feat: add metadata to the manifest (#960)

* Add metadata to the manifest * Remove useless import * Fix the openapi documentation * Use optional fields instead of default values * Remove testRestApi target * Return failure when the protobuf cannot get the field * Set download headers and fix cors headers when an error is returned * Add tests to verify the download headers * Try to adjust the content length header * Fix convertion to string * Remove the content length header * Remove testRestApi target * Removing debug messages
codex-storage · Oct 25, 2024 · 2fb7031 · 2fb7031
1 parent bcc1468
commit 2fb7031
Show file tree

Hide file tree

Showing 9 changed files with 344 additions and 39 deletions.
diff --git a/codex/manifest/coders.nim b/codex/manifest/coders.nim
@@ -10,6 +10,7 @@
 # This module implements serialization and deserialization of Manifest
 
 import pkg/upraises
+import times 
 
 push: {.upraises: [].}
 
@@ -59,6 +60,9 @@ proc encode*(manifest: Manifest): ?!seq[byte] =
   #     optional hcodec: MultiCodec = 5   # Multihash codec
   #     optional version: CidVersion = 6; # Cid version
   #     optional ErasureInfo erasure = 7; # erasure coding info
+  #     optional filename: ?string = 8;    # original filename
+  #     optional mimetype: ?string = 9;    # original mimetype
+  #     optional uploadedAt: ?int64 = 10;  # original uploadedAt
   #   }
   # ```
   #
@@ -70,6 +74,7 @@ proc encode*(manifest: Manifest): ?!seq[byte] =
   header.write(4, manifest.codec.uint32)
   header.write(5, manifest.hcodec.uint32)
   header.write(6, manifest.version.uint32)
+
   if manifest.protected:
     var erasureInfo = initProtoBuffer()
     erasureInfo.write(1, manifest.ecK.uint32)
@@ -90,6 +95,15 @@ proc encode*(manifest: Manifest): ?!seq[byte] =
     erasureInfo.finish()
     header.write(7, erasureInfo)
 
+  if manifest.filename.isSome:
+    header.write(8, manifest.filename.get())
+
+  if manifest.mimetype.isSome: 
+    header.write(9, manifest.mimetype.get())
+
+  if manifest.uploadedAt.isSome:
+    header.write(10, manifest.uploadedAt.get().uint64)
+
   pbNode.write(1, header) # set the treeCid as the data field
   pbNode.finish()
 
@@ -118,6 +132,9 @@ proc decode*(_: type Manifest, data: openArray[byte]): ?!Manifest =
     slotRoots: seq[seq[byte]]
     cellSize: uint32
     verifiableStrategy: uint32
+    filename: string
+    mimetype: string
+    uploadedAt: uint64
 
   # Decode `Header` message
   if pbNode.getField(1, pbHeader).isErr:
@@ -145,6 +162,15 @@ proc decode*(_: type Manifest, data: openArray[byte]): ?!Manifest =
   if pbHeader.getField(7, pbErasureInfo).isErr:
     return failure("Unable to decode `erasureInfo` from manifest!")
 
+  if pbHeader.getField(8, filename).isErr:
+    return failure("Unable to decode `filename` from manifest!")
+
+  if pbHeader.getField(9, mimetype).isErr:
+    return failure("Unable to decode `mimetype` from manifest!")
+
+  if pbHeader.getField(10, uploadedAt).isErr:
+    return failure("Unable to decode `uploadedAt` from manifest!")
+
   let protected = pbErasureInfo.buffer.len > 0
   var verifiable = false
   if protected:
@@ -183,6 +209,10 @@ proc decode*(_: type Manifest, data: openArray[byte]): ?!Manifest =
   let
     treeCid = ? Cid.init(treeCidBuf).mapFailure
 
