| 1 | //! String formatting and parsing. |
| 2 | |
| 3 | const std = @import("std"); |
| 4 | const builtin = @import("builtin"); |
| 5 | const extras = @import("extras"); |
| 6 | const nio = @import("./nio.zig"); |
| 7 | |
| 8 | /// Renders fmt string with args, calling `writer` with slices of bytes. |
| 9 | /// If `writer` returns an error, the error is returned from `format` and |
| 10 | /// `writer` is not called again. |
| 11 | /// |
| 12 | /// The format string must be comptime-known and may contain placeholders following |
| 13 | /// this format: |
| 14 | /// `{[argument][specifier]:[fill][alignment][width].[precision]}` |
| 15 | /// |
| 16 | /// Above, each word including its surrounding [ and ] is a parameter which you have to replace with something: |
| 17 | /// |
| 18 | /// - *argument* is either the numeric index or the field name of the argument that should be inserted |
| 19 | /// - when using a field name, you are required to enclose the field name (an identifier) in square |
| 20 | /// brackets, e.g. {[score]...} as opposed to the numeric index form which can be written e.g. {2...} |
| 21 | /// - *specifier* is a type-dependent formatting option that determines how a type should formatted (see below) |
| 22 | /// - *fill* is a single unicode codepoint which is used to pad the formatted text |
| 23 | /// - *alignment* is one of the three bytes '<', '^', or '>' to make the text left-, center-, or right-aligned, respectively |
| 24 | /// - *width* is the total width of the field in unicode codepoints |
| 25 | /// - *precision* specifies how many decimals a formatted number should have |
| 26 | /// |
| 27 | /// Note that most of the parameters are optional and may be omitted. Also you can leave out separators like `:` and `.` when |
| 28 | /// all parameters after the separator are omitted. |
| 29 | /// Only exception is the *fill* parameter. If a non-zero *fill* character is required at the same time as *width* is specified, |
| 30 | /// one has to specify *alignment* as well, as otherwise the digit following `:` is interpreted as *width*, not *fill*. |
| 31 | /// |
| 32 | /// The *specifier* has several options for types: |
| 33 | /// - `x` and `X`: output numeric value in hexadecimal notation |
| 34 | /// - `s`: |
| 35 | /// - for pointer-to-many and C pointers of u8, print as a C-string using zero-termination |
| 36 | /// - for slices of u8, print the entire slice as a string without zero-termination |
| 37 | /// - `e`: output floating point value in scientific notation |
| 38 | /// - `d`: output numeric value in decimal notation |
| 39 | /// - `b`: output integer value in binary notation |
| 40 | /// - `o`: output integer value in octal notation |
| 41 | /// - `c`: output integer as an ASCII character. Integer type must have 8 bits at max. |
| 42 | /// - `u`: output integer as an UTF-8 sequence. Integer type must have 21 bits at max. |
| 43 | /// - `?`: output optional value as either the unwrapped value, or `null`; may be followed by a format specifier for the underlying value. |
| 44 | /// - `!`: output error union value as either the unwrapped value, or the formatted error value; may be followed by a format specifier for the underlying value. |
| 45 | /// - `*`: output the address of the value instead of the value itself. |
| 46 | /// - `any`: output a value of any type using its default format. |
| 47 | /// |
| 48 | /// If a formatted user type contains a function of the type |
| 49 | /// ``` |
| 50 | /// fn format(value: ?, comptime fmt: []const u8, options: std.fmt.FormatOptions, writer: anytype) !void |
| 51 | /// ``` |
| 52 | /// with `?` being the type formatted, this function will be called instead of the default implementation. |
| 53 | /// This allows user types to be formatted in a logical manner instead of dumping all fields of the type. |
| 54 | /// |
| 55 | /// A user type may be a `struct`, `vector`, `union` or `enum` type. |
| 56 | /// |
| 57 | /// To print literal curly braces, escape them by writing them twice, e.g. `{{` or `}}`. |
| 58 | pub fn format(writer: anytype, comptime fmt: []const u8, args: anytype) !void { |
| 59 | const ArgsType = @TypeOf(args); |
| 60 | const args_type_info = @typeInfo(ArgsType); |
| 61 | if (args_type_info != .@"struct") { |
| 62 | @compileError("expected tuple or struct argument, found " ++ @typeName(ArgsType)); |
| 63 | } |
| 64 | |
| 65 | const fields_info = args_type_info.@"struct".fields; |
| 66 | if (fields_info.len > max_format_args) { |
| 67 | @compileError("32 arguments max are supported per format call"); |
| 68 | } |
| 69 | |
| 70 | @setEvalBranchQuota(std.math.maxInt(u32)); |
