Week 4: Production Hardening — Thread Safety, Fuzz Tests, Property-Based Tests, NimForum Smoke Test
CI / test (push) Has been cancelled
CI / verify (push) Has been cancelled
Clients CI / build-server (push) Has been cancelled
Clients CI / test-python (push) Has been cancelled
Clients CI / test-javascript (push) Has been cancelled
Clients CI / test-nim (push) Has been cancelled
Clients CI / test-rust (push) Has been cancelled

Thread Safety:
- Add lock to BTreeIndex (insert/get/scan/remove/contains)
- Add lock to InvertedIndex (addDocument/removeDocument/search/bm25/tfidf/fuzzy/regex)
- Add ctxLock to ExecutionContext for DDL/schema metadata, autoIncCounters, sequences
- Deep-copy sessionVars in cloneForConnection() to isolate per-connection state
- Fix activeConnections check-then-act race with Lock

Wire Protocol Hardening:
- Fix RangeDefect in deserializeValue for invalid FieldKind bytes
- Fix IndexDefect in fkFloat32/fkFloat64 by reusing readUint32/readUint64
- Fix int32/int64 cast from uint32/uint64 to use cast[] instead of constructor
- Fix parseHeader to validate MsgKind instead of holey enum cast

Property-Based Tests (tests/prop_test.nim):
- 8 invariants for evalExprValue: literal types, commutativity, identity,
  double negation, NULL propagation

Wire Protocol Fuzz (tests/fuzz_test.nim):
- 7 fuzz tests: random bytes deserialization, truncated buffers,
  MsgKind casts, parseHeader logic, message roundtrips

NimForum Adapter Smoke Test (tests/nimforum_smoke_test.nim):
- Full TCP server lifecycle with adapter CRUD and parameterized queries

Test Results: 423 tests pass, 0 failures
This commit is contained in:
2026-05-17 11:57:55 +03:00
parent 6021bfcb10
commit 19fa760604
9 changed files with 676 additions and 241 deletions
+30 -5
View File
@@ -7,6 +7,7 @@ import std/tables
import std/os
import std/endians
import std/monotimes
import std/locks
import config
import logging
import ../protocol/wire
@@ -37,6 +38,7 @@ type
gossipProtocol*: GossipProtocol
tls*: TLSContext
activeConnections*: int
activeConnectionsLock*: Lock
proc newServer*(config: BaraConfig): Server =
let dataDir = config.dataDir / "server"
@@ -83,13 +85,14 @@ proc newServer*(config: BaraConfig): Server =
gp.onSuspect = proc(nodeId: string) {.gcsafe.} =
cm.onNodeSuspect(nodeId)
Server(config: config, running: false, db: db, ctx: ctx,
result = Server(config: config, running: false, db: db, ctx: ctx,
txnManager: ctx.txnManager, distTxnManager: newDistTxnManager(),
replicationManager: newReplicationManager(),
shardRouter: shardRouter,
clusterMembership: cm,
gossipProtocol: gp,
tls: tls)
initLock(result.activeConnectionsLock)
# ----------------------------------------------------------------------
# Wire Protocol Helpers
@@ -105,9 +108,13 @@ proc parseHeader(data: string): (bool, MessageHeader) =
if data.len < 12:
return (false, MessageHeader())
let rawKind = readUint32BE(data, 0)
{.push warning[HoleEnumConv]: off.}
let kind = MsgKind(rawKind)
{.pop.}
let kind = cast[MsgKind](rawKind)
case kind
of mkClientHandshake, mkQuery, mkQueryParams, mkExecute, mkBatch, mkTransaction, mkClose, mkPing, mkAuth,
mkServerHandshake, mkReady, mkData, mkComplete, mkError, mkAuthChallenge, mkAuthOk, mkSchemaChange, mkPong, mkTransactionState:
discard
else:
return (false, MessageHeader())
let length = readUint32BE(data, 4)
let requestId = readUint32BE(data, 8)
return (true, MessageHeader(kind: kind, length: length, requestId: requestId))
@@ -523,8 +530,12 @@ proc handleClient(server: Server, client: AsyncSocket, clientId: int) {.async.}
except Exception as e:
errorMsg("Client " & $clientId & " error: " & e.msg)
finally:
acquire(server.activeConnectionsLock)
try:
if server.activeConnections > 0:
dec server.activeConnections
finally:
release(server.activeConnectionsLock)
info("Client " & $clientId & " disconnected")
client.close()
@@ -547,7 +558,16 @@ proc run*(server: Server) {.async.} =
info("BaraDB listening on " & server.config.address & ":" & $server.config.port)
while server.running:
let client = await sock.accept()
acquire(server.activeConnectionsLock)
