std.Io.Reader: use readVec for fill functions

readVec has two updated responsibilities:
1. it must respect any existing already buffered data.
2. it must write to the buffer if data is empty
This commit is contained in:
Andrew Kelley 2025-08-05 11:26:00 -07:00
parent e17a050bc6
commit 04fe1bfe3c
2 changed files with 123 additions and 65 deletions

View file

@ -70,13 +70,14 @@ pub const VTable = struct {
/// Returns number of bytes written to `data`. /// Returns number of bytes written to `data`.
/// ///
/// `data` may not have nonzero length. /// `data` must have nonzero length. `data[0]` may have zero length, in
/// which case the implementation must write to `Reader.buffer`.
/// ///
/// `data` may not contain an alias to `Reader.buffer`. /// `data` may not contain an alias to `Reader.buffer`.
/// ///
/// `data` is mutable because the implementation may to temporarily modify /// `data` is mutable because the implementation may temporarily modify the
/// the fields in order to handle partial reads. Implementations must /// fields in order to handle partial reads. Implementations must restore
/// restore the original value before returning. /// the original value before returning.
/// ///
/// Implementations may ignore `data`, writing directly to `Reader.buffer`, /// Implementations may ignore `data`, writing directly to `Reader.buffer`,
/// modifying `seek` and `end` accordingly, and returning 0 from this /// modifying `seek` and `end` accordingly, and returning 0 from this
@ -421,23 +422,29 @@ pub fn readVec(r: *Reader, data: [][]u8) Error!usize {
/// Writes to `Reader.buffer` or `data`, whichever has larger capacity. /// Writes to `Reader.buffer` or `data`, whichever has larger capacity.
pub fn defaultReadVec(r: *Reader, data: [][]u8) Error!usize { pub fn defaultReadVec(r: *Reader, data: [][]u8) Error!usize {
assert(r.seek == r.end);
r.seek = 0;
r.end = 0;
const first = data[0]; const first = data[0];
const direct = first.len >= r.buffer.len; if (r.seek == r.end and first.len >= r.buffer.len) {
var writer: Writer = .{
.buffer = first,
.end = 0,
.vtable = &.{ .drain = Writer.fixedDrain },
};
const limit: Limit = .limited(writer.buffer.len - writer.end);
return r.vtable.stream(r, &writer, limit) catch |err| switch (err) {
error.WriteFailed => unreachable,
else => |e| return e,
};
}
var writer: Writer = .{ var writer: Writer = .{
.buffer = if (direct) first else r.buffer, .buffer = r.buffer,
.end = 0, .end = r.end,
.vtable = &.{ .drain = Writer.fixedDrain }, .vtable = &.{ .drain = Writer.fixedDrain },
}; };
const limit: Limit = .limited(writer.buffer.len - writer.end); const limit: Limit = .limited(writer.buffer.len - writer.end);
const n = r.vtable.stream(r, &writer, limit) catch |err| switch (err) { r.end += r.vtable.stream(r, &writer, limit) catch |err| switch (err) {
error.WriteFailed => unreachable, error.WriteFailed => unreachable,
else => |e| return e, else => |e| return e,
}; };
if (direct) return n;
r.end += n;
return 0; return 0;
} }
@ -1059,17 +1066,8 @@ pub fn fill(r: *Reader, n: usize) Error!void {
/// increasing by a factor of 5 or more. /// increasing by a factor of 5 or more.
fn fillUnbuffered(r: *Reader, n: usize) Error!void { fn fillUnbuffered(r: *Reader, n: usize) Error!void {
try rebase(r, n); try rebase(r, n);
var writer: Writer = .{ var bufs: [1][]u8 = .{""};
.buffer = r.buffer, while (r.end < r.seek + n) _ = try r.vtable.readVec(r, &bufs);
.vtable = &.{ .drain = Writer.fixedDrain },
};
while (r.end < r.seek + n) {
writer.end = r.end;
r.end += r.vtable.stream(r, &writer, .limited(r.buffer.len - r.end)) catch |err| switch (err) {
error.WriteFailed => unreachable,
error.ReadFailed, error.EndOfStream => |e| return e,
};
}
} }
/// Without advancing the seek position, does exactly one underlying read, filling the buffer as /// Without advancing the seek position, does exactly one underlying read, filling the buffer as
@ -1079,15 +1077,8 @@ fn fillUnbuffered(r: *Reader, n: usize) Error!void {
/// Asserts buffer capacity is at least 1. /// Asserts buffer capacity is at least 1.
pub fn fillMore(r: *Reader) Error!void { pub fn fillMore(r: *Reader) Error!void {
try rebase(r, 1); try rebase(r, 1);
var writer: Writer = .{ var bufs: [1][]u8 = .{""};
.buffer = r.buffer, _ = try r.vtable.readVec(r, &bufs);
.end = r.end,
.vtable = &.{ .drain = Writer.fixedDrain },
};