| 71 | comptime var arg_state: ArgState = .{ .args_len = fields_info.len }; |
| 72 | comptime var i = 0; |
| 73 | comptime var literal: []const u8 = ""; |
| 74 | inline while (true) { |
| 75 | const start_index = i; |
| 76 | |
| 77 | inline while (i < fmt.len) : (i += 1) { |
| 78 | switch (fmt[i]) { |
| 79 | '{', '}' => break, |
| 80 | else => {}, |
| 81 | } |
| 82 | } |
| 83 | |
| 84 | comptime var end_index = i; |
| 85 | comptime var unescape_brace = false; |
| 86 | |
| 87 | // Handle {{ and }}, those are un-escaped as single braces |
| 88 | if (i + 1 < fmt.len and fmt[i + 1] == fmt[i]) { |
| 89 | unescape_brace = true; |
| 90 | // Make the first brace part of the literal... |
| 91 | end_index += 1; |
| 92 | // ...and skip both |
| 93 | i += 2; |
| 94 | } |
| 95 | |
| 96 | literal = literal ++ fmt[start_index..end_index]; |
| 97 | |
| 98 | // We've already skipped the other brace, restart the loop |
| 99 | if (unescape_brace) continue; |
| 100 | |
| 101 | // Write out the literal |
| 102 | if (literal.len != 0) { |
| 103 | try writer.writeAll(literal); |
| 104 | literal = ""; |
| 105 | } |
| 106 | |
| 107 | if (i >= fmt.len) break; |
| 108 | |
| 109 | if (fmt[i] == '}') { |
| 110 | @compileError("missing opening {"); |
| 111 | } |
| 112 | |
| 113 | // Get past the { |
| 114 | comptime std.debug.assert(fmt[i] == '{'); |
| 115 | i += 1; |
| 116 | |
| 117 | const fmt_begin = i; |
| 118 | // Find the closing brace |
| 119 | inline while (i < fmt.len and fmt[i] != '}') : (i += 1) {} |
| 120 | const fmt_end = i; |
| 121 | |
| 122 | if (i >= fmt.len) { |
| 123 | @compileError("missing closing }"); |
| 124 | } |
| 125 | |
| 126 | // Get past the } |
| 127 | comptime std.debug.assert(fmt[i] == '}'); |
| 128 | i += 1; |
| 129 | |
| 130 | const placeholder = comptime Placeholder.parse(fmt[fmt_begin..fmt_end].*); |
| 131 | const arg_pos = comptime switch (placeholder.arg) { |
| 132 | .none => null, |
| 133 | .number => |pos| pos, |
| 134 | .named => |arg_name| std.meta.fieldIndex(ArgsType, arg_name) orelse @compileError("no argument with name '" ++ arg_name ++ "'"), |
| 135 | }; |
| 136 | |
| 137 | const width = switch (placeholder.width) { |
| 138 | .none => null, |
| 139 | .number => |v| v, |
| 140 | .named => |arg_name| blk: { |
| 141 | const arg_i = comptime std.meta.fieldIndex(ArgsType, arg_name) orelse @compileError("no argument with name '" ++ arg_name ++ "'"); |
| 142 | _ = comptime arg_state.nextArg(arg_i) orelse @compileError("too few arguments"); |
| 143 | break :blk @field(args, arg_name); |
| 144 | }, |
| 145 | }; |
| 146 | |
| 147 | const precision = switch (placeholder.precision) { |
| 148 | .none => null, |
| 149 | .number => |v| v, |
| 150 | .named => |arg_name| blk: { |
| 151 | const arg_i = comptime std.meta.fieldIndex(ArgsType, arg_name) orelse @compileError("no argument with name '" ++ arg_name ++ "'"); |
| 152 | _ = comptime arg_state.nextArg(arg_i) orelse @compileError("too few arguments"); |
| 153 | break :blk @field(args, arg_name); |
| 154 | }, |
| 155 | }; |
| 156 | |
| 157 | const arg_to_print = comptime arg_state.nextArg(arg_pos) orelse @compileError("too few arguments"); |
| 158 | |
| 159 | try formatType( |
| 160 | @field(args, fields_info[arg_to_print].name), |
| 161 | placeholder.specifier_arg, |
| 162 | .{ |
| 163 | .fill = placeholder.fill, |
| 164 | .alignment = placeholder.alignment, |
| 165 | .width = width, |
| 166 | .precision = precision, |
| 167 | }, |
| 168 | writer, |
| 169 | std.options.fmt_max_depth, |
| 170 | ); |
| 171 | } |
| 172 | |
| 173 | if (comptime arg_state.hasUnusedArgs()) { |
| 174 | const missing_count = arg_state.args_len - @popCount(arg_state.used_args); |
| 175 | switch (missing_count) { |
| 176 | 0 => unreachable, |
| 177 | 1 => @compileError("unused argument in '" ++ fmt ++ "'"), |
| 178 | else => @compileError(comptimePrint("{d}", .{missing_count}) ++ " unused arguments in '" ++ fmt ++ "'"), |
| 179 | } |
| 180 | } |
| 181 | } |
| 182 | |
| 183 | pub const FormatOptions = struct { |
| 184 | precision: ?usize = null, |
| 185 | width: ?usize = null, |
| 186 | alignment: Alignment = default_alignment, |
| 187 | fill: u21 = default_fill_char, |
| 188 | }; |
| 189 | |
| 190 | pub const Alignment = enum { left, center, right }; |
| 191 | pub const Case = enum { lower, upper }; |
| 192 | |
| 193 | const default_max_depth = 3; |
| 194 | const default_alignment = .right; |
| 195 | const default_fill_char = ' '; |
| 196 | |
| 197 | const ArgSetType = u32; |
| 198 | const max_format_args = @typeInfo(ArgSetType).int.bits; |
| 199 | |
| 200 | pub const ArgState = struct { |
| 201 | next_arg: usize = 0, |