var shouldAccept = true
try:
if server.config.maxConnections > 0 and server.activeConnections >= server.config.maxConnections:
shouldAccept = false
else:
inc server.activeConnections
finally:
release(server.activeConnectionsLock)
if not shouldAccept:
client.close()
continue
if server.tls != nil:
@@ -555,10 +575,15 @@ proc run*(server: Server) {.async.} =
server.tls.wrapServer(client)
except Exception as e:
errorMsg("TLS handshake failed: " & e.msg)
acquire(server.activeConnectionsLock)
try:
if server.activeConnections > 0:
dec server.activeConnections
finally:
release(server.activeConnectionsLock)
client.close()
continue
inc clientId
inc server.activeConnections
asyncCheck server.handleClient(client, clientId)
proc stop*(server: Server) =
+60 -7
View File
@@ -4,6 +4,7 @@ import std/strutils
import std/unicode
import std/math
import std/algorithm
import std/locks
type
TermFreq* = Table[string, int]
@@ -20,6 +21,7 @@ type
docCount*: int
avgDocLen*: float64
totalTerms*: int
lock*: Lock
SearchResult* = object
docId*: uint64
@@ -111,16 +113,19 @@ proc tokenize*(text: string, config: TokenizerConfig = defaultTokenizerConfig())
result.add(token)
proc newInvertedIndex*(): InvertedIndex =
InvertedIndex(
result = InvertedIndex(
postings: initTable[string, seq[PostingEntry]](),
docLengths: initTable[uint64, int](),
docCount: 0,
avgDocLen: 0.0,
totalTerms: 0,
)
initLock(result.lock)
proc addDocument*(idx: InvertedIndex, docId: uint64, text: string,
config: TokenizerConfig = defaultTokenizerConfig()) =
acquire(idx.lock)
try:
let tokens = tokenize(text, config)
var termFreqs = initTable[string, int]()
var positions = initTable[string, seq[int]]()
@@ -145,8 +150,12 @@ proc addDocument*(idx: InvertedIndex, docId: uint64, text: string,
inc idx.docCount
idx.totalTerms += tokens.len
idx.avgDocLen = float64(idx.totalTerms) / float64(idx.docCount)
finally:
release(idx.lock)
proc removeDocument*(idx: InvertedIndex, docId: uint64) =
acquire(idx.lock)
try:
if docId notin idx.docLengths:
return
let docLen = idx.docLengths[docId]
@@ -162,8 +171,10 @@ proc removeDocument*(idx: InvertedIndex, docId: uint64) =
if entry.docId != docId:
newPostings.add(entry)
postings = newPostings
finally:
release(idx.lock)
proc bm25Score*(idx: InvertedIndex, term: string, docId: uint64,
proc bm25ScoreUnsafe(idx: InvertedIndex, term: string, docId: uint64,
k1: float64 = 1.2, b: float64 = 0.75): float64 =
if term notin idx.postings:
return 0.0
@@ -190,8 +201,18 @@ proc bm25Score*(idx: InvertedIndex, term: string, docId: uint64,
(float64(tf) + k1 * (1.0 - b + b * docLen / idx.avgDocLen))
return idf * tfNorm
proc bm25Score*(idx: InvertedIndex, term: string, docId: uint64,
k1: float64 = 1.2, b: float64 = 0.75): float64 =
acquire(idx.lock)
try:
result = bm25ScoreUnsafe(idx, term, docId, k1, b)
finally:
release(idx.lock)
proc search*(idx: InvertedIndex, query: string, limit: int = 10,
config: TokenizerConfig = defaultTokenizerConfig()): seq[SearchResult] =
acquire(idx.lock)
try:
let queryTokens = tokenize(query, config)
if queryTokens.len == 0:
return @[]
@@ -203,7 +224,7 @@ proc search*(idx: InvertedIndex, query: string, limit: int = 10,
if token notin idx.postings:
continue
for entry in idx.postings[token]:
let score = bm25Score(idx, token, entry.docId)
let score = bm25ScoreUnsafe(idx, token, entry.docId)
if entry.docId notin docScores:
docScores[entry.docId] = 0.0
docHighlights[entry.docId] = @[]
@@ -227,12 +248,25 @@ proc search*(idx: InvertedIndex, query: string, limit: int = 10,
results = results[0..<limit]
return results
finally:
release(idx.lock)
proc termCount*(idx: InvertedIndex): int = idx.postings.len
proc documentCount*(idx: InvertedIndex): int = idx.docCount
proc termCount*(idx: InvertedIndex): int =
acquire(idx.lock)
try:
result = idx.postings.len
finally:
release(idx.lock)
proc documentCount*(idx: InvertedIndex): int =
acquire(idx.lock)
try:
result = idx.docCount
finally:
release(idx.lock)
# TF-IDF ranking
proc tfidfScore*(idx: InvertedIndex, term: string, docId: uint64): float64 =
proc tfidfScoreUnsafe(idx: InvertedIndex, term: string, docId: uint64): float64 =
