1515use OCP \Files \File as NcFile ;
1616use OCP \Files \Folder as NcFolder ;
1717use OCP \Files \Node as NcNode ;
18+ use OCP \IConfig ;
1819use OCP \IDateTimeZone ;
20+ use OCP \IL10N ;
21+ use OCP \L10N \IFactory ;
1922use Psr \Log \LoggerInterface ;
2023use Sabre \DAV \Server ;
2124use Sabre \DAV \ServerPlugin ;
@@ -37,13 +40,25 @@ class ZipFolderPlugin extends ServerPlugin {
3740 * Reference to main server object
3841 */
3942 private ?Server $ servernull ;
43+ private bool $ reportMissingFiles
44+ private array $ missingInfo
45+ private IL10N $ l10n
4046
4147 public function __construct (
4248 private Tree $ tree
4349 private LoggerInterface $ logger
4450 private IEventDispatcher $ eventDispatcher
4551 private IDateTimeZone $ timezoneFactory
52+ private IConfig $ config
53+ private IFactory $ l10nFactory
4654 ) {
55+ $ this reportMissingFiles = $ this config ->getSystemValueBool ('archive_report_missing_files ' , false );
56+
57+ if ($ this reportMissingFiles ) {
58+ stream_filter_register ('count.bytes ' , ByteCounterFilter::class);
59+ }
60+
61+ $ this l10n = $ this l10nFactory ->get ('dav ' );
4762 }
4863
4964 /**
@@ -62,46 +77,78 @@ public function initialize(Server $server): void {
6277 }
6378
6479 /**
65- * @return iterable<NcNode>
80+ * Adding a node to the archive streamer.
81+ * @return ?string an error message if an error occurred and reporting is enabled, null otherwise
6682 */
67- protected function createIterator (array $ rootNodesiterable {
68- foreach ($ rootNodesas $ rootNode
69- yield from $ this iterateNodes ($ rootNode
83+ protected function streamNode (Streamer $ streamerNcNode $ nodestring $ rootPathstring {
84+ // Remove the root path from the filename to make it relative to the requested folder
85+ $ filenamestr_replace ($ rootPath'' , $ nodegetPath ());
86+
87+ $ mtime$ nodegetMTime ();
88+ if ($ nodeinstanceof NcFolder) {
89+ $ streameraddEmptyDir ($ filename$ mtime
90+ return null ;
7091 }
71- }
7292
73- /**
74- * Recursively iterate over all nodes in a folder.
75- * @return iterable<NcNode>
76- */
77- protected function iterateNodes (NcNode $ nodeiterable {
78- yield $ node
93+ if ($ nodeinstanceof NcFile) {
94+ $ nodeSize$ nodegetSize ();
95+ try {
96+ $ stream$ nodefopen ('rb ' );
97+ } catch (\Exception $ e
98+ // opening failed, log the failure as reason for the missing file
99+ if ($ this reportMissingFiles ) {
100+ $ exceptionClassget_class ($ e
101+ return $ this l10n ->t ('Error while opening the file: %s ' , [$ exceptionClass
102+ }
79103
80- if ($ nodeinstanceof NcFolder) {
81- foreach ($ nodegetDirectoryListing () as $ childNode
82- yield from $ this iterateNodes ($ childNode
104+ throw $ e
105+ }
106+
107+ if ($ this reportMissingFiles ) {
108+ if ($ streamfalse ) {
109+ return $ this l10n ->t ('File could not be opened (fopen). Please check the server logs for more information. ' );
110+ }
111+
112+ $ byteCounternew StreamByteCounter ();
113+ $ wrappedstream_filter_append ($ stream'count.bytes ' , STREAM_FILTER_READ , ['counter ' => $ byteCounter
114+ if ($ wrappedfalse ) {
115+ return $ this l10n ->t ('Unable to check file for consistency check ' );
116+ }
117+ }
118+
119+ $ fileAddedToStream$ streameraddFileFromStream ($ stream$ filename$ nodeSize$ mtime
120+ if ($ this reportMissingFiles ) {
121+ if (!$ fileAddedToStream
122+ return $ this l10n ->t ('The archive was already finalized ' );
123+ }
124+
125+ return $ this logStreamErrors ($ stream$ filename$ nodeSize$ byteCounterbytes );
83126 }
127+
128+ return null ;
84129 }
85130 }
86131
87132 /**
88- * Adding a node to the archive streamer.
133+ * Checks whether $stream was fully streamed or if there were other issues
134+ * with the stream, logging the error if necessary.
