pump-parser/cmd/analyze_rawtx_binary_size/main.go

package main

import (
	"flag"
	"fmt"
	"math/big"
	"os"
	"sort"

	pump_parser "github.com/thloyi/pump-parser"
)

type sizeStats struct {
	total uint64
	items map[string]uint64
}

type txInnerDataStat struct {
	TxOrdinal        int
	BlockIndex       int
	IndexWithinBlock uint32
	Slot             uint64
	Bytes            uint64
	InstructionCount int
}

func newSizeStats() *sizeStats {
	return &sizeStats{items: make(map[string]uint64)}
}

func (s *sizeStats) add(name string, n uint64) {
	s.items[name] += n
	s.total += n
}

func main() {
	filePath := flag.String("file", "testdata/rawtx-binary/rawtx-blocks-414696178-414696182.prbs", "path to RawTxBlocksBinary .prbs file")
	flag.Parse()

	raw, err := os.ReadFile(*filePath)
	if err != nil {
		fmt.Fprintf(os.Stderr, "read file: %v\n", err)
		os.Exit(1)
	}

	var blocks pump_parser.RawTxBlocksBinary
	if err := blocks.UnmarshalBinary(raw); err != nil {
		fmt.Fprintf(os.Stderr, "decode rawtx blocks binary: %v\n", err)
		os.Exit(1)
	}

	stats := analyzeRawTxBlocksBinary(&blocks)
	if stats.total != uint64(len(raw)) {
		fmt.Fprintf(os.Stderr, "size accounting mismatch: accounted=%d file=%d\n", stats.total, len(raw))
		os.Exit(1)
	}

	printReport(*filePath, len(raw), &blocks, stats)
	fmt.Println()
	printInnerInstructionDataDistribution(&blocks)
	fmt.Println()
	printBalanceAnalysis(&blocks)
}

func analyzeRawTxBlocksBinary(blocks *pump_parser.RawTxBlocksBinary) *sizeStats {
	stats := newSizeStats()
	stats.add("file.magic", 4)
	stats.add("file.schema_version", 2)
	stats.add("address_table.count", 4)
	stats.add("address_table.pubkeys", uint64(len(blocks.AddressTable))*32)
	stats.add("blocks.count", 4)
	stats.add("blocks.block_time", uint64(len(blocks.BlockTimes))*8)
	stats.add("blocks.tx_count", uint64(len(blocks.BlockTxCounts))*4)
	stats.add("txs.total_count", 4)

	for i := range blocks.Txs {
		addTx(stats, &blocks.Txs[i])
	}
	return stats
}

func addTx(stats *sizeStats, tx *pump_parser.RawTxBinary) {
	stats.add("tx.index_within_block", 4)
	stats.add("tx.slot", 8)
	stats.add("tx.version", 1)
	stats.add("tx.account_key_count", 4)
	stats.add("tx.account_list.count", 4)
	stats.add("tx.account_list.refs", uint64(len(tx.AccountList))*4)

	addMeta(stats, &tx.Meta)
	addTransaction(stats, &tx.Transaction)
}

func addMeta(stats *sizeStats, meta *pump_parser.RawTxMetaBinary) {
	addErr(stats, meta.Err)
	stats.add("meta.fee", 8)
	addInnerInstructions(stats, meta.InnerInstructions)
	addLamportBalances(stats, meta.PreBalances, meta.PostBalances)
	addTokenBalances(stats, "meta.token_balances", meta.TokenBalances)
	stats.add("meta.compute_units_consumed", 8)
}

func addErr(stats *sizeStats, errValue *pump_parser.TransactionParsedError) {
	stats.add("meta.err.present", 1)
	if errValue == nil {
		return
	}
	stats.add("meta.err.index", 1)
	stats.add("meta.err.variant", 4)
	stats.add("meta.err.enum", 4)
	stats.add("meta.err.custom_code", 4)
}

func addTransaction(stats *sizeStats, tx *pump_parser.RawTxTransactionBinary) {
	stats.add("transaction.signature.present", 1)
	if tx.HasSignature {
		stats.add("transaction.signature.first", 64)
	}
	addHeader(stats)
	addInstructions(stats, "transaction.instructions", tx.Message.Instructions)
	addAddressTableLookups(stats, tx.Message.AddressTableLookups)
}