if term notin idx.postings:
return 0.0
let df = idx.postings[term].len
@@ -249,8 +283,17 @@ proc tfidfScore*(idx: InvertedIndex, term: string, docId: uint64): float64 =
let idf = ln(float64(n) / float64(df))
return float64(tf) * idf
proc tfidfScore*(idx: InvertedIndex, term: string, docId: uint64): float64 =
acquire(idx.lock)
try:
result = tfidfScoreUnsafe(idx, term, docId)
finally:
release(idx.lock)
proc searchTfidf*(idx: InvertedIndex, query: string, limit: int = 10,
config: TokenizerConfig = defaultTokenizerConfig()): seq[SearchResult] =
acquire(idx.lock)
try:
let queryTokens = tokenize(query, config)
if queryTokens.len == 0:
return @[]
@@ -261,7 +304,7 @@ proc searchTfidf*(idx: InvertedIndex, query: string, limit: int = 10,
if token notin idx.postings:
continue
for entry in idx.postings[token]:
let score = idx.tfidfScore(token, entry.docId)
let score = tfidfScoreUnsafe(idx, token, entry.docId)
if entry.docId notin docScores:
docScores[entry.docId] = 0.0
docScores[entry.docId] += score
@@ -274,6 +317,8 @@ proc searchTfidf*(idx: InvertedIndex, query: string, limit: int = 10,
if results.len > limit:
results = results[0..<limit]
return results
finally:
release(idx.lock)
# Levenshtein distance for fuzzy matching
proc levenshtein*(a, b: string): int =
@@ -295,6 +340,8 @@ proc levenshtein*(a, b: string): int =
proc fuzzySearch*(idx: InvertedIndex, query: string, maxDistance: int = 2,
limit: int = 10, config: TokenizerConfig = defaultTokenizerConfig()): seq[SearchResult] =
acquire(idx.lock)
try:
let queryTokens = tokenize(query, config)
if queryTokens.len == 0:
return @[]
@@ -319,10 +366,14 @@ proc fuzzySearch*(idx: InvertedIndex, query: string, maxDistance: int = 2,
if results.len > limit:
results = results[0..<limit]
return results
finally:
release(idx.lock)
# Regex search
proc regexSearch*(idx: InvertedIndex, pattern: string,
limit: int = 10): seq[SearchResult] =
acquire(idx.lock)
try:
var docScores = initTable[uint64, float64]()
for term in idx.postings.keys:
@@ -360,3 +411,5 @@ proc regexSearch*(idx: InvertedIndex, pattern: string,
if results.len > limit:
results = results[0..<limit]
return results
finally:
release(idx.lock)
+10 -15
View File
@@ -205,7 +205,10 @@ proc deserializeValue*(buf: openArray[byte], pos: var int, depth: int = 0): Wire
raise newException(ValueError, "Wire protocol: max deserialization depth exceeded")
if pos >= buf.len:
raise newException(ValueError, "Wire protocol: unexpected end of buffer")
let kind = FieldKind(buf[pos])
let kindByte = buf[pos]
if kindByte > byte(high(FieldKind)):
raise newException(ValueError, "Wire protocol: invalid field kind")
let kind = FieldKind(kindByte)
inc pos
case kind
of fkNull: result = WireValue(kind: fkNull)
@@ -223,26 +226,18 @@ proc deserializeValue*(buf: openArray[byte], pos: var int, depth: int = 0): Wire
pos += 2
result = WireValue(kind: fkInt16, int16Val: val)
of fkInt32:
result = WireValue(kind: fkInt32, int32Val: int32(readUint32(buf, pos)))
result = WireValue(kind: fkInt32, int32Val: cast[int32](readUint32(buf, pos)))
of fkInt64:
result = WireValue(kind: fkInt64, int64Val: int64(readUint64(buf, pos)))
result = WireValue(kind: fkInt64, int64Val: cast[int64](readUint64(buf, pos)))
of fkFloat32:
var fl: float32
var bytes: array[4, byte]
for i in 0..3: bytes[i] = buf[pos + i]
var i32: int32
bigEndian32(addr i32, unsafeAddr bytes)
fl = cast[float32](i32)
pos += 4
let raw = readUint32(buf, pos)
fl = cast[float32](raw)
result = WireValue(kind: fkFloat32, float32Val: fl)
of fkFloat64:
var fl: float64
var bytes: array[8, byte]
for i in 0..7: bytes[i] = buf[pos + i]
var i64: int64
bigEndian64(addr i64, unsafeAddr bytes)
fl = cast[float64](i64)
pos += 8
let raw = readUint64(buf, pos)
fl = cast[float64](raw)
result = WireValue(kind: fkFloat64, float64Val: fl)
of fkString:
result = WireValue(kind: fkString, strVal: readString(buf, pos))
+65 -13
View File
@@ -11,6 +11,7 @@ import std/times
import std/json
import std/random
import std/monotimes
import std/locks
import lexer as qlex
import parser as qpar
import ast
@@ -74,6 +75,7 @@ type
