zig/lib/std / fmt.zig

String formatting and parsing.

//! String formatting and parsing.

default_max_depth

Renders fmt string with args, calling writer with slices of bytes. If writer returns an error, the error is returned from format and writer is not called again. The format string must be comptime-known and may contain placeholders following this format: {[argument][specifier]:[fill][alignment][width].[precision]} Above, each word including its surrounding [ and ] is a parameter which you have to replace with something: - *argument* is either the numeric index or the field name of the argument that should be inserted - when using a field name, you are required to enclose the field name (an identifier) in square brackets, e.g. {[score]...} as opposed to the numeric index form which can be written e.g. {2...} - *specifier* is a type-dependent formatting option that determines how a type should formatted (see below) - *fill* is a single unicode codepoint which is used to pad the formatted text - *alignment* is one of the three bytes '<', '^', or '>' to make the text left-, center-, or right-aligned, respectively - *width* is the total width of the field in unicode codepoints - *precision* specifies how many decimals a formatted number should have Note that most of the parameters are optional and may be omitted. Also you can leave out separators like : and . when all parameters after the separator are omitted. Only exception is the *fill* parameter. If *fill* is required, one has to specify *alignment* as well, as otherwise the digits after : is interpreted as *width*, not *fill*. The *specifier* has several options for types: - x and X: output numeric value in hexadecimal notation - s: - for pointer-to-many and C pointers of u8, print as a C-string using zero-termination - for slices of u8, print the entire slice as a string without zero-termination - e: output floating point value in scientific notation - d: output numeric value in decimal notation - b: output integer value in binary notation - o: output integer value in octal notation - c: output integer as an ASCII character. Integer type must have 8 bits at max. - u: output integer as an UTF-8 sequence. Integer type must have 21 bits at max. - ?: output optional value as either the unwrapped value, or null; may be followed by a format specifier for the underlying value. - !: 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. - *: output the address of the value instead of the value itself. - any: output a value of any type using its default format. If a formatted user type contains a function of the type pub fn format(value: ?, comptime fmt: []const u8, options: std.fmt.FormatOptions, writer: anytype) !void with ? being the type formatted, this function will be called instead of the default implementation. This allows user types to be formatted in a logical manner instead of dumping all fields of the type. A user type may be a struct, vector, union or enum type. To print literal curly braces, escape them by writing them twice, e.g. {{ or }}.


const std = @import("std.zig");
const builtin = @import("builtin");

Alignment

Return a Formatter for a []const u8 where every byte is formatted as a pair of lowercase hexadecimal digits.


const io = std.io;
const math = std.math;
const assert = std.debug.assert;
const mem = std.mem;
const unicode = std.unicode;
const meta = std.meta;
const lossyCast = std.math.lossyCast;
const expectFmt = std.testing.expectFmt;

FormatOptions

Return a Formatter for a []const u8 where every byte is formatted as pair of uppercase hexadecimal digits.


pub const default_max_depth = 3;

format()

Return a Formatter for a []const u8 where every non-printable ASCII character is escaped as \xNN, where NN is the character in lowercase hexadecimal notation.


pub const Alignment = enum {
    left,
    center,
    right,

Case

Return a Formatter for a []const u8 where every non-printable ASCII character is escaped as \xNN, where NN is the character in uppercase hexadecimal notation.

};

parse()

Return a Formatter for a u64 value representing a file size. This formatter represents the number as multiple of 1000 and uses the SI measurement units (kB, MB, GB, ...).


pub const FormatOptions = struct {
    precision: ?usize = null,
    width: ?usize = null,
    alignment: Alignment = .right,
    fill: u21 = ' ',

Case

Return a Formatter for a u64 value representing a file size. This formatter represents the number as multiple of 1024 and uses the IEC measurement units (KiB, MiB, GiB, ...).

};

Parser

Return a Formatter for number of nanoseconds according to its magnitude: [#y][#w][#d][#h][#m]#[.###][n|u|m]s


/// Renders fmt string with args, calling `writer` with slices of bytes.
/// If `writer` returns an error, the error is returned from `format` and
/// `writer` is not called again.
///
/// The format string must be comptime-known and may contain placeholders following
/// this format:
/// `{[argument][specifier]:[fill][alignment][width].[precision]}`
///
/// Above, each word including its surrounding [ and ] is a parameter which you have to replace with something:
///
/// - *argument* is either the numeric index or the field name of the argument that should be inserted
///   - when using a field name, you are required to enclose the field name (an identifier) in square
///     brackets, e.g. {[score]...} as opposed to the numeric index form which can be written e.g. {2...}
/// - *specifier* is a type-dependent formatting option that determines how a type should formatted (see below)
/// - *fill* is a single unicode codepoint which is used to pad the formatted text
/// - *alignment* is one of the three bytes '<', '^', or '>' to make the text left-, center-, or right-aligned, respectively
/// - *width* is the total width of the field in unicode codepoints
/// - *precision* specifies how many decimals a formatted number should have
///
/// Note that most of the parameters are optional and may be omitted. Also you can leave out separators like `:` and `.` when
/// all parameters after the separator are omitted.
/// Only exception is the *fill* parameter. If *fill* is required, one has to specify *alignment* as well, as otherwise
/// the digits after `:` is interpreted as *width*, not *fill*.
///
/// The *specifier* has several options for types:
/// - `x` and `X`: output numeric value in hexadecimal notation
/// - `s`:
///   - for pointer-to-many and C pointers of u8, print as a C-string using zero-termination
///   - for slices of u8, print the entire slice as a string without zero-termination
/// - `e`: output floating point value in scientific notation
/// - `d`: output numeric value in decimal notation
/// - `b`: output integer value in binary notation
/// - `o`: output integer value in octal notation
/// - `c`: output integer as an ASCII character. Integer type must have 8 bits at max.
/// - `u`: output integer as an UTF-8 sequence. Integer type must have 21 bits at max.
/// - `?`: output optional value as either the unwrapped value, or `null`; may be followed by a format specifier for the underlying value.
/// - `!`: 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.
/// - `*`: output the address of the value instead of the value itself.
/// - `any`: output a value of any type using its default format.
///
/// If a formatted user type contains a function of the type
/// ```
/// pub fn format(value: ?, comptime fmt: []const u8, options: std.fmt.FormatOptions, writer: anytype) !void
/// ```
/// with `?` being the type formatted, this function will be called instead of the default implementation.
/// This allows user types to be formatted in a logical manner instead of dumping all fields of the type.
///
/// A user type may be a `struct`, `vector`, `union` or `enum` type.
///
/// To print literal curly braces, escape them by writing them twice, e.g. `{{` or `}}`.

format()

Return a Formatter for number of nanoseconds according to its signed magnitude: [#y][#w][#d][#h][#m]#[.###][n|u|m]s

pub fn format(
    writer: anytype,
    comptime fmt: []const u8,
    args: anytype,

format()

The result cannot fit in the type specified

) !void {
    const ArgsType = @TypeOf(args);
    const args_type_info = @typeInfo(ArgsType);
    if (args_type_info != .Struct) {
        @compileError("expected tuple or struct argument, found " ++ @typeName(ArgsType));
    }

char()

The input was empty or contained an invalid character


    const fields_info = args_type_info.Struct.fields;
    if (fields_info.len > max_format_args) {
        @compileError("32 arguments max are supported per format call");
    }

maybe()

Creates a Formatter type from a format function. Wrapping data in Formatter(func) causes the data to be formatted using the given function func. func must be of the following form: fn formatExample( data: T, comptime fmt: []const u8, options: std.fmt.FormatOptions, writer: anytype, ) !void;


    @setEvalBranchQuota(2000000);
    comptime var arg_state: ArgState = .{ .args_len = fields_info.len };
    comptime var i = 0;
    inline while (i < fmt.len) {
        const start_index = i;

specifier()

Parses the string buf as signed or unsigned representation in the specified base of an integral value of type T. When base is zero the string prefix is examined to detect the true base: * A prefix of "0b" implies base=2, * A prefix of "0o" implies base=8, * A prefix of "0x" implies base=16, * Otherwise base=10 is assumed. Ignores '_' character in buf. See also parseUnsigned.


        inline while (i < fmt.len) : (i += 1) {
            switch (fmt[i]) {
                '{', '}' => break,
                else => {},
            }
        }

peek()

Parses the string buf as unsigned representation in the specified base of an integral value of type T. When base is zero the string prefix is examined to detect the true base: * A prefix of "0b" implies base=2, * A prefix of "0o" implies base=8, * A prefix of "0x" implies base=16, * Otherwise base=10 is assumed. Ignores '_' character in buf. See also parseInt.


        comptime var end_index = i;
        comptime var unescape_brace = false;

ArgSetType

Parses a number like '2G', '2Gi', or '2GiB'.


        // Handle {{ and }}, those are un-escaped as single braces
        if (i + 1 < fmt.len and fmt[i + 1] == fmt[i]) {
            unescape_brace = true;
            // Make the first brace part of the literal...
            end_index += 1;
            // ...and skip both
            i += 2;
        }

ArgState

As much as possible was written to the buffer, but it was too small to fit all the printed bytes.


        // Write out the literal
        if (start_index != end_index) {
            try writer.writeAll(fmt[start_index..end_index]);
        }

hasUnusedArgs()

Print a Formatter string into buf. Actually just a thin wrapper around format and fixedBufferStream. Returns a slice of the bytes printed to.


        // We've already skipped the other brace, restart the loop
        if (unescape_brace) continue;

nextArg()

Count the characters needed for format. Useful for preallocating memory


        if (i >= fmt.len) break;

formatAddress()

Encodes a sequence of bytes as hexadecimal digits. Returns an array containing the encoded bytes.


        if (fmt[i] == '}') {
            @compileError("missing opening {");
        }

defaultSpec()

Decodes the sequence of bytes represented by the specified string of hexadecimal characters. Returns a slice of the output buffer containing the decoded bytes.


        // Get past the {
        comptime assert(fmt[i] == '{');
        i += 1;

invalidFmtError()


        const fmt_begin = i;
        // Find the closing brace
        inline while (i < fmt.len and fmt[i] != '}') : (i += 1) {}
        const fmt_end = i;

formatType()


        if (i >= fmt.len) {
            @compileError("missing closing }");
        }

formatIntValue()


        // Get past the }
        comptime assert(fmt[i] == '}');
        i += 1;

format_float

fmt/format_float.zig

        const placeholder = comptime Placeholder.parse(fmt[fmt_begin..fmt_end].*);
        const arg_pos = comptime switch (placeholder.arg) {
            .none => null,
            .number => |pos| pos,
            .named => |arg_name| meta.fieldIndex(ArgsType, arg_name) orelse
                @compileError("no argument with name '" ++ arg_name ++ "'"),
        };

formatFloat


        const width = switch (placeholder.width) {
            .none => null,
            .number => |v| v,
            .named => |arg_name| blk: {
                const arg_i = comptime meta.fieldIndex(ArgsType, arg_name) orelse
                    @compileError("no argument with name '" ++ arg_name ++ "'");
                _ = comptime arg_state.nextArg(arg_i) orelse @compileError("too few arguments");
                break :blk @field(args, arg_name);
            },
        };

FormatFloatError


        const precision = switch (placeholder.precision) {
            .none => null,
            .number => |v| v,
            .named => |arg_name| blk: {
                const arg_i = comptime meta.fieldIndex(ArgsType, arg_name) orelse
                    @compileError("no argument with name '" ++ arg_name ++ "'");
                _ = comptime arg_state.nextArg(arg_i) orelse @compileError("too few arguments");
                break :blk @field(args, arg_name);
            },
        };

Case


        const arg_to_print = comptime arg_state.nextArg(arg_pos) orelse
            @compileError("too few arguments");

formatSliceHexImpl()


        try formatType(
            @field(args, fields_info[arg_to_print].name),
            placeholder.specifier_arg,
            FormatOptions{
                .fill = placeholder.fill,
                .alignment = placeholder.alignment,
                .width = width,
                .precision = precision,
            },
            writer,
            std.options.fmt_max_depth,
        );
    }

fmtSliceHexLower()


    if (comptime arg_state.hasUnusedArgs()) {
        const missing_count = arg_state.args_len - @popCount(arg_state.used_args);
        switch (missing_count) {
            0 => unreachable,
            1 => @compileError("unused argument in '" ++ fmt ++ "'"),
            else => @compileError(comptimePrint("{d}", .{missing_count}) ++ " unused arguments in '" ++ fmt ++ "'"),
        }
    }

charToDigit()

}

formatSliceEscapeImpl()


fn cacheString(str: anytype) []const u8 {
    return &str;

charToDigit()

}

fmtSliceEscapeUpper()


pub const Placeholder = struct {
    specifier_arg: []const u8,
    fill: u21,
    alignment: Alignment,
    arg: Specifier,
    width: Specifier,
    precision: Specifier,

fmtIntSizeDec()


    pub fn parse(comptime str: anytype) Placeholder {
        const view = std.unicode.Utf8View.initComptime(&str);
        comptime var parser = Parser{
            .buf = &str,
            .iter = view.iterator(),
        };

fmtIntSizeBin()


        // Parse the positional argument number
        const arg = comptime parser.specifier() catch |err|
            @compileError(@errorName(err));

formatText()


        // Parse the format specifier
        const specifier_arg = comptime parser.until(':');

formatAsciiChar()


        // Skip the colon, if present
        if (comptime parser.char()) |ch| {
            if (ch != ':') {
                @compileError("expected : or }, found '" ++ unicode.utf8EncodeComptime(ch) ++ "'");
            }
        }

formatUnicodeCodepoint()


        // Parse the fill character
        // The fill parameter requires the alignment parameter to be specified
        // too
        const fill = comptime if (parser.peek(1)) |ch|
            switch (ch) {
                '<', '^', '>' => parser.char().?,
                else => ' ',
            }
        else
            ' ';

formatBuf()


        // Parse the alignment parameter
        const alignment: Alignment = comptime if (parser.peek(0)) |ch| init: {
            switch (ch) {
                '<', '^', '>' => _ = parser.char(),
                else => {},
            }
            break :init switch (ch) {
                '<' => .left,
                '^' => .center,
                else => .right,
            };
        } else .right;

formatFloatHexadecimal()


        // Parse the width parameter
        const width = comptime parser.specifier() catch |err|
            @compileError(@errorName(err));

formatInt()


        // Skip the dot, if present
        if (comptime parser.char()) |ch| {
            if (ch != '.') {
                @compileError("expected . or }, found '" ++ unicode.utf8EncodeComptime(ch) ++ "'");
            }
        }

formatIntBuf()


        // Parse the precision parameter
        const precision = comptime parser.specifier() catch |err|
            @compileError(@errorName(err));

digits2()


        if (comptime parser.char()) |ch| {
            @compileError("extraneous trailing character '" ++ unicode.utf8EncodeComptime(ch) ++ "'");
        }

fmtDuration()


        return Placeholder{
            .specifier_arg = cacheString(specifier_arg[0..specifier_arg.len].*),
            .fill = fill,
            .alignment = alignment,
            .arg = arg,
            .width = width,
            .precision = precision,
        };
    }

AllocPrintError

};

fmtDurationSigned()


pub const Specifier = union(enum) {
    none,
    number: usize,
    named: []const u8,

AllocPrintError

};

ParseIntError


pub const Parser = struct {
    buf: []const u8,
    pos: usize = 0,
    iter: std.unicode.Utf8Iterator = undefined,

Formatter()


