Changelog¶

You can use this changelog to perform a safe update from an older version of TriplyETL to a newer one. See the documentation for Upgrading TriplyETL repositories for the advised approach, and how the changelog factors into that.

TriplyETL 4.4.6¶

Release data: 2024-06-24

[Fixed] Minor clean-ups¶

Replacing rdf-js for @rdfjs/types. Adding a resolution for what-wg and rimraf so that there are no deprecation warnings for those modules. Removing some old yarn specific commands, such as pinst.

TriplyETL 4.4.5¶

Release data: 2024-06-12

[Fixed] TriplyDB-JS instantiation¶

Ensure TriplyDB-JS is instantiated properly when proxy settings are passed to TriplyETL

TriplyETL 4.4.0¶

Release data: 2024-06-04

[Added] linked data event stream middleware (LDES)¶

TriplyETL 4.2.0¶

Release data: 2024-05-10

[Added] retry mechanism for OAI¶

Added retry-mechanism to from-oai. By default, all OAI requests now retry 3 times

TriplyETL 4.1.16¶

Release data: 2024-05-09

[Fixed] SHACL version update and fixing now() in SHACL rules engine¶

TriplyETL 4.1.15¶

Release data: 2024-05-08

[Fixed] large default timeout for OAI requests¶

Add (very) large default timeout for oai requests, to avoid possibly hanging when server does not respond

[Added] support for parsing multiple RDF/XML docs from OAI¶

TriplyETL 4.1.14¶

Release data: 2024-04-18

[Fixed] SHACL dependency version¶

Update SHACL dependency, with some performance improvements

TriplyETL 4.1.13¶

Release data: 2024-04-18

[Fixed] hanging ETL and preparations for a simplified eslint / prettier setup¶

TriplyETL 4.1.12¶

Release data: 2024-04-16

[Fixed] SHACL executeRules() and added Speedy as a peer dependency¶

Includes fix regarding SPARQL Functions used in executeRules(), the used Speedy SPARQL engine is now a peer dependency, which will use the same version as the one used as in the @triplyetl/etl package

TriplyETL 4.1.11¶

Release data: 2024-03-27

[Added] multiple hashing algorithms to hashedIri()¶

hashedIri() now supports SHA1, SHA256, SHA384 & SHA512 hashtypes, next to the existing (and still default) MD5 hashtype. See issue #390.

Improvements in the SHACL performance and usage.

TriplyETL 4.1.10¶

Release data: 2024-03-12

[Fixed] performance issue with validate()¶

Because of a previous fix, the validate middleware become slower. This fix makes it performant again.

TriplyETL 4.1.9¶

Release data: 2024-03-12

[Fixed] issue with fromOai() in combination with metadataYieldsRdf¶

Fixed issue where fromOai() middleware reported an error when using metadataYieldsRdf and OAI response contained exactly 1 record.

TriplyETL 4.1.8¶

Release date: 2024-03-10

[Added] preventServiceDowntime option to avoid Service downtime in toTriplyDb() function¶

You can now update services on TriplyDB without experiencing any downtime. Once data uploading is complete, each service will be recreated using a temporary name and the same configuration as the outdated service. Once the temporary service is up and running, the outdated one will be removed.

toTriplyDb({dataset: 'my-dataset',
  opts: {
    synchronizeServices: ['my-elastic-service', 'my-jena-service'],
    preventServiceDowntime: true
  }
})

The execution of the above snippet will result in the following console output:

Warning Service my-elastic-service of type elasticSearch with status running is out of sync.
Info Creating temporary elasticSearch service triplyetl-temp-1710169198327 for my-elastic-service.
Warning Service my-jena-service of type jena with status running is out of sync.
Info Creating temporary jena service triplyetl-temp-1710169198339 for my-jena-service.
Info Swapping service my-jena-service with triplyetl-temp-1710169198339
Info Service my-jena-service updated in 1 minute,
Info Swapping service my-elastic-service with triplyetl-temp-1710169198327
Info Service my-elastic-service updated in 2 minutes, 7 seconds

TriplyETL 4.1.7¶

Release date: 2024-03-09

[Enhanced] Using NamedNodes and/or Literal as content for addHashedIri()¶

The addHashedIri() function now considers whether a NamedNode and/or Literal object is utilized to generate a hash. In such cases, the internal JSON representation is no longer employed. Instead, we utilize the value property for a NamedNode or the combination of the value, language, and datatype value properties for a Literal. This enhancement aims to produce more consistent hashed IRIs over time.

