| ... | @@ -0,0 +1,339 @@ |
| 1 | const std = @import("std"); |
| 2 | const string = []const u8; |
| 3 | const extras = @import("extras"); |
| 4 | |
| 5 | const c = @cImport({ |
| 6 | @cInclude("yaml.h"); |
| 7 | }); |
| 8 | |
| 9 | // |
| 10 | // |
| 11 | |
| 12 | pub const Stream = struct { |
| 13 | docs: []const Document, |
| 14 | }; |
| 15 | |
| 16 | pub const Document = struct { |
| 17 | mapping: Mapping, |
| 18 | }; |
| 19 | |
| 20 | pub const Item = union(enum) { |
| 21 | event: Token, |
| 22 | kv: Key, |
| 23 | mapping: Mapping, |
| 24 | sequence: Sequence, |
| 25 | document: Document, |
| 26 | string: string, |
| 27 | stream: Stream, |
| 28 | |
| 29 | pub fn format(self: Item, comptime fmt: string, options: std.fmt.FormatOptions, writer: anytype) @TypeOf(writer).Error!void { |
| 30 | _ = fmt; |
| 31 | _ = options; |
| 32 | |
| 33 | try writer.writeAll("Item{"); |
| 34 | switch (self) { |
| 35 | .event => { |
| 36 | // try std.fmt.format(writer, "{s}", .{@tagName(self.event.type)}); |
| 37 | try std.fmt.format(writer, "event {d}", .{self.event}); |
| 38 | }, |
| 39 | .kv, .document, .stream => { |
| 40 | unreachable; |
| 41 | }, |
| 42 | .mapping => { |
| 43 | try std.fmt.format(writer, "{}", .{self.mapping}); |
| 44 | }, |
| 45 | .sequence => { |
| 46 | try writer.writeAll("[ "); |
| 47 | for (self.sequence) |it| { |
| 48 | try std.fmt.format(writer, "{}, ", .{it}); |
| 49 | } |
| 50 | try writer.writeAll("]"); |
| 51 | }, |
| 52 | .string => { |
| 53 | try std.fmt.format(writer, "{s}", .{self.string}); |
| 54 | }, |
| 55 | } |
| 56 | try writer.writeAll("}"); |
| 57 | } |
| 58 | }; |
| 59 | |
| 60 | pub const Sequence = []const Item; |
| 61 | |
| 62 | pub const Key = struct { |
| 63 | key: string, |
| 64 | value: Value, |
| 65 | }; |
| 66 | |
| 67 | pub const Value = union(enum) { |
| 68 | string: string, |
| 69 | mapping: Mapping, |
| 70 | sequence: Sequence, |
| 71 | |
| 72 | pub fn format(self: Value, comptime fmt: string, options: std.fmt.FormatOptions, writer: anytype) @TypeOf(writer).Error!void { |
| 73 | _ = fmt; |
| 74 | _ = options; |
| 75 | |
| 76 | try writer.writeAll("Value{"); |
| 77 | switch (self) { |
| 78 | .string => { |
| 79 | try std.fmt.format(writer, "{s}", .{self.string}); |
| 80 | }, |
| 81 | .mapping => { |
| 82 | try std.fmt.format(writer, "{}", .{self.mapping}); |
| 83 | }, |
| 84 | .sequence => { |
| 85 | try writer.writeAll("[ "); |
| 86 | for (self.sequence) |it| { |
| 87 | try std.fmt.format(writer, "{}, ", .{it}); |
| 88 | } |
| 89 | try writer.writeAll("]"); |
| 90 | }, |
| 91 | } |
| 92 | try writer.writeAll("}"); |
| 93 | } |
| 94 | }; |
| 95 | |
| 96 | pub const Mapping = struct { |
| 97 | items: []const Key, |
| 98 | |
| 99 | pub fn get(self: Mapping, k: string) ?Value { |
| 100 | for (self.items) |item| { |
| 101 | if (std.mem.eql(u8, item.key, k)) { |
| 102 | return item.value; |
| 103 | } |
| 104 | } |
| 105 | return null; |
| 106 | } |
| 107 | |
| 108 | pub fn getT(self: Mapping, k: string, comptime f: std.meta.FieldEnum(Value)) ?extras.FieldType(Value, f) { |
| 109 | for (self.items) |item| { |
| 110 | if (std.mem.eql(u8, item.key, k)) { |
| 111 | return @field(item.value, @tagName(f)); |
| 112 | } |
| 113 | } |
| 114 | return null; |
| 115 | } |
| 116 | |
| 117 | pub fn get_string(self: Mapping, k: string) string { |
| 118 | return if (self.get(k)) |v| v.string else ""; |
| 119 | } |
| 120 | |
| 121 | pub fn get_string_array(self: Mapping, alloc: std.mem.Allocator, k: string) ![]string { |
| 122 | var list = std.ArrayList(string).init(alloc); |