    // Returns a decimal number or null if the current character is not a
    // digit
    pub fn number(self: *@This()) ?usize {
        var r: ?usize = null;

format()


        while (self.peek(0)) |code_point| {
            switch (code_point) {
                '0'...'9' => {
                    if (r == null) r = 0;
                    r.? *= 10;
                    r.? += code_point - '0';
                },
                else => break,
            }
            _ = self.iter.nextCodepoint();
        }

parseInt()


        return r;
    }

Test: parseInt


    // Returns a substring of the input starting from the current position
    // and ending where `ch` is found or until the end if not found
    pub fn until(self: *@This(), ch: u21) []const u8 {
        var result: []const u8 = &[_]u8{};
        while (self.peek(0)) |code_point| {
            if (code_point == ch)
                break;
            result = result ++ (self.iter.nextCodepointSlice() orelse &[_]u8{});
        }
        return result;
    }

parseUnsigned()


    // Returns one character, if available
    pub fn char(self: *@This()) ?u21 {
        if (self.iter.nextCodepoint()) |code_point| {
            return code_point;
        }
        return null;
    }

Test: parseUnsigned


    pub fn maybe(self: *@This(), val: u21) bool {
        if (self.peek(0) == val) {
            _ = self.iter.nextCodepoint();
            return true;
        }
        return false;
    }

parseIntSizeSuffix()


    // Returns a decimal number or null if the current character is not a
    // digit
    pub fn specifier(self: *@This()) !Specifier {
        if (self.maybe('[')) {
            const arg_name = self.until(']');

Test: parseIntSizeSuffix


            if (!self.maybe(']'))
                return @field(anyerror, "Expected closing ]");

parseFloat

fmt/parse_float.zig

            return Specifier{ .named = arg_name };
        }
        if (self.number()) |i|
            return Specifier{ .number = i };

ParseFloatError

fmt/parse_float.zig

        return Specifier{ .none = {} };
    }

charToDigit()


    // Returns the n-th next character or null if that's past the end
    pub fn peek(self: *@This(), n: usize) ?u21 {
        const original_i = self.iter.i;
        defer self.iter.i = original_i;

digitToChar()


        var i = 0;
        var code_point: ?u21 = null;
        while (i <= n) : (i += 1) {
            code_point = self.iter.nextCodepoint();
            if (code_point == null) return null;
        }
        return code_point;
    }

AllocPrintError

};

bufPrint()


pub const ArgSetType = u32;
const max_format_args = @typeInfo(ArgSetType).Int.bits;

bufPrintZ()


pub const ArgState = struct {
    next_arg: usize = 0,
    used_args: ArgSetType = 0,
    args_len: usize,

count()


    pub fn hasUnusedArgs(self: *@This()) bool {
        return @popCount(self.used_args) != self.args_len;
    }

AllocPrintError


    pub fn nextArg(self: *@This(), arg_index: ?usize) ?usize {
        const next_index = arg_index orelse init: {
            const arg = self.next_arg;
            self.next_arg += 1;
            break :init arg;
        };

allocPrint()


        if (next_index >= self.args_len) {
            return null;
        }

allocPrintZ()


        // Mark this argument as used
        self.used_args |= @as(ArgSetType, 1) << @as(u5, @intCast(next_index));
        return next_index;
    }
};

Test: bufPrintIntToSlice


pub fn formatAddress(value: anytype, options: FormatOptions, writer: anytype) @TypeOf(writer).Error!void {
    _ = options;
    const T = @TypeOf(value);

bufPrintIntToSlice()


    switch (@typeInfo(T)) {
        .Pointer => |info| {
            try writer.writeAll(@typeName(info.child) ++ "@");
            if (info.size == .Slice)
                try formatInt(@intFromPtr(value.ptr), 16, .lower, FormatOptions{}, writer)
            else
                try formatInt(@intFromPtr(value), 16, .lower, FormatOptions{}, writer);
            return;
        },
        .Optional => |info| {
            if (@typeInfo(info.child) == .Pointer) {
                try writer.writeAll(@typeName(info.child) ++ "@");
                try formatInt(@intFromPtr(value), 16, .lower, FormatOptions{}, writer);
                return;
            }
        },
        else => {},
    }

comptimePrint()


    @compileError("cannot format non-pointer type " ++ @typeName(T) ++ " with * specifier");
}

Test: comptimePrint


// This ANY const is a workaround for: https://github.com/ziglang/zig/issues/7948
const ANY = "any";

Test:

parse u64 digit too big


pub fn defaultSpec(comptime T: type) [:0]const u8 {
    switch (@typeInfo(T)) {
        .Array => |_| return ANY,
        .Pointer => |ptr_info| switch (ptr_info.size) {
            .One => switch (@typeInfo(ptr_info.child)) {
                .Array => |_| return ANY,
                else => {},
            },
            .Many, .C => return "*",
            .Slice => return ANY,
        },
        .Optional => |info| return "?" ++ defaultSpec(info.child),
        .ErrorUnion => |info| return "!" ++ defaultSpec(info.payload),
        else => {},
    }
    return "";
}

Test:

parse unsigned comptime


fn stripOptionalOrErrorUnionSpec(comptime fmt: []const u8) []const u8 {
    return if (std.mem.eql(u8, fmt[1..], ANY))
        ANY
    else
        fmt[1..];
}

Test:

escaped braces


pub fn invalidFmtError(comptime fmt: []const u8, value: anytype) void {
    @compileError("invalid format string '" ++ fmt ++ "' for type '" ++ @typeName(@TypeOf(value)) ++ "'");
}

Test:

optional


pub fn formatType(
    value: anytype,
    comptime fmt: []const u8,
    options: FormatOptions,
    writer: anytype,
    max_depth: usize,
) @TypeOf(writer).Error!void {
    const T = @TypeOf(value);
    const actual_fmt = comptime if (std.mem.eql(u8, fmt, ANY))
        defaultSpec(@TypeOf(value))
    else if (fmt.len != 0 and (fmt[0] == '?' or fmt[0] == '!')) switch (@typeInfo(T)) {
        .Optional, .ErrorUnion => fmt,
        else => stripOptionalOrErrorUnionSpec(fmt),
    } else fmt;

Test:

error


    if (comptime std.mem.eql(u8, actual_fmt, "*")) {
        return formatAddress(value, options, writer);
    }

Test:

int.small


    if (std.meta.hasMethod(T, "format")) {
        return try value.format(actual_fmt, options, writer);
    }

Test:

int.specifier


    switch (@typeInfo(T)) {
        .ComptimeInt, .Int, .ComptimeFloat, .Float => {
            return formatValue(value, actual_fmt, options, writer);
        },
        .Void => {
            if (actual_fmt.len != 0) invalidFmtError(fmt, value);
            return formatBuf("void", options, writer);
        },
        .Bool => {
            if (actual_fmt.len != 0) invalidFmtError(fmt, value);
            return formatBuf(if (value) "true" else "false", options, writer);
        },
        .Optional => {
            if (actual_fmt.len == 0 or actual_fmt[0] != '?')
                @compileError("cannot format optional without a specifier (i.e. {?} or {any})");
            const remaining_fmt = comptime stripOptionalOrErrorUnionSpec(actual_fmt);
            if (value) |payload| {
                return formatType(payload, remaining_fmt, options, writer, max_depth);
            } else {
                return formatBuf("null", options, writer);
            }
        },
        .ErrorUnion => {
            if (actual_fmt.len == 0 or actual_fmt[0] != '!')
                @compileError("cannot format error union without a specifier (i.e. {!} or {any})");
            const remaining_fmt = comptime stripOptionalOrErrorUnionSpec(actual_fmt);
            if (value) |payload| {
                return formatType(payload, remaining_fmt, options, writer, max_depth);
            } else |err| {
                return formatType(err, "", options, writer, max_depth);
            }
        },
        .ErrorSet => {
            if (actual_fmt.len != 0) invalidFmtError(fmt, value);
            try writer.writeAll("error.");
            return writer.writeAll(@errorName(value));
        },
        .Enum => |enumInfo| {
            try writer.writeAll(@typeName(T));
            if (enumInfo.is_exhaustive) {
                if (actual_fmt.len != 0) invalidFmtError(fmt, value);
                try writer.writeAll(".");
                try writer.writeAll(@tagName(value));
                return;
            }

Test:

int.padded


            // Use @tagName only if value is one of known fields
            @setEvalBranchQuota(3 * enumInfo.fields.len);
            inline for (enumInfo.fields) |enumField| {
                if (@intFromEnum(value) == enumField.value) {
                    try writer.writeAll(".");
                    try writer.writeAll(@tagName(value));
                    return;
                }
            }

Test:

buffer


            try writer.writeAll("(");
            try formatType(@intFromEnum(value), actual_fmt, options, writer, max_depth);
            try writer.writeAll(")");
        },
        .Union => |info| {
            if (actual_fmt.len != 0) invalidFmtError(fmt, value);
            try writer.writeAll(@typeName(T));
            if (max_depth == 0) {
                return writer.writeAll("{ ... }");
            }
            if (info.tag_type) |UnionTagType| {
                try writer.writeAll("{ .");
                try writer.writeAll(@tagName(@as(UnionTagType, value)));
                try writer.writeAll(" = ");
                inline for (info.fields) |u_field| {
                    if (value == @field(UnionTagType, u_field.name)) {
                        try formatType(@field(value, u_field.name), ANY, options, writer, max_depth - 1);
                    }
                }
                try writer.writeAll(" }");
            } else {
                try format(writer, "@{x}", .{@intFromPtr(&value)});
            }
        },
        .Struct => |info| {
            if (actual_fmt.len != 0) invalidFmtError(fmt, value);
            if (info.is_tuple) {
                // Skip the type and field names when formatting tuples.
                if (max_depth == 0) {
                    return writer.writeAll("{ ... }");
                }
                try writer.writeAll("{");
                inline for (info.fields, 0..) |f, i| {
                    if (i == 0) {
                        try writer.writeAll(" ");
                    } else {
                        try writer.writeAll(", ");
                    }
                    try formatType(@field(value, f.name), ANY, options, writer, max_depth - 1);
                }
                return writer.writeAll(" }");
            }
            try writer.writeAll(@typeName(T));
            if (max_depth == 0) {
                return writer.writeAll("{ ... }");
            }
            try writer.writeAll("{");
            inline for (info.fields, 0..) |f, i| {
                if (i == 0) {
                    try writer.writeAll(" .");
                } else {
                    try writer.writeAll(", .");
                }
                try writer.writeAll(f.name);
                try writer.writeAll(" = ");
                try formatType(@field(value, f.name), ANY, options, writer, max_depth - 1);
            }
            try writer.writeAll(" }");
        },
        .Pointer => |ptr_info| switch (ptr_info.size) {
            .One => switch (@typeInfo(ptr_info.child)) {
                .Array, .Enum, .Union, .Struct => {
                    return formatType(value.*, actual_fmt, options, writer, max_depth);
                },
                else => return format(writer, "{s}@{x}", .{ @typeName(ptr_info.child), @intFromPtr(value) }),
            },
            .Many, .C => {
                if (actual_fmt.len == 0)
                    @compileError("cannot format pointer without a specifier (i.e. {s} or {*})");
                if (ptr_info.sentinel) |_| {
                    return formatType(mem.span(value), actual_fmt, options, writer, max_depth);
                }
                if (actual_fmt[0] == 's' and ptr_info.child == u8) {
                    return formatBuf(mem.span(value), options, writer);
                }
                invalidFmtError(fmt, value);
            },
            .Slice => {
                if (actual_fmt.len == 0)
                    @compileError("cannot format slice without a specifier (i.e. {s} or {any})");
                if (max_depth == 0) {
                    return writer.writeAll("{ ... }");
                }
                if (actual_fmt[0] == 's' and ptr_info.child == u8) {
                    return formatBuf(value, options, writer);
                }
                try writer.writeAll("{ ");
                for (value, 0..) |elem, i| {
                    try formatType(elem, actual_fmt, options, writer, max_depth - 1);
                    if (i != value.len - 1) {
                        try writer.writeAll(", ");
                    }
                }
                try writer.writeAll(" }");
            },
        },
        .Array => |info| {
            if (actual_fmt.len == 0)
                @compileError("cannot format array without a specifier (i.e. {s} or {any})");
            if (max_depth == 0) {
                return writer.writeAll("{ ... }");
            }
            if (actual_fmt[0] == 's' and info.child == u8) {
                return formatBuf(&value, options, writer);
            }
            try writer.writeAll("{ ");
            for (value, 0..) |elem, i| {
                try formatType(elem, actual_fmt, options, writer, max_depth - 1);
                if (i < value.len - 1) {
                    try writer.writeAll(", ");
                }
            }
            try writer.writeAll(" }");
        },
        .Vector => |info| {
            try writer.writeAll("{ ");
            var i: usize = 0;
            while (i < info.len) : (i += 1) {
                try formatValue(value[i], actual_fmt, options, writer);
                if (i < info.len - 1) {
                    try writer.writeAll(", ");
                }
            }
            try writer.writeAll(" }");
        },
        .Fn => @compileError("unable to format function body type, use '*const " ++ @typeName(T) ++ "' for a function pointer type"),
        .Type => {
            if (actual_fmt.len != 0) invalidFmtError(fmt, value);
            return formatBuf(@typeName(value), options, writer);
        },
        .EnumLiteral => {
            if (actual_fmt.len != 0) invalidFmtError(fmt, value);
            const buffer = [_]u8{'.'} ++ @tagName(value);
            return formatBuf(buffer, options, writer);
        },
        .Null => {
            if (actual_fmt.len != 0) invalidFmtError(fmt, value);
            return formatBuf("null", options, writer);
        },
        else => @compileError("unable to format type '" ++ @typeName(T) ++ "'"),
    }
}

Test:

array


fn formatValue(
    value: anytype,
    comptime fmt: []const u8,
    options: FormatOptions,
    writer: anytype,

format()

) !void {
    const T = @TypeOf(value);
    switch (@typeInfo(T)) {
        .Float, .ComptimeFloat => return formatFloatValue(value, fmt, options, writer),
        .Int, .ComptimeInt => return formatIntValue(value, fmt, options, writer),
        .Bool => return formatBuf(if (value) "true" else "false", options, writer),
        else => comptime unreachable,
    }
}

format()


pub fn formatIntValue(
    value: anytype,
    comptime fmt: []const u8,
    options: FormatOptions,
    writer: anytype,

format()

) !void {
    comptime var base = 10;
    comptime var case: Case = .lower;

Test:

pointer


    const int_value = if (@TypeOf(value) == comptime_int) blk: {
        const Int = math.IntFittingRange(value, value);
        break :blk @as(Int, value);
    } else value;

Test:

cstr


    if (fmt.len == 0 or comptime std.mem.eql(u8, fmt, "d")) {
        base = 10;
        case = .lower;
    } else if (comptime std.mem.eql(u8, fmt, "c")) {
        if (@typeInfo(@TypeOf(int_value)).Int.bits <= 8) {
            return formatAsciiChar(@as(u8, int_value), options, writer);
        } else {
            @compileError("cannot print integer that is larger than 8 bits as an ASCII character");
        }
    } else if (comptime std.mem.eql(u8, fmt, "u")) {
        if (@typeInfo(@TypeOf(int_value)).Int.bits <= 21) {
            return formatUnicodeCodepoint(@as(u21, int_value), options, writer);
        } else {
            @compileError("cannot print integer that is larger than 21 bits as an UTF-8 sequence");
        }
    } else if (comptime std.mem.eql(u8, fmt, "b")) {
        base = 2;
        case = .lower;
    } else if (comptime std.mem.eql(u8, fmt, "x")) {
        base = 16;
        case = .lower;
    } else if (comptime std.mem.eql(u8, fmt, "X")) {
        base = 16;
        case = .upper;
    } else if (comptime std.mem.eql(u8, fmt, "o")) {
        base = 8;
        case = .lower;
    } else {
        invalidFmtError(fmt, value);
    }

