validate.go 6.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261
  1. package btree
  2. import (
  3. "TsDB2/DB/core/page"
  4. "fmt"
  5. )
  6. // ValidateBTree validates all B+ tree invariants
  7. func ValidateBTree(rootPID uint32, handle page.HandleDb, isPrimary bool) error {
  8. if rootPID == 0 {
  9. return fmt.Errorf("empty tree")
  10. }
  11. validator := &btreeValidator{
  12. handle: handle,
  13. isPrimary: isPrimary,
  14. errors: []string{},
  15. }
  16. // 1. Validate root node
  17. root := validator.loadNode(rootPID)
  18. if root == nil {
  19. return fmt.Errorf("failed to load root page %d", rootPID)
  20. }
  21. // 2. Traverse and validate all nodes
  22. var leafKeys [][]byte
  23. validator.validateNode(root, nil, nil, &leafKeys)
  24. // 3. Validate leaf linked list
  25. validator.validateLeafList(leafKeys)
  26. if len(validator.errors) > 0 {
  27. return fmt.Errorf("B+ tree validation failed:\n%s",
  28. formatErrors(validator.errors))
  29. }
  30. return nil
  31. }
  32. type btreeValidator struct {
  33. handle page.HandleDb
  34. isPrimary bool
  35. errors []string
  36. }
  37. func (v *btreeValidator) loadNode(pid uint32) *btNode {
  38. if v.isPrimary {
  39. return btLoad(v.handle, pid)
  40. }
  41. // For secondary/string, we need a generic loader
  42. // For now, assume primary format
  43. return btLoad(v.handle, pid)
  44. }
  45. func (v *btreeValidator) validateNode(n *btNode, minKey, maxKey []byte, leafKeys *[][]byte) {
  46. if n == nil {
  47. v.errors = append(v.errors, "nil node encountered")
  48. return
  49. }
  50. el := int(n.info.EntryLen)
  51. cnt := int(n.info.Count)
  52. if cnt == 0 {
  53. v.errors = append(v.errors,
  54. fmt.Sprintf("page %d has 0 entries", n.pageID()))
  55. return
  56. }
  57. if n.isLeaf() {
  58. v.validateLeaf(n, minKey, maxKey, el, cnt, leafKeys)
  59. } else {
  60. v.validateInternal(n, minKey, maxKey, el, cnt, leafKeys)
  61. }
  62. }
  63. func (v *btreeValidator) validateLeaf(n *btNode, minKey, maxKey []byte, el, cnt int, leafKeys *[][]byte) {
  64. pid := n.pageID()
  65. for i := 0; i < cnt; i++ {
  66. entry := btEntry(n, i)
  67. key := btKey(entry)
  68. // Check ordering within leaf
  69. if i > 0 {
  70. prevKey := btKey(btEntry(n, i-1))
  71. if btCmp(prevKey, key) >= 0 {
  72. v.errors = append(v.errors,
  73. fmt.Sprintf("leaf page %d: entries not sorted at pos %d (%s >= %s)",
  74. pid, i, string(prevKey), string(key)))
  75. }
  76. }
  77. // Check range constraints
  78. if minKey != nil && btCmp(key, minKey) < 0 {
  79. v.errors = append(v.errors,
  80. fmt.Sprintf("leaf page %d: key %s < minKey %s at pos %d",
  81. pid, string(key), string(minKey), i))
  82. }
  83. if maxKey != nil && btCmp(key, maxKey) >= 0 {
  84. v.errors = append(v.errors,
  85. fmt.Sprintf("leaf page %d: key %s >= maxKey %s at pos %d",
  86. pid, string(key), string(maxKey), i))
  87. }
  88. *leafKeys = append(*leafKeys, key)
  89. }
  90. }
  91. func (v *btreeValidator) validateInternal(n *btNode, minKey, maxKey []byte, el, cnt int, leafKeys *[][]byte) {
  92. pid := n.pageID()
  93. for i := 0; i < cnt; i++ {
  94. entry := btEntry(n, i)
  95. sepKey := btKey(entry)
  96. childPID := btPID(entry)
  97. // Load child node
  98. child := v.loadNode(childPID)
  99. if child == nil {
  100. v.errors = append(v.errors,
  101. fmt.Sprintf("internal page %d: failed to load child %d at pos %d",
  102. pid, childPID, i))
  103. continue
  104. }
  105. // Determine child's key range
  106. var childMin, childMax []byte
  107. if i == 0 {