| 123 | defer list.deinit(); |
| 124 | if (self.get(k)) |val| { |
| 125 | if (val == .sequence) { |
| 126 | for (val.sequence) |item| { |
| 127 | if (item != .string) { |
| 128 | continue; |
| 129 | } |
| 130 | try list.append(item.string); |
| 131 | } |
| 132 | } |
| 133 | } |
| 134 | return list.toOwnedSlice(); |
| 135 | } |
| 136 | |
| 137 | pub fn format(self: Mapping, comptime fmt: string, options: std.fmt.FormatOptions, writer: anytype) @TypeOf(writer).Error!void { |
| 138 | _ = fmt; |
| 139 | _ = options; |
| 140 | |
| 141 | try writer.writeAll("{ "); |
| 142 | for (self.items) |it| { |
| 143 | try std.fmt.format(writer, "{s}: ", .{it.key}); |
| 144 | try std.fmt.format(writer, "{}, ", .{it.value}); |
| 145 | } |
| 146 | try writer.writeAll("}"); |
| 147 | } |
| 148 | }; |
| 149 | |
| 150 | pub const Token = c.yaml_event_t; |
| 151 | pub const TokenList = []const Token; |
| 152 | |
| 153 | // |
| 154 | // |
| 155 | |
| 156 | pub fn parse(alloc: std.mem.Allocator, input: string) !Document { |
| 157 | var parser: c.yaml_parser_t = undefined; |
| 158 | _ = c.yaml_parser_initialize(&parser); |
| 159 | defer c.yaml_parser_delete(&parser); |
| 160 | |
| 161 | const lines = try split(alloc, input, "\n"); |
| 162 | |
| 163 | _ = c.yaml_parser_set_input_string(&parser, input.ptr, input.len); |
| 164 | |
| 165 | var all_events = std.ArrayList(Token).init(alloc); |
| 166 | var event: Token = undefined; |
| 167 | while (true) { |
| 168 | const p = c.yaml_parser_parse(&parser, &event); |
| 169 | if (p == 0) { |
| 170 | break; |
| 171 | } |
| 172 | |
| 173 | const et = event.type; |
| 174 | try all_events.append(event); |
| 175 | c.yaml_event_delete(&event); |
| 176 | |
| 177 | if (et == c.YAML_STREAM_END_EVENT) { |
| 178 | break; |
| 179 | } |
| 180 | } |
| 181 | |
| 182 | const p = &Parser{ |
| 183 | .alloc = alloc, |
| 184 | .tokens = all_events.items, |
| 185 | .lines = lines, |
| 186 | .index = 0, |
| 187 | }; |
| 188 | const stream = try p.parse(); |
| 189 | return stream.docs[0]; |
| 190 | } |
| 191 | |
| 192 | pub const Parser = struct { |
| 193 | alloc: std.mem.Allocator, |
| 194 | tokens: TokenList, |
| 195 | lines: []const string, |
| 196 | index: usize, |
| 197 | |
| 198 | pub fn parse(self: *Parser) !Stream { |
| 199 | const item = try parse_item(self, null); |
| 200 | return item.stream; |
| 201 | } |
| 202 | |
| 203 | fn next(self: *Parser) ?Token { |
| 204 | if (self.index >= self.tokens.len) { |
| 205 | return null; |
| 206 | } |
| 207 | defer self.index += 1; |
| 208 | return self.tokens[self.index]; |
| 209 | } |
| 210 | }; |
| 211 | |
| 212 | pub const Error = |
| 213 | std.mem.Allocator.Error || |
| 214 | error{YamlUnexpectedToken}; |
| 215 | |
| 216 | fn parse_item(p: *Parser, start: ?Token) Error!Item { |
| 217 | const tok = start orelse p.next(); |
| 218 | return switch (tok.?.type) { |
| 219 | c.YAML_STREAM_START_EVENT => Item{ .stream = try parse_stream(p) }, |
| 220 | c.YAML_DOCUMENT_START_EVENT => Item{ .document = try parse_document(p) }, |
| 221 | c.YAML_MAPPING_START_EVENT => Item{ .mapping = try parse_mapping(p) }, |
| 222 | c.YAML_SEQUENCE_START_EVENT => Item{ .sequence = try parse_sequence(p) }, |
| 223 | c.YAML_SCALAR_EVENT => Item{ .string = try get_event_string(tok.?, p) }, |
| 224 | else => unreachable, |
| 225 | }; |
| 226 | } |
| 227 | |
| 228 | fn parse_stream(p: *Parser) Error!Stream { |
| 229 | var res = std.ArrayList(Document).init(p.alloc); |
| 230 | errdefer res.deinit(); |
| 231 | |
| 232 | while (true) { |
| 233 | const tok = p.next(); |