func addHeader(stats *sizeStats) {
	stats.add("transaction.header.num_readonly_signed", 4)
	stats.add("transaction.header.num_readonly_unsigned", 4)
	stats.add("transaction.header.num_required_signatures", 4)
}

func addInnerInstructions(stats *sizeStats, values []pump_parser.InnerInstructions) {
	stats.add("meta.inner_instructions.count", 4)
	for _, value := range values {
		stats.add("meta.inner_instructions.index", 4)
		addInstructions(stats, "meta.inner_instructions.instructions", value.Instructions)
	}
}

func addInstructions(stats *sizeStats, prefix string, values []pump_parser.Instruction) {
	stats.add(prefix+".count", 4)
	for _, value := range values {
		stats.add(prefix+".program_id_index", 2)
		stats.add(prefix+".accounts.count", 4)
		stats.add(prefix+".accounts.refs", uint64(len(value.Accounts))*2)
		stats.add(prefix+".data.length", 4)
		stats.add(prefix+".data.bytes", uint64(len(value.Data)))
		stats.add(prefix+".stack_height.present", 1)
		if value.StackHeight != nil {
			stats.add(prefix+".stack_height.value", 4)
		}
	}
}

func addAddressTableLookups(stats *sizeStats, values []pump_parser.RawTxAddressTableLookupBinary) {
	stats.add("transaction.address_table_lookups.count", 4)
	for _, value := range values {
		stats.add("transaction.address_table_lookups.account_key", 4)
		stats.add("transaction.address_table_lookups.writable.count", 4)
		stats.add("transaction.address_table_lookups.writable.indexes", uint64(len(value.WritableIndexes)))
		stats.add("transaction.address_table_lookups.readonly.count", 4)
		stats.add("transaction.address_table_lookups.readonly.indexes", uint64(len(value.ReadonlyIndexes)))
	}
}

func addUint64Slice(stats *sizeStats, prefix string, count int) {
	stats.add(prefix+".count", 4)
	stats.add(prefix+".values", uint64(count)*8)
}

func addLamportBalances(stats *sizeStats, preBalances []uint64, postBalances []uint64) {
	stats.add("meta.pre_balances.count_uvarint", uint64(uvarintLen(uint64(len(preBalances)))))
	for _, value := range preBalances {
		stats.add("meta.pre_balances.value_uvarint", uint64(uvarintLen(value)))
	}
	n := len(preBalances)
	if len(postBalances) < n {
		n = len(postBalances)
	}
	changed := 0
	for i := 0; i < n; i++ {
		if preBalances[i] != postBalances[i] {
			changed++
		}
	}
	stats.add("meta.post_balance_changes.count_uvarint", uint64(uvarintLen(uint64(changed))))
	for i := 0; i < n; i++ {
		if preBalances[i] == postBalances[i] {
			continue
		}
		stats.add("meta.post_balance_changes.index_uvarint", uint64(uvarintLen(uint64(i))))
		stats.add("meta.post_balance_changes.delta_uvarint", uint64(zigzagDeltaUvarintLen(preBalances[i], postBalances[i])))
	}
}

func addTokenBalances(stats *sizeStats, prefix string, values []pump_parser.RawTxTokenBalanceBinary) {
	stats.add(prefix+".count", 4)
	for _, value := range values {
		stats.add(prefix+".account_index", 2)
		stats.add(prefix+".mint_ref", 2)
		stats.add(prefix+".owner.present", 1)
		if value.HasOwnerAccount {
			stats.add(prefix+".owner_ref", 2)
		}
		stats.add(prefix+".program_id_ref", 2)
		stats.add(prefix+".decimals", 1)
		stats.add(prefix+".pre_amount.present", 1)
		if value.HasPreAmount {
			stats.add(prefix+".pre_amount.length", 1)
			stats.add(prefix+".pre_amount.bytes", uint64(uint256ByteLen(value.PreAmount)))
		}
		stats.add(prefix+".post_amount.present", 1)
		if value.HasPostAmount {
			stats.add(prefix+".post_amount.length", 1)
			stats.add(prefix+".post_amount.bytes", uint64(uint256ByteLen(value.PostAmount)))
		}
	}
}

func uint256ByteLen(value string) int {
	if value == "" || value == "0" {
		return 0
	}
	amount, ok := new(big.Int).SetString(value, 10)
	if !ok || amount.Sign() <= 0 {
		return 0
	}
	return len(amount.Bytes())
}

