index.zig 9.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300
  1. const std = @import("std");
  2. const storage = @import("storage.zig");
  3. var tree_arena = std.heap.ArenaAllocator.init(std.heap.page_allocator);
  4. const tree_allocator = tree_arena.allocator();
  5. var tree_mutex: std.Thread.Mutex = .{};
  6. const RadixNode = struct {
  7. edge: []const u8,
  8. children: std.StringHashMap(*RadixNode),
  9. is_terminal: bool,
  10. fn init(edge: []const u8) *RadixNode {
  11. const node = tree_allocator.create(RadixNode) catch unreachable;
  12. node.* = .{
  13. .edge = tree_allocator.dupe(u8, edge) catch unreachable,
  14. .children = std.StringHashMap(*RadixNode).init(tree_allocator),
  15. .is_terminal = false,
  16. };
  17. return node;
  18. }
  19. fn deinit(self: *RadixNode) void {
  20. var it = self.children.iterator();
  21. while (it.next()) |entry| {
  22. entry.value_ptr.*.deinit();
  23. }
  24. self.children.deinit();
  25. tree_allocator.free(self.edge);
  26. tree_allocator.destroy(self);
  27. }
  28. };
  29. var root: *RadixNode = undefined;
  30. var root_initialized = false;
  31. fn ensureRoot() void {
  32. if (!root_initialized) {
  33. root = RadixNode.init("");
  34. root_initialized = true;
  35. }
  36. }
  37. fn commonPrefixLen(a: []const u8, b: []const u8) usize {
  38. var i: usize = 0;
  39. while (i < a.len and i < b.len and a[i] == b[i]) {
  40. i += 1;
  41. }
  42. return i;
  43. }
  44. pub fn insert(key: []const u8) void {
  45. tree_mutex.lock();
  46. defer tree_mutex.unlock();
  47. ensureRoot();
  48. if (key.len == 0) return;
  49. var node = root;
  50. var remaining = key;
  51. while (remaining.len > 0) {
  52. var found = false;
  53. var it = node.children.iterator();
  54. while (it.next()) |entry| {
  55. const child = entry.value_ptr.*;
  56. const prefix_len = commonPrefixLen(child.edge, remaining);
  57. if (prefix_len > 0) {
  58. found = true;
  59. if (prefix_len == child.edge.len) {
  60. if (prefix_len == remaining.len) {
  61. child.is_terminal = true;
  62. return;
  63. }
  64. remaining = remaining[prefix_len..];
  65. node = child;
  66. break;
  67. } else {
  68. const old_edge = child.edge;
  69. const common = old_edge[0..prefix_len];
  70. const child_suffix = old_edge[prefix_len..];
  71. const key_suffix = remaining[prefix_len..];
  72. const intermediate = RadixNode.init(common);
  73. tree_allocator.free(child.edge);
  74. child.edge = tree_allocator.dupe(u8, child_suffix) catch unreachable;
  75. intermediate.children.put(child_suffix, child) catch unreachable;
  76. _ = node.children.remove(old_edge);
  77. node.children.put(common, intermediate) catch unreachable;
  78. if (key_suffix.len == 0) {
  79. intermediate.is_terminal = true;
  80. return;
  81. } else {
  82. const new_child = RadixNode.init(key_suffix);
  83. new_child.is_terminal = true;
  84. intermediate.children.put(key_suffix, new_child) catch unreachable;
  85. return;
  86. }
  87. }
  88. }
  89. }
  90. if (!found) {
  91. const new_child = RadixNode.init(remaining);
  92. new_child.is_terminal = true;
  93. node.children.put(remaining, new_child) catch unreachable;
  94. return;
  95. }
  96. }
  97. node.is_terminal = true;
  98. }
  99. pub fn delete(key: []const u8) void {
  100. tree_mutex.lock();
  101. defer tree_mutex.unlock();
  102. ensureRoot();
  103. if (key.len == 0) return;
  104. const node = findNode(root, key);
  105. if (node) |n| {
  106. n.is_terminal = false;
  107. }
  108. }
  109. fn findNodeForPrefix(node: *RadixNode, prefix: []const u8, path_buf: *[MAX_KEY_LENGTH]u8, path_len: *usize) ?*RadixNode {
  110. if (prefix.len == 0) {
  111. path_len.* = 0;
  112. return node;
  113. }
  114. var current = node;
  115. var remaining = prefix;
  116. path_len.* = 0;
  117. while (remaining.len > 0) {
  118. var found = false;
  119. var it = current.children.iterator();
  120. while (it.next()) |entry| {
  121. const child = entry.value_ptr.*;
  122. const prefix_match_len = commonPrefixLen(child.edge, remaining);
  123. if (prefix_match_len > 0) {
  124. if (prefix_match_len == remaining.len) {
  125. return child;
  126. }
  127. if (prefix_match_len == child.edge.len) {
  128. if (path_len.* + prefix_match_len > MAX_KEY_LENGTH) return null;
  129. @memcpy(path_buf[path_len.* .. path_len.* + prefix_match_len], child.edge[0..prefix_match_len]);
  130. path_len.* += prefix_match_len;
  131. remaining = remaining[prefix_match_len..];
  132. current = child;
  133. found = true;
  134. break;
  135. }
  136. return null;
  137. }
  138. }
  139. if (!found) {
  140. return null;
  141. }