| 234 | if (tok.?.type == c.YAML_STREAM_END_EVENT) { |
| 235 | return Stream{ .docs = try res.toOwnedSlice() }; |
| 236 | } |
| 237 | if (tok.?.type != c.YAML_DOCUMENT_START_EVENT) { |
| 238 | return error.YamlUnexpectedToken; |
| 239 | } |
| 240 | const item = try parse_item(p, tok); |
| 241 | try res.append(item.document); |
| 242 | } |
| 243 | } |
| 244 | |
| 245 | fn parse_document(p: *Parser) Error!Document { |
| 246 | const tok = p.next(); |
| 247 | if (tok.?.type != c.YAML_MAPPING_START_EVENT) { |
| 248 | return error.YamlUnexpectedToken; |
| 249 | } |
| 250 | const item = try parse_item(p, tok); |
| 251 | |
| 252 | if (p.next().?.type != c.YAML_DOCUMENT_END_EVENT) { |
| 253 | return error.YamlUnexpectedToken; |
| 254 | } |
| 255 | return Document{ .mapping = item.mapping }; |
| 256 | } |
| 257 | |
| 258 | fn parse_mapping(p: *Parser) Error!Mapping { |
| 259 | var res = std.ArrayList(Key).init(p.alloc); |
| 260 | errdefer res.deinit(); |
| 261 | |
| 262 | while (true) { |
| 263 | const tok = p.next(); |
| 264 | if (tok.?.type == c.YAML_MAPPING_END_EVENT) { |
| 265 | return Mapping{ .items = try res.toOwnedSlice() }; |
| 266 | } |
| 267 | if (tok.?.type != c.YAML_SCALAR_EVENT) { |
| 268 | return error.YamlUnexpectedToken; |
| 269 | } |
| 270 | try res.append(Key{ |
| 271 | .key = try get_event_string(tok.?, p), |
| 272 | .value = try parse_value(p), |
| 273 | }); |
| 274 | } |
| 275 | } |
| 276 | |
| 277 | fn parse_value(p: *Parser) Error!Value { |
| 278 | const item = try parse_item(p, null); |
| 279 | return switch (item) { |
| 280 | .mapping => |x| Value{ .mapping = x }, |
| 281 | .sequence => |x| Value{ .sequence = x }, |
| 282 | .string => |x| Value{ .string = x }, |
| 283 | else => unreachable, |
| 284 | }; |
| 285 | } |
| 286 | |
| 287 | fn parse_sequence(p: *Parser) Error!Sequence { |
| 288 | var res = std.ArrayList(Item).init(p.alloc); |
| 289 | errdefer res.deinit(); |
| 290 | |
| 291 | while (true) { |
| 292 | const tok = p.next(); |
| 293 | if (tok.?.type == c.YAML_SEQUENCE_END_EVENT) { |
| 294 | return try res.toOwnedSlice(); |
| 295 | } |
| 296 | try res.append(try parse_item(p, tok)); |
| 297 | } |
| 298 | } |
| 299 | |
| 300 | fn get_event_string(event: Token, p: *const Parser) !string { |
| 301 | const sm = event.start_mark; |
| 302 | const em = event.end_mark; |
| 303 | const lines = p.lines; |
| 304 | if (sm.line != em.line) { |
| 305 | const starter = lines[sm.line][sm.column..]; |
| 306 | std.debug.assert(starter.len == 1); |
| 307 | switch (starter[0]) { |
| 308 | '|' => { |
| 309 | var list = std.ArrayList(u8).init(p.alloc); |
| 310 | errdefer list.deinit(); |
| 311 | var i = sm.line + 1; |
| 312 | while (i <= em.line) : (i += 1) { |
| 313 | try list.appendSlice(std.mem.trimLeft(u8, lines[i], " ")); |
| 314 | try list.append('\n'); |
| 315 | } |
| 316 | return try list.toOwnedSlice(); |
| 317 | }, |
| 318 | else => @panic("TODO"), |
| 319 | } |
| 320 | } |
| 321 | const s = lines[sm.line][sm.column..em.column]; |
| 322 | if (s.len < 2) return s; |
| 323 | if (s[0] == '"' and s[s.len - 1] == '"') return std.mem.trim(u8, s, "\""); |
| 324 | return s; |
| 325 | } |
| 326 | |
| 327 | // |
| 328 | // |
| 329 | |
| 330 | fn split(alloc: std.mem.Allocator, in: string, delim: string) ![]string { |
| 331 | var list = std.ArrayList(string).init(alloc); |
| 332 | errdefer list.deinit(); |
| 333 | |
| 334 | var iter = std.mem.split(u8, in, delim); |
| 335 | while (iter.next()) |str| { |
| 336 | try list.append(str); |
| 337 | } |
| 338 | return try list.toOwnedSlice(); |
| 339 | } |