| 123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261 |
- package btree
- import (
- "TsDB2/DB/core/page"
- "fmt"
- )
- // ValidateBTree validates all B+ tree invariants
- func ValidateBTree(rootPID uint32, handle page.HandleDb, isPrimary bool) error {
- if rootPID == 0 {
- return fmt.Errorf("empty tree")
- }
- validator := &btreeValidator{
- handle: handle,
- isPrimary: isPrimary,
- errors: []string{},
- }
- // 1. Validate root node
- root := validator.loadNode(rootPID)
- if root == nil {
- return fmt.Errorf("failed to load root page %d", rootPID)
- }
- // 2. Traverse and validate all nodes
- var leafKeys [][]byte
- validator.validateNode(root, nil, nil, &leafKeys)
- // 3. Validate leaf linked list
- validator.validateLeafList(leafKeys)
- if len(validator.errors) > 0 {
- return fmt.Errorf("B+ tree validation failed:\n%s",
- formatErrors(validator.errors))
- }
- return nil
- }
- type btreeValidator struct {
- handle page.HandleDb
- isPrimary bool
- errors []string
- }
- func (v *btreeValidator) loadNode(pid uint32) *btNode {
- if v.isPrimary {
- return btLoad(v.handle, pid)
- }
- // For secondary/string, we need a generic loader
- // For now, assume primary format
- return btLoad(v.handle, pid)
- }
- func (v *btreeValidator) validateNode(n *btNode, minKey, maxKey []byte, leafKeys *[][]byte) {
- if n == nil {
- v.errors = append(v.errors, "nil node encountered")
- return
- }
- el := int(n.info.EntryLen)
- cnt := int(n.info.Count)
- if cnt == 0 {
- v.errors = append(v.errors,
- fmt.Sprintf("page %d has 0 entries", n.pageID()))
- return
- }
- if n.isLeaf() {
- v.validateLeaf(n, minKey, maxKey, el, cnt, leafKeys)
- } else {
- v.validateInternal(n, minKey, maxKey, el, cnt, leafKeys)
- }
- }
- func (v *btreeValidator) validateLeaf(n *btNode, minKey, maxKey []byte, el, cnt int, leafKeys *[][]byte) {
- pid := n.pageID()
- for i := 0; i < cnt; i++ {
- entry := btEntry(n, i)
- key := btKey(entry)
- // Check ordering within leaf
- if i > 0 {
- prevKey := btKey(btEntry(n, i-1))
- if btCmp(prevKey, key) >= 0 {
- v.errors = append(v.errors,
- fmt.Sprintf("leaf page %d: entries not sorted at pos %d (%s >= %s)",
- pid, i, string(prevKey), string(key)))
- }
- }
- // Check range constraints
- if minKey != nil && btCmp(key, minKey) < 0 {
- v.errors = append(v.errors,
- fmt.Sprintf("leaf page %d: key %s < minKey %s at pos %d",
- pid, string(key), string(minKey), i))
- }
- if maxKey != nil && btCmp(key, maxKey) >= 0 {
- v.errors = append(v.errors,
- fmt.Sprintf("leaf page %d: key %s >= maxKey %s at pos %d",
- pid, string(key), string(maxKey), i))
- }
- *leafKeys = append(*leafKeys, key)
- }
- }
- func (v *btreeValidator) validateInternal(n *btNode, minKey, maxKey []byte, el, cnt int, leafKeys *[][]byte) {
- pid := n.pageID()
- for i := 0; i < cnt; i++ {
- entry := btEntry(n, i)
- sepKey := btKey(entry)
- childPID := btPID(entry)
- // Load child node
- child := v.loadNode(childPID)
- if child == nil {
- v.errors = append(v.errors,
- fmt.Sprintf("internal page %d: failed to load child %d at pos %d",
- pid, childPID, i))
- continue
- }