sessionVars*: Table[string, string] # session-scoped key/value variables
autoIncCounters*: Table[string, int64] # table.col -> next auto-increment value
sequences*: Table[string, int64] # sequence name -> current value
ctxLock*: Lock # protects tables, views, btrees, ftsIndexes, users, policies, autoIncCounters, sequences
MigrationRecord* = object
name*: string
@@ -160,6 +162,7 @@ proc newExecutionContext*(db: LSMTree): ExecutionContext =
autoIncCounters: initTable[string, int64](),
sequences: initTable[string, int64](),
onChange: nil)
initLock(result.ctxLock)
restoreSchema(result)
# ----------------------------------------------------------------------
@@ -332,7 +335,10 @@ proc restoreSchema(ctx: ExecutionContext) =
else: discard
proc cloneForConnection*(ctx: ExecutionContext): ExecutionContext =
ExecutionContext(db: ctx.db, tables: ctx.tables,
var svCopy = initTable[string, string]()
for k, v in ctx.sessionVars:
svCopy[k] = v
result = ExecutionContext(db: ctx.db, tables: ctx.tables,
btrees: ctx.btrees, views: ctx.views,
cteTables: initTable[string, seq[Row]](),
ftsIndexes: ctx.ftsIndexes,
@@ -340,10 +346,11 @@ proc cloneForConnection*(ctx: ExecutionContext): ExecutionContext =
users: ctx.users, policies: ctx.policies,
txnManager: ctx.txnManager,
currentUser: ctx.currentUser, currentRole: ctx.currentRole,
sessionVars: ctx.sessionVars,
sessionVars: svCopy,
autoIncCounters: ctx.autoIncCounters,
sequences: ctx.sequences,
pendingTxn: nil, onChange: ctx.onChange)
initLock(result.ctxLock)
# ----------------------------------------------------------------------
# Migration Helpers
@@ -1059,18 +1066,27 @@ proc evalExpr*(expr: IRExpr, row: Table[string, string], ctx: ExecutionContext =
if ctx == nil: return "0"
let seqName = evalExpr(expr.irFuncArgs[0], row, ctx)
var val: int64 = 0
acquire(ctx.ctxLock)
try:
if seqName in ctx.sequences:
val = ctx.sequences[seqName]
val += 1
ctx.sequences[seqName] = val
finally:
release(ctx.ctxLock)
return $val
of "currval":
if expr.irFuncArgs.len < 1:
return "0"
if ctx == nil: return "0"
let seqName = evalExpr(expr.irFuncArgs[0], row, ctx)
acquire(ctx.ctxLock)
try:
if seqName in ctx.sequences:
return $ctx.sequences[seqName]
finally:
release(ctx.ctxLock)
return "0"
return "0"
of "snowflake_id":
# Snowflake ID: timestamp_ms(41 bits) | node_id(10 bits) | sequence(12 bits)
@@ -1457,6 +1473,7 @@ proc validateType*(colType: string, value: string): (bool, string) =
return (false, "Vector dimension mismatch: expected " & $expectedDim & " but got " & $vec.len)
return (true, "")
proc executeQueryImpl(ctx: ExecutionContext, astNode: Node, params: seq[WireValue] = @[]): ExecResult
proc executeQuery*(ctx: ExecutionContext, astNode: Node, params: seq[WireValue] = @[]): ExecResult
proc executeMigrationSql(ctx: ExecutionContext, sql: string): ExecResult
@@ -1468,7 +1485,7 @@ proc fireTriggers*(ctx: ExecutionContext, tableName: string, timing: string, eve
let tokens = qlex.tokenize(trig.action.strVal)
let astNode = qpar.parse(tokens)
if astNode.stmts.len > 0:
discard executeQuery(ctx, astNode)
discard executeQueryImpl(ctx, astNode)
proc validateConstraints*(ctx: ExecutionContext, tableName: string,
fields: seq[string], values: seq[seq[string]], skipPkCheck: bool = false): (bool, string) =
@@ -3075,7 +3092,21 @@ proc getSelectColumns(stmt: Node): seq[string] =
else:
result.add("col" & $i)
proc executeQuery*(ctx: ExecutionContext, astNode: Node, params: seq[WireValue] = @[]): ExecResult =
proc isDDL(stmt: Node): bool =
case stmt.kind
of nkCreateTable, nkDropTable, nkAlterTable,
nkCreateView, nkDropView,
nkCreateIndex, nkDropIndex,
nkCreateTrigger, nkDropTrigger,
nkCreateUser, nkDropUser,
nkCreatePolicy, nkDropPolicy,
nkGrant, nkRevoke,
nkEnableRLS, nkDisableRLS:
result = true
else:
result = false
proc executeQueryImpl(ctx: ExecutionContext, astNode: Node, params: seq[WireValue] = @[]): ExecResult =
if astNode == nil or astNode.stmts.len == 0:
return okResult()
@@ -3109,7 +3140,7 @@ proc executeQuery*(ctx: ExecutionContext, astNode: Node, params: seq[WireValue]