Bug fixes¶

Using skipRest() in ifElse() and switch() middlewares have caused unexpected ETL execution.

TriplyETL 4.1.6¶

Release date: 2024-03-07

[Fixed] SHACL validate() with SPARQL target returned incorrect results¶

Each shape undergoes conversion to a SHACL Validator object only once during ETL to avoid reloading shapes from disk or string for every record. This approach isn't feasible when SPARQL target nodes alter the model, as it would result in adding those targets for each record.

TriplyETL 4.1.2 through 4.1.5¶

Release date: 2024-03-01

[Enhanced] Improved the timeouts handling for fromOai() extractor¶

This enhancement resolves timeout errors that occurred with requests taking an extended period to respond. By utilizing a custom Fetch Agent from undici package, we've eliminated internal timeouts.

TriplyETL 4.1.1¶

Release date: 2024-02-18

[Changed] executeRules() supports only two arguments¶

To set a maximum number of iterations of the execution of the SHACL rules the maxIterations or errorOnMaxIterations needs to be specified in the executeRules() function.

[Enhanced] Increased stack size¶

Maximum call stack size has increased to load the large datasets without trowing an errors.

[Added] CLI flag --keep-tmp-dir to save temporary data directory¶

Introduced the cli flag --keep-tmp-dir in order to store all temporary files disregarding the completion status. This allows the user to debug ETL's by studying the intermediate files the ETL has created.

TriplyETL 4.0.0¶

Release date: 2024-01-29

[Changed] IRI-related middlewares no longer use skolem URLs¶

The following middlewares: addHashedIri(), addIri(), addRandomIri(), would no longer allow users to create URLs that have pathnames start with "/.well-known/genid/", since they would be consideres skolemised URLs.

[Changed] fromShapeFile() is now called fromShapefile()¶

The format is called ESRI Shapefile, hence our extractor function's name had to be changed from fromShapeFile() to fromShapefile().

[Removed] Function addRandomIri()¶

Since function addRandomIri() does not add anything beyond addSkolemIri(), the function has been removed from the TriplyETL library. Random IRIs should be skolem IRIs that can be readily replaced by blank nodes.

[Added] New variables added to ETL¶

New flag has been introduced when constructing an ETL:

  /**
   * Timeout ETL after set duration in milliseconds
   */
  timeout: number;

  /**
   * If set to TRUE, the ETL will do a hard exit, preventing uploads to TDB on timeouts
   */
  exitOnTimeout: boolean;

which can be set as following:

 const etl = new Etl({timeout: 1000, exitOnTimeout: true})

This will cause a hard exit when a timeout occurs and nothing will be executed after this timeout.

TriplyETL 3.1.0 && 3.1.1¶

Release date: 2024-01-15 && 2024-01-17

[Deprecated] Deprecated fromShapeFile() for fromShapefile()¶

[Deprecated] Deprecated addRandomIri() function.¶

Function addRandomIri() does not add anything beyond addSkolemIri(). Random IRIs should be skolem IRIs that can be readily replaced by blank nodes.

[Enhanced] Improved SHACL report.¶

When a SHACL shape is used to validate data does by itself not conform to the SHACL-SHACL shape, the report of that non-conforming shape is now printed.

[Enhanced] Improved objects() function¶

The objects() middleware now requires a minimum of 2 objects, deviating from its previous behavior, which was limited to functionality similar to the triple() function.

[Enhanced] RML middleware¶

RML map() middleware now allows a string Source and a string primitive as input.

[Enhanced] Static vocabularies¶

With the latest update, TriplyETL vocabularies are now represented as Vocabulary objects, replacing the previous usage of objects with the type IRI. This change may necessitate adjustments to existing ETLs that utilize static vocabularies, such as aat. In this case, the vocabulary would need to be updated to aat.toIri() to ensure compatibility with the correct type.

[Enhanced] NPM packages¶

All NPM packages are up to date with their latest version.