| 202 | used_args: ArgSetType = 0, |
| 203 | args_len: usize, |
| 204 | |
| 205 | fn hasUnusedArgs(self: *@This()) bool { |
| 206 | return @popCount(self.used_args) != self.args_len; |
| 207 | } |
| 208 | |
| 209 | fn nextArg(self: *@This(), arg_index: ?usize) ?usize { |
| 210 | const next_index = arg_index orelse init: { |
| 211 | const arg = self.next_arg; |
| 212 | self.next_arg += 1; |
| 213 | break :init arg; |
| 214 | }; |
| 215 | |
| 216 | if (next_index >= self.args_len) { |
| 217 | return null; |
| 218 | } |
| 219 | |
| 220 | // Mark this argument as used |
| 221 | self.used_args |= @as(ArgSetType, 1) << @as(u5, @intCast(next_index)); |
| 222 | return next_index; |
| 223 | } |
| 224 | }; |
| 225 | |
| 226 | pub const Placeholder = struct { |
| 227 | specifier_arg: []const u8, |
| 228 | fill: u21, |
| 229 | alignment: Alignment, |
| 230 | arg: Specifier, |
| 231 | width: Specifier, |
| 232 | precision: Specifier, |
| 233 | |
| 234 | fn parse(comptime str: anytype) Placeholder { |
| 235 | const view = std.unicode.Utf8View.initComptime(&str); |
| 236 | comptime var parser = Parser{ |
| 237 | .iter = view.iterator(), |
| 238 | }; |
| 239 | |
| 240 | // Parse the positional argument number |
| 241 | const arg = comptime parser.specifier() catch |err| @compileError(@errorName(err)); |
| 242 | |
| 243 | // Parse the format specifier |
| 244 | const specifier_arg = comptime parser.until(':'); |
| 245 | |
| 246 | // Skip the colon, if present |
| 247 | if (comptime parser.char()) |ch| { |
| 248 | if (ch != ':') { |
| 249 | @compileError("expected : or }, found '" ++ std.unicode.utf8EncodeComptime(ch) ++ "'"); |
| 250 | } |
| 251 | } |
| 252 | |
| 253 | // Parse the fill character, if present. |
| 254 | // When the width field is also specified, the fill character must |
| 255 | // be followed by an alignment specifier, unless it's '0' (zero) |
| 256 | // (in which case it's handled as part of the width specifier) |
| 257 | var fill: ?u21 = comptime if (parser.peek(1)) |ch| |
| 258 | switch (ch) { |
| 259 | '<', '^', '>' => parser.char(), |
| 260 | else => null, |
| 261 | } |
| 262 | else |
| 263 | null; |
| 264 | |
| 265 | // Parse the alignment parameter |
| 266 | const alignment: ?Alignment = comptime if (parser.peek(0)) |ch| init: { |
| 267 | switch (ch) { |
| 268 | '<', '^', '>' => { |
| 269 | // consume the character |
| 270 | break :init switch (parser.char().?) { |
| 271 | '<' => .left, |
| 272 | '^' => .center, |
| 273 | else => .right, |
| 274 | }; |
| 275 | }, |
| 276 | else => break :init null, |
| 277 | } |
| 278 | } else null; |
| 279 | |
| 280 | // When none of the fill character and the alignment specifier have |
| 281 | // been provided, check whether the width starts with a zero. |
| 282 | if (fill == null and alignment == null) { |
| 283 | fill = comptime if (parser.peek(0) == '0') '0' else null; |
| 284 | } |
| 285 | |
| 286 | // Parse the width parameter |
| 287 | const width = comptime parser.specifier() catch |err| @compileError(@errorName(err)); |
| 288 | |
| 289 | // Skip the dot, if present |
| 290 | if (comptime parser.char()) |ch| { |
| 291 | if (ch != '.') { |
| 292 | @compileError("expected . or }, found '" ++ std.unicode.utf8EncodeComptime(ch) ++ "'"); |
| 293 | } |
| 294 | } |
| 295 | |
| 296 | // Parse the precision parameter |
| 297 | const precision = comptime parser.specifier() catch |err| @compileError(@errorName(err)); |
| 298 | |
| 299 | if (comptime parser.char()) |ch| { |
| 300 | @compileError("extraneous trailing character '" ++ std.unicode.utf8EncodeComptime(ch) ++ "'"); |
| 301 | } |
| 302 | |
| 303 | return Placeholder{ |
| 304 | .specifier_arg = cacheString(specifier_arg[0..specifier_arg.len].*), |
| 305 | .fill = fill orelse default_fill_char, |
| 306 | .alignment = alignment orelse default_alignment, |
| 307 | .arg = arg, |
| 308 | .width = width, |
| 309 | .precision = precision, |
| 310 | }; |
| 311 | } |
| 312 | }; |
| 313 | |
| 314 | pub const Specifier = union(enum) { |
| 315 | none, |
| 316 | number: usize, |
| 317 | named: []const u8, |
| 318 | }; |
| 319 | |
| 320 | /// A stream based parser for format strings. |
| 321 | /// |
| 322 | /// Allows to implement formatters compatible with std.fmt without replicating |
| 323 | /// the standard library behavior. |
| 324 | pub const Parser = struct { |
| 325 | iter: std.unicode.Utf8Iterator, |
| 326 | |
| 327 | // Returns a decimal number or null if the current character is not a |
| 328 | // digit |
| 329 | fn number(self: *@This()) ?usize { |
| 330 | var r: ?usize = null; |
| 331 | |
| 332 | while (self.peek(0)) |code_point| { |
| 333 | switch (code_point) { |