Test:

filesize


    return formatInt(int_value, base, case, options, writer);
}

Test:

struct


pub const format_float = @import("fmt/format_float.zig");
pub const formatFloat = format_float.formatFloat;
pub const FormatFloatError = format_float.FormatError;

Test:

enum


fn formatFloatValue(
    value: anytype,
    comptime fmt: []const u8,
    options: FormatOptions,
    writer: anytype,

format()

) !void {
    var buf: [format_float.bufferSize(.decimal, f64)]u8 = undefined;

Test:

float.scientific


    if (fmt.len == 0 or comptime std.mem.eql(u8, fmt, "e")) {
        const s = formatFloat(&buf, value, .{ .mode = .scientific, .precision = options.precision }) catch |err| switch (err) {
            error.BufferTooSmall => "(float)",
        };
        return formatBuf(s, options, writer);
    } else if (comptime std.mem.eql(u8, fmt, "d")) {
        const s = formatFloat(&buf, value, .{ .mode = .decimal, .precision = options.precision }) catch |err| switch (err) {
            error.BufferTooSmall => "(float)",
        };
        return formatBuf(s, options, writer);
    } else if (comptime std.mem.eql(u8, fmt, "x")) {
        var buf_stream = std.io.fixedBufferStream(&buf);
        formatFloatHexadecimal(value, options, buf_stream.writer()) catch |err| switch (err) {
            error.NoSpaceLeft => unreachable,
        };
        return formatBuf(buf_stream.getWritten(), options, writer);
    } else {
        invalidFmtError(fmt, value);
    }
}

Test:

float.scientific.precision


test {
    _ = &format_float;
}

Test:

float.special


pub const Case = enum { lower, upper };

Test:

float.hexadecimal.special


fn formatSliceHexImpl(comptime case: Case) type {
    const charset = "0123456789" ++ if (case == .upper) "ABCDEF" else "abcdef";

Test:

float.hexadecimal


    return struct {
        pub fn formatSliceHexImpl(
            bytes: []const u8,
            comptime fmt: []const u8,
            options: std.fmt.FormatOptions,
            writer: anytype,
        ) !void {
            _ = fmt;
            _ = options;
            var buf: [2]u8 = undefined;

Test:

float.hexadecimal.precision


            for (bytes) |c| {
                buf[0] = charset[c >> 4];
                buf[1] = charset[c & 15];
                try writer.writeAll(&buf);
            }
        }
    };
}

Test:

float.decimal


const formatSliceHexLower = formatSliceHexImpl(.lower).formatSliceHexImpl;
const formatSliceHexUpper = formatSliceHexImpl(.upper).formatSliceHexImpl;

Test:

float.libc.sanity


/// Return a Formatter for a []const u8 where every byte is formatted as a pair
/// of lowercase hexadecimal digits.
pub fn fmtSliceHexLower(bytes: []const u8) std.fmt.Formatter(formatSliceHexLower) {
    return .{ .data = bytes };
}

Test:

custom


/// Return a Formatter for a []const u8 where every byte is formatted as pair
/// of uppercase hexadecimal digits.
pub fn fmtSliceHexUpper(bytes: []const u8) std.fmt.Formatter(formatSliceHexUpper) {
    return .{ .data = bytes };
}

format()


fn formatSliceEscapeImpl(comptime case: Case) type {
    const charset = "0123456789" ++ if (case == .upper) "ABCDEF" else "abcdef";

Test:

union


    return struct {
        pub fn formatSliceEscapeImpl(
            bytes: []const u8,
            comptime fmt: []const u8,
            options: std.fmt.FormatOptions,
            writer: anytype,
        ) !void {
            _ = fmt;
            _ = options;
            var buf: [4]u8 = undefined;

Test:

struct.self-referential


            buf[0] = '\\';
            buf[1] = 'x';

Test:

struct.zero-size


            for (bytes) |c| {
                if (std.ascii.isPrint(c)) {
                    try writer.writeByte(c);
                } else {
                    buf[2] = charset[c >> 4];
                    buf[3] = charset[c & 15];
                    try writer.writeAll(&buf);
                }
            }
        }
    };
}

Test:

bytes.hex


const formatSliceEscapeLower = formatSliceEscapeImpl(.lower).formatSliceEscapeImpl;
const formatSliceEscapeUpper = formatSliceEscapeImpl(.upper).formatSliceEscapeImpl;

bytesToHex()


/// Return a Formatter for a []const u8 where every non-printable ASCII
/// character is escaped as \xNN, where NN is the character in lowercase
/// hexadecimal notation.
pub fn fmtSliceEscapeLower(bytes: []const u8) std.fmt.Formatter(formatSliceEscapeLower) {
    return .{ .data = bytes };
}

hexToBytes()


/// Return a Formatter for a []const u8 where every non-printable ASCII
/// character is escaped as \xNN, where NN is the character in uppercase
/// hexadecimal notation.
pub fn fmtSliceEscapeUpper(bytes: []const u8) std.fmt.Formatter(formatSliceEscapeUpper) {
    return .{ .data = bytes };
}

Test: bytesToHex


fn formatSizeImpl(comptime base: comptime_int) type {
    return struct {
        fn formatSizeImpl(
            value: u64,
            comptime fmt: []const u8,
            options: FormatOptions,
            writer: anytype,
        ) !void {
            _ = fmt;
            if (value == 0) {
                return formatBuf("0B", options, writer);
            }
            // The worst case in terms of space needed is 32 bytes + 3 for the suffix.
            var buf: [format_float.min_buffer_size + 3]u8 = undefined;

Test: hexToBytes


            const mags_si = " kMGTPEZY";
            const mags_iec = " KMGTPEZY";

Test:

formatIntValue with comptime_int


            const log2 = math.log2(value);
            const magnitude = switch (base) {
                1000 => @min(log2 / comptime math.log2(1000), mags_si.len - 1),
                1024 => @min(log2 / 10, mags_iec.len - 1),
                else => unreachable,
            };
            const new_value = lossyCast(f64, value) / math.pow(f64, lossyCast(f64, base), lossyCast(f64, magnitude));
            const suffix = switch (base) {
                1000 => mags_si[magnitude],
                1024 => mags_iec[magnitude],
                else => unreachable,
            };

Test:

formatFloatValue with comptime_float


            const s = switch (magnitude) {
                0 => buf[0..formatIntBuf(&buf, value, 10, .lower, .{})],
                else => formatFloat(&buf, new_value, .{ .mode = .decimal, .precision = options.precision }) catch |err| switch (err) {
                    error.BufferTooSmall => unreachable,
                },
            };

Test:

formatType max_depth


            var i: usize = s.len;
            if (suffix == ' ') {
                buf[i] = 'B';
                i += 1;
            } else switch (base) {
                1000 => {
                    buf[i..][0..2].* = [_]u8{ suffix, 'B' };
                    i += 2;
                },
                1024 => {
                    buf[i..][0..3].* = [_]u8{ suffix, 'i', 'B' };
                    i += 3;
                },
                else => unreachable,
            }

format()


            return formatBuf(buf[0..i], options, writer);
        }
    };
}

Test:

positional


const formatSizeDec = formatSizeImpl(1000).formatSizeImpl;
const formatSizeBin = formatSizeImpl(1024).formatSizeImpl;

Test:

positional with specifier


/// Return a Formatter for a u64 value representing a file size.
/// This formatter represents the number as multiple of 1000 and uses the SI
/// measurement units (kB, MB, GB, ...).
pub fn fmtIntSizeDec(value: u64) std.fmt.Formatter(formatSizeDec) {
    return .{ .data = value };
}

Test:

positional/alignment/width/precision


/// Return a Formatter for a u64 value representing a file size.
/// This formatter represents the number as multiple of 1024 and uses the IEC
/// measurement units (KiB, MiB, GiB, ...).
pub fn fmtIntSizeBin(value: u64) std.fmt.Formatter(formatSizeBin) {
    return .{ .data = value };
}

Test:

vector


fn checkTextFmt(comptime fmt: []const u8) void {
    if (fmt.len != 1)
        @compileError("unsupported format string '" ++ fmt ++ "' when formatting text");
    switch (fmt[0]) {
        // Example of deprecation:
        // '[deprecated_specifier]' => @compileError("specifier '[deprecated_specifier]' has been deprecated, wrap your argument in `std.some_function` instead"),
        'x' => @compileError("specifier 'x' has been deprecated, wrap your argument in std.fmt.fmtSliceHexLower instead"),
        'X' => @compileError("specifier 'X' has been deprecated, wrap your argument in std.fmt.fmtSliceHexUpper instead"),
        else => {},
    }
}

Test:

enum-literal


pub fn formatText(
    bytes: []const u8,
    comptime fmt: []const u8,
    options: FormatOptions,
    writer: anytype,

format()

) !void {
    comptime checkTextFmt(fmt);
    return formatBuf(bytes, options, writer);
}

Test:

padding fill char utf


pub fn formatAsciiChar(
    c: u8,
    options: FormatOptions,
    writer: anytype,

format()

) !void {
    return formatBuf(@as(*const [1]u8, &c), options, writer);
}

Test:

sci float padding


pub fn formatUnicodeCodepoint(
    c: u21,
    options: FormatOptions,
    writer: anytype,

format()

) !void {
    var buf: [4]u8 = undefined;
    const len = unicode.utf8Encode(c, &buf) catch |err| switch (err) {
        error.Utf8CannotEncodeSurrogateHalf, error.CodepointTooLarge => {
            return formatBuf(&unicode.utf8EncodeComptime(unicode.replacement_character), options, writer);
        },
    };
    return formatBuf(buf[0..len], options, writer);
}

Test:

type


pub fn formatBuf(
    buf: []const u8,
    options: FormatOptions,
    writer: anytype,

format()

) !void {
    if (options.width) |min_width| {
        // In case of error assume the buffer content is ASCII-encoded
        const width = unicode.utf8CountCodepoints(buf) catch buf.len;
        const padding = if (width < min_width) min_width - width else 0;

Test:

runtime width specifier


        if (padding == 0)
            return writer.writeAll(buf);

Test:

runtime precision specifier


        var fill_buffer: [4]u8 = undefined;
        const fill_utf8 = if (unicode.utf8Encode(options.fill, &fill_buffer)) |len|
            fill_buffer[0..len]
        else |err| switch (err) {
            error.Utf8CannotEncodeSurrogateHalf,
            error.CodepointTooLarge,
            => &unicode.utf8EncodeComptime(unicode.replacement_character),
        };
        switch (options.alignment) {
            .left => {
                try writer.writeAll(buf);
                try writer.writeBytesNTimes(fill_utf8, padding);
            },
            .center => {
                const left_padding = padding / 2;
                const right_padding = (padding + 1) / 2;
                try writer.writeBytesNTimes(fill_utf8, left_padding);
                try writer.writeAll(buf);
                try writer.writeBytesNTimes(fill_utf8, right_padding);
            },
            .right => {
                try writer.writeBytesNTimes(fill_utf8, padding);
                try writer.writeAll(buf);
            },
        }
    } else {
        // Fast path, avoid counting the number of codepoints
        try writer.writeAll(buf);
    }
}

Test:

recursive format function


pub fn formatFloatHexadecimal(
    value: anytype,
    options: FormatOptions,
    writer: anytype,

format()

) !void {
    if (math.signbit(value)) {
        try writer.writeByte('-');
    }
    if (math.isNan(value)) {
        return writer.writeAll("nan");
    }
    if (math.isInf(value)) {
        return writer.writeAll("inf");
    }

    const T = @TypeOf(value);
    const TU = std.meta.Int(.unsigned, @bitSizeOf(T));

    const mantissa_bits = math.floatMantissaBits(T);
    const fractional_bits = math.floatFractionalBits(T);
    const exponent_bits = math.floatExponentBits(T);
    const mantissa_mask = (1 << mantissa_bits) - 1;
    const exponent_mask = (1 << exponent_bits) - 1;
    const exponent_bias = (1 << (exponent_bits - 1)) - 1;

    const as_bits = @as(TU, @bitCast(value));
    var mantissa = as_bits & mantissa_mask;
    var exponent: i32 = @as(u16, @truncate((as_bits >> mantissa_bits) & exponent_mask));

    const is_denormal = exponent == 0 and mantissa != 0;
    const is_zero = exponent == 0 and mantissa == 0;

    if (is_zero) {
        // Handle this case here to simplify the logic below.
        try writer.writeAll("0x0");
        if (options.precision) |precision| {
            if (precision > 0) {
                try writer.writeAll(".");
                try writer.writeByteNTimes('0', precision);
            }
        } else {
            try writer.writeAll(".0");
        }
        try writer.writeAll("p0");
        return;
    }

    if (is_denormal) {
        // Adjust the exponent for printing.
        exponent += 1;
    } else {
        if (fractional_bits == mantissa_bits)
            mantissa |= 1 << fractional_bits; // Add the implicit integer bit.
    }

    const mantissa_digits = (fractional_bits + 3) / 4;
    // Fill in zeroes to round the fraction width to a multiple of 4.
    mantissa <<= mantissa_digits * 4 - fractional_bits;

    if (options.precision) |precision| {
        // Round if needed.
        if (precision < mantissa_digits) {
            // We always have at least 4 extra bits.
            var extra_bits = (mantissa_digits - precision) * 4;
            // The result LSB is the Guard bit, we need two more (Round and
            // Sticky) to round the value.
            while (extra_bits > 2) {
                mantissa = (mantissa >> 1) | (mantissa & 1);
                extra_bits -= 1;
            }
            // Round to nearest, tie to even.
            mantissa |= @intFromBool(mantissa & 0b100 != 0);
            mantissa += 1;
            // Drop the excess bits.
            mantissa >>= 2;
            // Restore the alignment.
            mantissa <<= @as(math.Log2Int(TU), @intCast((mantissa_digits - precision) * 4));

            const overflow = mantissa & (1 << 1 + mantissa_digits * 4) != 0;
            // Prefer a normalized result in case of overflow.
            if (overflow) {
                mantissa >>= 1;
                exponent += 1;
            }
        }
    }

    // +1 for the decimal part.
    var buf: [1 + mantissa_digits]u8 = undefined;
    _ = formatIntBuf(&buf, mantissa, 16, .lower, .{ .fill = '0', .width = 1 + mantissa_digits });

    try writer.writeAll("0x");
    try writer.writeByte(buf[0]);
    const trimmed = mem.trimRight(u8, buf[1..], "0");
    if (options.precision) |precision| {
        if (precision > 0) try writer.writeAll(".");
    } else if (trimmed.len > 0) {
        try writer.writeAll(".");
    }
    try writer.writeAll(trimmed);
    // Add trailing zeros if explicitly requested.
    if (options.precision) |precision| if (precision > 0) {
        if (precision > trimmed.len)
            try writer.writeByteNTimes('0', precision - trimmed.len);
    };
    try writer.writeAll("p");
    try formatInt(exponent - exponent_bias, 10, .lower, .{}, writer);
}

pub fn formatInt(
    value: anytype,
    base: u8,
    case: Case,
    options: FormatOptions,
    writer: anytype,
) !void {
    assert(base >= 2);

    const int_value = if (@TypeOf(value) == comptime_int) blk: {
        const Int = math.IntFittingRange(value, value);
        break :blk @as(Int, value);
    } else value;

    const value_info = @typeInfo(@TypeOf(int_value)).Int;