func printReport(filePath string, fileSize int, blocks *pump_parser.RawTxBlocksBinary, stats *sizeStats) {
	type row struct {
		name  string
		bytes uint64
	}
	rows := make([]row, 0, len(stats.items))
	for name, bytes := range stats.items {
		rows = append(rows, row{name: name, bytes: bytes})
	}
	sort.Slice(rows, func(i, j int) bool {
		if rows[i].bytes == rows[j].bytes {
			return rows[i].name < rows[j].name
		}
		return rows[i].bytes > rows[j].bytes
	})

	fmt.Printf("file=%s\n", filePath)
	fmt.Printf("bytes=%d\n", fileSize)
	fmt.Printf("schema_version=%d\n", blocks.SchemaVersion)
	fmt.Printf("blocks=%d\n", len(blocks.BlockTxCounts))
	fmt.Printf("txs=%d\n", len(blocks.Txs))
	fmt.Printf("address_table_entries=%d\n", len(blocks.AddressTable))
	fmt.Println()
	fmt.Printf("%-56s %12s %8s\n", "field", "bytes", "pct")
	fmt.Printf("%-56s %12s %8s\n", "-----", "-----", "---")
	for _, row := range rows {
		fmt.Printf("%-56s %12d %7.2f%%\n", row.name, row.bytes, float64(row.bytes)*100/float64(fileSize))
	}
}

func printInnerInstructionDataDistribution(blocks *pump_parser.RawTxBlocksBinary) {
	stats := collectInnerInstructionDataStats(blocks)
	values := make([]uint64, 0, len(stats))
	var total uint64
	var nonZero int
	for _, stat := range stats {
		values = append(values, stat.Bytes)
		total += stat.Bytes
		if stat.Bytes > 0 {
			nonZero++
		}
	}
	sort.Slice(values, func(i, j int) bool { return values[i] < values[j] })

	fmt.Println("inner_instruction_data_bytes_per_tx")
	fmt.Printf("txs=%d nonzero_txs=%d total_bytes=%d avg=%.2f\n", len(stats), nonZero, total, avg(total, len(stats)))
	if len(values) > 0 {
		fmt.Printf("min=%d p50=%d p75=%d p90=%d p95=%d p99=%d max=%d\n",
			values[0],
			percentile(values, 0.50),
			percentile(values, 0.75),
			percentile(values, 0.90),
			percentile(values, 0.95),
			percentile(values, 0.99),
			values[len(values)-1],
		)
	}

	fmt.Println()
	fmt.Printf("%-16s %8s %8s\n", "bucket", "txs", "bytes")
	fmt.Printf("%-16s %8s %8s\n", "------", "---", "-----")
	for _, bucket := range innerDataBuckets(stats) {
		fmt.Printf("%-16s %8d %8d\n", bucket.label, bucket.count, bucket.bytes)
	}

	sort.Slice(stats, func(i, j int) bool {
		if stats[i].Bytes == stats[j].Bytes {
			return stats[i].TxOrdinal < stats[j].TxOrdinal
		}
		return stats[i].Bytes > stats[j].Bytes
	})
	fmt.Println()
	fmt.Printf("%-6s %-5s %-8s %-12s %-8s %-10s\n", "rank", "block", "tx_index", "slot", "bytes", "inner_ix")
	fmt.Printf("%-6s %-5s %-8s %-12s %-8s %-10s\n", "----", "-----", "--------", "----", "-----", "--------")
	limit := 20
	if len(stats) < limit {
		limit = len(stats)
	}
	for i := 0; i < limit; i++ {
		stat := stats[i]
		fmt.Printf("%-6d %-5d %-8d %-12d %-8d %-10d\n", i+1, stat.BlockIndex, stat.IndexWithinBlock, stat.Slot, stat.Bytes, stat.InstructionCount)
	}
}

func collectInnerInstructionDataStats(blocks *pump_parser.RawTxBlocksBinary) []txInnerDataStat {
	out := make([]txInnerDataStat, 0, len(blocks.Txs))
	txOffset := 0
	for blockIndex, count := range blocks.BlockTxCounts {
		for i := uint32(0); i < count; i++ {
			tx := &blocks.Txs[txOffset]
			var bytes uint64
			var instructionCount int
			for _, inner := range tx.Meta.InnerInstructions {
				for _, instruction := range inner.Instructions {
					bytes += uint64(len(instruction.Data))
					instructionCount++
				}
			}
			out = append(out, txInnerDataStat{
				TxOrdinal:        txOffset,
				BlockIndex:       blockIndex,
				IndexWithinBlock: tx.IndexWithinBlock,
				Slot:             tx.Slot,
				Bytes:            bytes,
				InstructionCount: instructionCount,
			})
			txOffset++
		}
	}
	return out
}