| 334 | '0'...'9' => { |
| 335 | if (r == null) r = 0; |
| 336 | r.? *= 10; |
| 337 | r.? += code_point - '0'; |
| 338 | }, |
| 339 | else => break, |
| 340 | } |
| 341 | _ = self.iter.nextCodepoint(); |
| 342 | } |
| 343 | |
| 344 | return r; |
| 345 | } |
| 346 | |
| 347 | // Returns a substring of the input starting from the current position |
| 348 | // and ending where `ch` is found or until the end if not found |
| 349 | fn until(self: *@This(), ch: u21) []const u8 { |
| 350 | const start = self.iter.i; |
| 351 | while (self.peek(0)) |code_point| { |
| 352 | if (code_point == ch) |
| 353 | break; |
| 354 | _ = self.iter.nextCodepoint(); |
| 355 | } |
| 356 | return self.iter.bytes[start..self.iter.i]; |
| 357 | } |
| 358 | |
| 359 | // Returns the character pointed to by the iterator if available, or |
| 360 | // null otherwise |
| 361 | fn char(self: *@This()) ?u21 { |
| 362 | if (self.iter.nextCodepoint()) |code_point| { |
| 363 | return code_point; |
| 364 | } |
| 365 | return null; |
| 366 | } |
| 367 | |
| 368 | // Returns true if the iterator points to an existing character and |
| 369 | // false otherwise |
| 370 | fn maybe(self: *@This(), val: u21) bool { |
| 371 | if (self.peek(0) == val) { |
| 372 | _ = self.iter.nextCodepoint(); |
| 373 | return true; |
| 374 | } |
| 375 | return false; |
| 376 | } |
| 377 | |
| 378 | // Returns a decimal number or null if the current character is not a |
| 379 | // digit |
| 380 | fn specifier(self: *@This()) !Specifier { |
| 381 | if (self.maybe('[')) { |
| 382 | const arg_name = self.until(']'); |
| 383 | |
| 384 | if (!self.maybe(']')) |
| 385 | return @field(anyerror, "Expected closing ]"); |
| 386 | |
| 387 | return Specifier{ .named = arg_name }; |
| 388 | } |
| 389 | if (self.number()) |i| |
| 390 | return Specifier{ .number = i }; |
| 391 | |
| 392 | return Specifier{ .none = {} }; |
| 393 | } |
| 394 | |
| 395 | // Returns the n-th next character or null if that's past the end |
| 396 | fn peek(self: *@This(), n: usize) ?u21 { |
| 397 | const original_i = self.iter.i; |
| 398 | defer self.iter.i = original_i; |
| 399 | |
| 400 | var i: usize = 0; |
| 401 | var code_point: ?u21 = null; |
| 402 | while (i <= n) : (i += 1) { |
| 403 | code_point = self.iter.nextCodepoint(); |
| 404 | if (code_point == null) return null; |
| 405 | } |
| 406 | return code_point; |
| 407 | } |
| 408 | }; |
| 409 | |
| 410 | fn cacheString(str: anytype) []const u8 { |
| 411 | return &str; |
| 412 | } |
| 413 | |
| 414 | fn formatType(value: anytype, comptime fmt: []const u8, options: FormatOptions, writer: anytype, max_depth: usize) extras.Pointee(@TypeOf(writer)).WriteError!void { |
| 415 | const T = @TypeOf(value); |
| 416 | const actual_fmt = comptime if (std.mem.eql(u8, fmt, "any")) |
| 417 | defaultSpec(T) |
| 418 | else if (fmt.len != 0 and (fmt[0] == '?' or fmt[0] == '!')) switch (@typeInfo(T)) { |
| 419 | .optional, .error_union => fmt, |
| 420 | else => stripOptionalOrErrorUnionSpec(fmt), |
| 421 | } else fmt; |
| 422 | |
| 423 | if (comptime std.mem.eql(u8, actual_fmt, "*")) { |
| 424 | return formatAddress(value, options, writer); |
| 425 | } |
| 426 | |
| 427 | if (comptime std.meta.hasMethod(T, "format") and @typeInfo(@TypeOf(T.format)).@"fn".params[1].type.? != *std.Io.Writer) { |
| 428 | @compileError("fix this: " ++ @typeName(T)); |
| 429 | } |
| 430 | if (comptime std.meta.hasMethod(T, "nprint")) { |
| 431 | return value.nprint(writer); |
| 432 | } |
| 433 | |
| 434 | switch (@typeInfo(T)) { |
| 435 | .comptime_int, .int, .comptime_float, .float => { |
| 436 | return formatValue(value, actual_fmt, options, writer); |
| 437 | }, |
| 438 | .void => { |
| 439 | if (actual_fmt.len != 0) invalidFmtError(fmt, value); |
| 440 | return formatBuf("void", options, writer); |
| 441 | }, |
| 442 | .bool => { |
| 443 | if (actual_fmt.len != 0) invalidFmtError(fmt, value); |
| 444 | return formatBuf(if (value) "true" else "false", options, writer); |
| 445 | }, |
| 446 | .optional => { |
| 447 | if (actual_fmt.len == 0 or actual_fmt[0] != '?') @compileError("cannot format optional without a specifier (i.e. {?} or {any})"); |
| 448 | const remaining_fmt = comptime stripOptionalOrErrorUnionSpec(actual_fmt); |
| 449 | if (value) |payload| { |
| 450 | return formatType(payload, remaining_fmt, options, writer, max_depth); |
| 451 | } else { |
| 452 | return formatBuf("null", options, writer); |
| 453 | } |
| 454 | }, |
| 455 | .error_union => { |
| 456 | if (actual_fmt.len == 0 or actual_fmt[0] != '!') @compileError("cannot format error union without a specifier (i.e. {!} or {any})"); |