    // The type must have the same size as `base` or be wider in order for the
    // division to work
    const min_int_bits = comptime @max(value_info.bits, 8);
    const MinInt = std.meta.Int(.unsigned, min_int_bits);

    const abs_value = @abs(int_value);
    // The worst case in terms of space needed is base 2, plus 1 for the sign
    var buf: [1 + @max(@as(comptime_int, value_info.bits), 1)]u8 = undefined;

    var a: MinInt = abs_value;
    var index: usize = buf.len;

    if (base == 10) {
        while (a >= 100) : (a = @divTrunc(a, 100)) {
            index -= 2;
            buf[index..][0..2].* = digits2(@as(usize, @intCast(a % 100)));
        }

        if (a < 10) {
            index -= 1;
            buf[index] = '0' + @as(u8, @intCast(a));
        } else {
            index -= 2;
            buf[index..][0..2].* = digits2(@as(usize, @intCast(a)));
        }
    } else {
        while (true) {
            const digit = a % base;
            index -= 1;
            buf[index] = digitToChar(@as(u8, @intCast(digit)), case);
            a /= base;
            if (a == 0) break;
        }
    }

    if (value_info.signedness == .signed) {
        if (value < 0) {
            // Negative integer
            index -= 1;
            buf[index] = '-';
        } else if (options.width == null or options.width.? == 0) {
            // Positive integer, omit the plus sign
        } else {
            // Positive integer
            index -= 1;
            buf[index] = '+';
        }
    }

    return formatBuf(buf[index..], options, writer);
}

pub fn formatIntBuf(out_buf: []u8, value: anytype, base: u8, case: Case, options: FormatOptions) usize {
    var fbs = std.io.fixedBufferStream(out_buf);
    formatInt(value, base, case, options, fbs.writer()) catch unreachable;
    return fbs.pos;
}

// Converts values in the range [0, 100) to a string.
pub fn digits2(value: usize) [2]u8 {
    return ("0001020304050607080910111213141516171819" ++
        "2021222324252627282930313233343536373839" ++
        "4041424344454647484950515253545556575859" ++
        "6061626364656667686970717273747576777879" ++
        "8081828384858687888990919293949596979899")[value * 2 ..][0..2].*;
}

const FormatDurationData = struct {
    ns: u64,
    negative: bool = false,
};

fn formatDuration(data: FormatDurationData, comptime fmt: []const u8, options: std.fmt.FormatOptions, writer: anytype) !void {
    _ = fmt;

    // worst case: "-XXXyXXwXXdXXhXXmXX.XXXs".len = 24
    var buf: [24]u8 = undefined;
    var fbs = std.io.fixedBufferStream(&buf);
    var buf_writer = fbs.writer();
    if (data.negative) {
        buf_writer.writeByte('-') catch unreachable;
    }

    var ns_remaining = data.ns;
    inline for (.{
        .{ .ns = 365 * std.time.ns_per_day, .sep = 'y' },
        .{ .ns = std.time.ns_per_week, .sep = 'w' },
        .{ .ns = std.time.ns_per_day, .sep = 'd' },
        .{ .ns = std.time.ns_per_hour, .sep = 'h' },
        .{ .ns = std.time.ns_per_min, .sep = 'm' },
    }) |unit| {
        if (ns_remaining >= unit.ns) {
            const units = ns_remaining / unit.ns;
            formatInt(units, 10, .lower, .{}, buf_writer) catch unreachable;
            buf_writer.writeByte(unit.sep) catch unreachable;
            ns_remaining -= units * unit.ns;
            if (ns_remaining == 0)
                return formatBuf(fbs.getWritten(), options, writer);
        }
    }

    inline for (.{
        .{ .ns = std.time.ns_per_s, .sep = "s" },
        .{ .ns = std.time.ns_per_ms, .sep = "ms" },
        .{ .ns = std.time.ns_per_us, .sep = "us" },
    }) |unit| {
        const kunits = ns_remaining * 1000 / unit.ns;
        if (kunits >= 1000) {
            formatInt(kunits / 1000, 10, .lower, .{}, buf_writer) catch unreachable;
            const frac = kunits % 1000;
            if (frac > 0) {
                // Write up to 3 decimal places
                var decimal_buf = [_]u8{ '.', 0, 0, 0 };
                _ = formatIntBuf(decimal_buf[1..], frac, 10, .lower, .{ .fill = '0', .width = 3 });
                var end: usize = 4;
                while (end > 1) : (end -= 1) {
                    if (decimal_buf[end - 1] != '0') break;
                }
                buf_writer.writeAll(decimal_buf[0..end]) catch unreachable;
            }
            buf_writer.writeAll(unit.sep) catch unreachable;
            return formatBuf(fbs.getWritten(), options, writer);
        }
    }

    formatInt(ns_remaining, 10, .lower, .{}, buf_writer) catch unreachable;
    buf_writer.writeAll("ns") catch unreachable;
    return formatBuf(fbs.getWritten(), options, writer);
}

/// Return a Formatter for number of nanoseconds according to its magnitude:
/// [#y][#w][#d][#h][#m]#[.###][n|u|m]s
pub fn fmtDuration(ns: u64) Formatter(formatDuration) {
    const data = FormatDurationData{ .ns = ns };
    return .{ .data = data };
}

test fmtDuration {
    var buf: [24]u8 = undefined;
    inline for (.{
        .{ .s = "0ns", .d = 0 },
        .{ .s = "1ns", .d = 1 },
        .{ .s = "999ns", .d = std.time.ns_per_us - 1 },
        .{ .s = "1us", .d = std.time.ns_per_us },
        .{ .s = "1.45us", .d = 1450 },
        .{ .s = "1.5us", .d = 3 * std.time.ns_per_us / 2 },
        .{ .s = "14.5us", .d = 14500 },
        .{ .s = "145us", .d = 145000 },
        .{ .s = "999.999us", .d = std.time.ns_per_ms - 1 },
        .{ .s = "1ms", .d = std.time.ns_per_ms + 1 },
        .{ .s = "1.5ms", .d = 3 * std.time.ns_per_ms / 2 },
        .{ .s = "1.11ms", .d = 1110000 },
        .{ .s = "1.111ms", .d = 1111000 },
        .{ .s = "1.111ms", .d = 1111100 },
        .{ .s = "999.999ms", .d = std.time.ns_per_s - 1 },
        .{ .s = "1s", .d = std.time.ns_per_s },
        .{ .s = "59.999s", .d = std.time.ns_per_min - 1 },
        .{ .s = "1m", .d = std.time.ns_per_min },
        .{ .s = "1h", .d = std.time.ns_per_hour },
        .{ .s = "1d", .d = std.time.ns_per_day },
        .{ .s = "1w", .d = std.time.ns_per_week },
        .{ .s = "1y", .d = 365 * std.time.ns_per_day },
        .{ .s = "1y52w23h59m59.999s", .d = 730 * std.time.ns_per_day - 1 }, // 365d = 52w1d
        .{ .s = "1y1h1.001s", .d = 365 * std.time.ns_per_day + std.time.ns_per_hour + std.time.ns_per_s + std.time.ns_per_ms },
        .{ .s = "1y1h1s", .d = 365 * std.time.ns_per_day + std.time.ns_per_hour + std.time.ns_per_s + 999 * std.time.ns_per_us },
        .{ .s = "1y1h999.999us", .d = 365 * std.time.ns_per_day + std.time.ns_per_hour + std.time.ns_per_ms - 1 },
        .{ .s = "1y1h1ms", .d = 365 * std.time.ns_per_day + std.time.ns_per_hour + std.time.ns_per_ms },
        .{ .s = "1y1h1ms", .d = 365 * std.time.ns_per_day + std.time.ns_per_hour + std.time.ns_per_ms + 1 },
        .{ .s = "1y1m999ns", .d = 365 * std.time.ns_per_day + std.time.ns_per_min + 999 },
        .{ .s = "584y49w23h34m33.709s", .d = math.maxInt(u64) },
    }) |tc| {
        const slice = try bufPrint(&buf, "{}", .{fmtDuration(tc.d)});
        try std.testing.expectEqualStrings(tc.s, slice);
    }

    inline for (.{
        .{ .s = "=======0ns", .f = "{s:=>10}", .d = 0 },
        .{ .s = "1ns=======", .f = "{s:=<10}", .d = 1 },
        .{ .s = "  999ns   ", .f = "{s:^10}", .d = std.time.ns_per_us - 1 },
    }) |tc| {
        const slice = try bufPrint(&buf, tc.f, .{fmtDuration(tc.d)});
        try std.testing.expectEqualStrings(tc.s, slice);
    }
}

fn formatDurationSigned(ns: i64, comptime fmt: []const u8, options: std.fmt.FormatOptions, writer: anytype) !void {
    if (ns < 0) {
        const data = FormatDurationData{ .ns = @as(u64, @intCast(-ns)), .negative = true };
        try formatDuration(data, fmt, options, writer);
    } else {
        const data = FormatDurationData{ .ns = @as(u64, @intCast(ns)) };
        try formatDuration(data, fmt, options, writer);
    }
}

/// Return a Formatter for number of nanoseconds according to its signed magnitude:
/// [#y][#w][#d][#h][#m]#[.###][n|u|m]s
pub fn fmtDurationSigned(ns: i64) Formatter(formatDurationSigned) {
    return .{ .data = ns };
}

test fmtDurationSigned {
    var buf: [24]u8 = undefined;
    inline for (.{
        .{ .s = "0ns", .d = 0 },
        .{ .s = "1ns", .d = 1 },
        .{ .s = "-1ns", .d = -(1) },
        .{ .s = "999ns", .d = std.time.ns_per_us - 1 },
        .{ .s = "-999ns", .d = -(std.time.ns_per_us - 1) },
        .{ .s = "1us", .d = std.time.ns_per_us },
        .{ .s = "-1us", .d = -(std.time.ns_per_us) },
        .{ .s = "1.45us", .d = 1450 },
        .{ .s = "-1.45us", .d = -(1450) },
        .{ .s = "1.5us", .d = 3 * std.time.ns_per_us / 2 },
        .{ .s = "-1.5us", .d = -(3 * std.time.ns_per_us / 2) },
        .{ .s = "14.5us", .d = 14500 },
        .{ .s = "-14.5us", .d = -(14500) },
        .{ .s = "145us", .d = 145000 },
        .{ .s = "-145us", .d = -(145000) },
        .{ .s = "999.999us", .d = std.time.ns_per_ms - 1 },
        .{ .s = "-999.999us", .d = -(std.time.ns_per_ms - 1) },
        .{ .s = "1ms", .d = std.time.ns_per_ms + 1 },
        .{ .s = "-1ms", .d = -(std.time.ns_per_ms + 1) },
        .{ .s = "1.5ms", .d = 3 * std.time.ns_per_ms / 2 },
        .{ .s = "-1.5ms", .d = -(3 * std.time.ns_per_ms / 2) },
        .{ .s = "1.11ms", .d = 1110000 },
        .{ .s = "-1.11ms", .d = -(1110000) },
        .{ .s = "1.111ms", .d = 1111000 },
        .{ .s = "-1.111ms", .d = -(1111000) },
        .{ .s = "1.111ms", .d = 1111100 },
        .{ .s = "-1.111ms", .d = -(1111100) },
        .{ .s = "999.999ms", .d = std.time.ns_per_s - 1 },
        .{ .s = "-999.999ms", .d = -(std.time.ns_per_s - 1) },
        .{ .s = "1s", .d = std.time.ns_per_s },
        .{ .s = "-1s", .d = -(std.time.ns_per_s) },
        .{ .s = "59.999s", .d = std.time.ns_per_min - 1 },
        .{ .s = "-59.999s", .d = -(std.time.ns_per_min - 1) },
        .{ .s = "1m", .d = std.time.ns_per_min },
        .{ .s = "-1m", .d = -(std.time.ns_per_min) },
        .{ .s = "1h", .d = std.time.ns_per_hour },
        .{ .s = "-1h", .d = -(std.time.ns_per_hour) },
        .{ .s = "1d", .d = std.time.ns_per_day },
        .{ .s = "-1d", .d = -(std.time.ns_per_day) },
        .{ .s = "1w", .d = std.time.ns_per_week },
        .{ .s = "-1w", .d = -(std.time.ns_per_week) },
        .{ .s = "1y", .d = 365 * std.time.ns_per_day },
        .{ .s = "-1y", .d = -(365 * std.time.ns_per_day) },
        .{ .s = "1y52w23h59m59.999s", .d = 730 * std.time.ns_per_day - 1 }, // 365d = 52w1d
        .{ .s = "-1y52w23h59m59.999s", .d = -(730 * std.time.ns_per_day - 1) }, // 365d = 52w1d
        .{ .s = "1y1h1.001s", .d = 365 * std.time.ns_per_day + std.time.ns_per_hour + std.time.ns_per_s + std.time.ns_per_ms },
        .{ .s = "-1y1h1.001s", .d = -(365 * std.time.ns_per_day + std.time.ns_per_hour + std.time.ns_per_s + std.time.ns_per_ms) },
        .{ .s = "1y1h1s", .d = 365 * std.time.ns_per_day + std.time.ns_per_hour + std.time.ns_per_s + 999 * std.time.ns_per_us },
        .{ .s = "-1y1h1s", .d = -(365 * std.time.ns_per_day + std.time.ns_per_hour + std.time.ns_per_s + 999 * std.time.ns_per_us) },
        .{ .s = "1y1h999.999us", .d = 365 * std.time.ns_per_day + std.time.ns_per_hour + std.time.ns_per_ms - 1 },
        .{ .s = "-1y1h999.999us", .d = -(365 * std.time.ns_per_day + std.time.ns_per_hour + std.time.ns_per_ms - 1) },
        .{ .s = "1y1h1ms", .d = 365 * std.time.ns_per_day + std.time.ns_per_hour + std.time.ns_per_ms },
        .{ .s = "-1y1h1ms", .d = -(365 * std.time.ns_per_day + std.time.ns_per_hour + std.time.ns_per_ms) },
        .{ .s = "1y1h1ms", .d = 365 * std.time.ns_per_day + std.time.ns_per_hour + std.time.ns_per_ms + 1 },
        .{ .s = "-1y1h1ms", .d = -(365 * std.time.ns_per_day + std.time.ns_per_hour + std.time.ns_per_ms + 1) },
        .{ .s = "1y1m999ns", .d = 365 * std.time.ns_per_day + std.time.ns_per_min + 999 },
        .{ .s = "-1y1m999ns", .d = -(365 * std.time.ns_per_day + std.time.ns_per_min + 999) },
        .{ .s = "292y24w3d23h47m16.854s", .d = math.maxInt(i64) },
        .{ .s = "-292y24w3d23h47m16.854s", .d = math.minInt(i64) + 1 },
    }) |tc| {
        const slice = try bufPrint(&buf, "{}", .{fmtDurationSigned(tc.d)});
        try std.testing.expectEqualStrings(tc.s, slice);
    }

    inline for (.{
        .{ .s = "=======0ns", .f = "{s:=>10}", .d = 0 },
        .{ .s = "1ns=======", .f = "{s:=<10}", .d = 1 },
        .{ .s = "-1ns======", .f = "{s:=<10}", .d = -(1) },
        .{ .s = "  -999ns  ", .f = "{s:^10}", .d = -(std.time.ns_per_us - 1) },
    }) |tc| {
        const slice = try bufPrint(&buf, tc.f, .{fmtDurationSigned(tc.d)});
        try std.testing.expectEqualStrings(tc.s, slice);
    }
}

pub const ParseIntError = error{
    /// The result cannot fit in the type specified
    Overflow,

