|
| 1 | +const std = @import("std"); |
| 2 | +const Allocator = std.mem.Allocator; |
| 3 | +const HashMap = std.AutoArrayHashMap; |
| 4 | + |
| 5 | +const TrieError = error{ |
| 6 | + InvalidNode, |
| 7 | +}; |
| 8 | + |
| 9 | +pub fn TrieNode(comptime T: type) type { |
| 10 | + return struct { |
| 11 | + const Self = @This(); |
| 12 | + node_data: T, |
| 13 | + children: HashMap(u8, *Self), |
| 14 | + parent: ?*Self, |
| 15 | + |
| 16 | + fn init(node_data: T, allocator: Allocator, parent: ?*Self) TrieNode(T) { |
| 17 | + return TrieNode(T){ |
| 18 | + .node_data = node_data, |
| 19 | + .children = HashMap(u8, *Self).init(allocator), |
| 20 | + .parent = parent, |
| 21 | + }; |
| 22 | + } |
| 23 | + }; |
| 24 | +} |
| 25 | + |
| 26 | +/// Interface to traverse the trie |
| 27 | +fn TrieIterator(comptime T: type) type { |
| 28 | + return struct { |
| 29 | + const Self = @This(); |
| 30 | + const NodeType = TrieNode(T); |
| 31 | + /// Can be dereferenced to modify node data |
| 32 | + node_at_iterator: *NodeType, |
| 33 | + fn init(node_ptr: *NodeType) Self { |
| 34 | + return Self{ |
| 35 | + .node_at_iterator = node_ptr, |
| 36 | + }; |
| 37 | + } |
| 38 | + |
| 39 | + /// Returns an optional iterator pointing to the child following the `char` edge |
| 40 | + pub fn go_to_child(self: Self, char: u8) ?Self { |
| 41 | + if (self.node_at_iterator.children.get(char)) |new_ptr| { |
| 42 | + return Self{ |
| 43 | + .node_at_iterator = new_ptr, |
| 44 | + }; |
| 45 | + } else { |
| 46 | + return null; |
| 47 | + } |
| 48 | + } |
| 49 | + |
| 50 | + /// Returns an optional iterator pointing to the parent |
| 51 | + pub fn go_to_parent(self: Self) ?Self { |
| 52 | + if (self.node_at_iterator.parent) |new_ptr| { |
| 53 | + return Self{ |
| 54 | + .node_at_iterator = new_ptr, |
| 55 | + }; |
| 56 | + } else { |
| 57 | + return null; |
| 58 | + } |
| 59 | + } |
| 60 | + }; |
| 61 | +} |
| 62 | + |
| 63 | +pub fn Trie(comptime T: type) type { |
| 64 | + return struct { |
| 65 | + const Self = @This(); |
| 66 | + const NodeType = TrieNode(T); |
| 67 | + const IteratorType = TrieIterator(T); |
| 68 | + trie_root: *NodeType, |
| 69 | + allocator: Allocator, |
| 70 | + |
| 71 | + /// Allocate a new node and return its pointer |
| 72 | + fn new_node(self: Self, node_data: T, parent: ?*NodeType) !*NodeType { |
| 73 | + const node_ptr = try self.allocator.create(NodeType); |
| 74 | + node_ptr.* = NodeType.init( |
| 75 | + node_data, |
| 76 | + self.allocator, |
| 77 | + parent, |
| 78 | + ); |
| 79 | + return node_ptr; |
| 80 | + } |
| 81 | + |
| 82 | + pub fn init(root_data: T, allocator: Allocator) !Self { |
| 83 | + const node_ptr = try allocator.create(NodeType); |
| 84 | + node_ptr.* = NodeType.init(root_data, allocator, null); |
| 85 | + return Self{ |
| 86 | + .trie_root = node_ptr, |
| 87 | + .allocator = allocator, |
| 88 | + }; |
| 89 | + } |
| 90 | + |
| 91 | + /// Returns an iterator pointing to the root |
| 92 | + pub fn get_root_iterator(self: Self) IteratorType { |
| 93 | + return IteratorType.init(self.trie_root); |
| 94 | + } |
| 95 | + |
| 96 | + /// Add a string to the trie, assigning newly created node's data with `new_value` |
| 97 | + pub fn add_string(self: Self, new_string: []const u8, new_value: T) !IteratorType { |
| 98 | + var iterator = self.get_root_iterator(); |
| 99 | + for (new_string) |char| { |
| 100 | + if (iterator.go_to_child(char)) |new_iterator| { |
| 101 | + iterator = new_iterator; |
| 102 | + } else { |
| 103 | + const node = try self.new_node( |
| 104 | + new_value, |
| 105 | + iterator.node_at_iterator, |
| 106 | + ); |
| 107 | + try iterator.node_at_iterator.children.put(char, node); |
| 108 | + iterator = iterator.go_to_child(char).?; |
