A companion library to Flint for the direct compilation of Flint data into Apache Jena-based SPARQL queries and updates.

Uses Flint version v0.3.0.

{com.yetanalytics/flint-jena {:mvn/version "0.1.3"
                              :exclusions [org.clojure/clojure]}}

Vanilla Flint API functions (e.g. format-query and format-updates) return SPARQL strings, which are useful for many applications; however, within Apache Jena it is more useful to work directly with Query and UpdateRequest objects. Parsing from Flint to SPARQL strings back to Jena incurs a performance hit compared to directly instantiating the Jena objects, which this library seeks to avoid.

There are two API functions:

• create-query accepts a single Flint query and returns a Query instance.
• create-updates accepts a collection of Flint updates and returns an UpdateRequest instance.

Both API functions have the following keyword arguments:

• spec-ed?: If true, then the entire Clojure spec error data is displayed upon a syntax error. Default false.
• iri->datatype: A function from SPARQL literal datatype IRIs to Jena datatype classes, for coercing typed literals. The map axiom/xsd-datatype-map is the default value.*
• aggregate-fns: A set of function IRI strings (not wrapped, so \"http://fn.com\" is a valid entry) to be treated as custom aggregate expressions, as opposed to run-of-the-mill non-aggregate expressions.

* There is also an axiom/xsd-datatype-map* map; the main difference is that there is a more one-to-one mapping of numeric datatype IRIs to classes, whereas axiom/xsd-datatype-map coerces them to either xsd:integer or xsd:decimal.

The flint-jena repo comes with benchmarking utilities to bench create-query and create-updates against the vanilla Flint API functions, specifically the format-* functions followed by Jena SPARQL parsing. To bench, clone this repo, then execute the following function:

com.yetanalytics.flint-jena-bench/bench

The bench function accepts the following arguments:

• :query-inputs: A vector of file paths from which to read query EDN. Default: ["dev-resources/test-fixtures/query"].
• :update-inputs: A vector of file paths from which to read update EDN. Default: ["dev-resources/test-fixtures/query"].
• :query-output: A file path string where to write the query bench output. Default: "target/bench/query.txt"
• :update-output: A file path string where to write the update bench output. Default: "target/bench/update.txt"

The namespace also contains bench-queries and bench-updates if one only wants to bench create-query and create-updates, respectively. There are also bench, bench-queries, and bench-updates Makefile targets for benching with default args.

An example of use:

clojure -X:bench
  com.yetanalytics.flint-jena-bench/bench-queries 
  :query-inputs '["dev-resources/test-fixtures/query/select/select-1.edn" 
                  "dev-resources/test-fixtures/query/select/select-2.edn"]'

Which then outputs the following to target/bench/query.txt:

************************ Queries creation bench results (in µs) ************************

|        :file | :format-query | :create-query |  :difference | :percent |
|--------------+---------------+---------------+--------------+----------|
| select-1.edn |  61.69 ± 0.00 |  25.38 ± 0.00 | 36.30 ± 0.00 |   58.85% |
| select-2.edn |  91.55 ± 0.00 |  37.61 ± 0.00 | 53.93 ± 0.00 |   58.91% |

Copyright © 2022-2025 Yet Analytics, Inc.

Distributed under the Apache License version 2.0.