| 457 | const remaining_fmt = comptime stripOptionalOrErrorUnionSpec(actual_fmt); |
| 458 | if (value) |payload| { |
| 459 | return formatType(payload, remaining_fmt, options, writer, max_depth); |
| 460 | } else |err| { |
| 461 | return formatType(err, "", options, writer, max_depth); |
| 462 | } |
| 463 | }, |
| 464 | .error_set => { |
| 465 | if (actual_fmt.len != 0) invalidFmtError(fmt, value); |
| 466 | try writer.writeAll("error."); |
| 467 | return writer.writeAll(@errorName(value)); |
| 468 | }, |
| 469 | .@"enum" => |enumInfo| { |
| 470 | if (comptime std.mem.eql(u8, actual_fmt, "d")) { |
| 471 | try formatInt(@intFromEnum(value), 10, .lower, options, writer); |
| 472 | return; |
| 473 | } |
| 474 | if (comptime std.mem.eql(u8, actual_fmt, "s") and enumInfo.is_exhaustive) { |
| 475 | try writer.writeAll(@tagName(value)); |
| 476 | return; |
| 477 | } |
| 478 | }, |
| 479 | .@"union" => |info| { |
| 480 | if (actual_fmt.len != 0) invalidFmtError(fmt, value); |
| 481 | try writer.writeAll(@typeName(T)); |
| 482 | if (max_depth == 0) { |
| 483 | return writer.writeAll("{ ... }"); |
| 484 | } |
| 485 | if (info.tag_type) |UnionTagType| { |
| 486 | try writer.writeAll("{ ."); |
| 487 | try writer.writeAll(@tagName(@as(UnionTagType, value))); |
| 488 | try writer.writeAll(" = "); |
| 489 | inline for (info.fields) |u_field| { |
| 490 | if (value == @field(UnionTagType, u_field.name)) { |
| 491 | try formatType(@field(value, u_field.name), "any", options, writer, max_depth - 1); |
| 492 | } |
| 493 | } |
| 494 | try writer.writeAll(" }"); |
| 495 | } else { |
| 496 | try format(writer, "@{x}", .{@intFromPtr(&value)}); |
| 497 | } |
| 498 | return; |
| 499 | }, |
| 500 | .pointer => |ptr_info| switch (ptr_info.size) { |
| 501 | .one => switch (@typeInfo(ptr_info.child)) { |
| 502 | .array, .@"enum", .@"union", .@"struct" => { |
| 503 | return formatType(value.*, actual_fmt, options, writer, max_depth); |
| 504 | }, |
| 505 | else => return format(writer, "{s}@{x}", .{ @typeName(ptr_info.child), @intFromPtr(value) }), |
| 506 | }, |
| 507 | .many, .c => { |
| 508 | if (actual_fmt.len == 0) |
| 509 | @compileError("cannot format pointer without a specifier (i.e. {s} or {*})"); |
| 510 | if (ptr_info.sentinel() != null) { |
| 511 | return formatType(std.mem.span(value), actual_fmt, options, writer, max_depth); |
| 512 | } |
| 513 | if (actual_fmt[0] == 's' and ptr_info.child == u8) { |
| 514 | return formatBuf(std.mem.span(value), options, writer); |
| 515 | } |
| 516 | }, |
| 517 | .slice => { |
| 518 | if (actual_fmt.len == 0) { |
| 519 | @compileError("cannot format slice without a specifier (i.e. {s} or {any})"); |
| 520 | } |
| 521 | if (max_depth == 0) { |
| 522 | return writer.writeAll("{ ... }"); |
| 523 | } |
| 524 | if (actual_fmt[0] == 's' and ptr_info.child == u8) { |
| 525 | return formatBuf(value, options, writer); |
| 526 | } |
| 527 | try writer.writeAll("{ "); |
| 528 | for (value, 0..) |elem, i| { |
| 529 | try formatType(elem, actual_fmt, options, writer, max_depth - 1); |
| 530 | if (i != value.len - 1) { |
| 531 | try writer.writeAll(", "); |
| 532 | } |
| 533 | } |
| 534 | try writer.writeAll(" }"); |
| 535 | return; |
| 536 | }, |
| 537 | }, |
| 538 | .array => |info| { |
| 539 | if (actual_fmt.len == 0) { |
| 540 | @compileError("cannot format array without a specifier (i.e. {s} or {any})"); |
| 541 | } |
| 542 | if (max_depth == 0) { |
| 543 | return writer.writeAll("{ ... }"); |
| 544 | } |
| 545 | if (actual_fmt[0] == 's' and info.child == u8) { |
| 546 | return formatBuf(&value, options, writer); |
| 547 | } |
| 548 | try writer.writeAll("{ "); |
| 549 | for (value, 0..) |elem, i| { |
| 550 | try formatType(elem, actual_fmt, options, writer, max_depth - 1); |
| 551 | if (i < value.len - 1) { |
| 552 | try writer.writeAll(", "); |
| 553 | } |
| 554 | } |
| 555 | try writer.writeAll(" }"); |
| 556 | return; |
| 557 | }, |
| 558 | .vector => |info| { |
| 559 | if (max_depth == 0) { |
| 560 | return writer.writeAll("{ ... }"); |
| 561 | } |
| 562 | try writer.writeAll("{ "); |
| 563 | var i: usize = 0; |
| 564 | while (i < info.len) : (i += 1) { |
| 565 | try formatType(value[i], actual_fmt, options, writer, max_depth - 1); |
| 566 | if (i < info.len - 1) { |
| 567 | try writer.writeAll(", "); |
| 568 | } |
| 569 | } |
| 570 | try writer.writeAll(" }"); |
| 571 | return; |
| 572 | }, |
| 573 | .@"fn" => @compileError("unable to format function body type, use '*const " ++ @typeName(T) ++ "' for a function pointer type"), |
| 574 | .type => { |
| 575 | if (actual_fmt.len != 0) invalidFmtError(fmt, value); |
| 576 | return formatBuf(@typeName(value), options, writer); |