[Fixed] Base IRI when using loadRdf()¶

There were some inconsistency between the expected base IRI. For example, the following snippet:

import { logQuads } from '@triplyetl/etl/debug'
import { Etl, loadRdf, Source } from '@triplyetl/etl/generic'

export default async function (): Promise<Etl> {
  const etl = new Etl()
  etl.use(
    loadRdf(Source.string('<s><p><o>.')),
    logQuads(),
  )
  return etl
}

would result in:

<https://triplydb.com/graph/default> {
<https://triplydb.com/graph/s> <https://triplydb.com/graph/p> <https://triplydb.com/graph/o>
}

rather than:

<https://triplydb.com/graph/default> {
<https://triplydb.com/s> <https://triplydb.com/p> <https://triplydb.com/o>
}

This issue has been fixed.

[Fixed] String encoding for IRIs¶

It is now possible to check whether a value of a key used to create an IRI contains valid characters. A previous warning incorrectly flagged a space (' ') as an invalid character in the IRI, but that has been taken care of that. Now, when you run the script, you won't encounter the misleading warning, providing a more accurate and hassle-free execution.

In this case, [1] is resulting in [2] instead of invalid [3]:

[1] a b
[2] http://ex.com/a%20b
[3] http://ex.com/ a b

As well as [4] being encoded as [5]:

[4] a&b
[5] a&b

Or [6] can be legitimately encoded in CSV using [7]:

[6] a,b
[7] "a,b"

[Fixed] New datatype added to addPoint() middleware¶

Datatype wktLiteral has been added to the addPoint() middleware.

TriplyETL 3.0.20¶

Release date: 2024-01-04

[Enhanced] Improved copySource() function¶

Function etl.copySource() accepts the same destination format as toTriplyDB(), so that the same destination does not need to be specified twice.

[Enhanced] Prefix uploading¶

Prefixes are no longer uploaded by default, only explicit prefixes that are defined when constructing an ETL with new Etl({ prefixes }).

TriplyETL 3.0.15 through 3.0.18¶

Release date: 2023-12-07 through 2023-12-28

[Enhanced] RDF compression before upload¶

It is now possible to enable compression of RDF data before being uploaded to TriplyDB. See the toRdf() function for more information.

[Enhanced] Skolem IRI prefix use¶

TriplyETL now emits an error when a Skolem IRI prefix is used with addHashedIri().

Bug fixes¶

This release provides bug fixes to XSLT support (see XSLT Transformations and XSLT Assertions).

TriplyETL 3.0.14¶

Release date: 2023-12-04

[Added] Support for RML¶

This release introduces support for the RML transformation and assertion language. RML is an ETL configuration language that has gained traction in the linked data community over the last couple of years. See the following pages for more information:

• RML Transformations
• RML Assertions

[Build] Environments beyond the DTAP defaults¶

It is now possible to extend the standard environments offered by TriplyETL.

Bug fixes¶

This release fixes a URL/request-related error in the fromOai extractor.

TriplyETL 3.0.7 through 3.0.9¶

Release date: 2023-11-29

[Added] CLI flag to skip version check¶

Introduced the cli flag --skip-version-check because some users can not use remote connections because of security policies.

[Added] Support for JPath expressions¶

toJson() middleware now uses path selectors just as fromXml(), but also JPath expressions.

[Added] Authentication for the OAI-PMH extractor¶

fromOai() now accepts a Request object as the value for the url option, allowing more fine graded use of the HTTP request (including authentication information).

[Added] XSLT support for the OAI-PMH extractor¶

fromOai() now accepts an optional parameter stylesheet. The argument must be an object with 1 required key source pointing to the source of the stylesheet and 1 optional argument yieldsRdf. When provided the XSL Stylesheet source is processed using the OAI response. The result of the transformation must still be a valid OAI response.

Bug fixes¶

Fixes issue where keys that are used for internal administration were shown in logged records, after using the ifElse() and switch() control structures.

TriplyETL 3.0.6¶

Release date: 2023-11-14

Bug fixes¶

This release fixes issues with upstream packages that contained breaking changes.

TriplyETL 3.0.5¶

Release date: 2023-11-09

Bug fixes¶

The progress bar would sometimes terminate at 99% or 101%, instead of the expected 100%.