r.end += r.vtable.stream(r, &writer, .limited(r.buffer.len - r.end)) catch |err| switch (err) {
error.WriteFailed => unreachable,
else => |e| return e,
};
} }
/// Returns the next byte from the stream or returns `error.EndOfStream`. /// Returns the next byte from the stream or returns `error.EndOfStream`.
@ -1796,18 +1787,26 @@ pub fn Hashed(comptime Hasher: type) type {
fn readVec(r: *Reader, data: [][]u8) Error!usize { fn readVec(r: *Reader, data: [][]u8) Error!usize {
const this: *@This() = @alignCast(@fieldParentPtr("reader", r)); const this: *@This() = @alignCast(@fieldParentPtr("reader", r));
const n = try this.in.readVec(data); var vecs: [8][]u8 = undefined; // Arbitrarily chosen amount.
const dest_n, const data_size = try r.writableVector(&vecs, data);
const dest = vecs[0..dest_n];
const n = try this.in.readVec(dest);
var remaining: usize = n; var remaining: usize = n;
for (data) |slice| { for (dest) |slice| {
if (remaining < slice.len) { if (remaining < slice.len) {
this.hasher.update(slice[0..remaining]); this.hasher.update(slice[0..remaining]);
return n; remaining = 0;
break;
} else { } else {
remaining -= slice.len; remaining -= slice.len;
this.hasher.update(slice); this.hasher.update(slice);
} }
} }
assert(remaining == 0); assert(remaining == 0);
if (n > data_size) {
r.end += n - data_size;
return data_size;
}
return n; return n;
} }
@ -1824,17 +1823,24 @@ pub fn Hashed(comptime Hasher: type) type {
pub fn writableVectorPosix(r: *Reader, buffer: []std.posix.iovec, data: []const []u8) Error!struct { usize, usize } { pub fn writableVectorPosix(r: *Reader, buffer: []std.posix.iovec, data: []const []u8) Error!struct { usize, usize } {
var i: usize = 0; var i: usize = 0;
var n: usize = 0; var n: usize = 0;
for (data) |buf| { if (r.seek == r.end) {
if (buffer.len - i == 0) return .{ i, n }; for (data) |buf| {
if (buffer.len - i == 0) return .{ i, n };
if (buf.len != 0) {
buffer[i] = .{ .base = buf.ptr, .len = buf.len };
i += 1;
n += buf.len;
}
}
const buf = r.buffer;
if (buf.len != 0) { if (buf.len != 0) {
r.seek = 0;
r.end = 0;
buffer[i] = .{ .base = buf.ptr, .len = buf.len }; buffer[i] = .{ .base = buf.ptr, .len = buf.len };
i += 1; i += 1;
n += buf.len;
} }
} } else {
assert(r.seek == r.end); const buf = r.buffer[r.end..];
const buf = r.buffer;
if (buf.len != 0) {
buffer[i] = .{ .base = buf.ptr, .len = buf.len }; buffer[i] = .{ .base = buf.ptr, .len = buf.len };
i += 1; i += 1;
} }
@ -1848,28 +1854,62 @@ pub fn writableVectorWsa(
) Error!struct { usize, usize } { ) Error!struct { usize, usize } {
var i: usize = 0; var i: usize = 0;
var n: usize = 0; var n: usize = 0;
for (data) |buf| { if (r.seek == r.end) {
if (buffer.len - i == 0) return .{ i, n }; for (data) |buf| {
if (buf.len == 0) continue; if (buffer.len - i == 0) return .{ i, n };
if (std.math.cast(u32, buf.len)) |len| { if (buf.len == 0) continue;
buffer[i] = .{ .buf = buf.ptr, .len = len }; if (std.math.cast(u32, buf.len)) |len| {
i += 1; buffer[i] = .{ .buf = buf.ptr, .len = len };
n += len; i += 1;
continue; n += len;
} continue;
buffer[i] = .{ .buf = buf.ptr, .len = std.math.maxInt(u32) }; }
i += 1;
n += std.math.maxInt(u32);
return .{ i, n };
}
assert(r.seek == r.end);
const buf = r.buffer;
if (buf.len != 0) {
if (std.math.cast(u32, buf.len)) |len| {
buffer[i] = .{ .buf = buf.ptr, .len = len };
} else {
buffer[i] = .{ .buf = buf.ptr, .len = std.math.maxInt(u32) }; buffer[i] = .{ .buf = buf.ptr, .len = std.math.maxInt(u32) };
i += 1;
n += std.math.maxInt(u32);
return .{ i, n };
} }
const buf = r.buffer;
if (buf.len != 0) {
r.seek = 0;
r.end = 0;
if (std.math.cast(u32, buf.len)) |len| {
buffer[i] = .{ .buf = buf.ptr, .len = len };
} else {
buffer[i] = .{ .buf = buf.ptr, .len = std.math.maxInt(u32) };
}
i += 1;
}
} else {
buffer[i] = .{
.buf = r.buffer.ptr + r.end,
.len = @min(std.math.maxInt(u32), r.buffer.len - r.end),
};
i += 1;
}
return .{ i, n };
}
pub fn writableVector(r: *Reader, buffer: [][]u8, data: []const []u8) Error!struct { usize, usize } {
var i: usize = 0;
var n: usize = 0;
if (r.seek == r.end) {
for (data) |buf| {
if (buffer.len - i == 0) return .{ i, n };
if (buf.len != 0) {
buffer[i] = buf;
i += 1;
n += buf.len;
}
}
if (r.buffer.len != 0) {
r.seek = 0;
r.end = 0;
buffer[i] = r.buffer;
i += 1;
}
} else {
buffer[i] = r.buffer[r.end..];
i += 1; i += 1;
} }
return .{ i, n }; return .{ i, n };