| 577 | }, |
| 578 | .enum_literal => { |
| 579 | if (actual_fmt.len != 0) invalidFmtError(fmt, value); |
| 580 | const buffer = [_]u8{'.'} ++ @tagName(value); |
| 581 | return formatBuf(buffer, options, writer); |
| 582 | }, |
| 583 | .null => { |
| 584 | if (actual_fmt.len != 0) invalidFmtError(fmt, value); |
| 585 | return formatBuf("null", options, writer); |
| 586 | }, |
| 587 | else => @compileError("unable to format type '" ++ @typeName(T) ++ "'"), |
| 588 | } |
| 589 | @compileError("unable to format type '" ++ @typeName(T) ++ "' with specifier '" ++ actual_fmt ++ "'"); |
| 590 | } |
| 591 | |
| 592 | pub inline fn comptimePrint(comptime fmt: []const u8, args: anytype) *const [count(fmt, args):0]u8 { |
| 593 | comptime { |
| 594 | var buf: [count(fmt, args):0]u8 = undefined; |
| 595 | _ = bufPrint(&buf, fmt, args) catch unreachable; |
| 596 | buf[buf.len] = 0; |
| 597 | const final = buf; |
| 598 | return &final; |
| 599 | } |
| 600 | } |
| 601 | |
| 602 | fn defaultSpec(comptime T: type) [:0]const u8 { |
| 603 | switch (@typeInfo(T)) { |
| 604 | .array, .vector => return "any", |
| 605 | .pointer => |ptr_info| switch (ptr_info.size) { |
| 606 | .one => switch (@typeInfo(ptr_info.child)) { |
| 607 | .array => return "any", |
| 608 | else => {}, |
| 609 | }, |
| 610 | .many, .c => return "*", |
| 611 | .slice => return "any", |
| 612 | }, |
| 613 | .optional => |info| return "?" ++ defaultSpec(info.child), |
| 614 | .error_union => |info| return "!" ++ defaultSpec(info.payload), |
| 615 | else => {}, |
| 616 | } |
| 617 | return ""; |
| 618 | } |
| 619 | |
| 620 | fn stripOptionalOrErrorUnionSpec(comptime fmt: []const u8) []const u8 { |
| 621 | return if (std.mem.eql(u8, fmt[1..], "any")) |
| 622 | "any" |
| 623 | else |
| 624 | fmt[1..]; |
| 625 | } |
| 626 | |
| 627 | fn formatAddress(value: anytype, options: FormatOptions, writer: anytype) @TypeOf(writer).Error!void { |
| 628 | _ = options; |
| 629 | const T = @TypeOf(value); |
| 630 | |
| 631 | switch (@typeInfo(T)) { |
| 632 | .pointer => |info| { |
| 633 | try writer.writeAll(@typeName(info.child) ++ "@"); |
| 634 | if (info.size == .slice) |
| 635 | try formatInt(@intFromPtr(value.ptr), 16, .lower, FormatOptions{}, writer) |
| 636 | else |
| 637 | try formatInt(@intFromPtr(value), 16, .lower, FormatOptions{}, writer); |
| 638 | return; |
| 639 | }, |
| 640 | .optional => |info| { |
| 641 | if (@typeInfo(info.child) == .pointer) { |
| 642 | try writer.writeAll(@typeName(info.child) ++ "@"); |
| 643 | try formatInt(@intFromPtr(value), 16, .lower, FormatOptions{}, writer); |
| 644 | return; |
| 645 | } |
| 646 | }, |
| 647 | else => {}, |
| 648 | } |
| 649 | |
| 650 | @compileError("cannot format non-pointer type " ++ @typeName(T) ++ " with * specifier"); |
| 651 | } |
| 652 | |
| 653 | pub fn formatInt(value: anytype, base: u8, case: Case, options: FormatOptions, writer: anytype) !void { |
| 654 | std.debug.assert(base >= 2); |
| 655 | |
| 656 | const int_value = if (@TypeOf(value) == comptime_int) blk: { |
| 657 | const Int = std.math.IntFittingRange(value, value); |
| 658 | break :blk @as(Int, value); |
| 659 | } else value; |
| 660 | |
| 661 | const value_info = @typeInfo(@TypeOf(int_value)).int; |
| 662 | |
| 663 | // The type must have the same size as `base` or be wider in order for the |
| 664 | // division to work |
| 665 | const min_int_bits = comptime @max(value_info.bits, 8); |
| 666 | const MinInt = std.meta.Int(.unsigned, min_int_bits); |
| 667 | |
| 668 | const abs_value = @abs(int_value); |
| 669 | // The worst case in terms of space needed is base 2, plus 1 for the sign |
| 670 | var buf: [1 + @max(@as(comptime_int, value_info.bits), 1)]u8 = undefined; |
| 671 | |
| 672 | var a: MinInt = abs_value; |
| 673 | var index: usize = buf.len; |
| 674 | |
| 675 | if (base == 10) { |
| 676 | while (a >= 100) : (a = @divTrunc(a, 100)) { |
| 677 | index -= 2; |
| 678 | buf[index..][0..2].* = digits2(@intCast(a % 100)); |
| 679 | } |
| 680 | |
| 681 | if (a < 10) { |
| 682 | index -= 1; |
| 683 | buf[index] = '0' + @as(u8, @intCast(a)); |
| 684 | } else { |
| 685 | index -= 2; |
| 686 | buf[index..][0..2].* = digits2(@intCast(a)); |
| 687 | } |
| 688 | } else { |
| 689 | while (true) { |
| 690 | const digit = a % base; |
| 691 | index -= 1; |
| 692 | buf[index] = digitToChar(@intCast(digit), case); |
| 693 | a /= base; |
| 694 | if (a == 0) break; |
| 695 | } |
| 696 | } |
| 697 | |
| 698 | if (value_info.signedness == .signed) { |
| 699 | if (value < 0) { |
| 700 | // Negative integer |
| 701 | index -= 1; |
| 702 | buf[index] = '-'; |
| 703 | } else if (options.width == null or options.width.? == 0) { |