TriplyETL 3.0.4¶

Release date: 2023-11-07

[Added] Dataset metadata specification¶

It is now possible to use metadata when creating datasets. If a new dataset is created then the metadata is always used, for existing datasets you can choose to ignore, merge or replace existing metadata. This feature is to prevent ETLs accidentally overwriting metadata might have changed in the UI/Dashboard of their TriplyDB instance.

The following options are available for this new feature in the opts.existingMetadata parameter:

• ignore: no metadata will be changed even if no metadata is present for this dataset (this is the default value)
• merge: only properties that have no value will be overwritten bij the provided metadata
• replace: all existing metadata will be replaced, even if the provided metadata contains empty keys

[CLI] Reverse logic for creating error traces¶

Before this release, running an ETL would always create an error trace file. It was possible to disable this behavior with CLI flag --skip-error-trace. Starting in this release, the error trace file is no longer created by default, and a newly added CLI flag --create-error-trace must now be specified in order ot create the error trace file.

Bug fixes¶

The following bugs were fixed:

• The number of synchronized services was not always reported correctly in CLI output.
• A package we depend on introduced a breaking change causing a function not to be there anymore.

TriplyETL 3.0.3¶

Release date: 2023-11-01

[Changed] Support for the NDE Dataset Register¶

The code to submit datasets to the NDE Dataset Register has been moved to TriplyDB-JS. The way to publish a dataset now is to add an option to the toTriplyDb() function: { submitToNDEDatasetRegister: true }.

Example:

toTriplyDb({dataset: 'nde', opts: {submitToNDEDatasetRegister: true}})

TriplyETL 3.0.2¶

Release date: 2023-10-23

[Added] Static statement assertion¶

export default async function(): Promise<Etl> {
  const etl = new Etl({baseIri: Iri('https://example.com/')})
  await etl.staticAssertions(
    pairs(
      iri(etl.standardGraphs.default),
      [a, dcat.Dataset],
      [skos.prefLabel, literal(str("Family Doe"), lang.en)],
      [dct.created, literal(str(new Date().toISOString()), xsd.dateTime)],
    ),
  );
  await etl.staticAssertions(
    pairs(iri(etl.standardGraphs.default), [skos.prefLabel, literal(str("Familie Doe"), lang.nl)]),
  );
  etl.use(
    fromJson([{ name: "John Doe" }, { name: "Jane Doe" }]),
    triple(iri(etl.standardPrefixes.id, "$recordId"), sdo.name, "name"),
    logQuads(),
  );
  return etl
}

You can now assert so called "static triples": triples that are not related to the source extractors but should only be asserted once per ETL.

Bug fixes¶

There was an error in the ifElse() control structure, that caused ETLs to not use the fallback 'else' block in some situations.

TriplyETL 3.0.1¶

Release date: 2023-10-19

[Enhanced] Source string validation¶

The addLiteral() function can now validate string data that occurs in the Record. Such validation can be used in addition to validation in the Internal Store (graph comparison and SHACL validation). Look at the documentation of addLiteral() for more information.

[Enhanced] Synchronize specific services¶

When publishing linked data to TriplyDB, it is now possible to synchronize one specific service. This is specifically useful in case an Acceptance and a Production service are used, and only the former should be synchronized. See the documentation for publishing to remote data destinations for more information.

[Fixed] Bug fixes¶

The following bugs have been fixed:

• The progress bar would sometimes go over 100%.
• the error report file (etl.err) would sometimes contain sentinel keys like $sentinel-${MD5-HASH}. These sentinel keys are used for internal bookkeeping in TriplyETL, and are no longer part of the Record.
• Some XSLT transformations failed on Windows, because of incorrect redirecting of error messages.

TriplyETL 3.0.0¶

Release date: 2023-10-12

[Added] Support for XSLT¶

Support was added for the Extensible Stylesheet Language Transformations (XSLT) configuration language. This can be used in the following two functions:

1. The fromXml() extractor function, for XML sources that transformed to regular XML.
2. The loadRdf() function, for XML sources that are transformed into RDF/XML.

In both cases, this functionality is used by configuring the stylesheet parameter with an XSLT Stylesheet (e.g. using Source.file()).

