excel-d

Build Status Build Status Coverage

Excel API bindings and wrapper API for D

This dub package contains D declarations for the Excel SDK as well as a D wrapper API. This allows programmers to write Excel worksheet functions in D.

Motivation and background for the project can be found here. See also the DConf 2017 lightning talk about excel-d.

Run makedocs.sh to generate the documentation Generated documentation - a work in progress - is available at dpldocs.

A working XLL example can be found in the example directory. Running dub build there will create an XLL (myxll32.xll) that can be loaded in Excel making all of the functions in test/xlld/test_d_funcs.d available to be used in Excel cells. The types are automatically converted between D native types and Excel ones. To build the example: dub build -c example [--arch=x86_mscoff|--arch=x86_64].

For this package to build you will need the Excel SDK xlcall32.lib that can be downloaded from Microsoft. Copying it to the build directory should be sufficient (i.e. when building the example, to the example directory). The library file should be useable as-is, as long as on 32-bit Excel dub build is run with --arch=x86_mscoff to use Microsoft’s binary format. If linking with optlink, the file must be converted first. We recommend using link.exe to not need the conversion. On 64 bit Excel just use --arch=x86_64 - no questions of different library formats.

As part of the build a .def file is generated with all functions to be exported by the XLL.

Excel won’t load the XLL automatically: this must be done manually in File->Tools->Add-Ins. Click on “Go” for “Excel Add-Ins” (the default) and select your XLL there after clicking on “Browse”.

The only difference between building for 32-bit or 64-bit Excel is the arch= option passed to dub. A 32-bit XLL will only work on 32-bit Excel and similarly for 64-bit. You will also need the appropriate 32/64 xlcall32.lib from the Excel SDK to link.

Sample code (see the example directory for more):

    import xlld;

    @Excel(ArgumentText("Array to add"),
           HelpTopic("Adds all cells in an array"),
           FunctionHelp("Adds all cells in an array"),
           ArgumentHelp(["The array to add"]))
    double FuncAddEverything(double[][] args) nothrow @nogc { // nothrow and @nogc are optional
        import std.algorithm: fold;
        import std.math: isNaN;

        double ret = 0;
        foreach(row; args)
            ret += row.fold!((a, b) => b.isNaN ? a : a + b)(0.0);
        return ret;
    }

and then in Excel:

=FuncAddEverything(A1:D20)

Future functionality will include creating menu items and dialogue boxes. Pull requests welcomed.

WARNING: Memory for parameters passed to D functions

Any parameters with indirections (pointers, slices) should NOT be escaped. The memory for those parameters WILL be reused and might cause crashes.

There is support to fail at compile-time if user-written D functions attempt to escape their arguments but unfortunately given the current D defaults requires user intervention. Annotate all D code to be called by Excel with @safe and compile with -dip1000 - all parameters will then need to be scope or the code will not compile.

It is strongly advised to compile with -dip1000 and to make all your functions @safe, or your add-ins could cause Excel to crash.

Function spelling

excel-d will always convert the first character in the D function being wrapped to uppercase since that is the Excel convention.

Variant type Any

Sometimes it is useful for a D function to take in any type that Excel supports. Typically this will happen when receiving a matrix of values where the types might differ (e.g. the columns are date, string, double). To get the expected D type from an Any value, use xlld.wrap.fromXlOper. An example:

double Func(Any[][] values) {
    import xlld.wrap: fromXlOper;
    import std.experimental.allocator: theAllocator;
    foreach(row; values) {
        auto date = row[0].fromXlOper!DateTime(theAllocator);
        auto string_ = row[1].fromXlOper!DateTime(theAllocator);
        auto double_ = row[2].fromXlOper!double(theAllocator);
        // ...
    }
    return ret;
}

Asynchronous functions

A D function can be decorated with the @Async UDA and will be executed asynchronously:

@Async
double AsyncFunc(double d) {
    // long-running task
}

Please see the Microsoft documentation.

Custom enum coversions

If the usual conversions between strings and enums don’t work for the user, it is possible to register custom coversions by calling the functions registerConversionTo and registerConversionFrom.

Structs

D structs can be returned by functions. They are transformed into a string representation.

D structs can also be passed to functions. To do so, pass in a 1D array with the same number of elements as the struct in question.

Optional custom memory allocation and @nogc

If you are not familiar with questions of memory allocation, the below may seem intimidating. However it’s entirely optional and unless performance and latency are critical to you (or possibly if you are interfacing with C or C++ code) then you do not need to worry about the extra complexity introduced by using allocators. The code in the previous section will simply work.

excel-d uses a custom allocator for all allocations that are needed when doing the conversions between D and Excel types. It uses a different one for allocations of XLOPER12s that are returned to Excel, which are then freed in xlAutoFree12 with the same allocator. D functions that are @nogc are wrapped by @nogc Excel functions and similarly for @safe. However, if returning a value that is dynamically allocated from a D function and not using the GC (such as an array of doubles), it is necessary to specify how that memory is to be freed. An example:

// @Dispose is used to tell the framework how to free memory that is dynamically
// allocated by the D function. After returning, the value is converted to an
// Excel type and the D value is freed using the lambda defined here.
@Dispose!((ret) {
    import std.experimental.allocator.mallocator: Mallocator;
    import std.experimental.allocator: dispose;
    Mallocator.instance.dispose(ret);
})
double[] FuncReturnArrayNoGc(double[] numbers) @nogc @safe nothrow {
    import std.experimental.allocator.mallocator: Mallocator;
    import std.experimental.allocator: makeArray;
    import std.algorithm: map;

    try {
        // Allocate memory here in order to return an array of doubles.
        // The memory will be freed after the call by calling the
        // function in `@Dispose` above
        return Mallocator.instance.makeArray(numbers.map!(a => a * 2));
    } catch(Exception _) {
        return [];
    }
}

This allows for @nogc functions to be called from Excel without memory leaks.

Registering code to run when the XLL is unloaded

Since this library automatically writes xlAutoClose it is not possible to use it to run custom code at XLL unloading. As an alternative XLL writers can use xlld.xll.registerAutoCloseFunc passing it a function or a delegate to be executed when xlAutoClose is called.

About Kaleidic Associates

We are a boutique consultancy that advises a small number of hedge fund clients. We are not accepting new clients currently, but if you are interested in working either remotely or locally in London or Hong Kong, and if you are a talented hacker with a moral compass who aspires to excellence then feel free to drop me a line: laeeth at kaleidic.io

We work with our partner Symmetry Investments, and some background on the firm can be found here:

http://symmetryinvestments.com/about-us/