func percentile(values []uint64, p float64) uint64 {
	if len(values) == 0 {
		return 0
	}
	idx := int(float64(len(values)-1) * p)
	return values[idx]
}

func avg(total uint64, count int) float64 {
	if count == 0 {
		return 0
	}
	return float64(total) / float64(count)
}

type innerDataBucket struct {
	label string
	count int
	bytes uint64
}

func innerDataBuckets(stats []txInnerDataStat) []innerDataBucket {
	buckets := []innerDataBucket{
		{label: "0"},
		{label: "1-63"},
		{label: "64-127"},
		{label: "128-255"},
		{label: "256-511"},
		{label: "512-1023"},
		{label: "1024-2047"},
		{label: "2048-4095"},
		{label: "4096+"},
	}
	for _, stat := range stats {
		index := 0
		switch {
		case stat.Bytes == 0:
			index = 0
		case stat.Bytes < 64:
			index = 1
		case stat.Bytes < 128:
			index = 2
		case stat.Bytes < 256:
			index = 3
		case stat.Bytes < 512:
			index = 4
		case stat.Bytes < 1024:
			index = 5
		case stat.Bytes < 2048:
			index = 6
		case stat.Bytes < 4096:
			index = 7
		default:
			index = 8
		}
		buckets[index].count++
		buckets[index].bytes += stat.Bytes
	}
	return buckets
}

type balanceValueStats struct {
	name           string
	count          int
	unique         int
	zeroCount      int
	topValues      []balanceTopValue
	fixedBytes     uint64
	uvarintBytes   uint64
	duplicateCount int
}

type balanceTopValue struct {
	value uint64
	count int
}

type balancePairStats struct {
	txCount                 int
	pairCount               int
	lengthMismatchTxs       int
	unchangedCount          int
	changedCount            int
	currentFixedValueBytes  uint64
	bothUvarintBytes        uint64
	preUvarintPostDelta     uint64
	preUvarintChangedDeltas uint64
}

func printBalanceAnalysis(blocks *pump_parser.RawTxBlocksBinary) {
	preValues := make([]uint64, 0)
	postValues := make([]uint64, 0)
	pairs := balancePairStats{}
	pairs.txCount = len(blocks.Txs)

	for _, tx := range blocks.Txs {
		preValues = append(preValues, tx.Meta.PreBalances...)
		postValues = append(postValues, tx.Meta.PostBalances...)
		preLen := len(tx.Meta.PreBalances)
		postLen := len(tx.Meta.PostBalances)
		if preLen != postLen {
			pairs.lengthMismatchTxs++
		}
		n := preLen
		if postLen < n {
			n = postLen
		}
		pairs.currentFixedValueBytes += uint64(preLen+postLen) * 8
		pairs.preUvarintChangedDeltas += uint64(uvarintLen(uint64(n)))
		for i := 0; i < preLen; i++ {
			pairs.bothUvarintBytes += uint64(uvarintLen(tx.Meta.PreBalances[i]))
			pairs.preUvarintPostDelta += uint64(uvarintLen(tx.Meta.PreBalances[i]))
			pairs.preUvarintChangedDeltas += uint64(uvarintLen(tx.Meta.PreBalances[i]))
		}
		for i := 0; i < postLen; i++ {
			pairs.bothUvarintBytes += uint64(uvarintLen(tx.Meta.PostBalances[i]))
		}
		for i := 0; i < n; i++ {
			pre := tx.Meta.PreBalances[i]
			post := tx.Meta.PostBalances[i]
			pairs.pairCount++
			pairs.preUvarintPostDelta += uint64(zigzagDeltaUvarintLen(pre, post))
			if pre == post {
				pairs.unchangedCount++
				continue
			}
			pairs.changedCount++
			pairs.preUvarintChangedDeltas += uint64(uvarintLen(uint64(i)))
			pairs.preUvarintChangedDeltas += uint64(zigzagDeltaUvarintLen(pre, post))
		}
	}