# Step 1: Execute the non-recursive anchor (left side of UNION)
var innerLeft = Node(kind: nkStatementList, stmts: @[])
innerLeft.stmts.add(cteQuery.setOpLeft)
let anchorRes = executeQuery(ctx, innerLeft)
let anchorRes = executeQueryImpl(ctx, innerLeft)
for row in anchorRes.rows:
allRows.add(row)
@@ -3125,7 +3156,7 @@ proc executeQuery*(ctx: ExecutionContext, astNode: Node, params: seq[WireValue]
var innerRight = Node(kind: nkStatementList, stmts: @[])
innerRight.stmts.add(cteQuery.setOpRight)
let rightRes = executeQuery(ctx, innerRight)
let rightRes = executeQueryImpl(ctx, innerRight)
ctx.cteTables = savedCte
@@ -3159,7 +3190,7 @@ proc executeQuery*(ctx: ExecutionContext, astNode: Node, params: seq[WireValue]
# Recursive CTE without UNION — treat as non-recursive fallback
var inner = Node(kind: nkStatementList, stmts: @[])
inner.stmts.add(cteQuery)
let cteRes = executeQuery(ctx, inner)
let cteRes = executeQueryImpl(ctx, inner)
var cteRows: seq[Row] = @[]
for row in cteRes.rows:
cteRows.add(row)
@@ -3167,7 +3198,7 @@ proc executeQuery*(ctx: ExecutionContext, astNode: Node, params: seq[WireValue]
else:
var inner = Node(kind: nkStatementList, stmts: @[])
inner.stmts.add(cteQuery)
let cteRes = executeQuery(ctx, inner)
let cteRes = executeQueryImpl(ctx, inner)
var cteRows: seq[Row] = @[]
for row in cteRes.rows:
cteRows.add(row)
@@ -3180,7 +3211,7 @@ proc executeQuery*(ctx: ExecutionContext, astNode: Node, params: seq[WireValue]
# Execute the view's underlying query
var inner = Node(kind: nkStatementList, stmts: @[])
inner.stmts.add(viewQuery)
let innerResult = executeQuery(ctx, inner)
let innerResult = executeQueryImpl(ctx, inner)
# Now filter and project with outer query constraints
var filteredRows = innerResult.rows
var cols = innerResult.columns
@@ -3405,11 +3436,11 @@ proc executeQuery*(ctx: ExecutionContext, astNode: Node, params: seq[WireValue]
# Execute left and right queries
var innerLeft = Node(kind: nkStatementList, stmts: @[])
innerLeft.stmts.add(stmt.setOpLeft)
let leftRes = executeQuery(ctx, innerLeft)
let leftRes = executeQueryImpl(ctx, innerLeft)
var innerRight = Node(kind: nkStatementList, stmts: @[])
innerRight.stmts.add(stmt.setOpRight)
let rightRes = executeQuery(ctx, innerRight)
let rightRes = executeQueryImpl(ctx, innerRight)
# Derive columns from left side
var cols = leftRes.columns
@@ -3493,9 +3524,13 @@ proc executeQuery*(ctx: ExecutionContext, astNode: Node, params: seq[WireValue]
if col.autoIncrement and col.name notin mutableFields:
let counterKey = stmt.insTarget & "." & col.name
var nextVal: int64 = 1
acquire(ctx.ctxLock)
try:
if counterKey in ctx.autoIncCounters:
nextVal = ctx.autoIncCounters[counterKey]
ctx.autoIncCounters[counterKey] = nextVal + int64(mutableValues.len)
finally:
release(ctx.ctxLock)
# Insert at position 0 so it becomes the primary storage key
mutableFields.insert(col.name, 0)
for i in 0..<mutableValues.len:
@@ -3510,8 +3545,12 @@ proc executeQuery*(ctx: ExecutionContext, astNode: Node, params: seq[WireValue]
try:
let intVal = parseInt(providedVal)
let counterKey = stmt.insTarget & "." & col.name
acquire(ctx.ctxLock)
try:
if counterKey notin ctx.autoIncCounters or intVal >= ctx.autoIncCounters[counterKey]:
ctx.autoIncCounters[counterKey] = intVal + 1
finally:
release(ctx.ctxLock)
except: discard
applyDefaultValues(tbl, mutableFields, mutableValues)
@@ -3657,7 +3696,7 @@ proc executeQuery*(ctx: ExecutionContext, astNode: Node, params: seq[WireValue]
var sourceRows: seq[Row] = @[]
if stmt.mergeSource != nil:
if stmt.mergeSource.kind == nkSelect:
let srcRes = executeQuery(ctx, Node(kind: nkStatementList, stmts: @[stmt.mergeSource]))
let srcRes = executeQueryImpl(ctx, Node(kind: nkStatementList, stmts: @[stmt.mergeSource]))
sourceRows = srcRes.rows
elif stmt.mergeSource.kind == nkIdent:
sourceRows = execScan(ctx, stmt.mergeSource.identName)
@@ -4236,9 +4275,22 @@ proc executeQuery*(ctx: ExecutionContext, astNode: Node, params: seq[WireValue]
return errResult("Unsupported statement type: " & $stmt.kind)
proc executeQuery*(ctx: ExecutionContext, astNode: Node, params: seq[WireValue] = @[]): ExecResult =
if astNode == nil or astNode.stmts.len == 0:
return okResult()
let stmt = astNode.stmts[0]
if isDDL(stmt):
acquire(ctx.ctxLock)
try:
result = executeQueryImpl(ctx, astNode, params)
finally:
release(ctx.ctxLock)
else:
result = executeQueryImpl(ctx, astNode, params)
proc executeMigrationSql(ctx: ExecutionContext, sql: string): ExecResult =
let tokens = qlex.tokenize(sql)
let astNode = qpar.parse(tokens)
if astNode.stmts.len > 0:
return executeQuery(ctx, astNode)
return executeQueryImpl(ctx, astNode)
return okResult(msg="Empty migration body")
+32 -4
View File
@@ -1,5 +1,6 @@
## B-Tree Index — ordered key-value index
import std/tables
import std/locks
const
DefaultBTreeOrder* = 32
@@ -16,6 +17,7 @@ type
root: BTreeNode[K, V]
order: int
size: int
lock*: Lock
proc newBTreeNode[K, V](isLeaf: bool = true): BTreeNode[K, V] =
BTreeNode[K, V](
@@ -24,7 +26,8 @@ proc newBTreeNode[K, V](isLeaf: bool = true): BTreeNode[K, V] =
)
proc newBTreeIndex*[K, V](order: int = DefaultBTreeOrder): BTreeIndex[K, V] =
BTreeIndex[K, V](root: newBTreeNode[K, V](), order: order, size: 0)
result = BTreeIndex[K, V](root: newBTreeNode[K, V](), order: order, size: 0)
initLock(result.lock)
proc search[K, V](node: BTreeNode[K, V], key: K): seq[V] =
var i = 0
@@ -84,6 +87,8 @@ proc insertNonFull[K, V](node: BTreeNode[K, V], key: K, value: V, order: int) =
insertNonFull(node.children[i], key, value, order)
proc insert*[K, V](btree: var BTreeIndex[K, V], key: K, value: V) =
acquire(btree.lock)
try:
if btree.root.keys.len == btree.order - 1:
var newRoot = newBTreeNode[K, V](isLeaf = false)
newRoot.children.add(btree.root)
@@ -93,14 +98,26 @@ proc insert*[K, V](btree: var BTreeIndex[K, V], key: K, value: V) =
else:
insertNonFull(btree.root, key, value, btree.order)
inc btree.size
finally:
release(btree.lock)
proc get*[K, V](btree: BTreeIndex[K, V], key: K): seq[V] =
search(btree.root, key)
acquire(btree.lock)
try:
result = search(btree.root, key)
finally:
release(btree.lock)
proc contains*[K, V](btree: BTreeIndex[K, V], key: K): bool =
return btree.get(key).len > 0
acquire(btree.lock)
try:
result = search(btree.root, key).len > 0
finally:
release(btree.lock)
proc scan*[K, V](btree: BTreeIndex[K, V], startKey, endKey: K): seq[(K, seq[V])] =
acquire(btree.lock)
try:
result = @[]
var node = btree.root
while not node.isLeaf:
@@ -117,10 +134,19 @@ proc scan*[K, V](btree: BTreeIndex[K, V], startKey, endKey: K): seq[(K, seq[V])]
else:
return
node = node.next
finally:
release(btree.lock)
proc len*[K, V](btree: BTreeIndex[K, V]): int = btree.size
proc len*[K, V](btree: BTreeIndex[K, V]): int =
acquire(btree.lock)
try:
result = btree.size
finally:
release(btree.lock)
proc remove*[K, V](btree: var BTreeIndex[K, V], key: K, value: V) =
acquire(btree.lock)
try:
proc removeRec(node: BTreeNode[K, V]): bool =
var i = 0
while i < node.keys.len and key > node.keys[i]:
@@ -147,3 +173,5 @@ proc remove*[K, V](btree: var BTreeIndex[K, V], key: K, value: V) =
if removeRec(btree.root):
dec btree.size
finally:
release(btree.lock)
+87
View File
@@ -0,0 +1,87 @@
## Wire Protocol Fuzz Tests
import std/unittest
import std/random
import std/strutils
import barabadb/protocol/wire
suite "Wire Protocol Fuzz":
test "deserializeValue survives random bytes":
var rng = initRand(12345)
for i in 0..<500:
var buf = newSeq[byte](rng.rand(1..256))
for b in buf.mitems:
b = byte(rng.rand(0..255))
var pos = 0
try:
discard deserializeValue(buf, pos)
except:
discard # Exceptions are expected for garbage input
test "deserializeValue survives truncated buffers":
var rng = initRand(12346)
for i in 0..<200:
# Start with a valid-ish prefix then truncate
var buf: seq[byte] = @[byte(rng.rand(0..12))] # random FieldKind
let extra = rng.rand(0..64)
for j in 0..<extra:
buf.add(byte(rng.rand(0..255)))
var pos = 0
try:
discard deserializeValue(buf, pos)
except:
discard
test "MsgKind cast with random uint32 values":
var rng = initRand(12347)
for i in 0..<1000:
let raw = uint32(rng.rand(int64(high(uint32))))
let kind = cast[MsgKind](raw)
# Should not crash; cast is unsafe but does not range-check
check true