+  var filenameOption = if filename.len == 0: string.none else: filename.some
+  var mimetypeOption = if mimetype.len == 0: string.none else: mimetype.some
+  var uploadedAtOption = if uploadedAt == 0: int64.none else: uploadedAt.int64.some
+
   let
     self = if protected:
       Manifest.new(
@@ -196,15 +226,21 @@ proc decode*(_: type Manifest, data: openArray[byte]): ?!Manifest =
         ecM = ecM.int,
         originalTreeCid = ? Cid.init(originalTreeCid).mapFailure,
         originalDatasetSize = originalDatasetSize.NBytes,
-        strategy = StrategyType(protectedStrategy))
+        strategy = StrategyType(protectedStrategy),
+        filename = filenameOption,
+        mimetype = mimetypeOption,
+        uploadedAt = uploadedAtOption)
       else:
         Manifest.new(
           treeCid = treeCid,
           datasetSize = datasetSize.NBytes,
           blockSize = blockSize.NBytes,
           version = CidVersion(version),
           hcodec = hcodec.MultiCodec,
-          codec = codec.MultiCodec)
+          codec = codec.MultiCodec,
+          filename = filenameOption,
+          mimetype = mimetypeOption,
+          uploadedAt = uploadedAtOption)
 
   ? self.verify()
 

diff --git a/codex/manifest/manifest.nim b/codex/manifest/manifest.nim
@@ -36,6 +36,9 @@ type
     codec: MultiCodec                       # Dataset codec
     hcodec: MultiCodec                      # Multihash codec
     version: CidVersion                     # Cid version
+    filename {.serialize.}: ?string          # The filename of the content uploaded (optional)
+    mimetype {.serialize.}: ?string          # The mimetype of the content uploaded (optional)
+    uploadedAt {.serialize.}: ?int64          # The UTC creation timestamp in seconds
     case protected {.serialize.}: bool      # Protected datasets have erasure coded info
     of true:
       ecK: int                              # Number of blocks to encode
@@ -121,6 +124,14 @@ func verifiableStrategy*(self: Manifest): StrategyType =
 func numSlotBlocks*(self: Manifest): int =
   divUp(self.blocksCount, self.numSlots)
 
+func filename*(self: Manifest): ?string =
+  self.filename
+
+func mimetype*(self: Manifest): ?string =
+  self.mimetype
+
+func uploadedAt*(self: Manifest): ?int64 =
+  self.uploadedAt
 ############################################################
 # Operations on block list
 ############################################################
@@ -163,6 +174,9 @@ func `==`*(a, b: Manifest): bool =
   (a.hcodec == b.hcodec) and
   (a.codec == b.codec) and
   (a.protected == b.protected) and
+  (a.filename == b.filename) and
+  (a.mimetype == b.mimetype) and
+  (a.uploadedAt == b.uploadedAt) and
     (if a.protected:
       (a.ecK == b.ecK) and
       (a.ecM == b.ecM) and
@@ -181,26 +195,38 @@ func `==`*(a, b: Manifest): bool =
       true)
 
 func `$`*(self: Manifest): string =
-  "treeCid: " & $self.treeCid &
+  result = "treeCid: " & $self.treeCid &
     ", datasetSize: " & $self.datasetSize &
     ", blockSize: " & $self.blockSize &
     ", version: " & $self.version &
     ", hcodec: " & $self.hcodec &
     ", codec: " & $self.codec &
-    ", protected: " & $self.protected &
-    (if self.protected:
-      ", ecK: " & $self.ecK &
-      ", ecM: " & $self.ecM &
-      ", originalTreeCid: " & $self.originalTreeCid &
-      ", originalDatasetSize: " & $self.originalDatasetSize &
-      ", verifiable: " & $self.verifiable &
-      (if self.verifiable:
-        ", verifyRoot: " & $self.verifyRoot &
-        ", slotRoots: " & $self.slotRoots
-      else:
-        "")
+    ", protected: " & $self.protected
+
+  if self.filename.isSome:
+    result &= ", filename: " & $self.filename
+
+  if self.mimetype.isSome:
+    result &= ", mimetype: " & $self.mimetype
+
+  if self.uploadedAt.isSome:
+    result &= ", uploadedAt: " & $self.uploadedAt
+
+  result &= (if self.protected:
+    ", ecK: " & $self.ecK &
+    ", ecM: " & $self.ecM &
+    ", originalTreeCid: " & $self.originalTreeCid &
+    ", originalDatasetSize: " & $self.originalDatasetSize &
+    ", verifiable: " & $self.verifiable &
+    (if self.verifiable:
+      ", verifyRoot: " & $self.verifyRoot &
+      ", slotRoots: " & $self.slotRoots
     else:
       "")
+  else:
+    "")
+
+  return result
 