- // Determine child's key range
- var childMin, childMax []byte
- if i == 0 {
- childMin = minKey
- } else {
- // Child's min should be >= previous separator
- prevSep := btKey(btEntry(n, i-1))
- childMin = prevSep
- }
- if i == cnt-1 {
- childMax = maxKey
- } else {
- // Child's max should be < next separator
- nextSep := sepKey
- childMax = nextSep
- }
- // Validate separator ordering
- if i > 0 {
- prevSep := btKey(btEntry(n, i-1))
- if btCmp(prevSep, sepKey) >= 0 {
- v.errors = append(v.errors,
- fmt.Sprintf("internal page %d: separators not sorted at pos %d (%s >= %s)",
- pid, i, string(prevSep), string(sepKey)))
- }
- }
- // Recursively validate child
- v.validateNode(child, childMin, childMax, leafKeys)
- }
- }
- func (v *btreeValidator) validateLeafList(allKeys [][]byte) {
- // Check that all keys are globally sorted
- for i := 1; i < len(allKeys); i++ {
- if btCmp(allKeys[i-1], allKeys[i]) >= 0 {
- v.errors = append(v.errors,
- fmt.Sprintf("global key order violation: %s >= %s",
- string(allKeys[i-1]), string(allKeys[i])))
- }
- }
- }
- func formatErrors(errors []string) string {
- result := ""
- for i, err := range errors {
- if i > 0 {
- result += "\n"
- }
- result += fmt.Sprintf(" [%d] %s", i+1, err)
- }
- return result
- }
- // DumpBTree prints the B+ tree structure for debugging
- func DumpBTree(rootPID uint32, handle page.HandleDb, isPrimary bool, maxDepth int) {
- if rootPID == 0 {
- fmt.Println("Empty tree")
- return
- }
- dumper := &btreeDumper{
- handle: handle,
- isPrimary: isPrimary,
- maxDepth: maxDepth,
- }
- fmt.Println("=== B+ Tree Structure ===")
- dumper.dumpNode(rootPID, 0)
- fmt.Println("=========================")
- }
- type btreeDumper struct {
- handle page.HandleDb
- isPrimary bool
- maxDepth int
- }
- func (d *btreeDumper) dumpNode(pid uint32, depth int) {
- if depth > d.maxDepth {
- fmt.Printf("%s... (max depth reached)\n", indent(depth))
- return
- }
- n := btLoad(d.handle, pid)
- if n == nil {
- fmt.Printf("%sPage %d: FAILED TO LOAD\n", indent(depth), pid)
- return
- }
- el := int(n.info.EntryLen)
- cnt := int(n.info.Count)
- nodeType := "LEAF"
- if !n.isLeaf() {
- nodeType = "INTERNAL"
- }
- fmt.Printf("%sPage %d [%s] count=%d entryLen=%d\n",
- indent(depth), pid, nodeType, cnt, el)
- if n.isLeaf() {
- for i := 0; i < cnt && i < 5; i++ { // Show first 5 entries
- entry := btEntry(n, i)
- key := btKey(entry)
- dataPID := btPID(entry)
- fmt.Printf("%s [%d] key='%s' dataPage=%d\n",
- indent(depth+1), i, string(key), dataPID)
- }
- if cnt > 5 {
- fmt.Printf("%s ... and %d more entries\n", indent(depth+1), cnt-5)
- }
- } else {
- for i := 0; i < cnt && i < 5; i++ { // Show first 5 children
- entry := btEntry(n, i)
- sepKey := btKey(entry)
- childPID := btPID(entry)
- fmt.Printf("%s [%d] sep='%s' -> child=%d\n",
- indent(depth+1), i, string(sepKey), childPID)
- d.dumpNode(childPID, depth+1)
- }
- if cnt > 5 {
- fmt.Printf("%s ... and %d more children\n", indent(depth+1), cnt-5)
- }
- }
- }
- func indent(depth int) string {
- s := ""
- for i := 0; i < depth; i++ {
- s += " "
- }
- return s
- }
|