test "parseHeader-like logic with random 12-byte chunks":
var rng = initRand(12348)
for i in 0..<500:
var data = ""
for j in 0..<12:
data.add(char(rng.rand(0..255)))
let rawKind = uint32(ord(data[0])) shl 24 or uint32(ord(data[1])) shl 16 or
uint32(ord(data[2])) shl 8 or uint32(ord(data[3]))
let kind = cast[MsgKind](rawKind)
let length = uint32(ord(data[4])) shl 24 or uint32(ord(data[5])) shl 16 or
uint32(ord(data[6])) shl 8 or uint32(ord(data[7]))
let requestId = uint32(ord(data[8])) shl 24 or uint32(ord(data[9])) shl 16 or
uint32(ord(data[10])) shl 8 or uint32(ord(data[11]))
let header = MessageHeader(kind: kind, length: length, requestId: requestId)
check header.length == length
test "makeQueryMessage roundtrip with random queries":
var rng = initRand(12349)
for i in 0..<100:
var query = ""
let len = rng.rand(0..200)
for j in 0..<len:
query.add(char(rng.rand(32..126)))
let msg = makeQueryMessage(uint32(i), query)
check msg.len >= 12
test "serializeMessage roundtrip with random payloads":
var rng = initRand(12350)
for i in 0..<100:
let reqId = uint32(rng.rand(int64(high(uint32))))
var msg = WireMessage(
header: MessageHeader(kind: mkQuery, length: uint32(rng.rand(0..1000)), requestId: reqId),
payload: newSeq[byte](rng.rand(0..200))
)
for j in 0..<msg.payload.len:
msg.payload[j] = byte(rng.rand(0..255))
let s = serializeMessage(msg)
check s.len >= 12
test "makeErrorMessage with large IDs":
var rng = initRand(12351)
for i in 0..<100:
let reqId = uint32(rng.rand(int64(high(uint32))))
let err = makeErrorMessage(reqId, uint32(rng.rand(0..255)), "fuzz")
check err.len > 0
+69
View File
@@ -0,0 +1,69 @@
## NimForum Adapter Smoke Test
import std/unittest
import std/osproc
import std/os
import std/strutils
import std/strtabs
import std/times
import ../adaptors/nim/baradb_sqlite as sqlite
suite "NimForum Adapter Smoke Test":
var serverProcess: Process
var port: int
var dataDir: string
setup:
port = 35000 + (getTime().toUnix.int mod 10000)
dataDir = getTempDir() / "baradb_nimforum_" & $port
createDir(dataDir)
var env = newStringTable()
for key, val in envPairs():
env[key] = val
env["BARADB_PORT"] = $port
env["BARADB_DATA_DIR"] = dataDir
env["BARADB_LOG_LEVEL"] = "error"
serverProcess = startProcess("./build/baradadb", env=env, options={poStdErrToStdOut, poDaemon})
sleep(800)
teardown:
if serverProcess != nil:
serverProcess.terminate()
discard serverProcess.waitForExit()
removeDir(dataDir)
test "Adapter basic CRUD over TCP":
var db = open("127.0.0.1:" & $port, "", "", "default")
db.exec(sql"CREATE TABLE nf_test (id INT PRIMARY KEY, name STRING)")
db.exec(sql"INSERT INTO nf_test (id, name) VALUES (1, 'hello')")
db.exec(sql"INSERT INTO nf_test (id, name) VALUES (2, 'world')")
let rows = db.getAllRows(sql"SELECT * FROM nf_test")
check rows.len == 2
let row = db.getRow(sql"SELECT * FROM nf_test WHERE id = 1")
check row.len == 2
check row[0] == "1"
check row[1] == "hello"
let val = db.getValue(sql"SELECT name FROM nf_test WHERE id = 2")
check val == "world"
let cnt = db.getValue(sql"SELECT count(*) FROM nf_test")
check cnt == "2"
db.close()
test "Adapter parameterized queries":
var db = open("127.0.0.1:" & $port, "", "", "default")
db.exec(sql"CREATE TABLE nf_params (id INT PRIMARY KEY, val STRING)")
db.exec(sql"INSERT INTO nf_params (id, val) VALUES (?, ?)", 10, "ten")
db.exec(sql"INSERT INTO nf_params (id, val) VALUES (?, ?)", 20, "twenty")
let row = db.getRow(sql"SELECT val FROM nf_params WHERE id = ?", 10)
check row.len >= 1
check row[row.len - 1] == "ten"
db.close()
+123
View File
@@ -0,0 +1,123 @@
## Property-Based Tests — evalExprValue invariants
import std/unittest
import std/tables
import std/random
import std/os
import std/monotimes
import std/math
import barabadb/core/types
import barabadb/storage/lsm
import barabadb/query/ir as qir
import barabadb/query/executor as qexec
suite "Property-Based — evalExprValue Invariants":
setup:
var testDir = getTempDir() / "baradb_prop_test_" & $getCurrentProcessId() & "_" & $getMonoTime().ticks
createDir(testDir)
var db = newLSMTree(testDir)
var ctx = qexec.newExecutionContext(db)
teardown:
removeDir(testDir)
