Track transferred bytes/progress of longer PUT request with std.http

Help
,
21

You got the wrong userdata. Basically you are passing pointer to your own Writer interface, to the underlying writer which expects pointer to its own Writer interface. Should be

        const n = try self.inner.vtable.drain(self.inner, data, splat);

Otherwise looking good. Took me a while to spot the issue.

2 Likes
22

Hey, thank you very much. It totally makes sense, but is hard to get while reading through the source code.

It now works with this code:

pub const CountingWriter = struct {
    inner: *std.Io.Writer,
    pos: u64 = 0,
    interface: std.Io.Writer,

    pub fn init(inner: *std.Io.Writer, buffer: []u8) CountingWriter {
        return .{
            .inner = inner,
            .interface = .{
                .vtable = &.{
                    .drain = drainFn,
                },
                .buffer = buffer,
                .end = 0,
            },
        };
    }

    fn drainFn(w: *std.Io.Writer, data: []const []const u8, splat: usize) std.Io.Writer.Error!usize {
        const self: *CountingWriter = @fieldParentPtr("interface", w);
        const n = try self.inner.vtable.drain(self.inner, data, splat);
        self.pos += n;
        std.debug.print("{d}\n", .{self.logicalPos()});
        return n;
    }

    pub fn logicalPos(self: *const CountingWriter) u64 {
        return self.pos;
    }
};

Thanks @psznm , @xeubie and @lalinsky for all the good examples and advices. I’ll mark this as solution, because its working code; so anyone who is looking to solve a similar task finds it. But of course, I would not have worked it out (this fast or maybe ever?) without help of the named users!!

1 Like
23

I took the inspirations and tipps from this thread and implemented wrapping Reader and Writer for the http requests of the S3 lib z3. I hope this is might help somebody who has a similar task to solve. And of course thanks again to anybody who answered in this thread!

1 Like