| 704 | // Positive integer, omit the plus sign |
| 705 | } else { |
| 706 | // Positive integer |
| 707 | index -= 1; |
| 708 | buf[index] = '+'; |
| 709 | } |
| 710 | } |
| 711 | |
| 712 | return formatBuf(buf[index..], options, writer); |
| 713 | } |
| 714 | |
| 715 | fn formatValue(value: anytype, comptime fmt: []const u8, options: FormatOptions, writer: anytype) !void { |
| 716 | const T = @TypeOf(value); |
| 717 | switch (@typeInfo(T)) { |
| 718 | .float, .comptime_float => return formatFloatValue(value, fmt, options, writer), |
| 719 | .int, .comptime_int => return formatIntValue(value, fmt, options, writer), |
| 720 | .bool => return formatBuf(if (value) "true" else "false", options, writer), |
| 721 | else => comptime unreachable, |
| 722 | } |
| 723 | } |
| 724 | |
| 725 | fn invalidFmtError(comptime fmt: []const u8, value: anytype) void { |
| 726 | @compileError("invalid format string '" ++ fmt ++ "' for type '" ++ @typeName(@TypeOf(value)) ++ "'"); |
| 727 | } |
| 728 | |
| 729 | fn formatBuf(buf: []const u8, options: FormatOptions, writer: anytype) !void { |
| 730 | if (options.width) |min_width| { |
| 731 | // In case of error assume the buffer content is ASCII-encoded |
| 732 | const width = std.unicode.utf8CountCodepoints(buf) catch buf.len; |
| 733 | const padding = if (width < min_width) min_width - width else 0; |
| 734 | |
| 735 | if (padding == 0) return writer.writeAll(buf); |
| 736 | |
| 737 | var fill_buffer: [4]u8 = undefined; |
| 738 | const fill_utf8 = if (std.unicode.utf8Encode(options.fill, &fill_buffer)) |len| |
| 739 | fill_buffer[0..len] |
| 740 | else |err| switch (err) { |
| 741 | error.Utf8CannotEncodeSurrogateHalf, |
| 742 | error.CodepointTooLarge, |
| 743 | => &std.unicode.utf8EncodeComptime(std.unicode.replacement_character), |
| 744 | }; |
| 745 | switch (options.alignment) { |
| 746 | .left => { |
| 747 | try writer.writeAll(buf); |
| 748 | try writer.writeNTimes(fill_utf8, padding); |
| 749 | }, |
| 750 | .center => { |
| 751 | const left_padding = padding / 2; |
| 752 | const right_padding = (padding + 1) / 2; |
| 753 | try writer.writeNTimes(fill_utf8, left_padding); |
| 754 | try writer.writeAll(buf); |
| 755 | try writer.writeNTimes(fill_utf8, right_padding); |
| 756 | }, |
| 757 | .right => { |
| 758 | try writer.writeNTimes(fill_utf8, padding); |
| 759 | try writer.writeAll(buf); |
| 760 | }, |
| 761 | } |
| 762 | } else { |
| 763 | // Fast path, avoid counting the number of codepoints |
| 764 | try writer.writeAll(buf); |
| 765 | } |
| 766 | } |
| 767 | |
| 768 | /// Count the characters needed for format. Useful for preallocating memory |
| 769 | fn count(comptime fmt: []const u8, args: anytype) u64 { |
| 770 | var counting_writer: nio.CountingWriter(nio.NullWriter) = .init(.{}); |
| 771 | format(&counting_writer, fmt, args) catch unreachable; |
| 772 | return counting_writer.bytes_written; |
| 773 | } |
| 774 | |
| 775 | /// Print a Formatter string into `buf`. Actually just a thin wrapper around `format` and `fixedBufferStream`. |
| 776 | /// Returns a slice of the bytes printed to. |
| 777 | pub fn bufPrint(buf: []u8, comptime fmt: []const u8, args: anytype) ![]u8 { |
| 778 | var fbs: nio.FixedBufferStream([]u8) = .init(buf); |
| 779 | format(&fbs, fmt, args) catch |err| switch (err) { |
| 780 | error.NoSpaceLeft => return error.NoSpaceLeft, |
| 781 | else => unreachable, |
| 782 | }; |
| 783 | return fbs.written(); |
| 784 | } |
| 785 | pub fn bufPrintZ(buf: []u8, comptime fmt: []const u8, args: anytype) ![:0]u8 { |
| 786 | const result = try bufPrint(buf, fmt ++ "\x00", args); |
| 787 | return result[0 .. result.len - 1 :0]; |
| 788 | } |
| 789 | |
| 790 | const digits2_alphabet = blk: { |
| 791 | var data: [200]u8 = @splat(0); |
| 792 | for (0..10) |m| { |
| 793 | for (0..10) |n| { |
| 794 | data[(m * 10 + n) * 2 + 0] = m + '0'; |
| 795 | data[(m * 10 + n) * 2 + 1] = n + '0'; |
| 796 | } |
| 797 | } |
| 798 | const result = data; |
| 799 | break :blk result; |
| 800 | }; |
| 801 | |
| 802 | /// Converts values in the range [0, 100) to a base 10 string. |
| 803 | pub fn digits2(value: u8) [2]u8 { |
| 804 | return digits2_alphabet[value * 2 ..][0..2].*; |
| 805 | } |
| 806 | |
| 807 | pub fn digitToChar(digit: u8, case: Case) u8 { |
| 808 | return switch (digit) { |
| 809 | 0...9 => digit + '0', |
| 810 | 10...35 => digit + (if (case == .upper) @as(u8, 'A') else @as(u8, 'a')) - 10, |
| 811 | else => unreachable, |
| 812 | }; |
| 813 | } |
| 814 | |
| 815 | fn formatIntValue(value: anytype, comptime fmt: []const u8, options: FormatOptions, writer: anytype) !void { |
