atescomp / rdf-transform Goto Github PK

When in record mode, any literal declared at the 2nd tier or later (beyond the root level) will not process into the preview or export.

I found this extension very useful. We have a very large set of XML files that share same structure, so one mapping file can apply to them all. However, it would be tedious to manually load them one by one to OpenRefine. I'd like to ask whether there is a batch mode that allows us to run the transformation in command line. Thanks.

Currently, all properties are "constant" IRIs. Modify the UI Property code to allow for column expressions.

Hi,

I was downloading latest version of rdf-transform-2.2.1 but got "Not Found" error message on screen.

Checked the download url:
https://github.com/AtesComp/rdf-transform/releases/download/v2.2.0/rdf-transform-2.2.1.zip

Version is not correctly mapped.

The correct url is :
https://github.com/AtesComp/rdf-transform/releases/download/v2.2.1/rdf-transform-2.2.1.zip

Currently, when exporting an RDF file, the exporter uses OpenRefine's record indicator to start processing Root Node subject maps as records starting on non-empty cells of the first column and all subsequent empty cells in the column indicating the rest of the record--a following non-empty cell beings a new record. Once a record is identified, as it processes other columns in the transform, the record processing changes to row processing. This is done to process data as independent lines in the record. Otherwise, the record elements would be considered one long row (or rather squashed row) and complimentary subjects of a column acquire properties and objects belonging to other otherwise separate subjects.

However, there may be sub-record structures in a given record designated by the same first column logic but applied to other columns and the initial record processing would be desirable on these columns. The transform should reset the row processing on these sub-records to record mode via a setting on an object mapping to act as a subject record processor by detecting the object column's non-blank and blank cells--forming sub-record ranges mirroring OpenRefine's reported record range--and processing them accordingly.

This is predicated on the initial record or row setting for the data--it should process sub-records only when the data has been set to record mode in OpenRefine.

I installed the RDF Transform according to your instructions on my Mac where I already had a working OpenRefine (v. 3.5.2) but it doesn't show at the OpenRefine GUI. I stopped the OpenRefine service, installed the RDF Transform extension and started OpenRefine again with no success.
I tried tried to install on both the ~/Library/Application Support/OpenRefine/extensions/rdf-transform and on the OpenRefine package (Contents/Resources/webapp/extensions/rdf-transform), also with no success. Is there anything else I should do to make it work?

Hi @AtesComp,

Again, thank you so much for working on this! This new RDF extension looks really great.
Before you published it, we were in talks with the current maintainer of the older RDF extension to transfer it to OpenRefine's GitHub organization, as a means to give it more visibility (this has not progressed because the maintainer has not responded to this invitation yet).

If that is interest to you, we could do this for your extension instead. We would of course give you all admin rights on the repository.

I still have in mind your request for more documentation about the migration process for OpenRefine 4.0 - I hope to find the time to work on this soon.

If there is anything else we can do to support your work, let us know :)

To recreate:

• Installed RDF Transform 2.2.0 with OpenRefine 3.6.2 (Linux build)
• Ran OpenRefine, accessed the UI using Safari, and tried to open a project
• Project doesn't display, and in developer console errors can be seen

Error is:

[Error] SyntaxError: Unexpected token '{'
	(anonymous function) (project-bundle.js:54566)

which references this part of the code

    // Setup default Master Root Node (copy as needed)...
    static gnodeMasterRoot = {};
    static {
        this.gnodeMasterRoot.valueSource = {};
        this.gnodeMasterRoot.valueSource.source = null; // ...to be replaced with row / record index
        this.gnodeMasterRoot.expression = {};
        this.gnodeMasterRoot.expression.language = RDFTransform.gstrDefaultExpLang;
        this.gnodeMasterRoot.expression.code = null; // ...to be replaced with default language expression
        this.gnodeMasterRoot.propertyMappings = [];
    }

specifically the error occurs at

If I instead access the UI via Chrome, the project loads and no errors are reported in the console

• Windows 10 Pro v 21H2 build 19044.1645
• OpenRefine 3.5.2
• Java runtime 1.8.0_331
• RDF Transform 2.0.3-alpha

FYI: This issue isn't really that urgent (to me). After running into it, I managed to install & run OR with RDF Transform under WSL Ubuntu, and this issue does not appear there. However, see issue #5.

