Status: Working POC
Heavily inspired by https://github.com/tinysearch/tinysearch.
- emsdk
- cereal
- nlohmann json
- boost
Tested on gcc-9 + linux.
Currently this runs in two stages.
First compile and execute the main program, passing in a json file and an output directory.
./main --input ../fixtures/index.json --output /home/g/tmp/static_test/
Then navigate to the output directory you specifed a use cmake to compile the program there.
Finally, test your search in <your-output-dir>/dist
.
-
Consider a more efficient data structure than
unordred_set
. -
stop words, better clean up of post text
-
Come up with a scoring/ranking mechanism for results
-
Can this work end to end?
-
Can you make the file sizes smaller here?