MeTA Search App

This project is a very simple search engine that shows how you can use MeTA in the backend of a web-based search application.

Initial setup

Dependencies

• Coffeescript is nice to use, and the project uses this to generate the Javascript files. If you don't like Coffeescript, you can slightly tweak the project to not use it.
• NodeJS is used for the proxy server.

MeTA API Server

cd cpp/
git submodule update --init --recursive

Next, we can compile the API server.

Mac OS X

On Mac OS X, you will need to explicitly provide the ICU root directory:

mkdir build/
cd build/
CXX=clang++ cmake ../ -DCMAKE_BUILD_TYPE=Release -DICU_ROOT=/usr/local/opt/icu4c
make -j4

Linux

mkdir build/
cd build/
CXX=clang++ cmake ../ -DCMAKE_BUILD_TYPE=Release
make -j4

The API server should now be built. You can run it from inside the build/ directory with the command

./api-server ../config.toml

This assumes you have config.toml set up to point to a dataset. For more information, see the Analyzers, Tokenizers, and Filters tutorial.

After you make modifications to the source files, you should only have to run make and restart the server.

Web Frontend

cd web/
make

The make command compiles the Coffeescript into Javascript and places it in the javascript/ directory. You're now ready to start the server:

python2 -m SimpleHTTPServer

Whenever you modify the Coffeescript files, you'll need to run make and reload the page.

Node Proxy

The node proxy is needed for local development. It directs traffic to either the C++ server or the frontend site itself. The following command sets up the necessary node modules.

cd node/
npm install

Now, you can start the proxy server:

node proxy.js

View the page at http://localhost:9001, and enter a query! The first time you run this, the index needs to be created. Subsequent times, it will simply be loaded from disk just like in MeTA.