  108. childMin = minKey
  109. } else {
  110. // Child's min should be >= previous separator
  111. prevSep := btKey(btEntry(n, i-1))
  112. childMin = prevSep
  113. }
  114. if i == cnt-1 {
  115. childMax = maxKey
  116. } else {
  117. // Child's max should be < next separator
  118. nextSep := sepKey
  119. childMax = nextSep
  120. }
  121. // Validate separator ordering
  122. if i > 0 {
  123. prevSep := btKey(btEntry(n, i-1))
  124. if btCmp(prevSep, sepKey) >= 0 {
  125. v.errors = append(v.errors,
  126. fmt.Sprintf("internal page %d: separators not sorted at pos %d (%s >= %s)",
  127. pid, i, string(prevSep), string(sepKey)))
  128. }
  129. }
  130. // Recursively validate child
  131. v.validateNode(child, childMin, childMax, leafKeys)
  132. }
  133. }
  134. func (v *btreeValidator) validateLeafList(allKeys [][]byte) {
  135. // Check that all keys are globally sorted
  136. for i := 1; i < len(allKeys); i++ {
  137. if btCmp(allKeys[i-1], allKeys[i]) >= 0 {
  138. v.errors = append(v.errors,
  139. fmt.Sprintf("global key order violation: %s >= %s",
  140. string(allKeys[i-1]), string(allKeys[i])))
  141. }
  142. }
  143. }
  144. func formatErrors(errors []string) string {
  145. result := ""
  146. for i, err := range errors {
  147. if i > 0 {
  148. result += "\n"
  149. }
  150. result += fmt.Sprintf(" [%d] %s", i+1, err)
  151. }
  152. return result
  153. }
  154. // DumpBTree prints the B+ tree structure for debugging
  155. func DumpBTree(rootPID uint32, handle page.HandleDb, isPrimary bool, maxDepth int) {
  156. if rootPID == 0 {
  157. fmt.Println("Empty tree")
  158. return
  159. }
  160. dumper := &btreeDumper{
  161. handle: handle,
  162. isPrimary: isPrimary,
  163. maxDepth: maxDepth,
  164. }
  165. fmt.Println("=== B+ Tree Structure ===")
  166. dumper.dumpNode(rootPID, 0)
  167. fmt.Println("=========================")
  168. }
  169. type btreeDumper struct {
  170. handle page.HandleDb
  171. isPrimary bool
  172. maxDepth int
  173. }
  174. func (d *btreeDumper) dumpNode(pid uint32, depth int) {
  175. if depth > d.maxDepth {
  176. fmt.Printf("%s... (max depth reached)\n", indent(depth))
  177. return
  178. }
  179. n := btLoad(d.handle, pid)
  180. if n == nil {
  181. fmt.Printf("%sPage %d: FAILED TO LOAD\n", indent(depth), pid)
  182. return
  183. }
  184. el := int(n.info.EntryLen)
  185. cnt := int(n.info.Count)
  186. nodeType := "LEAF"
  187. if !n.isLeaf() {
  188. nodeType = "INTERNAL"
  189. }
  190. fmt.Printf("%sPage %d [%s] count=%d entryLen=%d\n",
  191. indent(depth), pid, nodeType, cnt, el)
  192. if n.isLeaf() {
  193. for i := 0; i < cnt && i < 5; i++ { // Show first 5 entries
  194. entry := btEntry(n, i)
  195. key := btKey(entry)
  196. dataPID := btPID(entry)
  197. fmt.Printf("%s [%d] key='%s' dataPage=%d\n",
  198. indent(depth+1), i, string(key), dataPID)
  199. }
  200. if cnt > 5 {
  201. fmt.Printf("%s ... and %d more entries\n", indent(depth+1), cnt-5)
  202. }
  203. } else {
  204. for i := 0; i < cnt && i < 5; i++ { // Show first 5 children
  205. entry := btEntry(n, i)
  206. sepKey := btKey(entry)
  207. childPID := btPID(entry)
  208. fmt.Printf("%s [%d] sep='%s' -> child=%d\n",
  209. indent(depth+1), i, string(sepKey), childPID)
  210. d.dumpNode(childPID, depth+1)
  211. }
  212. if cnt > 5 {
  213. fmt.Printf("%s ... and %d more children\n", indent(depth+1), cnt-5)
  214. }
  215. }
  216. }
  217. func indent(depth int) string {
  218. s := ""
  219. for i := 0; i < depth; i++ {
  220. s += " "
  221. }
  222. return s
  223. }