From 82753a6cda4ae96802ec6f9ca5988fb6ef6a8f89 Mon Sep 17 00:00:00 2001
From: Alex Sharov <AskAlexSharov@gmail.com>
Date: Tue, 21 Dec 2021 10:48:34 +0700
Subject: [PATCH] add words count in .seg (breaking change in snapshot format)
 (#3140)

---
 cmd/hack/hack.go | 22 ++++++++++++++--------
 go.mod           |  2 +-
 go.sum           |  4 ++--
 3 files changed, 17 insertions(+), 11 deletions(-)

diff --git a/cmd/hack/hack.go b/cmd/hack/hack.go
index c7a41d1bf..6c6c98828 100644
--- a/cmd/hack/hack.go
+++ b/cmd/hack/hack.go
@@ -2171,19 +2171,19 @@ func reducedict(name string, segmentFileName string) error {
 		wg.Add(1)
 		go reduceDictWorker(ch, &wg, &pt, collector, inputSize, outputSize, posMap)
 	}
-	i := 0
+	var wordsCount uint64
 	if err := snapshotsync.ReadSimpleFile(name+".dat", func(v []byte) error {
 		input := make([]byte, 8+int(len(v)))
-		binary.BigEndian.PutUint64(input, uint64(i))
+		binary.BigEndian.PutUint64(input, wordsCount)
 		copy(input[8:], v)
 		ch <- input
-		i++
+		wordsCount++
 		select {
 		default:
 		case <-logEvery.C:
 			var m runtime.MemStats
 			runtime.ReadMemStats(&m)
-			log.Info("Replacement preprocessing", "processed", fmt.Sprintf("%dK", i/1_000), "input", common.StorageSize(inputSize.Load()), "output", common.StorageSize(outputSize.Load()), "alloc", common.StorageSize(m.Alloc), "sys", common.StorageSize(m.Sys))
+			log.Info("Replacement preprocessing", "processed", fmt.Sprintf("%dK", wordsCount/1_000), "input", common.StorageSize(inputSize.Load()), "output", common.StorageSize(outputSize.Load()), "alloc", common.StorageSize(m.Alloc), "sys", common.StorageSize(m.Sys))
 		}
 		return nil
 	}); err != nil {
@@ -2191,6 +2191,7 @@ func reducedict(name string, segmentFileName string) error {
 	}
 	close(ch)
 	wg.Wait()
+
 	var m runtime.MemStats
 	runtime.ReadMemStats(&m)
 	log.Info("Done", "input", common.StorageSize(inputSize.Load()), "output", common.StorageSize(outputSize.Load()), "alloc", common.StorageSize(m.Alloc), "sys", common.StorageSize(m.Sys))
@@ -2223,7 +2224,7 @@ func reducedict(name string, segmentFileName string) error {
 		offset += uint64(n + len(p.w))
 	}
 	patternCutoff := offset // All offsets below this will be considered patterns
-	i = 0
+	i := 0
 	log.Info("Effective dictionary", "size", patternList.Len())
 	// Build Huffman tree for codes
 	var codeHeap PatternHeap
@@ -2284,17 +2285,22 @@ func reducedict(name string, segmentFileName string) error {
 		return err
 	}
 	cw := bufio.NewWriterSize(cf, etl.BufIOSize)
-	// First, output dictionary
+	// 1-st, output dictionary
+	binary.BigEndian.PutUint64(numBuf, wordsCount) // Dictionary size
+	if _, err = cw.Write(numBuf[:8]); err != nil {
+		return err
+	}
+	// 2-nd, output dictionary
 	binary.BigEndian.PutUint64(numBuf, offset) // Dictionary size
 	if _, err = cw.Write(numBuf[:8]); err != nil {
 		return err
 	}
-	// Secondly, output directory root
+	// 3-rd, output directory root
 	binary.BigEndian.PutUint64(numBuf, root.offset)
 	if _, err = cw.Write(numBuf[:8]); err != nil {
 		return err
 	}
-	// Thirdly, output pattern cutoff offset
+	// 4-th, output pattern cutoff offset
 	binary.BigEndian.PutUint64(numBuf, patternCutoff)
 	if _, err = cw.Write(numBuf[:8]); err != nil {
 		return err
diff --git a/go.mod b/go.mod
index 1b84228ff..7c7c10ef0 100644
--- a/go.mod
+++ b/go.mod
@@ -37,7 +37,7 @@ require (
 	github.com/json-iterator/go v1.1.12
 	github.com/julienschmidt/httprouter v1.3.0
 	github.com/kevinburke/go-bindata v3.21.0+incompatible
-	github.com/ledgerwatch/erigon-lib v0.0.0-20211217093546-8d06531e4ed3
+	github.com/ledgerwatch/erigon-lib v0.0.0-20211221034520-583c3f9b5b6f
 	github.com/ledgerwatch/log/v3 v3.4.0
 	github.com/ledgerwatch/secp256k1 v1.0.0
 	github.com/logrusorgru/aurora/v3 v3.0.0
diff --git a/go.sum b/go.sum
index 2ef60e92e..5eb9a4e59 100644
--- a/go.sum
+++ b/go.sum
@@ -617,8 +617,8 @@ github.com/kylelemons/godebug v0.0.0-20170224010052-a616ab194758 h1:0D5M2HQSGD3P
 github.com/kylelemons/godebug v0.0.0-20170224010052-a616ab194758/go.mod h1:B69LEHPfb2qLo0BaaOLcbitczOKLWTsrBG9LczfCD4k=
 github.com/leanovate/gopter v0.2.9 h1:fQjYxZaynp97ozCzfOyOuAGOU4aU/z37zf/tOujFk7c=
 github.com/leanovate/gopter v0.2.9/go.mod h1:U2L/78B+KVFIx2VmW6onHJQzXtFb+p5y3y2Sh+Jxxv8=
-github.com/ledgerwatch/erigon-lib v0.0.0-20211217093546-8d06531e4ed3 h1:8qDZvisP+6pFiVFd20BTD2y8/rYAe4go//HdBnk6CX8=
-github.com/ledgerwatch/erigon-lib v0.0.0-20211217093546-8d06531e4ed3/go.mod h1:lyGP3i0x4CeabdKZ4beycD5xZfHWZwJsAX+70OfGj4Y=
+github.com/ledgerwatch/erigon-lib v0.0.0-20211221034520-583c3f9b5b6f h1:MCIljelbCsLcgMzNTsrRg2Nu5DFyNlLxf5ZSWdy3CiM=
+github.com/ledgerwatch/erigon-lib v0.0.0-20211221034520-583c3f9b5b6f/go.mod h1:lyGP3i0x4CeabdKZ4beycD5xZfHWZwJsAX+70OfGj4Y=
 github.com/ledgerwatch/log/v3 v3.4.0 h1:SEIOcv5a2zkG3PmoT5jeTU9m/0nEUv0BJS5bzsjwKCI=
 github.com/ledgerwatch/log/v3 v3.4.0/go.mod h1:VXcz6Ssn6XEeU92dCMc39/g1F0OYAjw1Mt+dGP5DjXY=
 github.com/ledgerwatch/secp256k1 v1.0.0 h1:Usvz87YoTG0uePIV8woOof5cQnLXGYa162rFf3YnwaQ=