The sparql.anything from sparql-anything

[html] Invalid local name

An exception occurs with details:

Exception in thread "main" org.apache.jena.shared.InvalidPropertyURIException: http://www.w3.org/1999/xhtml#http:
	at org.apache.jena.rdf.model.impl.PropertyImpl.checkLocalName(PropertyImpl.java:66)
	at org.apache.jena.rdf.model.impl.PropertyImpl.<init>(PropertyImpl.java:55)
	at org.apache.jena.rdf.model.ResourceFactory$Impl.createProperty(ResourceFactory.java:296)
	at org.apache.jena.rdf.model.ResourceFactory.createProperty(ResourceFactory.java:144)

Support embedding content to triplify as IRI argument

SELECT ?o {

SERVICE <facade-x:content=Text to triplify with txt triplifier, media-type=application/text> {?s rdf:_1 ?o}

}

This query should return "Text to triplify with txt triplifier".

XML triplifier to use default fx: namespace

Instead of using the file location.

Support CSV first row as header

If my CSV has a first row as header, as in the example:

email,id,first name,last name
[email protected],2070,Laura,Grey
[email protected],4081,Craig,Johnson
[email protected],9346,Mary,Jenkins
[email protected],5079,Jamie,Smith

could I supply an option for the headers to provide additional property bindings (escaping spaces and all) instead of rdf:_x?

Based on the code at CSVTriplifier.java, I tried to pass -Dcsv.headers=true, but it doesn't seem to be picked up by the way the engine passes Java properties to the triplifier.

Currently, triplifiers transform all the data before executing the query. We could use the query expression and extract triple patterns to limit the triples added to the model. The resulting graph could be a subset of the full graph, including all the triples useful to evaluate the query. This approach is easy to implement and may improve performance in some (but not all) cases.

Exception when reading remote CSV

It seems that CSV Triplifier has a problem when reading remote resources and throws an exception here.
The issue can be reproduced with the query:

PREFIX xyz: <http://sparql.xyz/facade-x/data/>
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>

SELECT *
WHERE {

    SERVICE <x-sparql-anything:csv.headers=true,location=https://raw.githubusercontent.com/pcm-dpc/COVID-19/master/dati-andamento-nazionale/dpc-covid19-ita-andamento-nazionale-20200409.csv> {
        ?s ?p ?o
    }

}

Add CLI parameter 'strategy'

We are developing alternative approaches to query execution, these should be also available as options in the CLI.

Guess output format from output file extension

At the moment, the CLI wants you to specify the output format as a separate parameter:

fx -q titles.rq -o titles.xml -f xml

While doing the following generates a JSON file (named .XML) - because the format falls back to defalut/JSON:

fx -q titles.rq  -o titles.xml

Support Excel and Google spreadsheets

As the title says

RDF files as resources

It could be handy to allow users to SPARQL static RDF files using the SERVICE operator. I think SPARQL.anything should include this feature although it is not related to Facade-X so probably we don't want to use the same protocol handler?

CLI: support multiple output formats

And to pick the right serialisation format, accordingly to the following serialisations: JSON, XML, CSV, TEXT, TTL, NT

In the future, we may reuse the io.github.basilapi:rendering library which is based on mime types as format identifiers and supports a large variety of formats.

Use input parameters to customise the output filename

When iterating over input parameters, when multiple files are generated, it would be useful to customise the output filename.
This is useful for creating a collection of RDF files each one identified nicely from the parameters.

For example, having an input including tuples such as

<?xml version="1.0"?>
<sparql xmlns="http://www.w3.org/2005/sparql-results#">
  <head>
    <variable name="artistUrl"/>
    <variable name="artistNickname"/>
  </head>
  <results>
    <result>
      <binding name="artistUrl">
        <literal>https://imma.ie/artists/william-leech/</literal>
      </binding>
      <binding name="artistNickname">
        <literal>leech-william</literal>
      </binding>
    </result>
    <result>
      <binding name="artistUrl">
        <literal>https://imma.ie/artists/marie-foley/</literal>
      </binding>
      <binding name="artistNickname">
        <literal>foley-marie</literal>
      </binding>
    </result>

being able to instruct the tool to use artistNickname as (part of the) output filename.

[html] incorrect placement of text nodes

Currently, text nodes are misplaced, the content being appended to the last child item.

Support traversing a local directory

It could be useful to just traverse a local directory and explore its content with sparql.anything

Support file formats based on mime-types

We can use io.github.basilapi:rendering library which is based on mime types as format identifiers and supports a large variety of formats.

CLI option --output not working

The cli ignores the parameter and prints to STDOUT

JSON:API / Drupal

Any idea on how to integrate data exposed using a JSON:API based API ?