Example code that uses fromXml():

fromXml(Source.file(XMLFile), {
        selectors: "rdf:RDF.sdo:Person",
        stylesheet: Source.file(XSLTStylesheet),
      }),

Example code that uses loadRdf():

loadRdf(Source.file(XMLFile), {
        contentType: "application/rdf+xml",
        stylesheet: Source.file(XSLTStylesheet),
      }),

[Added] Support for the SPARQL Select and SPARQL Ask queries¶

The extractors fromCsv(), fromJson(), fromTsv() and fromXml() now support SPARQL Select queries.

The extractors fromJson() and fromXml() also support SPARQL Ask queries.

The example below hows how to use a SPARQL ask query in the fromJson() extractor:

fromJson(Source.TriplyDb.query('account', 'query-name', { triplyDb: { url: 'https://api.triplydb.com' } }))

[Enhanced] Simplified usage of 'nestedPairs()'¶

The nestedPairs() middleware can be used without providing the subject node that connects the pairs to the object/predicate. This will automatically create a Skolem IRI for the subject:

nestedPairs(S, P, [a, sdo.Person])

For example:

fromJson([{ id: '1', height: 15 }]),
addSkolemIri({
  prefix: prefix.skolem,
  key: '_height',
}),
nestedPairs(iri(prefix.product, 'id'), sdo.height,
  [qudt.unit, unit.CentiM],
  [rdf.value, 'height'],
),

Will result in the following linked data assertions:

product:1
  sdo:height
    [ qudt:unit unit:CentiM;
      rdf:value 15 ].

[Changed] Automatic prefix handling in TriplyDB using 'toRdf()'¶

Manually specified and standard prefixes are automatically added to TriplyDB when toRdf() is used. The middleware uploadPrefixes() is removed.

[Changed] New approach to prefix handling in TriplyETL¶

Prefixes are no longer defined as function that concatenates a value to an Iri. The Iri is a new type of Object in TriplyETL, that has a concat() method which allows you to add a value to the first part of an Iri. For example:

const baseIri = Iri('https://example.com/')
const prefixId = baseIri.concat('id/')
const johnDoe = prefixId.concat('john-doe')

[Changed] New package '@triplyetl/vocabularies'¶

The vocabularies and languages are no longer part of @triplyetl/etl package. A new module has been released: @triplyetl/vocabularies:

Individual vocabularies like rdf and schema.org can be imported in the following way:

import { a, rdf, sdo } from '@triplyetl/vocabularies'

To import all vocabularies, use:

import * as vocab from "@triplyetl/vocabularies"

Some vocabularies are too large to include, but they can still be used like this:

import { aat } from '@triplyetl/vocabularies'
const moustache = aat.concat('300379271')

or

import { aat } from '@triplyetl/vocabularies'
addIri({prefix: aat, content: str('300379271'), key: 'moustache'})

To use the RATT lang tools:

Import languages:

import { languages } from '@triplyetl/vocabularies'

Import languages and region:

import { region, language } from '@triplyetl/vocabularies'

const nl_BE = language.nl.addRegion(region.BE)

[Changed] RDF serialization parsing with 'loadRdf()'¶

The loadRdf() function is able to parse known RDF serializations (Turtle, TriG, N-Triples, N-Quads) provided as a string without specifying mimetype.

const data = Source.string('...')
loadRdf(data)

[Changed] Extended log and terminal output for ETL debugging¶

The output of the logfile and terminal output is changed. It contains more information to help users debugging ETL's. The format of time representation is now H:i:s.u where:

• H: 24-hour format of an hour with leading zeros (00 through 23)
• i: Minutes with leading zeros (00 to 59)
• s: Seconds with leading zeros (00 through 59)
• u: Microseconds (example: 654321)

[Changed] 'toRdf()' for account-based token access¶

The toRdf() middleware now accepts "me" as account name based on the token. Below are some examples of this being used.

toTriplyDb({account: "me", dataset: "myDataset"})

loadRdf(Source.TriplyDb.rdf("me", datasetName))

Destination.TriplyDb.rdf("me", datasetName)

[Changed] Relocation middleware: 'resetStore()' and 'randomKey()'¶

The resetStore() middleware is now moved from ratt to the generic namespace . The randomKey() middleware moved from generic to ratt.

