← Semantechs home The tarka Manual EPUBA4 PDFGitHub
Contents
  1. Introduction

Part I — Setting out

  1. 1What tarka does
  2. 2Installing
  3. 3The Bookshop Trail

Part II — TARQL

  1. 4A first mapping
  2. 5TARQL in depth

Part III — OTTR

  1. 6Templates
  2. 7OTTR over tables
  3. 8bOTTR

Part IV — Data frames

  1. 9Python and Polars

Part V — Checking the transformation

  1. 10Shapes and SHACL
  2. 11SPARQL checks
  3. 12The Ontology Quality Suite

Part VI — In production

  1. 13Into a store
  2. 14A pipeline

Appendices

  1. ACommand reference
  2. BPython reference
  3. CDifferences
  4. DTroubleshooting
  5. EThe example files

Appendix C. Differences

tarka gives the RDF of three tools. Where they disagree with each other, or tarka with one of them, it is deliberate, and here.

From TARQL (TARQL 1.3, on Jena 4.8)

TARQL tarka
?ROWNUM counts from 1 counts from 0, as oxi-gen does
A backslash in a quoted CSV field taken literally escapes the next character, as in oxi-gen; change it with -p
File encoding the platform's (on Windows, not UTF-8) UTF-8, everywhere
Typed values as Jena writes them: "2.0e0"^^xsd:double canonical: "2"^^xsd:double (the same value)
afn:sprintf formats as Java's String.format refused, with a message
afn:bnode a blank node's label not available (tarka's labels are made afresh)
Property functions that read the data (list:member, rdfs:member, apf:bag …) match nothing in a row refused, with a message
A property function inside OPTIONAL, UNION, a nested group evaluated refused, with a message

Jena's other property functions and functions give what TARQL 1.3 gives, which tests/fixtures/jena records and the tests check.

From oxi-gen

oxi-gen tarka
BOUND(?column) always false: cells are substituted into the query, so FILTER(BOUND(?column)) aborts it true when the cell has a value, as in TARQL
A function the engine does not have a crash a message, before any row is read
A malformed row a crash a message naming the row; the file is not written
-o FILE when a run fails a partial file the earlier file, kept
Blank node labels random per run stable: the same input gives the same output
Jena's apf: and afn: functions none as TARQL has them

From Lutra

tarka's OTTR, bOTTR and expand output is Lutra's on every fixture in the repository, and the linter reports what Lutra's does. The differences are in what is reported, not what is made:

Lutra tarka
A literal of a datatype OTTR does not have, read with ottr:datatypeSep written, with a warning that the datatype is unusual written
ottr:type naming a type OTTR does not have an error per row; the values written as plain strings a warning, once; the values written as plain strings
A relative path in an H2 query from the working directory from the bOTTR file's directory (@@THIS_DIR@@ works in both)
SQL beyond SELECT … FROM CSVREAD(…) run by H2 refused (tarka does not embed H2)
JDBC sources read not read
tarka run -l (OTTR over a table, by column name) no such feature tarka's own

From the RDF it reads and writes

tarka works with RDF 1.1 and SPARQL 1.1. Its engine (HOLOS's) parses RDF 1.2 and SPARQL 1.2, so tarka refuses them explicitly: a triple term, a SPARQL 1.2 function, a literal with a base direction, in a query, a template or an endpoint's answer. A source that writes a base direction, such as the Bookshop Trail's three right-to-left titles ("…"@ar--rtl), can be mapped with the language and not the direction.