proc randIntLit(rng: var Rand, minVal: int = -1000, maxVal: int = 1000): IRExpr =
result = IRExpr(kind: irekLiteral, valueKind: vkInt64)
result.literal = IRLiteral(kind: vkInt64, int64Val: int64(rng.rand(minVal..maxVal)))
proc randFloatLit(rng: var Rand, minVal: float = -1000.0, maxVal: float = 1000.0): IRExpr =
result = IRExpr(kind: irekLiteral, valueKind: vkFloat64)
result.literal = IRLiteral(kind: vkFloat64, float64Val: minVal + rng.rand(maxVal - minVal))
proc randBinaryExpr(rng: var Rand, left, right: IRExpr, op: IROperator): IRExpr =
result = IRExpr(kind: irekBinary)
result.binOp = op
result.binLeft = left
result.binRight = right
if left.valueKind == vkInt64 and right.valueKind == vkInt64 and op != irDiv:
result.valueKind = vkInt64
else:
result.valueKind = vkFloat64
proc randUnaryExpr(rng: var Rand, expr: IRExpr, op: IROperator): IRExpr =
result = IRExpr(kind: irekUnary)
result.unOp = op
result.unExpr = expr
result.valueKind = expr.valueKind
test "Literal eval returns correct ValueKind (INT)":
var rng = initRand(42)
for i in 0..<100:
let lit = randIntLit(rng)
let v = evalExprValue(lit, initTable[string, string](), nil)
check v.kind == vkInt64
test "Literal eval returns correct ValueKind (FLOAT)":
var rng = initRand(43)
for i in 0..<100:
let lit = randFloatLit(rng)
let v = evalExprValue(lit, initTable[string, string](), nil)
check v.kind == vkFloat64
test "INT addition is commutative":
var rng = initRand(44)
for i in 0..<100:
let a = randIntLit(rng)
let b = randIntLit(rng)
let sum1 = evalExprValue(randBinaryExpr(rng, a, b, irAdd), initTable[string, string](), nil)
let sum2 = evalExprValue(randBinaryExpr(rng, b, a, irAdd), initTable[string, string](), nil)
if sum1.kind == vkInt64 and sum2.kind == vkInt64:
check sum1.int64Val == sum2.int64Val
test "FLOAT addition is commutative":
var rng = initRand(45)
for i in 0..<100:
let a = randFloatLit(rng)
let b = randFloatLit(rng)
let sum1 = evalExprValue(randBinaryExpr(rng, a, b, irAdd), initTable[string, string](), nil)
let sum2 = evalExprValue(randBinaryExpr(rng, b, a, irAdd), initTable[string, string](), nil)
if sum1.kind == vkFloat64 and sum2.kind == vkFloat64:
check abs(sum1.float64Val - sum2.float64Val) < 1e-9
test "INT multiplication by 1 is identity":
var rng = initRand(46)
for i in 0..<100:
let a = randIntLit(rng)
let one = IRExpr(kind: irekLiteral, valueKind: vkInt64)
one.literal = IRLiteral(kind: vkInt64, int64Val: 1)
let prod = evalExprValue(randBinaryExpr(rng, a, one, irMul), initTable[string, string](), nil)
if prod.kind == vkInt64:
check prod.int64Val == a.literal.int64Val
test "Double negation of INT is identity":
var rng = initRand(47)
for i in 0..<100:
let a = randIntLit(rng)
let neg = randUnaryExpr(rng, a, irNeg)
let negNeg = randUnaryExpr(rng, neg, irNeg)
let v = evalExprValue(negNeg, initTable[string, string](), nil)
if v.kind == vkInt64:
check v.int64Val == a.literal.int64Val
test "Double negation of FLOAT is identity":
var rng = initRand(48)
for i in 0..<100:
let a = randFloatLit(rng)
let neg = randUnaryExpr(rng, a, irNeg)
let negNeg = randUnaryExpr(rng, neg, irNeg)
let v = evalExprValue(negNeg, initTable[string, string](), nil)
if v.kind == vkFloat64:
check abs(v.float64Val - a.literal.float64Val) < 1e-9
test "NULL literal propagates through arithmetic":
let nullLit = IRExpr(kind: irekLiteral, valueKind: vkNull)
nullLit.literal = IRLiteral(kind: vkNull)
let intLit = IRExpr(kind: irekLiteral, valueKind: vkInt64)
intLit.literal = IRLiteral(kind: vkInt64, int64Val: 5)
for op in [irAdd, irSub, irMul, irDiv]:
let expr = IRExpr(kind: irekBinary)
expr.binOp = op
expr.binLeft = nullLit
expr.binRight = intLit
let v = evalExprValue(expr, initTable[string, string](), nil)
check v.kind == vkNull
+3
View File
@@ -6,6 +6,7 @@ import std/os
import std/asyncdispatch
import std/monotimes
import std/base64
import std/random
import barabadb/core/types
import barabadb/core/mvcc
@@ -3728,3 +3729,5 @@ suite "Join Performance — Hash Join & Index Nested Loop":
check r.rows[0]["name"] == "Alice"
check r.rows[3]["name"] == "Carol"
check r.rows[3]["total"] == "\\N"