That would help to integrate data contained in Drupal-systems. See https://www.drupal.org/docs/core-modules-and-themes/core-modules/jsonapi-module

java -jar issue: constructing graphs without -f parameter leads to an error

small issue: the help says TTL is the default but without -f TTL I get this exception:

Exception in thread "main" org.apache.jena.riot.RiotException: No graph writer for 'Lang:RDF/XML'

automatic detection mime-type

https://docs.oracle.com/javase/7/docs/api/java/nio/file/Files.html#probeContentType%28java.nio.file.Path%29 ??

Include simple logger in cli artefact

We can use slf4j:slf4j-simple

Handling namespaces for properties and entities

At the moment, the namespace option parameter is used as default prefix fro both schema elements and named entities. This may create problems, for example, when joining two CSVs. Users should declare different namespaces to avoid clashes. An easier way may be to use the root entity (file name) as default prefix.

Relative location seems not working properly for paths like "file.ext"

Facade-X options as basic graph patterns

It could be useful to support facade-x configuration from within the SERVICE query with magic properties.

facade-x IRI not compliant with IETF scheme

Jena complains as follows:

Bad IRI: <facade-x:media-type=text/html,html.selector=#az-group,location=https://imma.ie/artists/> Code: 45/UNREGISTERED_NONIETF_SCHEME_TREE in SCHEME: The scheme name has a "-" in it, but it does not start in "x-" and the prefix is not known as the prefix of an alternative tree for URI schemes.

We may change the URI scheme, discussion open in alternatives.

Refactor artefacts and package names

to com.github.sparqlanything.*

[html] Include dom components innerText and dom innerHTML

It would be very useful in addition to the HTML source data, to also have values from those two elements. However, they should not be represented with the same namespace we use for the content (xhtml:). One option is to use the location of the DOM specification: https://dom.spec.whatwg.org/#.

Performance: slicing

A simple way of improving usability with large files is to support a general option to 'slice' content in some ways. Some ideas:

by indicating the number of containers to be produced, or the maximum number of values
by indicating the top N slots, or the bottom N
by indicating a limit in the triples produced.

We may think of others.

New parameter: input

This feature will allow chaining the output of a previous query to a new one.

When the input is a sparql result set, it will be used as a source of query parameters, following the BASIL convention.

Support blank space in option values

CLI: default format depending on query type or output file extension

ASK / SELECT: application/sparql-results+json
CONSTRUCT: text/turtle

Consistent parameter names in URIs

I suggest using location, namespace (applied to domain specific properties/types), media-type for general-purpose parameters and specify type-specific ones using, for example, csv.format=DEFAULT or json.root ...

Collaborate with Ontop

Please collaborate with Ontop.

See

This was allready suggested in #6

New parameter: load

This is useful to chain the output of one query into another. RDF files can be loaded and queried along with the output of the service clause(s).

JSON iterator breaks with Unicode escaped characters

Unicode escaped characters should be expected in JSON serialisations. However, JsonIterator does not seem to support them and horribly breaks. Could not find a configuration parameter or anything else related to this issue so far.

Improve description of URI schema, removing BNF-like syntax

As the summary says, currently the documentation gives a false impression that the IRI schema shown is in a formal BNF, which is not.

Support file archives (zip, tar, ...)

It would be useful to be able to traverse the content of file archives.

Support triplifications of HTTP Headers when resolving HTTP URLs

SHACL definition of Facade-X

It would be a useful tool for validating the output of sparql.anything transformers.

Support bibliographic formats e.g. bibtex

#127
Document

Default facade-x ns uri

Currently we are using urn:facade-x:ns# but it would be better to use a public one, e.g. http://sparql.xyz/facade-x/ns/

Add default type for root element

It would be useful to be able to distinguish root elements from normal containers.

What about adding, for example rdf:type <urn:facade-x:default#root>?

Implement blank-node=false for CSV, HTML, and XML

This should be straight-forward for all the currently supported formats.

Audit graph

We can add an option audit=1 to include a graph with information about the generated, queried graphs, using the SPARQL Service Description Vocabulary and VoID. This would be an optional meta-graph, useful for debugging and troubleshooting.

We can use a new boolean option 'audit=1' (defaults off)

Cleanup logging framework

Multiple SLF4J bindings, references to log4j12 could be avoided.

Avoid executing the same transformation multiple times when facade-x IRI built from variable bindings

facade-x SERVICE IRIs can be dynamically generated from variable bindings. However, the same variables may be evaluated multiple times with the same value, resulting in repeated calls to the SERVICE clause with the exact same parameters!

The executor may remember service calls already performed and avoid repeating the same operation multiple times.

HTML profiles: DOM, Microdata, RDFa, ...

HTML can be a data source in many different ways. Currently, we support a CSS-selector and generate a Facade-X representation of the related DOM portions (each solution in a separate graph). We should consider also alternative approaches, for example, referring to well-known approaches to embed data, such as Microdata or RDFa.

sparql-anything / sparql.anything Goto Github PK

sparql.anything's People

Contributors

Stargazers

Watchers

Forkers

sparql.anything's Issues

Recommend Projects

Recommend Topics

Recommend Org