| 816 | comptime var base = 10; |
| 817 | comptime var case: Case = .lower; |
| 818 | |
| 819 | const int_value = if (@TypeOf(value) == comptime_int) blk: { |
| 820 | const Int = std.math.IntFittingRange(value, value); |
| 821 | break :blk @as(Int, value); |
| 822 | } else value; |
| 823 | |
| 824 | if (fmt.len == 0 or comptime std.mem.eql(u8, fmt, "d")) { |
| 825 | base = 10; |
| 826 | case = .lower; |
| 827 | } else if (comptime std.mem.eql(u8, fmt, "c")) { |
| 828 | if (@typeInfo(@TypeOf(int_value)).int.bits <= 8) { |
| 829 | return formatAsciiChar(@as(u8, int_value), options, writer); |
| 830 | } else { |
| 831 | @compileError("cannot print integer that is larger than 8 bits as an ASCII character"); |
| 832 | } |
| 833 | } else if (comptime std.mem.eql(u8, fmt, "u")) { |
| 834 | if (@typeInfo(@TypeOf(int_value)).int.bits <= 21) { |
| 835 | return formatUnicodeCodepoint(@as(u21, int_value), options, writer); |
| 836 | } else { |
| 837 | @compileError("cannot print integer that is larger than 21 bits as an UTF-8 sequence"); |
| 838 | } |
| 839 | } else if (comptime std.mem.eql(u8, fmt, "b")) { |
| 840 | base = 2; |
| 841 | case = .lower; |
| 842 | } else if (comptime std.mem.eql(u8, fmt, "x")) { |
| 843 | base = 16; |
| 844 | case = .lower; |
| 845 | } else if (comptime std.mem.eql(u8, fmt, "X")) { |
| 846 | base = 16; |
| 847 | case = .upper; |
| 848 | } else if (comptime std.mem.eql(u8, fmt, "o")) { |
| 849 | base = 8; |
| 850 | case = .lower; |
| 851 | } else { |
| 852 | invalidFmtError(fmt, value); |
| 853 | } |
| 854 | |
| 855 | return formatInt(int_value, base, case, options, writer); |
| 856 | } |
| 857 | |
| 858 | fn formatAsciiChar(c: u8, options: FormatOptions, writer: anytype) !void { |
| 859 | return formatBuf(@as(*const [1]u8, &c), options, writer); |
| 860 | } |
| 861 | |
| 862 | fn formatUnicodeCodepoint(c: u21, options: FormatOptions, writer: anytype) !void { |
| 863 | var buf: [4]u8 = undefined; |
| 864 | const len = std.unicode.utf8Encode(c, &buf) catch |err| switch (err) { |
| 865 | error.Utf8CannotEncodeSurrogateHalf, error.CodepointTooLarge => { |
| 866 | return formatBuf(&std.unicode.utf8EncodeComptime(std.unicode.replacement_character), options, writer); |
| 867 | }, |
| 868 | }; |
| 869 | return formatBuf(buf[0..len], options, writer); |
| 870 | } |
| 871 | |
| 872 | fn formatFloatValue(value: anytype, comptime fmt: []const u8, options: FormatOptions, writer: anytype) !void { |
| 873 | const v = switch (@TypeOf(value)) { |
| 874 | comptime_float => @as(f128, value), // comptime_float internally is a f128; this preserves precision. |
| 875 | else => value, |
| 876 | }; |
| 877 | const T = @TypeOf(v); |
| 878 | comptime std.debug.assert(@typeInfo(T) == .float); |
| 879 | const I = @Int(.unsigned, @bitSizeOf(T)); |
| 880 | const DT = if (@bitSizeOf(T) <= 64) u64 else u128; |
| 881 | const tables = switch (DT) { |
| 882 | u64 => &std.fmt.float.Backend64_TablesFull, |
| 883 | u128 => &std.fmt.float.Backend128_Tables, |
| 884 | else => unreachable, |
| 885 | }; |
| 886 | const has_explicit_leading_bit = std.math.floatMantissaBits(T) - std.math.floatFractionalBits(T) != 0; |
| 887 | const d = std.fmt.float.binaryToDecimal(DT, @as(I, @bitCast(v)), std.math.floatMantissaBits(T), std.math.floatExponentBits(T), has_explicit_leading_bit, tables); |
| 888 | var buf: [std.fmt.float.bufferSize(.decimal, T)]u8 = undefined; |
| 889 | |
| 890 | if (fmt.len == 0 or comptime std.mem.eql(u8, fmt, "e")) { |
| 891 | try writer.writeAll(std.fmt.float.formatScientific(DT, &buf, d, options.precision) catch unreachable); |
| 892 | } else if (comptime std.mem.eql(u8, fmt, "d")) { |
| 893 | try writer.writeAll(std.fmt.float.formatDecimal(DT, &buf, d, options.precision) catch unreachable); |
| 894 | } else if (comptime std.mem.eql(u8, fmt, "x")) { |
| 895 | var w: std.Io.Writer = .fixed(&buf); |
| 896 | w.printFloatHex(v, .lower, options.precision) catch unreachable; |
| 897 | try writer.writeAll(buf[0..w.end]); |
| 898 | } else { |
| 899 | invalidFmtError(fmt, value); |
| 900 | } |
| 901 | } |
| 902 | |
| 903 | pub fn allocPrint(allocator: std.mem.Allocator, comptime fmt: []const u8, args: anytype) ![]u8 { |
| 904 | var aw: nio.AllocatingWriter = .init(allocator); |
| 905 | errdefer aw.deinit(); |
| 906 | try aw.ensureUnusedCapacity(fmt.len); |
| 907 | try format(&aw, fmt, args); |
| 908 | return aw.toOwnedSlice(); |
| 909 | } |
| 910 | pub fn allocPrintZ(allocator: std.mem.Allocator, comptime fmt: []const u8, args: anytype) ![:0]u8 { |
| 911 | const result = try allocPrint(allocator, fmt ++ "\x00", args); |
| 912 | return result[0 .. result.len - 1 :0]; |
| 913 | } |