[Changed] Record selection with '--offset' and '--limit'¶

You can now use --offset and --limit instead of --from-record-id and --head, e.g. LIMIT=1 OFFSET=8 npx etl. The old arguments can still be used for backwards compatibility.

[Changed] Removal of 'mapQuads()'¶

The mapQuads() function was removed.

[Changed] Warning for old Node.JS versions¶

If the users Node.JS version is older that the recommended version (currently >=18.0.0) a warning is shown.

[Changed] SHACL Validation Engine¶

A SHACL Validation Engine improved performance.

[Changed] Trace for large records¶

A new flag now bypasses generating the trace for very large records: ---skip-error-trace. Thus, no trace file is created.

[Changed] Transition to in-memory engine Speedy¶

Comunica is no longer part of TriplyETL, the in-memory engine is now Triply's Speedy.

[Enhanced] Improvements to ETL logs¶

The logging format was improved by including the following information:

• the TriplyETL version
• the Node.js version
• the DTAP mode
• the start date and time
• the end date and time

[Enhanced] Prevent using multiple extractors¶

TriplyETL only supports one extractor per ETL configuration object. In the past, it was possible to use multiple extractors, which would result in faulty behavior during ETL runs. Starting in this release, TriplyETL will emit an error when multiple extractors are used.

[Enhanced] Better error reporting for CSV, TSV, and XML sources.¶

In previous releases, the extractor functions fromCsv(), fromTsv(), and fromXml() would not emit the file name in case an error occurred. This was specifically problematic when a large number of data source files were used. Starting in this release, the file name is included in error message.

[Enhanced] Default CRS for 'wkt.addPoint()'¶

In previous releases, the Coordinate Reference System (CRS) was a required attribute for transformation function wkt.addPoint(). Starting in this release, the CRS argument has become optional. When not specified, the default CRS http://www.opengis.net/def/crs/OGC/1.3/CRS84 is used.

[Enhanced] Handle conflicting TriplyDB instance specifications¶

In previous releases, it was possible to introduce an ambiguity in specify the TriplyDB instance to publish data to. This was possible by (1) specifying a TriplyDB API Token in the environment (e.g. though an .env file), and (2) by configuring the triplyDb option in the loadRdf() function. Starting in this release, TriplyETL will emit an error if the TriplyDB instance in the API Token differs from the TriplyDB instance configured in the triplyDb option.

[Enhanced] More information for failing HTTP calls¶

In previous releases, when a failing HTTP call resulted in an error message, only the body of that HTTP call would be included in the error message. Starting in this release, the HTTP status code of the failing HTTP call is included in the error message as well.

Bug fixes¶

This release fixes several out-of-memory bugs in the SHACL validation function.

TriplyETL 2.0.7 through 2.0.19¶

Release dates: 2023-06-17 through 2023-09-29

Bug fixes¶

The following bugs were fixed:

• Processing an Excel sheet with fromXml() would sometimes consume too much memory.
• Several installation issues on Windows have been resolved.
• The async-saxophone library for XML processing was adjusted to support the current LTS version of Node.js (v18).

TriplyETL 2.0.6¶

Release date: 2023-06-07

[Added] Support for the PREMIS vocabulary¶

Support was added for the PREMIS 3.0.0 vocabulary. This vocabulary is published by the Library of Congress and can be used to publish metadata about the preservation of digital objects. See the PREMIS documentation for more information.

The vocabulary can be imported from the 'vocab' module:

import { premis } from '@triplyetl/vocabularies'

See the documentation on external vocabulary declarations for more information.

[Added] New debug function logMemory()¶

A new debug function logMemory() is added. This function prints an overview of the current memory usage of TriplyETL. This allows users to detect fluctuations in memory consumption inside their pipelines.

[Added] Support for the 'ListIdentifiers' verb in the OAI-PMH extractor¶

The fromOai() extractor already supported the ListRecords verb. This release adds support for the ListIdentifiers verb as well. This new verb allows users to stream through the headers of all records in an OAI-PMH collection, without requiring the full record (i.e. body) to be retrieved.

TriplyETL 2.0.5¶

Release date: 2023-05-25

[Changed] New default engine for SPARQL Construct¶

The default engine for evaluating SPARQL Construct queries (function construct()) has changed from Comunica to Speedy. Speedy is a new SPARQL engine that is developed by Triply. Comunica is an open source engine that is developed by the open source community. Since SPARQL is a standardized query language, this change should not cause a difference in behavior for your ETL pipelines.