 ############################################################
 # Constructors
@@ -214,7 +240,10 @@ func new*(
   version: CidVersion = CIDv1,
   hcodec = Sha256HashCodec,
   codec = BlockCodec,
-  protected = false): Manifest =
+  protected = false,
+  filename: ?string = string.none,
+  mimetype: ?string = string.none,
+  uploadedAt: ?int64 = int64.none): Manifest =
 
   T(
     treeCid: treeCid,
@@ -223,7 +252,10 @@ func new*(
     version: version,
     codec: codec,
     hcodec: hcodec,
-    protected: protected)
+    protected: protected,
+    filename: filename,
+    mimetype: mimetype,
+    uploadedAt: uploadedAt)
 
 func new*(
   T: type Manifest,
@@ -247,7 +279,11 @@ func new*(
     ecK: ecK, ecM: ecM,
     originalTreeCid: manifest.treeCid,
     originalDatasetSize: manifest.datasetSize,
-    protectedStrategy: strategy)
+    protectedStrategy: strategy,
+    filename: manifest.filename,
+    mimetype: manifest.mimetype,
+    uploadedAt: manifest.uploadedAt
+    )
 
 func new*(
   T: type Manifest,
@@ -263,7 +299,10 @@ func new*(
     codec: manifest.codec,
     hcodec: manifest.hcodec,
     blockSize: manifest.blockSize,
-    protected: false)
+    protected: false,
+    filename: manifest.filename,
+    mimetype: manifest.mimetype,
+    uploadedAt: manifest.uploadedAt)
 
 func new*(
   T: type Manifest,
@@ -277,7 +316,10 @@ func new*(
   ecM: int,
   originalTreeCid: Cid,
   originalDatasetSize: NBytes,
-  strategy = SteppedStrategy): Manifest =
+  strategy = SteppedStrategy,
+  filename: ?string = string.none,
+  mimetype: ?string = string.none,
+  uploadedAt: ?int64 = int64.none): Manifest =
 
   Manifest(
     treeCid: treeCid,
@@ -291,7 +333,10 @@ func new*(
     ecM: ecM,
     originalTreeCid: originalTreeCid,
     originalDatasetSize: originalDatasetSize,
-    protectedStrategy: strategy)
+    protectedStrategy: strategy,
+    filename: filename, 
+    mimetype: mimetype,
+    uploadedAt: uploadedAt)
 
 func new*(
   T: type Manifest,
@@ -329,7 +374,11 @@ func new*(
     verifyRoot: verifyRoot,
     slotRoots: @slotRoots,
     cellSize: cellSize,
-    verifiableStrategy: strategy)
+    verifiableStrategy: strategy,
+    filename: manifest.filename,
+    mimetype: manifest.mimetype,
+    uploadedAt: manifest.uploadedAt
+    )
 
 func new*(
   T: type Manifest,

diff --git a/codex/node.nim b/codex/node.nim
@@ -14,6 +14,7 @@ import std/sequtils
 import std/strformat
 import std/sugar
 import std/cpuinfo
+import times
 
 import pkg/questionable
 import pkg/questionable/results
@@ -297,6 +298,8 @@ proc retrieve*(
 proc store*(
   self: CodexNodeRef,
   stream: LPStream,
+  filename: ?string = string.none,
+  mimetype: ?string = string.none,
   blockSize = DefaultBlockSize): Future[?!Cid] {.async.} =
   ## Save stream contents as dataset with given blockSize
   ## to nodes's BlockStore, and return Cid of its manifest
@@ -355,7 +358,10 @@ proc store*(
     datasetSize = NBytes(chunker.offset),
     version = CIDv1,
     hcodec = hcodec,
-    codec = dataCodec)
+    codec = dataCodec,
+    filename = filename,
+    mimetype = mimetype,
+    uploadedAt = now().utc.toTime.toUnix.some)
 
   without manifestBlk =? await self.storeManifest(manifest), err:
     error "Unable to store manifest"
@@ -364,7 +370,9 @@ proc store*(
   info "Stored data", manifestCid = manifestBlk.cid,
                       treeCid = treeCid,
                       blocks = manifest.blocksCount,
-                      datasetSize = manifest.datasetSize
+                      datasetSize = manifest.datasetSize,
+                      filename = manifest.filename,
+                      mimetype = manifest.mimetype
 
   return manifestBlk.cid.success