    /// The input was empty or contained an invalid character
    InvalidCharacter,
};

/// Creates a Formatter type from a format function. Wrapping data in Formatter(func) causes
/// the data to be formatted using the given function `func`.  `func` must be of the following
/// form:
///
///     fn formatExample(
///         data: T,
///         comptime fmt: []const u8,
///         options: std.fmt.FormatOptions,
///         writer: anytype,
///     ) !void;
///
pub fn Formatter(comptime format_fn: anytype) type {
    const Data = @typeInfo(@TypeOf(format_fn)).Fn.params[0].type.?;
    return struct {
        data: Data,
        pub fn format(
            self: @This(),
            comptime fmt: []const u8,
            options: std.fmt.FormatOptions,
            writer: anytype,
        ) @TypeOf(writer).Error!void {
            try format_fn(self.data, fmt, options, writer);
        }
    };
}

/// Parses the string `buf` as signed or unsigned representation in the
/// specified base of an integral value of type `T`.
///
/// When `base` is zero the string prefix is examined to detect the true base:
///  * A prefix of "0b" implies base=2,
///  * A prefix of "0o" implies base=8,
///  * A prefix of "0x" implies base=16,
///  * Otherwise base=10 is assumed.
///
/// Ignores '_' character in `buf`.
/// See also `parseUnsigned`.
pub fn parseInt(comptime T: type, buf: []const u8, base: u8) ParseIntError!T {
    if (buf.len == 0) return error.InvalidCharacter;
    if (buf[0] == '+') return parseWithSign(T, buf[1..], base, .pos);
    if (buf[0] == '-') return parseWithSign(T, buf[1..], base, .neg);
    return parseWithSign(T, buf, base, .pos);
}

test parseInt {
    try std.testing.expect((try parseInt(i32, "-10", 10)) == -10);
    try std.testing.expect((try parseInt(i32, "+10", 10)) == 10);
    try std.testing.expect((try parseInt(u32, "+10", 10)) == 10);
    try std.testing.expectError(error.Overflow, parseInt(u32, "-10", 10));
    try std.testing.expectError(error.InvalidCharacter, parseInt(u32, " 10", 10));
    try std.testing.expectError(error.InvalidCharacter, parseInt(u32, "10 ", 10));
    try std.testing.expectError(error.InvalidCharacter, parseInt(u32, "_10_", 10));
    try std.testing.expectError(error.InvalidCharacter, parseInt(u32, "0x_10_", 10));
    try std.testing.expectError(error.InvalidCharacter, parseInt(u32, "0x10_", 10));
    try std.testing.expectError(error.InvalidCharacter, parseInt(u32, "0x_10", 10));
    try std.testing.expect((try parseInt(u8, "255", 10)) == 255);
    try std.testing.expectError(error.Overflow, parseInt(u8, "256", 10));

    // +0 and -0 should work for unsigned
    try std.testing.expect((try parseInt(u8, "-0", 10)) == 0);
    try std.testing.expect((try parseInt(u8, "+0", 10)) == 0);

    // ensure minInt is parsed correctly
    try std.testing.expect((try parseInt(i1, "-1", 10)) == math.minInt(i1));
    try std.testing.expect((try parseInt(i8, "-128", 10)) == math.minInt(i8));
    try std.testing.expect((try parseInt(i43, "-4398046511104", 10)) == math.minInt(i43));

    // empty string or bare +- is invalid
    try std.testing.expectError(error.InvalidCharacter, parseInt(u32, "", 10));
    try std.testing.expectError(error.InvalidCharacter, parseInt(i32, "", 10));
    try std.testing.expectError(error.InvalidCharacter, parseInt(u32, "+", 10));
    try std.testing.expectError(error.InvalidCharacter, parseInt(i32, "+", 10));
    try std.testing.expectError(error.InvalidCharacter, parseInt(u32, "-", 10));
    try std.testing.expectError(error.InvalidCharacter, parseInt(i32, "-", 10));

    // autodectect the base
    try std.testing.expect((try parseInt(i32, "111", 0)) == 111);
    try std.testing.expect((try parseInt(i32, "1_1_1", 0)) == 111);
    try std.testing.expect((try parseInt(i32, "1_1_1", 0)) == 111);
    try std.testing.expect((try parseInt(i32, "+0b111", 0)) == 7);
    try std.testing.expect((try parseInt(i32, "+0B111", 0)) == 7);
    try std.testing.expect((try parseInt(i32, "+0b1_11", 0)) == 7);
    try std.testing.expect((try parseInt(i32, "+0o111", 0)) == 73);
    try std.testing.expect((try parseInt(i32, "+0O111", 0)) == 73);
    try std.testing.expect((try parseInt(i32, "+0o11_1", 0)) == 73);
    try std.testing.expect((try parseInt(i32, "+0x111", 0)) == 273);
    try std.testing.expect((try parseInt(i32, "-0b111", 0)) == -7);
    try std.testing.expect((try parseInt(i32, "-0b11_1", 0)) == -7);
    try std.testing.expect((try parseInt(i32, "-0o111", 0)) == -73);
    try std.testing.expect((try parseInt(i32, "-0x111", 0)) == -273);
    try std.testing.expect((try parseInt(i32, "-0X111", 0)) == -273);
    try std.testing.expect((try parseInt(i32, "-0x1_11", 0)) == -273);

    // bare binary/octal/decimal prefix is invalid
    try std.testing.expectError(error.InvalidCharacter, parseInt(u32, "0b", 0));
    try std.testing.expectError(error.InvalidCharacter, parseInt(u32, "0o", 0));
    try std.testing.expectError(error.InvalidCharacter, parseInt(u32, "0x", 0));

    // edge cases which previously errored due to base overflowing T
    try std.testing.expectEqual(@as(i2, -2), try std.fmt.parseInt(i2, "-10", 2));
    try std.testing.expectEqual(@as(i4, -8), try std.fmt.parseInt(i4, "-10", 8));
    try std.testing.expectEqual(@as(i5, -16), try std.fmt.parseInt(i5, "-10", 16));
}

fn parseWithSign(
    comptime T: type,
    buf: []const u8,
    base: u8,
    comptime sign: enum { pos, neg },
) ParseIntError!T {
    if (buf.len == 0) return error.InvalidCharacter;

    var buf_base = base;
    var buf_start = buf;
    if (base == 0) {
        // Treat is as a decimal number by default.
        buf_base = 10;
        // Detect the base by looking at buf prefix.
        if (buf.len > 2 and buf[0] == '0') {
            switch (std.ascii.toLower(buf[1])) {
                'b' => {
                    buf_base = 2;
                    buf_start = buf[2..];
                },
                'o' => {
                    buf_base = 8;
                    buf_start = buf[2..];
                },
                'x' => {
                    buf_base = 16;
                    buf_start = buf[2..];
                },
                else => {},
            }
        }
    }

    const add = switch (sign) {
        .pos => math.add,
        .neg => math.sub,
    };

    // accumulate into U which is always 8 bits or larger.  this prevents
    // `buf_base` from overflowing T.
    const info = @typeInfo(T);
    const U = std.meta.Int(info.Int.signedness, @max(8, info.Int.bits));
    var x: U = 0;

    if (buf_start[0] == '_' or buf_start[buf_start.len - 1] == '_') return error.InvalidCharacter;

    for (buf_start) |c| {
        if (c == '_') continue;
        const digit = try charToDigit(c, buf_base);
        if (x != 0) {
            x = try math.mul(U, x, math.cast(U, buf_base) orelse return error.Overflow);
        } else if (sign == .neg) {
            // The first digit of a negative number.
            // Consider parsing "-4" as an i3.
            // This should work, but positive 4 overflows i3, so we can't cast the digit to T and subtract.
            x = math.cast(U, -@as(i8, @intCast(digit))) orelse return error.Overflow;
            continue;
        }
        x = try add(U, x, math.cast(U, digit) orelse return error.Overflow);
    }

    return if (T == U)
        x
    else
        math.cast(T, x) orelse return error.Overflow;
}

/// Parses the string `buf` as unsigned representation in the specified base
/// of an integral value of type `T`.
///
/// When `base` is zero the string prefix is examined to detect the true base:
///  * A prefix of "0b" implies base=2,
///  * A prefix of "0o" implies base=8,
///  * A prefix of "0x" implies base=16,
///  * Otherwise base=10 is assumed.
///
/// Ignores '_' character in `buf`.
/// See also `parseInt`.
pub fn parseUnsigned(comptime T: type, buf: []const u8, base: u8) ParseIntError!T {
    return parseWithSign(T, buf, base, .pos);
}

test parseUnsigned {
    try std.testing.expect((try parseUnsigned(u16, "050124", 10)) == 50124);
    try std.testing.expect((try parseUnsigned(u16, "65535", 10)) == 65535);
    try std.testing.expect((try parseUnsigned(u16, "65_535", 10)) == 65535);
    try std.testing.expectError(error.Overflow, parseUnsigned(u16, "65536", 10));

    try std.testing.expect((try parseUnsigned(u64, "0ffffffffffffffff", 16)) == 0xffffffffffffffff);
    try std.testing.expect((try parseUnsigned(u64, "0f_fff_fff_fff_fff_fff", 16)) == 0xffffffffffffffff);
    try std.testing.expectError(error.Overflow, parseUnsigned(u64, "10000000000000000", 16));

    try std.testing.expect((try parseUnsigned(u32, "DeadBeef", 16)) == 0xDEADBEEF);

    try std.testing.expect((try parseUnsigned(u7, "1", 10)) == 1);
    try std.testing.expect((try parseUnsigned(u7, "1000", 2)) == 8);

    try std.testing.expectError(error.InvalidCharacter, parseUnsigned(u32, "f", 10));
    try std.testing.expectError(error.InvalidCharacter, parseUnsigned(u8, "109", 8));

    try std.testing.expect((try parseUnsigned(u32, "NUMBER", 36)) == 1442151747);

    // these numbers should fit even though the base itself doesn't fit in the destination type
    try std.testing.expect((try parseUnsigned(u1, "0", 10)) == 0);
    try std.testing.expect((try parseUnsigned(u1, "1", 10)) == 1);
    try std.testing.expectError(error.Overflow, parseUnsigned(u1, "2", 10));
    try std.testing.expect((try parseUnsigned(u1, "001", 16)) == 1);
    try std.testing.expect((try parseUnsigned(u2, "3", 16)) == 3);
    try std.testing.expectError(error.Overflow, parseUnsigned(u2, "4", 16));

    // parseUnsigned does not expect a sign
    try std.testing.expectError(error.InvalidCharacter, parseUnsigned(u8, "+0", 10));
    try std.testing.expectError(error.InvalidCharacter, parseUnsigned(u8, "-0", 10));

    // test empty string error
    try std.testing.expectError(error.InvalidCharacter, parseUnsigned(u8, "", 10));
}

/// Parses a number like '2G', '2Gi', or '2GiB'.
pub fn parseIntSizeSuffix(buf: []const u8, digit_base: u8) ParseIntError!usize {
    var without_B = buf;
    if (mem.endsWith(u8, buf, "B")) without_B.len -= 1;
    var without_i = without_B;
    var magnitude_base: usize = 1000;
    if (mem.endsWith(u8, without_B, "i")) {
        without_i.len -= 1;
        magnitude_base = 1024;
    }
    if (without_i.len == 0) return error.InvalidCharacter;
    const orders_of_magnitude: usize = switch (without_i[without_i.len - 1]) {
        'k', 'K' => 1,
        'M' => 2,
        'G' => 3,
        'T' => 4,
        'P' => 5,
        'E' => 6,
        'Z' => 7,
        'Y' => 8,
        'R' => 9,
        'Q' => 10,
        else => 0,
    };
    var without_suffix = without_i;
    if (orders_of_magnitude > 0) {
        without_suffix.len -= 1;
    } else if (without_i.len != without_B.len) {
        return error.InvalidCharacter;
    }
    const multiplier = math.powi(usize, magnitude_base, orders_of_magnitude) catch |err| switch (err) {
        error.Underflow => unreachable,
        error.Overflow => return error.Overflow,
    };
    const number = try std.fmt.parseInt(usize, without_suffix, digit_base);
    return math.mul(usize, number, multiplier);
}

test parseIntSizeSuffix {
    try std.testing.expect(try parseIntSizeSuffix("2", 10) == 2);
    try std.testing.expect(try parseIntSizeSuffix("2B", 10) == 2);
    try std.testing.expect(try parseIntSizeSuffix("2kB", 10) == 2000);
    try std.testing.expect(try parseIntSizeSuffix("2k", 10) == 2000);
    try std.testing.expect(try parseIntSizeSuffix("2KiB", 10) == 2048);
    try std.testing.expect(try parseIntSizeSuffix("2Ki", 10) == 2048);
    try std.testing.expect(try parseIntSizeSuffix("aKiB", 16) == 10240);
    try std.testing.expect(parseIntSizeSuffix("", 10) == error.InvalidCharacter);
    try std.testing.expect(parseIntSizeSuffix("2iB", 10) == error.InvalidCharacter);
}

pub const parseFloat = @import("fmt/parse_float.zig").parseFloat;
pub const ParseFloatError = @import("fmt/parse_float.zig").ParseFloatError;

test {
    _ = &parseFloat;
}

pub fn charToDigit(c: u8, base: u8) (error{InvalidCharacter}!u8) {
    const value = switch (c) {
        '0'...'9' => c - '0',
        'A'...'Z' => c - 'A' + 10,
        'a'...'z' => c - 'a' + 10,
        else => return error.InvalidCharacter,
    };

    if (value >= base) return error.InvalidCharacter;

    return value;
}

pub fn digitToChar(digit: u8, case: Case) u8 {
    return switch (digit) {
        0...9 => digit + '0',
        10...35 => digit + ((if (case == .upper) @as(u8, 'A') else @as(u8, 'a')) - 10),
        else => unreachable,
    };
}

pub const BufPrintError = error{
    /// As much as possible was written to the buffer, but it was too small to fit all the printed bytes.
    NoSpaceLeft,
};

/// Print a Formatter string into `buf`. Actually just a thin wrapper around `format` and `fixedBufferStream`.
/// Returns a slice of the bytes printed to.
pub fn bufPrint(buf: []u8, comptime fmt: []const u8, args: anytype) BufPrintError![]u8 {
    var fbs = std.io.fixedBufferStream(buf);
    format(fbs.writer().any(), fmt, args) catch |err| switch (err) {
        error.NoSpaceLeft => return error.NoSpaceLeft,
        else => unreachable,
    };
    return fbs.getWritten();
}

pub fn bufPrintZ(buf: []u8, comptime fmt: []const u8, args: anytype) BufPrintError![:0]u8 {
    const result = try bufPrint(buf, fmt ++ "\x00", args);
    return result[0 .. result.len - 1 :0];
}