135+ *
89136 */
90- protected function streamNode (Streamer $ streamerNcNode $ nodestring $ rootPathvoid {
91- // Remove the root path from the filename to make it relative to the requested folder
92- $ filenamestr_replace ($ rootPath'' , $ nodegetPath ());
137+ private function logStreamErrors (mixed $ streamstring $ pathfloat |int $ expectedFileSizefloat |int $ readFileSizestring {
138+ $ streamMetadatastream_get_meta_data ($ stream
139+ if (!is_resource ($ streamget_resource_type ($ stream'stream ' ) {
140+ return $ this l10n ->t ('Resource is not a stream or is closed. ' );
141+ }
93142
94- $ mtime$ nodegetMTime ();
95- if ($ nodeinstanceof NcFile) {
96- $ resource$ nodefopen ('rb ' );
97- if ($ resourcefalse ) {
98- $ this logger ->info ('Cannot read file for zip stream ' , ['filePath ' => $ nodegetPath ()]);
99- throw new \Sabre \DAV \Exception \ServiceUnavailable ('Requested file can currently not be accessed. ' );
100- }
101- $ streameraddFileFromStream ($ resource$ filename$ nodegetSize (), $ mtime
102- } elseif ($ nodeinstanceof NcFolder) {
103- $ streameraddEmptyDir ($ filename$ mtime
143+ if ($ streamMetadata'timed_out ' ] ?? false ) {
144+ return $ this l10n ->t ('Timeout while reading from stream. ' );
145+ }
146+
147+ if (!($ streamMetadata'eof ' ] ?? true ) || $ readFileSize$ expectedFileSize
148+ return $ this l10n ->t ('Read %d out of %d bytes from storage. This means the connection may have been closed due to a network/storage error. ' , [$ expectedFileSize$ readFileSize
104149 }
150+
151+ return null ;
105152 }
106153
107154 /**
@@ -155,14 +202,7 @@ public function handleDownload(Request $request, Response $response): ?bool {
155202 }
156203
157204 $ folder$ nodegetNode ();
158- $ rootNodesempty ($ files$ foldergetDirectoryListing () : [];
159- foreach ($ filesas $ path
160- $ child$ nodegetChild ($ path
161- assert ($ childinstanceof Node);
162- $ rootNodes$ childgetNode ();
163- }
164-
165- $ eventnew BeforeZipCreatedEvent ($ folder$ files$ this createIterator ($ rootNodes
205+ $ eventnew BeforeZipCreatedEvent ($ folder$ files$ this reportMissingFiles );
166206 $ this eventDispatcher ->dispatchTyped ($ event
167207 if ((!$ eventisSuccessful ()) || $ eventgetErrorMessage () !== null ) {
168208 $ errorMessage$ eventgetErrorMessage ();
@@ -175,6 +215,17 @@ public function handleDownload(Request $request, Response $response): ?bool {
175215 throw new Forbidden ($ errorMessage
176216 }
177217
218+ // At this point either the event handlers did not block the download
219+ // or they support the new mechanism that filters out nodes that are not
220+ // downloadable, in either case we can use the new API to set the iterator
221+ $ contentempty ($ files$ foldergetDirectoryListing () : [];
222+ foreach ($ filesas $ path
223+ $ child$ nodegetChild ($ path
224+ assert ($ childinstanceof Node);
225+ $ content$ childgetNode ();
226+ }
227+ $ eventsetNodesIterable ($ this getIterableFromNodes ($ content
228+
178229 $ archiveName$ foldergetName ();
179230 if (count (explode ('/ ' , trim ($ foldergetPath (), '/ ' ), 3 )) === 2 ) {
180231 // this is a download of the root folder
@@ -187,31 +238,71 @@ public function handleDownload(Request $request, Response $response): ?bool {
187238 $ rootPathdirname ($ foldergetPath ());
188239 }
189240
190- $ streamernew Streamer ($ tarRequest1 , count ($ rootNodes$ this timezoneFactory );
241+ // numberOfFiles is irrelevant as size=-1 forces the use of zip64 already
242+ $ streamernew Streamer ($ tarRequest1 , 0 , $ this timezoneFactory );
191243 $ streamersendHeaders ($ archiveName
192244 // For full folder downloads we also add the folder itself to the archive
193245 if (empty ($ files
194246 $ streameraddEmptyDir ($ archiveName
195247 }
196- foreach ($ eventgetNodes () as $ node
197- $ this streamNode ($ streamer$ node$ rootPath
248+
249+ foreach ($ eventgetNodes () as $ path$ node$ reason
250+ $ filenamestr_replace ($ rootPath'' , $ path
251+ if ($ nodenull ) {
252+ if ($ this reportMissingFiles ) {
253+ $ this missingInfo [$ filename$ reason
254+ }
255+ continue ;
256+ }
257+
258+ $ streamError$ this streamNode ($ streamer$ node$ rootPath
259+ if ($ this reportMissingFiles && $ streamErrornull ) {
260+ $ this missingInfo [$ filename$ streamError
261+ }
262+ }
263+
264+ if ($ this reportMissingFiles && !empty ($ this missingInfo )) {
265+ $ jsonjson_encode ($ this missingInfo , JSON_UNESCAPED_SLASHES | JSON_PRETTY_PRINT );
266+ $ streamfopen ('php://temp ' , 'r+ ' );
267+ fwrite ($ stream$ json
268+ rewind ($ stream
269+ $ streameraddFileFromStream ($ stream'missing_files.json ' , (float )strlen ($ jsonfalse );
198270 }
199271 $ streamerfinalize ();
200272 return false ;
201273 }
202274
203275 /**
204- * Tell sabre/dav not to trigger it's own response sending logic as the handleDownload will have already send the response
276+ * Given a set of nodes, produces a list of all nodes contained in them
277+ * recursively.
278+ *
279+ * @param NcNode[] $nodes
280+ * @return iterable<NcNode>
281+ */
282+ private function getIterableFromNodes (array $ nodesiterable {
283+ foreach ($ nodesas $ node
284+ yield $ node
285+
286+ if ($ nodeinstanceof NcFolder) {
287+ foreach ($ nodegetDirectoryListing () as $ child
288+ yield from $ this getIterableFromNodes ([$ child
289+ }
290+ }
291+ }
292+ }
293+
294+ /**
295+ * Tell sabre/dav not to trigger its own response sending logic as the handleDownload will have already send the response
205296 *
206297 * @return false|null
207298 */
208299 public function afterDownload (Request $ requestResponse $ responsebool {
209300 $ node$ this tree ->getNodeForPath ($ requestgetPath ());
210- if (!( $ nodeinstanceof Directory) ) {
301+ if ($ nodeinstanceof Directory) {
211302 // only handle directories
212- return null ;
213- } else {
214303 return false ;
215304 }
305+
306+ return null ;
216307 }
217308}
0 commit comments