Steps to reproduce

Start OpenRefine, select any project, click RDF Transform -> Edit RDF Transform -> Message appears ERROR: Could not retrieve default namespaces

Click OK. RDF Transform pane appears. Click Add, in Prefix box type ex, in IRI box type http://www.example.com/, click OK -> Message appears An IRI is required! The given IRI is invalid: http://www.example.com/

config info

RDF Transform is in C:\Users<myname>\AppData\Roaming\OpenRefine\extensions\rdf-transform

I normally run OR from here: C:\Users<myname>\Downloads\openrefine-3.5.2\openrefine.exe
I also tried to run a copy here: C:\Program Files\OpenRefine\openrefine-3.5.2\openrefine.exe
and here: C:\Users<myname>\AppData\Roaming\OpenRefine\openrefine-3.5.2\openrefine.exe

Same problem always. I think it's looking for these:

"C:\Users<myname>\AppData\Roaming\OpenRefine\extensions\rdf-transform\module\MOD-INF\classes\files\PredefinedVocabs"
"C:\Users<myname>\AppData\Roaming\OpenRefine\extensions\rdf-transform\module\MOD-INF\classes\files\Namespaces"

Any tweaks I could try would be welcome.

using RDF transform v2.0.3-alpha with OR 3.5.2 and Java 17 in WSL Ubuntu 20.04

I've added abbrev ex for http://example.com/ as namespace and this is also my base IRI. It does not retrieve any real namespace, it doesn't need to.

The problem is that the abbrev does not show up in the preview nor the turtle. It looks like this:

@prefix : <http://example.com/> .
@prefix skos: <http://www.w3.org/2004/02/skos/core#> .

:mammals a skos:Concept;
  skos:narrower :felines, :bovines .

It should look like this:

@prefix ex: <http://example.com/> .
@prefix skos: <http://www.w3.org/2004/02/skos/core#> .

ex:mammals a skos:Concept;
  skos:narrower ex:felines, ex:bovines .

Should you have any hints as to how I can fix this myself by tweaking something (I don't know Java), please feel free to share.

In addition, I have to say I'm really happy that RDF Transform came along. Repeated occurrences (e.g. the two narrowers here) are a real pain in the neck to deal with in the "old" RDF-extension, see here.

Provide an enhancement to examine the project data size and system memory to determine an optimal buffer size and allocations.

I'm using OpenRefine v.3.5.2 and its extension "rdf-extension v.1.3.1". I'm trying to import ontologies like dbpedia and schema. When I add prefix and insert the URI, it gives me no problem. However, when i try to access its properties it says: 'Not in the imported vocabulary definition.' I tried ti upload the ontology.owl as a local file and the same issue still exist. Could you help with that?

When the user selects any of the RDF Node types (index, column, constant) from another types, change the expression to the default expression of that type unless it's changing back to the original type. If changing back to the original, restore the original expression.

Also, consider column expressions may be compatible between columns, so preserve a column expression if just changing columns.

For index types, a change between record and row mode already does this:

• Changes between "row.record.index" and "row.index".

Support HTML files containing the new license text as a comment, "<!-- -->", adds an additional HTML element that confuses the bind processing after the loadHTML() call. The JQuery object no longer points to the dialog as element 0 (the only element).

Provide functionality to edit the sample record / row count for preview. Provide for first row or record to start from and the number of following. Or, design it like page printing (1-4, 7-9, 11). Something like that.

Looking at LICENSE.txt it appears to be reporting on the licenses of dependencies only, but not your code under the src tree. Since there are no license headers, it is unclear whether your code is also licensed under Apache 2.0 or is unlicensed.

If you intend your software to be permissively licensed, do you mind clarifying LICENSE.txt with your desired license (preferably Apache 2.0) and/or adding license headers to your source files?

Much appreciated!

The Prefixed Namespace Adder fails to process file imports resulting in a hung / locked condition.

The Namespace Adder has three modes:

1. Add a simple Prefix and Namespace
2. Add a Prefix and Namespace along with its ontology class and property hints from the web
3. Add a Prefix and Namespace along with its ontology class and property hints from a file

Currently, the RDF Preview tab uses Jena's Stream processing to present the data. Modify the preview to use a whole graph pretty printer approach with a Jena Model. The code uses a Stream as its modeled on the Export process. However, the Export will also need to export using a pretty printer as well to support other output types that don't stream well.

The pretty printer approach presents an optimized output as it can examine the entire graph for the most condensed format. Streams only get smaller chunks of the entire graph for output, so cannot optimize.