	preStats := collectBalanceValueStats("pre_balances", preValues)
	postStats := collectBalanceValueStats("post_balances", postValues)
	combined := append(append([]uint64(nil), preValues...), postValues...)
	combinedStats := collectBalanceValueStats("pre+post_balances", combined)

	fmt.Println("balance_values_analysis")
	printBalanceValueStats(preStats)
	printBalanceValueStats(postStats)
	printBalanceValueStats(combinedStats)

	fmt.Println()
	fmt.Println("balance_encoding_estimates")
	fmt.Printf("txs=%d pairs=%d length_mismatch_txs=%d unchanged_pairs=%d changed_pairs=%d unchanged_pct=%.2f%%\n",
		pairs.txCount,
		pairs.pairCount,
		pairs.lengthMismatchTxs,
		pairs.unchangedCount,
		pairs.changedCount,
		float64(pairs.unchangedCount)*100/float64(maxInt(pairs.pairCount, 1)),
	)
	printEstimate("current_fixed_uint64_values", pairs.currentFixedValueBytes, pairs.currentFixedValueBytes)
	printEstimate("both_values_uvarint", pairs.bothUvarintBytes, pairs.currentFixedValueBytes)
	printEstimate("pre_uvarint_post_delta_each_index", pairs.preUvarintPostDelta, pairs.currentFixedValueBytes)
	printEstimate("pre_uvarint_post_changed_delta_pairs", pairs.preUvarintChangedDeltas, pairs.currentFixedValueBytes)
}

func collectBalanceValueStats(name string, values []uint64) balanceValueStats {
	freq := make(map[uint64]int)
	var zeroCount int
	var uvarintBytes uint64
	for _, value := range values {
		freq[value]++
		if value == 0 {
			zeroCount++
		}
		uvarintBytes += uint64(uvarintLen(value))
	}
	top := make([]balanceTopValue, 0, len(freq))
	var duplicateCount int
	for value, count := range freq {
		top = append(top, balanceTopValue{value: value, count: count})
		if count > 1 {
			duplicateCount += count - 1
		}
	}
	sort.Slice(top, func(i, j int) bool {
		if top[i].count == top[j].count {
			return top[i].value < top[j].value
		}
		return top[i].count > top[j].count
	})
	if len(top) > 10 {
		top = top[:10]
	}
	return balanceValueStats{
		name:           name,
		count:          len(values),
		unique:         len(freq),
		zeroCount:      zeroCount,
		topValues:      top,
		fixedBytes:     uint64(len(values)) * 8,
		uvarintBytes:   uvarintBytes,
		duplicateCount: duplicateCount,
	}
}

func printBalanceValueStats(stats balanceValueStats) {
	fmt.Printf("%s: count=%d unique=%d duplicate_values=%d zero=%d zero_pct=%.2f%% fixed_bytes=%d uvarint_bytes=%d uvarint_saved=%.2f%%\n",
		stats.name,
		stats.count,
		stats.unique,
		stats.duplicateCount,
		stats.zeroCount,
		float64(stats.zeroCount)*100/float64(maxInt(stats.count, 1)),
		stats.fixedBytes,
		stats.uvarintBytes,
		savedPct(stats.fixedBytes, stats.uvarintBytes),
	)
	fmt.Printf("%-22s %-8s %-8s\n", "value", "count", "pct")
	for _, item := range stats.topValues {
		fmt.Printf("%-22d %-8d %7.2f%%\n", item.value, item.count, float64(item.count)*100/float64(maxInt(stats.count, 1)))
	}
}

func printEstimate(name string, bytes uint64, baseline uint64) {
	fmt.Printf("%-38s %10d saved=%7.2f%%\n", name, bytes, savedPct(baseline, bytes))
}

func savedPct(baseline uint64, value uint64) float64 {
	if baseline == 0 {
		return 0
	}
	return (float64(baseline) - float64(value)) * 100 / float64(baseline)
}

func uvarintLen(value uint64) int {
	n := 1
	for value >= 0x80 {
		value >>= 7
		n++
	}
	return n
}

func zigzagDeltaUvarintLen(pre uint64, post uint64) int {
	if post >= pre {
		return uvarintLen((post - pre) << 1)
	}
	return uvarintLen(((pre - post) << 1) - 1)
}

func maxInt(a int, b int) int {
	if a > b {
		return a
	}
	return b
}