| 109 | + } |
| 110 | + } |
| 111 | + return iterator; |
| 112 | + } |
| 113 | + |
| 114 | + /// traverse and write u8 to `output_buffer` for each edge traversed in dfs order |
| 115 | + pub fn traverse(self: Self, iterator: IteratorType, output_buffer: []u8) usize { |
| 116 | + var it = iterator.node_at_iterator.children.iterator(); |
| 117 | + var write_position: usize = 0; |
| 118 | + while (it.next()) |entry| { |
| 119 | + output_buffer[write_position] = entry.key_ptr.*; |
| 120 | + write_position += 1; |
| 121 | + write_position += self.traverse( |
| 122 | + IteratorType.init(entry.value_ptr.*), |
| 123 | + output_buffer[write_position..output_buffer.len], |
| 124 | + ); |
| 125 | + } |
| 126 | + return write_position; |
| 127 | + } |
| 128 | + |
| 129 | + /// Apply a function to every node in the trie. |
| 130 | + /// If `top_down = true`, apply the function before applying to children, and vice versa. |
| 131 | + pub fn apply( |
| 132 | + self: Self, |
| 133 | + iterator: IteratorType, |
| 134 | + func: ?*fn (node: *NodeType) void, |
| 135 | + top_down: bool, |
| 136 | + ) void { |
| 137 | + if (func) |f| { |
| 138 | + if (top_down) { |
| 139 | + f(iterator.node_at_iterator); |
| 140 | + } else { |
| 141 | + defer f(iterator.node_at_iterator); |
| 142 | + } |
| 143 | + } |
| 144 | + var it = iterator.node_at_iterator.children.iterator(); |
| 145 | + while (it.next()) |entry| { |
| 146 | + self.apply( |
| 147 | + IteratorType.init(entry.value_ptr.*), |
| 148 | + func, |
| 149 | + top_down, |
| 150 | + ); |
| 151 | + } |
| 152 | + } |
| 153 | + |
| 154 | + fn recursive_free(self: Self, iterator: IteratorType) void { |
| 155 | + var it = iterator.node_at_iterator.children.iterator(); |
| 156 | + while (it.next()) |entry| { |
| 157 | + self.recursive_free(IteratorType.init(entry.value_ptr.*)); |
| 158 | + } |
| 159 | + iterator.node_at_iterator.children.deinit(); |
| 160 | + self.allocator.destroy(iterator.node_at_iterator); |
| 161 | + } |
| 162 | + |
| 163 | + pub fn deinit(self: Self) void { |
| 164 | + self.recursive_free(self.get_root_iterator()); |
| 165 | + } |
| 166 | + }; |
| 167 | +} |
| 168 | + |
| 169 | +test "basic traverse" { |
| 170 | + var gpa = std.heap.GeneralPurposeAllocator(.{}){}; |
| 171 | + const debug_allocator = gpa.allocator(); |
| 172 | + const trie = try Trie(i32).init(0, debug_allocator); |
| 173 | + defer trie.deinit(); |
| 174 | + _ = try trie.add_string("aaa", 0); |
| 175 | + _ = try trie.add_string("abb", 0); |
| 176 | + _ = try trie.add_string("abc", 0); |
| 177 | + // a |
| 178 | + // / \ |
| 179 | + // a b |
| 180 | + // / / \ |
| 181 | + // a b c |
| 182 | + const answer = "aaabbc"; |
| 183 | + var buffer: [6]u8 = undefined; |
| 184 | + _ = trie.traverse(trie.get_root_iterator(), buffer[0..6]); |
| 185 | + try std.testing.expectEqualSlices(u8, buffer[0..6], answer[0..6]); |
| 186 | +} |
| 187 | + |
| 188 | +test "iterator traverse" { |
| 189 | + var gpa = std.heap.GeneralPurposeAllocator(.{}){}; |
| 190 | + const debug_allocator = gpa.allocator(); |
| 191 | + const trie = try Trie(i32).init(0, debug_allocator); |
| 192 | + defer trie.deinit(); |
| 193 | + var it = try trie.add_string("abc", 0); // "abc" |
| 194 | + try std.testing.expectEqual(null, it.go_to_child('a')); |
| 195 | + it = it.go_to_parent().?; // "ab" |
| 196 | + it = it.go_to_parent().?; // "a" |
| 197 | + it = it.go_to_parent().?; // "" |
| 198 | + try std.testing.expectEqual(null, it.go_to_parent()); |
| 199 | + const it2 = try trie.add_string("ae", 0); // "ae" |
| 200 | + it = it.go_to_child('a').?; // "a" |
| 201 | + it = it.go_to_child('e').?; // "ae" |
| 202 | + try std.testing.expectEqual(it, it2); |
| 203 | +} |
0 commit comments