/// Count the characters needed for format. Useful for preallocating memory
pub fn count(comptime fmt: []const u8, args: anytype) u64 {
    var counting_writer = std.io.countingWriter(std.io.null_writer);
    format(counting_writer.writer().any(), fmt, args) catch unreachable;
    return counting_writer.bytes_written;
}

pub const AllocPrintError = error{OutOfMemory};

pub fn allocPrint(allocator: mem.Allocator, comptime fmt: []const u8, args: anytype) AllocPrintError![]u8 {
    const size = math.cast(usize, count(fmt, args)) orelse return error.OutOfMemory;
    const buf = try allocator.alloc(u8, size);
    return bufPrint(buf, fmt, args) catch |err| switch (err) {
        error.NoSpaceLeft => unreachable, // we just counted the size above
    };
}

pub fn allocPrintZ(allocator: mem.Allocator, comptime fmt: []const u8, args: anytype) AllocPrintError![:0]u8 {
    const result = try allocPrint(allocator, fmt ++ "\x00", args);
    return result[0 .. result.len - 1 :0];
}

test bufPrintIntToSlice {
    var buffer: [100]u8 = undefined;
    const buf = buffer[0..];

    try std.testing.expectEqualSlices(u8, "-1", bufPrintIntToSlice(buf, @as(i1, -1), 10, .lower, FormatOptions{}));

    try std.testing.expectEqualSlices(u8, "-101111000110000101001110", bufPrintIntToSlice(buf, @as(i32, -12345678), 2, .lower, FormatOptions{}));
    try std.testing.expectEqualSlices(u8, "-12345678", bufPrintIntToSlice(buf, @as(i32, -12345678), 10, .lower, FormatOptions{}));
    try std.testing.expectEqualSlices(u8, "-bc614e", bufPrintIntToSlice(buf, @as(i32, -12345678), 16, .lower, FormatOptions{}));
    try std.testing.expectEqualSlices(u8, "-BC614E", bufPrintIntToSlice(buf, @as(i32, -12345678), 16, .upper, FormatOptions{}));

    try std.testing.expectEqualSlices(u8, "12345678", bufPrintIntToSlice(buf, @as(u32, 12345678), 10, .upper, FormatOptions{}));

    try std.testing.expectEqualSlices(u8, "   666", bufPrintIntToSlice(buf, @as(u32, 666), 10, .lower, FormatOptions{ .width = 6 }));
    try std.testing.expectEqualSlices(u8, "  1234", bufPrintIntToSlice(buf, @as(u32, 0x1234), 16, .lower, FormatOptions{ .width = 6 }));
    try std.testing.expectEqualSlices(u8, "1234", bufPrintIntToSlice(buf, @as(u32, 0x1234), 16, .lower, FormatOptions{ .width = 1 }));

    try std.testing.expectEqualSlices(u8, "+42", bufPrintIntToSlice(buf, @as(i32, 42), 10, .lower, FormatOptions{ .width = 3 }));
    try std.testing.expectEqualSlices(u8, "-42", bufPrintIntToSlice(buf, @as(i32, -42), 10, .lower, FormatOptions{ .width = 3 }));
}

pub fn bufPrintIntToSlice(buf: []u8, value: anytype, base: u8, case: Case, options: FormatOptions) []u8 {
    return buf[0..formatIntBuf(buf, value, base, case, options)];
}

pub inline fn comptimePrint(comptime fmt: []const u8, args: anytype) *const [count(fmt, args):0]u8 {
    comptime {
        var buf: [count(fmt, args):0]u8 = undefined;
        _ = bufPrint(&buf, fmt, args) catch unreachable;
        buf[buf.len] = 0;
        const final = buf;
        return &final;
    }
}

test comptimePrint {
    @setEvalBranchQuota(2000);
    try std.testing.expectEqual(*const [3:0]u8, @TypeOf(comptimePrint("{}", .{100})));
    try std.testing.expectEqualSlices(u8, "100", comptimePrint("{}", .{100}));
    try std.testing.expectEqualStrings("30", comptimePrint("{d}", .{30.0}));
    try std.testing.expectEqualStrings("30.0", comptimePrint("{d:3.1}", .{30.0}));
    try std.testing.expectEqualStrings("0.05", comptimePrint("{d}", .{0.05}));
    try std.testing.expectEqualStrings("5e-2", comptimePrint("{e}", .{0.05}));
}

test "parse u64 digit too big" {
    _ = parseUnsigned(u64, "123a", 10) catch |err| {
        if (err == error.InvalidCharacter) return;
        unreachable;
    };
    unreachable;
}

test "parse unsigned comptime" {
    comptime {
        try std.testing.expect((try parseUnsigned(usize, "2", 10)) == 2);
    }
}

test "escaped braces" {
    try expectFmt("escaped: {{foo}}\n", "escaped: {{{{foo}}}}\n", .{});
    try expectFmt("escaped: {foo}\n", "escaped: {{foo}}\n", .{});
}

test "optional" {
    {
        const value: ?i32 = 1234;
        try expectFmt("optional: 1234\n", "optional: {?}\n", .{value});
        try expectFmt("optional: 1234\n", "optional: {?d}\n", .{value});
        try expectFmt("optional: 4d2\n", "optional: {?x}\n", .{value});
    }
    {
        const value: ?[]const u8 = "string";
        try expectFmt("optional: string\n", "optional: {?s}\n", .{value});
    }
    {
        const value: ?i32 = null;
        try expectFmt("optional: null\n", "optional: {?}\n", .{value});
    }
    {
        const value = @as(?*i32, @ptrFromInt(0xf000d000));
        try expectFmt("optional: *i32@f000d000\n", "optional: {*}\n", .{value});
    }
}

test "error" {
    {
        const value: anyerror!i32 = 1234;
        try expectFmt("error union: 1234\n", "error union: {!}\n", .{value});
        try expectFmt("error union: 1234\n", "error union: {!d}\n", .{value});
        try expectFmt("error union: 4d2\n", "error union: {!x}\n", .{value});
    }
    {
        const value: anyerror![]const u8 = "string";
        try expectFmt("error union: string\n", "error union: {!s}\n", .{value});
    }
    {
        const value: anyerror!i32 = error.InvalidChar;
        try expectFmt("error union: error.InvalidChar\n", "error union: {!}\n", .{value});
    }
}

test "int.small" {
    {
        const value: u3 = 0b101;
        try expectFmt("u3: 5\n", "u3: {}\n", .{value});
    }
}

test "int.specifier" {
    {
        const value: u8 = 'a';
        try expectFmt("u8: a\n", "u8: {c}\n", .{value});
    }
    {
        const value: u8 = 0b1100;
        try expectFmt("u8: 0b1100\n", "u8: 0b{b}\n", .{value});
    }
    {
        const value: u16 = 0o1234;
        try expectFmt("u16: 0o1234\n", "u16: 0o{o}\n", .{value});
    }
    {
        const value: u8 = 'a';
        try expectFmt("UTF-8: a\n", "UTF-8: {u}\n", .{value});
    }
    {
        const value: u21 = 0x1F310;
        try expectFmt("UTF-8: 🌐\n", "UTF-8: {u}\n", .{value});
    }
    {
        const value: u21 = 0xD800;
        try expectFmt("UTF-8: �\n", "UTF-8: {u}\n", .{value});
    }
    {
        const value: u21 = 0x110001;
        try expectFmt("UTF-8: �\n", "UTF-8: {u}\n", .{value});
    }
}

test "int.padded" {
    try expectFmt("u8: '   1'", "u8: '{:4}'", .{@as(u8, 1)});
    try expectFmt("u8: '1000'", "u8: '{:0<4}'", .{@as(u8, 1)});
    try expectFmt("u8: '0001'", "u8: '{:0>4}'", .{@as(u8, 1)});
    try expectFmt("u8: '0100'", "u8: '{:0^4}'", .{@as(u8, 1)});
    try expectFmt("i8: '-1  '", "i8: '{:<4}'", .{@as(i8, -1)});
    try expectFmt("i8: '  -1'", "i8: '{:>4}'", .{@as(i8, -1)});
    try expectFmt("i8: ' -1 '", "i8: '{:^4}'", .{@as(i8, -1)});
    try expectFmt("i16: '-1234'", "i16: '{:4}'", .{@as(i16, -1234)});
    try expectFmt("i16: '+1234'", "i16: '{:4}'", .{@as(i16, 1234)});
    try expectFmt("i16: '-12345'", "i16: '{:4}'", .{@as(i16, -12345)});
    try expectFmt("i16: '+12345'", "i16: '{:4}'", .{@as(i16, 12345)});
    try expectFmt("u16: '12345'", "u16: '{:4}'", .{@as(u16, 12345)});

    try expectFmt("UTF-8: 'ü   '", "UTF-8: '{u:<4}'", .{'ü'});
    try expectFmt("UTF-8: '   ü'", "UTF-8: '{u:>4}'", .{'ü'});
    try expectFmt("UTF-8: ' ü  '", "UTF-8: '{u:^4}'", .{'ü'});
}

test "buffer" {
    {
        var buf1: [32]u8 = undefined;
        var fbs = std.io.fixedBufferStream(&buf1);
        try formatType(1234, "", FormatOptions{}, fbs.writer(), std.options.fmt_max_depth);
        try std.testing.expect(mem.eql(u8, fbs.getWritten(), "1234"));

        fbs.reset();
        try formatType('a', "c", FormatOptions{}, fbs.writer(), std.options.fmt_max_depth);
        try std.testing.expect(mem.eql(u8, fbs.getWritten(), "a"));

        fbs.reset();
        try formatType(0b1100, "b", FormatOptions{}, fbs.writer(), std.options.fmt_max_depth);
        try std.testing.expect(mem.eql(u8, fbs.getWritten(), "1100"));
    }
}

// Test formatting of arrays by value, by single-item pointer, and as a slice
fn expectArrayFmt(expected: []const u8, comptime template: []const u8, comptime array_value: anytype) !void {
    try expectFmt(expected, template, .{array_value});
    try expectFmt(expected, template, .{&array_value});
    var runtime_zero: usize = 0;
    _ = &runtime_zero;
    try expectFmt(expected, template, .{array_value[runtime_zero..]});
}

test "array" {
    {
        const value: [3]u8 = "abc".*;
        try expectArrayFmt("array: abc\n", "array: {s}\n", value);
        try expectArrayFmt("array: { 97, 98, 99 }\n", "array: {d}\n", value);
        try expectArrayFmt("array: { 61, 62, 63 }\n", "array: {x}\n", value);
        try expectArrayFmt("array: { 97, 98, 99 }\n", "array: {any}\n", value);

        var buf: [100]u8 = undefined;
        try expectFmt(
            try bufPrint(buf[0..], "array: [3]u8@{x}\n", .{@intFromPtr(&value)}),
            "array: {*}\n",
            .{&value},
        );
    }

    {
        const value = [2][3]u8{ "abc".*, "def".* };

        try expectArrayFmt("array: { abc, def }\n", "array: {s}\n", value);
        try expectArrayFmt("array: { { 97, 98, 99 }, { 100, 101, 102 } }\n", "array: {d}\n", value);
        try expectArrayFmt("array: { { 61, 62, 63 }, { 64, 65, 66 } }\n", "array: {x}\n", value);
    }
}

test "slice" {
    {
        const value: []const u8 = "abc";
        try expectFmt("slice: abc\n", "slice: {s}\n", .{value});
        try expectFmt("slice: { 97, 98, 99 }\n", "slice: {d}\n", .{value});
        try expectFmt("slice: { 61, 62, 63 }\n", "slice: {x}\n", .{value});
        try expectFmt("slice: { 97, 98, 99 }\n", "slice: {any}\n", .{value});
    }
    {
        var runtime_zero: usize = 0;
        _ = &runtime_zero;
        const value = @as([*]align(1) const []const u8, @ptrFromInt(0xdeadbeef))[runtime_zero..runtime_zero];
        try expectFmt("slice: []const u8@deadbeef\n", "slice: {*}\n", .{value});
    }
    {
        const null_term_slice: [:0]const u8 = "\x00hello\x00";
        try expectFmt("buf: \x00hello\x00\n", "buf: {s}\n", .{null_term_slice});
    }

    try expectFmt("buf:  Test\n", "buf: {s:5}\n", .{"Test"});
    try expectFmt("buf: Test\n Other text", "buf: {s}\n Other text", .{"Test"});

    {
        var int_slice = [_]u32{ 1, 4096, 391891, 1111111111 };
        var runtime_zero: usize = 0;
        _ = &runtime_zero;
        try expectFmt("int: { 1, 4096, 391891, 1111111111 }", "int: {any}", .{int_slice[runtime_zero..]});
        try expectFmt("int: { 1, 4096, 391891, 1111111111 }", "int: {d}", .{int_slice[runtime_zero..]});
        try expectFmt("int: { 1, 1000, 5fad3, 423a35c7 }", "int: {x}", .{int_slice[runtime_zero..]});
        try expectFmt("int: { 00001, 01000, 5fad3, 423a35c7 }", "int: {x:0>5}", .{int_slice[runtime_zero..]});
    }
    {
        const S1 = struct {
            x: u8,
        };
        const struct_slice: []const S1 = &[_]S1{ S1{ .x = 8 }, S1{ .x = 42 } };
        try expectFmt("slice: { fmt.test.slice.S1{ .x = 8 }, fmt.test.slice.S1{ .x = 42 } }", "slice: {any}", .{struct_slice});
    }
    {
        const S2 = struct {
            x: u8,

            pub fn format(s: @This(), comptime _: []const u8, _: std.fmt.FormatOptions, writer: anytype) !void {
                try writer.print("S2({})", .{s.x});
            }
        };
        const struct_slice: []const S2 = &[_]S2{ S2{ .x = 8 }, S2{ .x = 42 } };
        try expectFmt("slice: { S2(8), S2(42) }", "slice: {any}", .{struct_slice});
    }
}

test "escape non-printable" {
    try expectFmt("abc 123", "{s}", .{fmtSliceEscapeLower("abc 123")});
    try expectFmt("ab\\xffc", "{s}", .{fmtSliceEscapeLower("ab\xffc")});
    try expectFmt("abc 123", "{s}", .{fmtSliceEscapeUpper("abc 123")});
    try expectFmt("ab\\xFFc", "{s}", .{fmtSliceEscapeUpper("ab\xffc")});
}

test "pointer" {
    {
        const value = @as(*align(1) i32, @ptrFromInt(0xdeadbeef));
        try expectFmt("pointer: i32@deadbeef\n", "pointer: {}\n", .{value});
        try expectFmt("pointer: i32@deadbeef\n", "pointer: {*}\n", .{value});
    }
    const FnPtr = *align(1) const fn () void;
    {
        const value = @as(FnPtr, @ptrFromInt(0xdeadbeef));
        try expectFmt("pointer: fn () void@deadbeef\n", "pointer: {}\n", .{value});
    }
    {
        const value = @as(FnPtr, @ptrFromInt(0xdeadbeef));
        try expectFmt("pointer: fn () void@deadbeef\n", "pointer: {}\n", .{value});
    }
}

test "cstr" {
    try expectFmt(
        "cstr: Test C\n",
        "cstr: {s}\n",
        .{@as([*c]const u8, @ptrCast("Test C"))},
    );
    try expectFmt(
        "cstr:     Test C\n",
        "cstr: {s:10}\n",
        .{@as([*c]const u8, @ptrCast("Test C"))},
    );
}