In the unexpected case where an ETL pipeline is negatively affected by this change, the old situation can be restored by explicitly configuring the Comunica engine:

import { construct } from '@triplyetl/etl/sparql'

construct(Source.TriplyDb.query('my-query'), { sparqlEngine: 'comunica' }),

The benefit of switching to the Speedy engine is that this engine is expected to be faster for most queries. Overall, this change will therefore result in speed improvements for your TriplyETL pipelines.

[Added] New CLI tool for comparing graphs¶

The new CLI tool compare allows graph comparison to be performed from the command-line. This uses the same algorithm that is used by the compareGraphs() validation function.

Bug fixes¶

This release fixes the following bugs:

• fromXlsx() did not remove trailing whitespace in cell values.
• When a SHACL result was printed, an incorrect message about a faulty SHACL model was shown.
• Some RDF processors did not handle empty RDF inputs correctly.

TriplyETL 2.0.4¶

Release date: 2023-05-11

[Enhanced] Better output for graph comparison¶

Before this release, when two graphs were not isomorph and their difference consisted of a mapping from blank nodes onto blank nodes exclusively, an empty difference message was communicated.

From this release onwards, the difference message is non-empty, and specifically indicates the difference between the non-isomorphic graphs in terms of the mismatching blank nodes.

Look at this example from the graph comparison documentation, which emits such a difference message.

TriplyETL 2.0.3¶

Release date: 2023-05-10

Bug fixes¶

This release includes the following bug fixes:

• Error location information is not shown in TriplyETL Runner.

• Issue when a URL data source (Source.url()) includes an HTTP body.

TriplyETL 2.0.2¶

Release date: 2023-05-09

Bug fixes¶

This release fixes bugs related to the recent switch from CommonJS to ESM:

• Dynamic import bug on Windows.

• Error reporting issues due to ESM imports.

TriplyETL 2.0.1¶

Release date: 2023-05-03

[Added] Timeout flag for TriplyETL Runner¶

The TriplyETL Runner is the CLI tool that is used to run ETL pipelines. Starting with this version, you can specify a --timeout flag when using the TriplyETL Runner.

When the indicated timeout is reached before the pipeline finishes, the TriplyETL Runner will gracefully terminate the ETL by acting as if there are no more incoming records.

See the TriplyETL Runner documentation page for more information.

TriplyETL 2.0.0¶

Release date: 2023-05-01

[Changed] Modules infrastructure moves from CommonJS to ESM¶

Before this release, TriplyETL used CommonJS modules to modularize its functionality into different components. Starting in this release, ECMAScript Modules (ESM) are used to modularize TriplyETL functionality into different modules.

ESM is a more modern approach for modularizing ECMAScript (JavaScript, TypeScript, and Node.js) code. While CommonJS imports are evaluated at runtime, ESM imports are evaluated at compile time. TriplyETL users benefit from this change, since error messages related to module imports will be detected much earlier in the development process.

All documentation examples were update to use ESM syntax for module imports, for example:

import { logRecord } from '@triplyetl/etl/debug'

[Changed] Debug functions move to a new module¶

Before this release, debug functions like logRecord() and startTrace() were part of the RATT module. Since debug functions can be used in combination with any ETL configuration approach, they were moved to a new module.

The debug functions are imported from their new module in the following way:

import { logRecord, traceEnd, traceStart } from '@triplyetl/etl/debug'

[Enhanced] Better error messages when things go wrong¶

This release introduces a new approach for communicating errors back to the user. When TriplyETL functionality detects an error condition, a unified 'trace middleware' is now used to retrieve information from the environment in which the error occurred. This information is then printed to the error output stream for communication with the user.

Bug fixes¶

The following bug fixes are included in this release:

• Incorrect behavior of the _switch() control function.

• The fromOai() extractor now communicates clearer when the accessed OAI-PMH endpoint encounters any issues.

• When a key with a NULL value was accessed, the name of that key is now included in the error message. This makes it easier for users to find the NULL value in their source data.

TriplyETL 1.0.x¶

TriplyETL 1.0.0 was released on 2023-03-20.