Directly provide "const unsigned char **" using CGO

[aren...@gmail.com]

Hello,

I'm in the process of writing a wkhtmltox library wrapper. I got most parts working despite my small C experience.

My problem at the moment is retrieving the output PDF. Since I don't want to let the library write the PDF file into some kind of temp dir, I plan to use the wkhtmltopdf_get_output method[1].
This method writes the PDF data into a const unsigned char **

My question is: Is there a possibility to directly provide a fitting variable or do I need to write a C wrapper function that calls a exported Go method. I would like to minimize data copies.

Jan

[1] https://github.com/wkhtmltopdf/wkhtmltopdf/blob/master/src/lib/pdf.h#L75

[Tamás Gulácsi]

AFAIK a **char means you have to allocate a void pointer and will get the *char's address in there. From there ön you can
var pdf []byte = *((*[1<<31]byte)(unsafe.Pointer(p)))

where p is the returned pointer.

[James Bardin]

On Friday, April 1, 2016 at 8:47:59 AM UTC-4, Tamás Gulácsi wrote:

AFAIK a **char means you have to allocate a void pointer and will get the *char's address in there. From there ön you can
var pdf []byte = *((*[1<<31]byte)(unsafe.Pointer(p)))

This isn't a valid assignment, but you should also slice the array immediately so that you aren't left with out of bounds indexes

    pdf := (*(*[1<<31 - 1]byte)(unsafe.Pointer(p)))[:length:length]

 

[aren...@gmail.com]

Thanks for your help, but this only helps with dereferencing a returned pointer. However, the C function takes a pointer that i need to provide.
Sorry i should have pointed that out more clearly. Here is the function definition:

CAPI(long) wkhtmltopdf_get_output(wkhtmltopdf_converter * converter, const unsigned char **);

[James Bardin]

On Friday, April 1, 2016 at 9:35:24 AM UTC-4, aren...@gmail.com wrote:

Thanks for your help, but this only helps with dereferencing a returned pointer. However, the C function takes a pointer that i need to provide.
Sorry i should have pointed that out more clearly. Here is the function definition:

CAPI(long) wkhtmltopdf_get_output(wkhtmltopdf_converter * converter, const unsigned char **);

Give it a pointer to a nil pointer of the correct type:

var p *C.uchar

ret := wkhtmltopdf_get_output(converter, &p)

[Bryan Matsuo]

In lmdb-go I have been making slices from using the reflect package.

https://github.com/bmatsuo/lmdb-go/blob/40ebaca9ecf9cd77b0b70fc5eb6a89c47a839187/lmdb/val.go#L99-L104

I can't remember off the top of my head where I first saw that trick. But I  remember reading about a different project doing the same thing.

The [1<<31 - 1]byte trick is interesting though. I wonder if it is faster.

[Tamás Gulácsi]

The reflect.SliceHeader solution is deprecated.

[Bryan Matsuo]

On Saturday, April 2, 2016 at 5:25:05 AM UTC-7, Tamás Gulácsi wrote:

The reflect.SliceHeader solution is deprecated.

Can you please give a better explanation than that? Ideally one that has a link. I get tat the docs say it's unsafe, but to me it's not obviously less safe than the other conversion being discussed.

[James Bardin]

I assume because the docs for the SliceHeader say: "[SliceHeader]
cannot be used safely or portably and its representation may change in
a later release.". Whereas the array-to-slice conversion happens
correctly and safely regardless.

https://github.com/golang/go/wiki/cgo#turning-c-arrays-into-go-slices

[Bryan Matsuo]

Fair enough. I didn't really remember that warning from before.

So three sources provide 3 choices of "a very big array". Using 1<<31-1 would be the largest portable choice, right?