analytical

Analytical example

Small end-to-end demo: tiny edge CSVs → dense IDs → serialized columnar tables → one query_eval_exec . Run query_eval_exec with no arguments to print query form examples.

What the data is

Input files are relationship CSVs. setup_data.py normalizes them to dense IDs and serializes them.

The query chains seven such relations (country → city → person → forum → post → comment → tag → tag class) and counts matching tuples.

Prerequisites

Build the C++ tools from the repo root (see the main README) so these exist:

• build/table_serializer (or DEBUG/ / RELEASE/ if you use those build dirs)
• build/query_eval_exec

run_query.py searches build/, then DEBUG/, then RELEASE/ for query_eval_exec.

Pipeline

From examples/analytical/ (or prefix paths with examples/analytical/ if you run from the repo root):

Step	Script	What it does
1	setup_data.py	Normalizes data to dense IDs from 0, and serializes it using config.yaml.
2	run_query.py	Runs query_eval_exec <serialized_dir> "<query from query.txt>" "<ordering from ordering.txt>".

cd examples/analytical
python3 setup_data.py
python3 run_query.py

query.txt and ordering.txt

• query.txt — A single line in rule form: Q(COUNT(*)) := <atoms>. Each atom is a serialized table name with variables matching its projection, e.g. City_isPartOf_Country(city,country). Commas separate atoms; variable names that repeat across atoms are join keys.

• ordering.txt — The join order.

Edit those two files to experiment.

Other files

• config.yaml — Defines table names, source paths, and column/projection configurations for the serializer.