Rhyme

Appearance

Rhyme

An Expressive Data-Centric Query Language

Query nested data, produce nested data as result

Overview
Getting Started
Examples
Rhyme

Tree-to-Tree Queries

Process nested structures (JSON, Tensors) as input, produce nested structures as result.

Easy to Metaprogram

Compose query fragments in JS, following the structure of the input or output.

Compile to JS (or C)

Queries are optimized and translated to low-level code for maximum performance.

From Query to Code ​

A Rhyme query is compiled through an intermediate representation (IR) that makes data dependencies explicit, optimized, and finally translated to efficient JavaScript (or C).

1 · Source Query
js
// share of total population, per country
rh`{
  data.*A.country:
      sum(data.*A.population)
    / sum(data.*B.population)
}`
2 · Intermediate Representation
IR graph: t2 accumulates the total population, t1 the per-country sum, t0 divides them
3 · Generated Code
js
inp => {
  let tmp = {}
  tmp.t1 ??= {}
  tmp.t2 ??= 0
  for (let xB in inp.data) {
    tmp.t2 += inp.data[xB].population
  }
  tmp.t0 ??= {}
  for (let xA in inp.data) {
    let k = inp.data[xA].country
    tmp.t1[k] ??= 0
    tmp.t1[k] += inp.data[xA].population
    tmp.t0[k] = tmp.t1[k] / tmp.t2
  }
  return tmp.t0
}