colblk: expose block reader Describe methods

Expose Describe methods from the IndexBlockReader, DataBlockReader and KeyspanReader types allowing a caller to write a description of the binary representation of the block to a binfmt.Formatter. Additionally, ensure we format the trailing 0x00 block padding byte.
cockroachdb · Sep 9, 2024 · 934a223 · 934a223
1 parent 376d455
commit 934a223
Show file tree

Hide file tree

Showing 10 changed files with 32 additions and 4 deletions.
diff --git a/sstable/colblk/data_block.go b/sstable/colblk/data_block.go
@@ -562,13 +562,16 @@ func (r *DataBlockReader) Init(schema KeySchema, data []byte) {
 	r.maximumKeyLength = binary.LittleEndian.Uint32(data[:dataBlockCustomHeaderSize])
 }
 
-func (r *DataBlockReader) toFormatter(f *binfmt.Formatter) {
+// Describe descirbes the binary format of the data block, assuming f.Offset()
+// is positioned at the beginning of the same data block described by r.
+func (r *DataBlockReader) Describe(f *binfmt.Formatter) {
 	f.CommentLine("data block header")
 	f.HexBytesln(4, "maximum key length: %d", r.maximumKeyLength)
 	r.r.headerToBinFormatter(f)
 	for i := 0; i < int(r.r.header.Columns); i++ {
 		r.r.columnToBinFormatter(f, i, int(r.r.header.Rows))
 	}
+	f.HexBytesln(1, "block padding byte")
 }
 
 // DataBlockIter iterates over a columnar data block.

diff --git a/sstable/colblk/data_block_test.go b/sstable/colblk/data_block_test.go
@@ -61,7 +61,7 @@ func TestDataBlock(t *testing.T) {
 				block := w.Finish()
 				r.Init(testKeysSchema, block)
 				f := binfmt.New(r.r.data).LineWidth(20)
-				r.toFormatter(f)
+				r.Describe(f)
 
 				return f.String()
 			case "iter":

diff --git a/sstable/colblk/index_block.go b/sstable/colblk/index_block.go
@@ -142,12 +142,19 @@ func (r *IndexReader) Init(data []byte) {
 // representation.
 func (r *IndexReader) DebugString() string {
 	f := binfmt.New(r.br.data).LineWidth(20)
+	r.Describe(f)
+	return f.String()
+}
+
+// Describe describes the binary format of the index block, assuming f.Offset()
+// is positioned at the beginning of the same index block described by r.
+func (r *IndexReader) Describe(f *binfmt.Formatter) {
 	f.CommentLine("index block header")
 	r.br.headerToBinFormatter(f)
 	for i := 0; i < indexBlockColumnCount; i++ {
 		r.br.columnToBinFormatter(f, i, int(r.br.header.Rows))
 	}
-	return f.String()
+	f.HexBytesln(1, "block padding byte")
 }
 
 // IndexIter is an iterator over the block entries in an index block.

diff --git a/sstable/colblk/keyspan.go b/sstable/colblk/keyspan.go
@@ -231,6 +231,14 @@ func (r *KeyspanReader) Init(data []byte) {
 // representation.
 func (r *KeyspanReader) DebugString() string {
 	f := binfmt.New(r.blockReader.data).LineWidth(20)
+	r.Describe(f)
+	return f.String()
+}
+
+// Describe describes the binary format of the keyspan block, assuming
+// f.Offset() is positioned at the beginning of the same keyspan block described
+// by r.
+func (r *KeyspanReader) Describe(f *binfmt.Formatter) {
 	f.CommentLine("keyspan block header")
 	f.HexBytesln(4, "user key count: %d", r.boundaryKeysCount)
 	r.blockReader.headerToBinFormatter(f)
@@ -246,7 +254,7 @@ func (r *KeyspanReader) DebugString() string {
 		}
 		r.blockReader.columnToBinFormatter(f, i, rows)
 	}
-	return f.String()
+	f.HexBytesln(1, "block padding byte")
 }
 