test "filesize" {
    try expectFmt("file size: 42B\n", "file size: {}\n", .{fmtIntSizeDec(42)});
    try expectFmt("file size: 42B\n", "file size: {}\n", .{fmtIntSizeBin(42)});
    try expectFmt("file size: 63MB\n", "file size: {}\n", .{fmtIntSizeDec(63 * 1000 * 1000)});
    try expectFmt("file size: 63MiB\n", "file size: {}\n", .{fmtIntSizeBin(63 * 1024 * 1024)});
    try expectFmt("file size: 42B\n", "file size: {:.2}\n", .{fmtIntSizeDec(42)});
    try expectFmt("file size:       42B\n", "file size: {:>9.2}\n", .{fmtIntSizeDec(42)});
    try expectFmt("file size: 66.06MB\n", "file size: {:.2}\n", .{fmtIntSizeDec(63 * 1024 * 1024)});
    try expectFmt("file size: 60.08MiB\n", "file size: {:.2}\n", .{fmtIntSizeBin(63 * 1000 * 1000)});
    try expectFmt("file size: =66.06MB=\n", "file size: {:=^9.2}\n", .{fmtIntSizeDec(63 * 1024 * 1024)});
    try expectFmt("file size:   66.06MB\n", "file size: {: >9.2}\n", .{fmtIntSizeDec(63 * 1024 * 1024)});
    try expectFmt("file size: 66.06MB  \n", "file size: {: <9.2}\n", .{fmtIntSizeDec(63 * 1024 * 1024)});
    try expectFmt("file size: 0.01844674407370955ZB\n", "file size: {}\n", .{fmtIntSizeDec(math.maxInt(u64))});
}

test "struct" {
    {
        const Struct = struct {
            field: u8,
        };
        const value = Struct{ .field = 42 };
        try expectFmt("struct: fmt.test.struct.Struct{ .field = 42 }\n", "struct: {}\n", .{value});
        try expectFmt("struct: fmt.test.struct.Struct{ .field = 42 }\n", "struct: {}\n", .{&value});
    }
    {
        const Struct = struct {
            a: u0,
            b: u1,
        };
        const value = Struct{ .a = 0, .b = 1 };
        try expectFmt("struct: fmt.test.struct.Struct{ .a = 0, .b = 1 }\n", "struct: {}\n", .{value});
    }

    const S = struct {
        a: u32,
        b: anyerror,
    };

    const inst = S{
        .a = 456,
        .b = error.Unused,
    };

    try expectFmt("fmt.test.struct.S{ .a = 456, .b = error.Unused }", "{}", .{inst});
    // Tuples
    try expectFmt("{ }", "{}", .{.{}});
    try expectFmt("{ -1 }", "{}", .{.{-1}});
    try expectFmt("{ -1, 42, 2.5e4 }", "{}", .{.{ -1, 42, 0.25e5 }});
}

test "enum" {
    const Enum = enum {
        One,
        Two,
    };
    const value = Enum.Two;
    try expectFmt("enum: fmt.test.enum.Enum.Two\n", "enum: {}\n", .{value});
    try expectFmt("enum: fmt.test.enum.Enum.Two\n", "enum: {}\n", .{&value});
    try expectFmt("enum: fmt.test.enum.Enum.One\n", "enum: {}\n", .{Enum.One});
    try expectFmt("enum: fmt.test.enum.Enum.Two\n", "enum: {}\n", .{Enum.Two});

    // test very large enum to verify ct branch quota is large enough
    // TODO: https://github.com/ziglang/zig/issues/15609
    if (!((builtin.cpu.arch == .wasm32) and builtin.mode == .Debug)) {
        try expectFmt("enum: os.windows.win32error.Win32Error.INVALID_FUNCTION\n", "enum: {}\n", .{std.os.windows.Win32Error.INVALID_FUNCTION});
    }

    const E = enum {
        One,
        Two,
        Three,
    };

    const inst = E.Two;

    try expectFmt("fmt.test.enum.E.Two", "{}", .{inst});
}

test "non-exhaustive enum" {
    const Enum = enum(u16) {
        One = 0x000f,
        Two = 0xbeef,
        _,
    };
    try expectFmt("enum: fmt.test.non-exhaustive enum.Enum.One\n", "enum: {}\n", .{Enum.One});
    try expectFmt("enum: fmt.test.non-exhaustive enum.Enum.Two\n", "enum: {}\n", .{Enum.Two});
    try expectFmt("enum: fmt.test.non-exhaustive enum.Enum(4660)\n", "enum: {}\n", .{@as(Enum, @enumFromInt(0x1234))});
    try expectFmt("enum: fmt.test.non-exhaustive enum.Enum.One\n", "enum: {x}\n", .{Enum.One});
    try expectFmt("enum: fmt.test.non-exhaustive enum.Enum.Two\n", "enum: {x}\n", .{Enum.Two});
    try expectFmt("enum: fmt.test.non-exhaustive enum.Enum.Two\n", "enum: {X}\n", .{Enum.Two});
    try expectFmt("enum: fmt.test.non-exhaustive enum.Enum(1234)\n", "enum: {x}\n", .{@as(Enum, @enumFromInt(0x1234))});
}

test "float.scientific" {
    try expectFmt("f32: 1.34e0", "f32: {e}", .{@as(f32, 1.34)});
    try expectFmt("f32: 1.234e1", "f32: {e}", .{@as(f32, 12.34)});
    try expectFmt("f64: -1.234e11", "f64: {e}", .{@as(f64, -12.34e10)});
    try expectFmt("f64: 9.99996e-40", "f64: {e}", .{@as(f64, 9.999960e-40)});
}

test "float.scientific.precision" {
    try expectFmt("f64: 1.40971e-42", "f64: {e:.5}", .{@as(f64, 1.409706e-42)});
    try expectFmt("f64: 1.00000e-9", "f64: {e:.5}", .{@as(f64, @as(f32, @bitCast(@as(u32, 814313563))))});
    try expectFmt("f64: 7.81250e-3", "f64: {e:.5}", .{@as(f64, @as(f32, @bitCast(@as(u32, 1006632960))))});
    // libc rounds 1.000005e5 to 1.00000e5 but zig does 1.00001e5.
    // In fact, libc doesn't round a lot of 5 cases up when one past the precision point.
    try expectFmt("f64: 1.00001e5", "f64: {e:.5}", .{@as(f64, @as(f32, @bitCast(@as(u32, 1203982400))))});
}

test "float.special" {
    try expectFmt("f64: nan", "f64: {}", .{math.nan(f64)});
    // negative nan is not defined by IEE 754,
    // and ARM thus normalizes it to positive nan
    if (builtin.target.cpu.arch != .arm) {
        try expectFmt("f64: -nan", "f64: {}", .{-math.nan(f64)});
    }
    try expectFmt("f64: inf", "f64: {}", .{math.inf(f64)});
    try expectFmt("f64: -inf", "f64: {}", .{-math.inf(f64)});
}

test "float.hexadecimal.special" {
    try expectFmt("f64: nan", "f64: {x}", .{math.nan(f64)});
    // negative nan is not defined by IEE 754,
    // and ARM thus normalizes it to positive nan
    if (builtin.target.cpu.arch != .arm) {
        try expectFmt("f64: -nan", "f64: {x}", .{-math.nan(f64)});
    }
    try expectFmt("f64: inf", "f64: {x}", .{math.inf(f64)});
    try expectFmt("f64: -inf", "f64: {x}", .{-math.inf(f64)});

    try expectFmt("f64: 0x0.0p0", "f64: {x}", .{@as(f64, 0)});
    try expectFmt("f64: -0x0.0p0", "f64: {x}", .{-@as(f64, 0)});
}

test "float.hexadecimal" {
    try expectFmt("f16: 0x1.554p-2", "f16: {x}", .{@as(f16, 1.0 / 3.0)});
    try expectFmt("f32: 0x1.555556p-2", "f32: {x}", .{@as(f32, 1.0 / 3.0)});
    try expectFmt("f64: 0x1.5555555555555p-2", "f64: {x}", .{@as(f64, 1.0 / 3.0)});
    try expectFmt("f80: 0x1.5555555555555556p-2", "f80: {x}", .{@as(f80, 1.0 / 3.0)});
    try expectFmt("f128: 0x1.5555555555555555555555555555p-2", "f128: {x}", .{@as(f128, 1.0 / 3.0)});

    try expectFmt("f16: 0x1p-14", "f16: {x}", .{math.floatMin(f16)});
    try expectFmt("f32: 0x1p-126", "f32: {x}", .{math.floatMin(f32)});
    try expectFmt("f64: 0x1p-1022", "f64: {x}", .{math.floatMin(f64)});
    try expectFmt("f80: 0x1p-16382", "f80: {x}", .{math.floatMin(f80)});
    try expectFmt("f128: 0x1p-16382", "f128: {x}", .{math.floatMin(f128)});

    try expectFmt("f16: 0x0.004p-14", "f16: {x}", .{math.floatTrueMin(f16)});
    try expectFmt("f32: 0x0.000002p-126", "f32: {x}", .{math.floatTrueMin(f32)});
    try expectFmt("f64: 0x0.0000000000001p-1022", "f64: {x}", .{math.floatTrueMin(f64)});
    try expectFmt("f80: 0x0.0000000000000002p-16382", "f80: {x}", .{math.floatTrueMin(f80)});
    try expectFmt("f128: 0x0.0000000000000000000000000001p-16382", "f128: {x}", .{math.floatTrueMin(f128)});

    try expectFmt("f16: 0x1.ffcp15", "f16: {x}", .{math.floatMax(f16)});
    try expectFmt("f32: 0x1.fffffep127", "f32: {x}", .{math.floatMax(f32)});
    try expectFmt("f64: 0x1.fffffffffffffp1023", "f64: {x}", .{math.floatMax(f64)});
    try expectFmt("f80: 0x1.fffffffffffffffep16383", "f80: {x}", .{math.floatMax(f80)});
    try expectFmt("f128: 0x1.ffffffffffffffffffffffffffffp16383", "f128: {x}", .{math.floatMax(f128)});
}

test "float.hexadecimal.precision" {
    try expectFmt("f16: 0x1.5p-2", "f16: {x:.1}", .{@as(f16, 1.0 / 3.0)});
    try expectFmt("f32: 0x1.555p-2", "f32: {x:.3}", .{@as(f32, 1.0 / 3.0)});
    try expectFmt("f64: 0x1.55555p-2", "f64: {x:.5}", .{@as(f64, 1.0 / 3.0)});
    try expectFmt("f80: 0x1.5555555p-2", "f80: {x:.7}", .{@as(f80, 1.0 / 3.0)});
    try expectFmt("f128: 0x1.555555555p-2", "f128: {x:.9}", .{@as(f128, 1.0 / 3.0)});

    try expectFmt("f16: 0x1.00000p0", "f16: {x:.5}", .{@as(f16, 1.0)});
    try expectFmt("f32: 0x1.00000p0", "f32: {x:.5}", .{@as(f32, 1.0)});
    try expectFmt("f64: 0x1.00000p0", "f64: {x:.5}", .{@as(f64, 1.0)});
    try expectFmt("f80: 0x1.00000p0", "f80: {x:.5}", .{@as(f80, 1.0)});
    try expectFmt("f128: 0x1.00000p0", "f128: {x:.5}", .{@as(f128, 1.0)});
}

test "float.decimal" {
    try expectFmt("f64: 152314000000000000000000000000", "f64: {d}", .{@as(f64, 1.52314e29)});
    try expectFmt("f32: 0", "f32: {d}", .{@as(f32, 0.0)});
    try expectFmt("f32: 0", "f32: {d:.0}", .{@as(f32, 0.0)});
    try expectFmt("f32: 1.1", "f32: {d:.1}", .{@as(f32, 1.1234)});
    try expectFmt("f32: 1234.57", "f32: {d:.2}", .{@as(f32, 1234.567)});
    // -11.1234 is converted to f64 -11.12339... internally (errol3() function takes f64).
    // -11.12339... is rounded back up to -11.1234
    try expectFmt("f32: -11.1234", "f32: {d:.4}", .{@as(f32, -11.1234)});
    try expectFmt("f32: 91.12345", "f32: {d:.5}", .{@as(f32, 91.12345)});
    try expectFmt("f64: 91.1234567890", "f64: {d:.10}", .{@as(f64, 91.12345678901235)});
    try expectFmt("f64: 0.00000", "f64: {d:.5}", .{@as(f64, 0.0)});
    try expectFmt("f64: 6", "f64: {d:.0}", .{@as(f64, 5.700)});
    try expectFmt("f64: 10.0", "f64: {d:.1}", .{@as(f64, 9.999)});
    try expectFmt("f64: 1.000", "f64: {d:.3}", .{@as(f64, 1.0)});
    try expectFmt("f64: 0.00030000", "f64: {d:.8}", .{@as(f64, 0.0003)});
    try expectFmt("f64: 0.00000", "f64: {d:.5}", .{@as(f64, 1.40130e-45)});
    try expectFmt("f64: 0.00000", "f64: {d:.5}", .{@as(f64, 9.999960e-40)});
    try expectFmt("f64: 10000000000000.00", "f64: {d:.2}", .{@as(f64, 9999999999999.999)});
    try expectFmt("f64: 10000000000000000000000000000000000000", "f64: {d}", .{@as(f64, 1e37)});
    try expectFmt("f64: 100000000000000000000000000000000000000", "f64: {d}", .{@as(f64, 1e38)});
}

test "float.libc.sanity" {
    try expectFmt("f64: 0.00001", "f64: {d:.5}", .{@as(f64, @as(f32, @bitCast(@as(u32, 916964781))))});
    try expectFmt("f64: 0.00001", "f64: {d:.5}", .{@as(f64, @as(f32, @bitCast(@as(u32, 925353389))))});
    try expectFmt("f64: 0.10000", "f64: {d:.5}", .{@as(f64, @as(f32, @bitCast(@as(u32, 1036831278))))});
    try expectFmt("f64: 1.00000", "f64: {d:.5}", .{@as(f64, @as(f32, @bitCast(@as(u32, 1065353133))))});
    try expectFmt("f64: 10.00000", "f64: {d:.5}", .{@as(f64, @as(f32, @bitCast(@as(u32, 1092616192))))});

    // libc differences
    //
    // This is 0.015625 exactly according to gdb. We thus round down,
    // however glibc rounds up for some reason. This occurs for all
    // floats of the form x.yyyy25 on a precision point.
    try expectFmt("f64: 0.01563", "f64: {d:.5}", .{@as(f64, @as(f32, @bitCast(@as(u32, 1015021568))))});
    // errol3 rounds to ... 630 but libc rounds to ...632. Grisu3
    // also rounds to 630 so I'm inclined to believe libc is not
    // optimal here.
    try expectFmt("f64: 18014400656965630.00000", "f64: {d:.5}", .{@as(f64, @as(f32, @bitCast(@as(u32, 1518338049))))});
}

test "custom" {
    const Vec2 = struct {
        const SelfType = @This();
        x: f32,
        y: f32,

        pub fn format(
            self: SelfType,
            comptime fmt: []const u8,
            options: FormatOptions,
            writer: anytype,
        ) !void {
            _ = options;
            if (fmt.len == 0 or comptime std.mem.eql(u8, fmt, "p")) {
                return std.fmt.format(writer, "({d:.3},{d:.3})", .{ self.x, self.y });
            } else if (comptime std.mem.eql(u8, fmt, "d")) {
                return std.fmt.format(writer, "{d:.3}x{d:.3}", .{ self.x, self.y });
            } else {
                @compileError("unknown format character: '" ++ fmt ++ "'");
            }
        }
    };

    var value = Vec2{
        .x = 10.2,
        .y = 2.22,
    };
    try expectFmt("point: (10.200,2.220)\n", "point: {}\n", .{&value});
    try expectFmt("dim: 10.200x2.220\n", "dim: {d}\n", .{&value});