  142. }
  143. return current;
  144. }
  145. fn findNode(node: *RadixNode, key: []const u8) ?*RadixNode {
  146. var dummy_buf: [MAX_KEY_LENGTH]u8 = undefined;
  147. var dummy_len: usize = 0;
  148. return findNodeForPrefix(node, key, &dummy_buf, &dummy_len);
  149. }
  150. pub fn searchByPrefix(prefix: []const u8) ?*RadixNode {
  151. ensureRoot();
  152. if (prefix.len == 0) return root;
  153. return findNode(root, prefix);
  154. }
  155. fn countKeys(node: *RadixNode) usize {
  156. var count: usize = 0;
  157. if (node.is_terminal) {
  158. count += 1;
  159. }
  160. var it = node.children.iterator();
  161. while (it.next()) |entry| {
  162. count += countKeys(entry.value_ptr.*);
  163. }
  164. return count;
  165. }
  166. const MAX_KEYS_RETURN = 100_000_000;
  167. const MAX_KEY_LENGTH = 1024;
  168. fn collectKeysWithBuffer(node: *RadixNode, prefix_buffer: []u8, prefix_len: usize, keys: *std.ArrayListUnmanaged([]const u8), max_keys: usize, search_prefix: []const u8, include_node_edge: bool, allocator: std.mem.Allocator) void {
  169. if (keys.items.len >= max_keys) return;
  170. var current_len = prefix_len;
  171. if (include_node_edge and node.edge.len > 0) {
  172. if (current_len + node.edge.len > MAX_KEY_LENGTH) return;
  173. @memcpy(prefix_buffer[current_len .. current_len + node.edge.len], node.edge);
  174. current_len += node.edge.len;
  175. }
  176. if (node.is_terminal) {
  177. const key = prefix_buffer[0..current_len];
  178. if (key.len >= search_prefix.len and std.mem.eql(u8, key[0..search_prefix.len], search_prefix)) {
  179. const key_copy = allocator.dupe(u8, key) catch return;
  180. keys.append(allocator, key_copy) catch return;
  181. }
  182. }
  183. var it = node.children.iterator();
  184. while (it.next()) |entry| {
  185. if (keys.items.len >= max_keys) break;
  186. const child = entry.value_ptr.*;
  187. collectKeysWithBuffer(child, prefix_buffer, current_len, keys, max_keys, search_prefix, true, allocator);
  188. }
  189. }
  190. fn collectKeys(node: *RadixNode, prefix: []const u8, keys: *std.ArrayListUnmanaged([]const u8), max_keys: usize, search_prefix: []const u8, allocator: std.mem.Allocator) void {
  191. var prefix_buffer: [MAX_KEY_LENGTH]u8 = undefined;
  192. if (prefix.len > MAX_KEY_LENGTH) return;
  193. @memcpy(prefix_buffer[0..prefix.len], prefix);
  194. collectKeysWithBuffer(node, &prefix_buffer, prefix.len, keys, max_keys, search_prefix, true, allocator);
  195. }
  196. pub fn getKeysFromNode(node: *RadixNode, prefix: []const u8, allocator: std.mem.Allocator) [][]const u8 {
  197. var keys_list = std.ArrayListUnmanaged([]const u8){};
  198. collectKeys(node, prefix, &keys_list, MAX_KEYS_RETURN, prefix, allocator);
  199. return keys_list.toOwnedSlice(allocator) catch &[_][]const u8{};
  200. }
  201. pub fn getKeysByPrefix(prefix: []const u8, allocator: std.mem.Allocator) []const u8 {
  202. tree_mutex.lock();
  203. defer tree_mutex.unlock();
  204. ensureRoot();
  205. const node = searchByPrefix(prefix) orelse return "";
  206. const keys = getKeysFromNode(node, prefix, allocator);
  207. if (keys.len == 0) return "";
  208. return std.mem.join(allocator, "\n", keys) catch "";
  209. }
  210. pub fn getValuesByPrefix(prefix: []const u8, allocator: std.mem.Allocator) []const u8 {
  211. // Phase 1: Collect matching keys under tree_mutex
  212. var keys: [][]const u8 = &[_][]const u8{};
  213. {
  214. tree_mutex.lock();
  215. defer tree_mutex.unlock();
  216. ensureRoot();
  217. var path_buf: [MAX_KEY_LENGTH]u8 = undefined;
  218. var path_len: usize = 0;
  219. const node = findNodeForPrefix(root, prefix, &path_buf, &path_len) orelse {
  220. return "";
  221. };
  222. var keys_list = std.ArrayListUnmanaged([]const u8){};
  223. collectKeys(node, path_buf[0..path_len], &keys_list, MAX_KEYS_RETURN, prefix, allocator);
  224. keys = keys_list.toOwnedSlice(allocator) catch &[_][]const u8{};
  225. }
  226. if (keys.len == 0) return "";
  227. // Phase 2: Read values without tree_mutex to avoid deadlock with write/delete
  228. const values = allocator.alloc([]const u8, keys.len) catch return "";
  229. for (keys, 0..) |key, i| {
  230. const value = storage.read(key) orelse "";
  231. values[i] = value;
  232. }
  233. return std.mem.join(allocator, "\n", values) catch "";
  234. }
  235. pub fn getAllKeys(allocator: std.mem.Allocator) []const u8 {
  236. tree_mutex.lock();
  237. defer tree_mutex.unlock();
  238. ensureRoot();
  239. const keys = getKeysFromNode(root, &[_]u8{}, allocator);
  240. if (keys.len == 0) return "";
  241. return std.mem.join(allocator, "\n", keys) catch "";
  242. }