Equivalent function to RcppAnnoy's a$getItemsVector(i)

[d4tum]

Thanks for the great package!

There doesn't appear to be an equivalent function to a get the vector given an item's index number similar to RcppAnnoy's a$getItemsVector(i) where i is the item integer mapped to the vector within the index during build time. It appears the Python bindings has this feature but it's not been exposed in the R package -

From the readme https://github.com/nmslib/hnswlib :

get_items(ids) - returns a numpy array (shape:N*dim) of vectors that have integer identifiers specified in ids numpy vector (shape:N). Note that for cosine similarity it currently returns normalized vectors.

Would you please expose this function in the R package?

[jlmelville]

I'll see what I can do @d4tum.

Note to self: the Python binding C++ code is at:
https://github.com/nmslib/hnswlib/blob/359b2ba87358224963986f709e593d799064ace6/python_bindings/bindings.cpp#L304

[jlmelville]

@d4tum the master branch now contains an implementation of the getItems method. If you want to get back a matrix containing the first and tenth vectors that were added to the index, call e.g. ann$getItems(c(1, 10)).

I have not yet parallelized this function so it might be slow if returning large number of vectors.

Also, the hnsw library has recently had some stability issues with valgrind failures, other memory problems or compiler warnings, all of which prevent me from submitting a new version of the library to CRAN. I was hoping there would be a clean release of the library, because I really don't want to maintain a separate patched version of it internally to this package. I may take that step (or see what scope there is for submitting some more fixes upstream), but for now there is not likely to be an imminent CRAN submission.

[d4tum]

Fair enough and thank you for your efforts @jlmelville

[jlmelville]

This feature is part of the CRAN 0.5.0 release. Sorry it took so long.