These are the NodeJS bindings to libpostal, a fast, multilingual NLP library for parsing/normalizing physical addresses. libpostal is written in C because virtually every other language can bind to C.
var postal = require('node-postal'); // Expansion API postal.expand.expand_address('V XX Settembre, 20'); // Parser API postal.parser.parse_address('Barboncino 781 Franklin Ave, Crown Heights, Brooklyn, NY 11238');
Before using the Node bindings, you must install the libpostal C library. Make sure you have the following prerequisites:
sudo apt-get install curl autoconf automake libtool pkg-config
sudo yum install curl autoconf automake libtool pkgconfig
On Mac OSX
brew install curl autoconf automake libtool pkg-config
git clone https://github.com/openvenues/libpostal cd libpostal ./bootstrap.sh ./configure --datadir=[...some dir with a few GB of space...] make sudo make install # On Linux it's probably a good idea to run sudo ldconfig
Then install node-gyp system-wide:
sudo npm install -g node-gyp
Installing node-gyp on Linux (Ubuntu)
First make sure you have:
- gcc, g++ and make (
sudo apt-get install build-essentialon Ubuntu)
- NodeJS with all the development headers (if using a package manager like apt-get, follow the instructions here: https://nodejs.org/en/download/package-manager/)
- python 2.7
Installing the Node bindings
npm install node-postal
node-postal is tested on most major versions of Node (0.10+) in continuous integration. The C++ bindings are written using Nan, which abstracts the API changes across various versions of V8, so the latest versions of Node will likely work.
If you're having trouble compiling, post a Github issue with your OS and Node versions along with the output of npm install / node-gyp rebuild.
To run the unit tests:
node-postal is a thin wrapper around libpostal, which is written in C. Installing node-postal with npm will not install libpostal as a dependency. The install steps described above will install it in standard locations for your OS. For OS X and Linux, that's
/usr/local/lib. node-postal will look in these locations for the C libraries. If you have problems installing node-postal, check there for recently installed libpostal libraries.
Downloading Large Files
make process downloads ~750MB of data, ~2GB unzipped (note: these are production-quality machine learning models trained on over a billion addresses, that's how much disk/memory it takes). Make sure the --datadir argument to
configure, which requires an absolute path, has enough room to accommodate the models. If you get errors after downloading, e.g.:
make: *** [all-local] Error 1 make: *** [all-recursive] Error 1 make: *** [all-recursive] Error 1 make: *** [all] Error 2
there may be problems with subsequent install steps. Occasionally something will go wrong and it will give up downloading. Once downloading is complete, the make script will unpack the langauge_classifier.tar.gz.X files. If you see these files, your download may have stopped. Try running the make script again.
Lucas Hrabovsky (https://github.com/imlucas) for information and examples of repos using Nan, node-gyp, etc.