    // same thing but not passing a pointer
    try expectFmt("point: (10.200,2.220)\n", "point: {}\n", .{value});
    try expectFmt("dim: 10.200x2.220\n", "dim: {d}\n", .{value});
}

test "union" {
    const TU = union(enum) {
        float: f32,
        int: u32,
    };

    const UU = union {
        float: f32,
        int: u32,
    };

    const EU = extern union {
        float: f32,
        int: u32,
    };

    const tu_inst = TU{ .int = 123 };
    const uu_inst = UU{ .int = 456 };
    const eu_inst = EU{ .float = 321.123 };

    try expectFmt("fmt.test.union.TU{ .int = 123 }", "{}", .{tu_inst});

    var buf: [100]u8 = undefined;
    const uu_result = try bufPrint(buf[0..], "{}", .{uu_inst});
    try std.testing.expect(mem.eql(u8, uu_result[0..18], "fmt.test.union.UU@"));

    const eu_result = try bufPrint(buf[0..], "{}", .{eu_inst});
    try std.testing.expect(mem.eql(u8, eu_result[0..18], "fmt.test.union.EU@"));
}

test "struct.self-referential" {
    const S = struct {
        const SelfType = @This();
        a: ?*SelfType,
    };

    var inst = S{
        .a = null,
    };
    inst.a = &inst;

    try expectFmt("fmt.test.struct.self-referential.S{ .a = fmt.test.struct.self-referential.S{ .a = fmt.test.struct.self-referential.S{ .a = fmt.test.struct.self-referential.S{ ... } } } }", "{}", .{inst});
}

test "struct.zero-size" {
    const A = struct {
        fn foo() void {}
    };
    const B = struct {
        a: A,
        c: i32,
    };

    const a = A{};
    const b = B{ .a = a, .c = 0 };

    try expectFmt("fmt.test.struct.zero-size.B{ .a = fmt.test.struct.zero-size.A{ }, .c = 0 }", "{}", .{b});
}

test "bytes.hex" {
    const some_bytes = "\xCA\xFE\xBA\xBE";
    try expectFmt("lowercase: cafebabe\n", "lowercase: {x}\n", .{fmtSliceHexLower(some_bytes)});
    try expectFmt("uppercase: CAFEBABE\n", "uppercase: {X}\n", .{fmtSliceHexUpper(some_bytes)});
    //Test Slices
    try expectFmt("uppercase: CAFE\n", "uppercase: {X}\n", .{fmtSliceHexUpper(some_bytes[0..2])});
    try expectFmt("lowercase: babe\n", "lowercase: {x}\n", .{fmtSliceHexLower(some_bytes[2..])});
    const bytes_with_zeros = "\x00\x0E\xBA\xBE";
    try expectFmt("lowercase: 000ebabe\n", "lowercase: {x}\n", .{fmtSliceHexLower(bytes_with_zeros)});
}

/// Encodes a sequence of bytes as hexadecimal digits.
/// Returns an array containing the encoded bytes.
pub fn bytesToHex(input: anytype, case: Case) [input.len * 2]u8 {
    if (input.len == 0) return [_]u8{};
    comptime assert(@TypeOf(input[0]) == u8); // elements to encode must be unsigned bytes

    const charset = "0123456789" ++ if (case == .upper) "ABCDEF" else "abcdef";
    var result: [input.len * 2]u8 = undefined;
    for (input, 0..) |b, i| {
        result[i * 2 + 0] = charset[b >> 4];
        result[i * 2 + 1] = charset[b & 15];
    }
    return result;
}

/// Decodes the sequence of bytes represented by the specified string of
/// hexadecimal characters.
/// Returns a slice of the output buffer containing the decoded bytes.
pub fn hexToBytes(out: []u8, input: []const u8) ![]u8 {
    // Expect 0 or n pairs of hexadecimal digits.
    if (input.len & 1 != 0)
        return error.InvalidLength;
    if (out.len * 2 < input.len)
        return error.NoSpaceLeft;

    var in_i: usize = 0;
    while (in_i < input.len) : (in_i += 2) {
        const hi = try charToDigit(input[in_i], 16);
        const lo = try charToDigit(input[in_i + 1], 16);
        out[in_i / 2] = (hi << 4) | lo;
    }

    return out[0 .. in_i / 2];
}

test bytesToHex {
    const input = "input slice";
    const encoded = bytesToHex(input, .lower);
    var decoded: [input.len]u8 = undefined;
    try std.testing.expectEqualSlices(u8, input, try hexToBytes(&decoded, &encoded));
}

test hexToBytes {
    var buf: [32]u8 = undefined;
    try expectFmt("90" ** 32, "{s}", .{fmtSliceHexUpper(try hexToBytes(&buf, "90" ** 32))});
    try expectFmt("ABCD", "{s}", .{fmtSliceHexUpper(try hexToBytes(&buf, "ABCD"))});
    try expectFmt("", "{s}", .{fmtSliceHexUpper(try hexToBytes(&buf, ""))});
    try std.testing.expectError(error.InvalidCharacter, hexToBytes(&buf, "012Z"));
    try std.testing.expectError(error.InvalidLength, hexToBytes(&buf, "AAA"));
    try std.testing.expectError(error.NoSpaceLeft, hexToBytes(buf[0..1], "ABAB"));
}

test "formatIntValue with comptime_int" {
    const value: comptime_int = 123456789123456789;

    var buf: [20]u8 = undefined;
    var fbs = std.io.fixedBufferStream(&buf);
    try formatIntValue(value, "", FormatOptions{}, fbs.writer());
    try std.testing.expect(mem.eql(u8, fbs.getWritten(), "123456789123456789"));
}

test "formatFloatValue with comptime_float" {
    const value: comptime_float = 1.0;

    var buf: [20]u8 = undefined;
    var fbs = std.io.fixedBufferStream(&buf);
    try formatFloatValue(value, "", FormatOptions{}, fbs.writer());
    try std.testing.expectEqualStrings(fbs.getWritten(), "1e0");

    try expectFmt("1e0", "{}", .{value});
    try expectFmt("1e0", "{}", .{1.0});
}

test "formatType max_depth" {
    const Vec2 = struct {
        const SelfType = @This();
        x: f32,
        y: f32,

        pub fn format(
            self: SelfType,
            comptime fmt: []const u8,
            options: FormatOptions,
            writer: anytype,
        ) !void {
            _ = options;
            if (fmt.len == 0) {
                return std.fmt.format(writer, "({d:.3},{d:.3})", .{ self.x, self.y });
            } else {
                @compileError("unknown format string: '" ++ fmt ++ "'");
            }
        }
    };
    const E = enum {
        One,
        Two,
        Three,
    };
    const TU = union(enum) {
        const SelfType = @This();
        float: f32,
        int: u32,
        ptr: ?*SelfType,
    };
    const S = struct {
        const SelfType = @This();
        a: ?*SelfType,
        tu: TU,
        e: E,
        vec: Vec2,
    };

    var inst = S{
        .a = null,
        .tu = TU{ .ptr = null },
        .e = E.Two,
        .vec = Vec2{ .x = 10.2, .y = 2.22 },
    };
    inst.a = &inst;
    inst.tu.ptr = &inst.tu;

    var buf: [1000]u8 = undefined;
    var fbs = std.io.fixedBufferStream(&buf);
    try formatType(inst, "", FormatOptions{}, fbs.writer(), 0);
    try std.testing.expect(mem.eql(u8, fbs.getWritten(), "fmt.test.formatType max_depth.S{ ... }"));

    fbs.reset();
    try formatType(inst, "", FormatOptions{}, fbs.writer(), 1);
    try std.testing.expect(mem.eql(u8, fbs.getWritten(), "fmt.test.formatType max_depth.S{ .a = fmt.test.formatType max_depth.S{ ... }, .tu = fmt.test.formatType max_depth.TU{ ... }, .e = fmt.test.formatType max_depth.E.Two, .vec = (10.200,2.220) }"));

    fbs.reset();
    try formatType(inst, "", FormatOptions{}, fbs.writer(), 2);
    try std.testing.expect(mem.eql(u8, fbs.getWritten(), "fmt.test.formatType max_depth.S{ .a = fmt.test.formatType max_depth.S{ .a = fmt.test.formatType max_depth.S{ ... }, .tu = fmt.test.formatType max_depth.TU{ ... }, .e = fmt.test.formatType max_depth.E.Two, .vec = (10.200,2.220) }, .tu = fmt.test.formatType max_depth.TU{ .ptr = fmt.test.formatType max_depth.TU{ ... } }, .e = fmt.test.formatType max_depth.E.Two, .vec = (10.200,2.220) }"));

    fbs.reset();
    try formatType(inst, "", FormatOptions{}, fbs.writer(), 3);
    try std.testing.expect(mem.eql(u8, fbs.getWritten(), "fmt.test.formatType max_depth.S{ .a = fmt.test.formatType max_depth.S{ .a = fmt.test.formatType max_depth.S{ .a = fmt.test.formatType max_depth.S{ ... }, .tu = fmt.test.formatType max_depth.TU{ ... }, .e = fmt.test.formatType max_depth.E.Two, .vec = (10.200,2.220) }, .tu = fmt.test.formatType max_depth.TU{ .ptr = fmt.test.formatType max_depth.TU{ ... } }, .e = fmt.test.formatType max_depth.E.Two, .vec = (10.200,2.220) }, .tu = fmt.test.formatType max_depth.TU{ .ptr = fmt.test.formatType max_depth.TU{ .ptr = fmt.test.formatType max_depth.TU{ ... } } }, .e = fmt.test.formatType max_depth.E.Two, .vec = (10.200,2.220) }"));
}

test "positional" {
    try expectFmt("2 1 0", "{2} {1} {0}", .{ @as(usize, 0), @as(usize, 1), @as(usize, 2) });
    try expectFmt("2 1 0", "{2} {1} {}", .{ @as(usize, 0), @as(usize, 1), @as(usize, 2) });
    try expectFmt("0 0", "{0} {0}", .{@as(usize, 0)});
    try expectFmt("0 1", "{} {1}", .{ @as(usize, 0), @as(usize, 1) });
    try expectFmt("1 0 0 1", "{1} {} {0} {}", .{ @as(usize, 0), @as(usize, 1) });
}

test "positional with specifier" {
    try expectFmt("10.0", "{0d:.1}", .{@as(f64, 9.999)});
}

test "positional/alignment/width/precision" {
    try expectFmt("10.0", "{0d: >3.1}", .{@as(f64, 9.999)});
}

test "vector" {
    if (builtin.target.cpu.arch == .riscv64) {
        // https://github.com/ziglang/zig/issues/4486
        return error.SkipZigTest;
    }

    const vbool: @Vector(4, bool) = [_]bool{ true, false, true, false };
    const vi64: @Vector(4, i64) = [_]i64{ -2, -1, 0, 1 };
    const vu64: @Vector(4, u64) = [_]u64{ 1000, 2000, 3000, 4000 };

    try expectFmt("{ true, false, true, false }", "{}", .{vbool});
    try expectFmt("{ -2, -1, 0, 1 }", "{}", .{vi64});
    try expectFmt("{    -2,    -1,    +0,    +1 }", "{d:5}", .{vi64});
    try expectFmt("{ 1000, 2000, 3000, 4000 }", "{}", .{vu64});
    try expectFmt("{ 3e8, 7d0, bb8, fa0 }", "{x}", .{vu64});
}

test "enum-literal" {
    try expectFmt(".hello_world", "{}", .{.hello_world});
}

test "padding" {
    try expectFmt("Simple", "{s}", .{"Simple"});
    try expectFmt("      true", "{:10}", .{true});
    try expectFmt("      true", "{:>10}", .{true});
    try expectFmt("======true", "{:=>10}", .{true});
    try expectFmt("true======", "{:=<10}", .{true});
    try expectFmt("   true   ", "{:^10}", .{true});
    try expectFmt("===true===", "{:=^10}", .{true});
    try expectFmt("           Minimum width", "{s:18} width", .{"Minimum"});
    try expectFmt("==================Filled", "{s:=>24}", .{"Filled"});
    try expectFmt("        Centered        ", "{s:^24}", .{"Centered"});
    try expectFmt("-", "{s:-^1}", .{""});
    try expectFmt("==crêpe===", "{s:=^10}", .{"crêpe"});
    try expectFmt("=====crêpe", "{s:=>10}", .{"crêpe"});
    try expectFmt("crêpe=====", "{s:=<10}", .{"crêpe"});
    try expectFmt("====a", "{c:=>5}", .{'a'});
    try expectFmt("==a==", "{c:=^5}", .{'a'});
    try expectFmt("a====", "{c:=<5}", .{'a'});
}

test "padding fill char utf" {
    try expectFmt("──crêpe───", "{s:─^10}", .{"crêpe"});
    try expectFmt("─────crêpe", "{s:─>10}", .{"crêpe"});
    try expectFmt("crêpe─────", "{s:─<10}", .{"crêpe"});
    try expectFmt("────a", "{c:─>5}", .{'a'});
    try expectFmt("──a──", "{c:─^5}", .{'a'});
    try expectFmt("a────", "{c:─<5}", .{'a'});
}

test "decimal float padding" {
    const number: f32 = 3.1415;
    try expectFmt("left-pad:   **3.142\n", "left-pad:   {d:*>7.3}\n", .{number});
    try expectFmt("center-pad: *3.142*\n", "center-pad: {d:*^7.3}\n", .{number});
    try expectFmt("right-pad:  3.142**\n", "right-pad:  {d:*<7.3}\n", .{number});
}

test "sci float padding" {
    const number: f32 = 3.1415;
    try expectFmt("left-pad:   ****3.142e0\n", "left-pad:   {e:*>11.3}\n", .{number});
    try expectFmt("center-pad: **3.142e0**\n", "center-pad: {e:*^11.3}\n", .{number});
    try expectFmt("right-pad:  3.142e0****\n", "right-pad:  {e:*<11.3}\n", .{number});
}

test "null" {
    const inst = null;
    try expectFmt("null", "{}", .{inst});
}

test "type" {
    try expectFmt("u8", "{}", .{u8});
    try expectFmt("?f32", "{}", .{?f32});
    try expectFmt("[]const u8", "{}", .{[]const u8});
}

test "named arguments" {
    try expectFmt("hello world!", "{s} world{c}", .{ "hello", '!' });
    try expectFmt("hello world!", "{[greeting]s} world{[punctuation]c}", .{ .punctuation = '!', .greeting = "hello" });
    try expectFmt("hello world!", "{[1]s} world{[0]c}", .{ '!', "hello" });
}

test "runtime width specifier" {
    const width: usize = 9;
    try expectFmt("~~hello~~", "{s:~^[1]}", .{ "hello", width });
    try expectFmt("~~hello~~", "{s:~^[width]}", .{ .string = "hello", .width = width });
    try expectFmt("    hello", "{s:[1]}", .{ "hello", width });
    try expectFmt("42     hello", "{d} {s:[2]}", .{ 42, "hello", width });
}

test "runtime precision specifier" {
    const number: f32 = 3.1415;
    const precision: usize = 2;
    try expectFmt("3.14e0", "{:1.[1]}", .{ number, precision });
    try expectFmt("3.14e0", "{:1.[precision]}", .{ .number = number, .precision = precision });
}

test "recursive format function" {
    const R = union(enum) {
        const R = @This();
        Leaf: i32,
        Branch: struct { left: *const R, right: *const R },

        pub fn format(self: R, comptime _: []const u8, _: std.fmt.FormatOptions, writer: anytype) !void {
            return switch (self) {
                .Leaf => |n| std.fmt.format(writer, "Leaf({})", .{n}),
                .Branch => |b| std.fmt.format(writer, "Branch({}, {})", .{ b.left, b.right }),
            };
        }
    };

    var r = R{ .Leaf = 1 };
    try expectFmt("Leaf(1)\n", "{}\n", .{&r});
}