 // searchBoundaryKeys returns the index of the first boundary key greater than

diff --git a/sstable/colblk/testdata/data_block/bundle_search b/sstable/colblk/testdata/data_block/bundle_search
@@ -509,6 +509,7 @@ finish
 717-718: x 76                                                               # data[65]: v
 # data for column 5
 718-719: x 01                                                               # bitmap encoding
+719-720: x 00                                                               # block padding byte
 
 iter
 seek-ge backache

diff --git a/sstable/colblk/testdata/data_block/external_value b/sstable/colblk/testdata/data_block/external_value
@@ -258,6 +258,7 @@ finish
 623-624: x 00                                                               # bitmap encoding
 624-632: b 1100111011111011000001110000000000000000000000000000000000000000 # bitmap word 0
 632-640: b 0000000100000000000000000000000000000000000000000000000000000000 # bitmap summary word 0-63
+640-641: x 00                                                               # block padding byte
 
 # Scan across the block using next.
 iter

diff --git a/sstable/colblk/testdata/data_block/next_prefix b/sstable/colblk/testdata/data_block/next_prefix
@@ -233,6 +233,7 @@ finish
 406-406: x                                                                  # data[19]:
 # data for column 5
 406-407: x 01                                                               # bitmap encoding
+407-408: x 00                                                               # block padding byte
 
 # Scan across the block using next prefix.
 

diff --git a/sstable/colblk/testdata/data_block/simple b/sstable/colblk/testdata/data_block/simple
@@ -135,6 +135,7 @@ finish
 167-167: x                                                                  # data[6]:
 # data for column 5
 167-168: x 01                                                               # bitmap encoding
+168-169: x 00                                                               # block padding byte
 
 iter
 first
@@ -484,6 +485,7 @@ finish
 331-332: x 61                                                               # data[16]: a
 # data for column 5
 332-333: x 01                                                               # bitmap encoding
+333-334: x 00                                                               # block padding byte
 
 iter
 seek-ge aaa

diff --git a/sstable/colblk/testdata/index_block b/sstable/colblk/testdata/index_block
@@ -74,6 +74,7 @@ UnsafeSeparator(Rows()-1) = "bonifide"
 121-121: x                      # data[3]:
 121-124: x 627035               # data[4]: bp5
 124-127: x 627036               # data[5]: bp6
+127-128: x 00                   # block padding byte
 
 iter
 seek-ge zoo
@@ -193,6 +194,7 @@ UnsafeSeparator(Rows()-1) = "coat"
 114-114: x                      # data[2]:
 114-117: x 627034               # data[3]: bp4
 117-117: x                      # data[4]:
+117-118: x 00                   # block padding byte
 
 iter
 first

diff --git a/sstable/colblk/testdata/keyspan_block b/sstable/colblk/testdata/keyspan_block
@@ -123,6 +123,7 @@ Boundaries: a#0,RANGEDEL — e#inf,RANGEDEL
 72-72: x          # data[4]:
 72-72: x          # data[5]:
 72-72: x          # data[6]:
+72-73: x 00       # block padding byte
 
 # Test iterating over the block's spans.
 
@@ -277,6 +278,7 @@ Boundaries: b#4,RANGEKEYSET — d#inf,RANGEKEYSET
 61-62: x 07               # data[1] = 7 [69 overall]
 # data
 62-69: x 636f636f6e7574   # data[0]: coconut
+69-70: x 00               # block padding byte
 
 iter
 seek-ge a
@@ -374,6 +376,7 @@ Boundaries: b#4,RANGEKEYSET — g#inf,RANGEKEYSET
 # data
 68-75: x 636f636f6e7574 # data[0]: coconut
 75-79: x 74726565       # data[1]: tree
+79-80: x 00             # block padding byte
 
 iter
 seek-ge dog