It's not done yet! But it can do some things:
Download https://www.dropbox.com/s/lv44vyl8ia46llx/sample.tar.bz2?dl=0
and unzip it next to this directory.
It creates a directory named sample (that should be a sibling directory of this repo)
with about 8,476 files in it.
Run ./build_index.py to try to build the index.
This creates a few big files in the parent directory. With our sample from step 1, they take up half a GB of disk space. (It would be pretty easy to cut that in half; haven't bothered yet.)
Run ./read_index.py to read raw entries out of the index.
This is something less than a real search engine. You can only query one term at a time, and the output is bare-bones. But it works! With our sample, it takes a few seconds to start up; after that, queries are fast.