View file

@ -1312,7 +1312,16 @@ pub const Reader = struct {
if (is_windows) { if (is_windows) {
// Unfortunately, `ReadFileScatter` cannot be used since it // Unfortunately, `ReadFileScatter` cannot be used since it
// requires page alignment. // requires page alignment.
return readPositional(r, data[0]); assert(io_reader.seek == io_reader.end);
io_reader.seek = 0;
io_reader.end = 0;
const first = data[0];
if (first.len >= io_reader.buffer.len) {
return readPositional(r, first);
} else {
io_reader.end += try readPositional(r, io_reader.buffer);
return 0;
}
} }
var iovecs_buffer: [max_buffers_len]posix.iovec = undefined; var iovecs_buffer: [max_buffers_len]posix.iovec = undefined;
const dest_n, const data_size = try io_reader.writableVectorPosix(&iovecs_buffer, data); const dest_n, const data_size = try io_reader.writableVectorPosix(&iovecs_buffer, data);
@ -1352,7 +1361,16 @@ pub const Reader = struct {
if (is_windows) { if (is_windows) {
// Unfortunately, `ReadFileScatter` cannot be used since it // Unfortunately, `ReadFileScatter` cannot be used since it
// requires page alignment. // requires page alignment.
return readStreaming(r, data[0]); assert(io_reader.seek == io_reader.end);
io_reader.seek = 0;
io_reader.end = 0;
const first = data[0];
if (first.len >= io_reader.buffer.len) {
return readStreaming(r, first);
} else {
io_reader.end += try readStreaming(r, io_reader.buffer);
return 0;
}
} }
var iovecs_buffer: [max_buffers_len]posix.iovec = undefined; var iovecs_buffer: [max_buffers_len]posix.iovec = undefined;
const dest_n, const data_size = try io_reader.writableVectorPosix(&iovecs_buffer, data); const dest_n, const data_size = try io_reader.writableVectorPosix(&iovecs_buffer, data);