file with id "chunking-random"
+		Then the HTTP status code should be "500"
+
+	@s3-multipart
+	Scenario: Upload chunked file with special characters with new chunking v2
+		Given using new dav path
+		And user "user0" exists
+		And user "user0" creates a file locally with "3" x 5 MB chunks
+		And user "user0" creates a new chunking v2 upload with id "chunking-42" and destination "/äöü.txt"
+		And user "user0" uploads new chunk v2 file "1" to id "chunking-42"
+		And user "user0" uploads new chunk v2 file "2" to id "chunking-42"
+		And user "user0" uploads new chunk v2 file "3" to id "chunking-42"
+		And user "user0" moves new chunk v2 file with id "chunking-42"
+		Then the S3 multipart upload was successful with status "201"
+		When As an "user0"
+		And Downloading file "/äöü.txt"
+		Then Downloaded content should be the created file
+
+	@s3-multipart
+	Scenario: Upload chunked file with special characters in path with new chunking v2
+		Given using new dav path
+		And user "user0" exists
+		And User "user0" created a folder "üäöé"
+		And user "user0" creates a file locally with "3" x 5 MB chunks
+		And user "user0" creates a new chunking v2 upload with id "chunking-42" and destination "/üäöé/äöü.txt"
+		And user "user0" uploads new chunk v2 file "1" to id "chunking-42"
+		And user "user0" uploads new chunk v2 file "2" to id "chunking-42"
+		And user "user0" uploads new chunk v2 file "3" to id "chunking-42"
+		And user "user0" moves new chunk v2 file with id "chunking-42"
+		Then the S3 multipart upload was successful with status "201"
+		When As an "user0"
+		And Downloading file "/üäöé/äöü.txt"
+		Then Downloaded content should be the created file