Feat/rdf: Add optional module for ROCrate <-> rdf interop by St4NNi · Pull Request #3 · arunaengine/ro-crate-rs

St4NNi · 2025-12-23T12:24:11Z

This feature adds a RDF to ROCrate interoperability module that allows to export ROCrates as RDF graph and import ROCrates from RDF graphs. All conversion is funneled through the ROCrate struct.

Add RDF serialization/deserialization support via new rdf feature flag
Implement bidirectional conversion: ROCrate struct ↔ RDF triples
Support multiple RDF formats for export and import: Turtle, N-Triples, RDF/XML (powered by oxrdf/oxrdfio)
Add JSON-LD context resolution with embedded RO-Crate 1.1/1.2 schema support
Compaction: Contexts are used to compact expanded terms back to valid ROCrate
Testing: Add Python (rdflib) compatibility testing to verify semantic equivalence between Rust and Python RDF representations

Future implications: The full ROCrate expansion to RDF triples can be used to improve semantic validation of the ROCrate.

Completes: https://github.com/arunaengine/project-orga/issues/26

This requires the changes from: #2 and should be rebased when this is merged.

lfbrehm

I do not understand all of it, especially RDF semantic-related stuff, but did my best to provide a somewhat deep review. I also skipped tests and examples for the most part, because I trust that you ran the tests before.

lfbrehm · 2026-01-09T12:49:41Z

src/ro_crate/rdf/context.rs

+        let base_rebuilt = base_parts.join("/");
+        if base_rebuilt.is_empty() {
+            return rel_remaining.to_string();
+        }


Does this mean, that if relative part is for example ../../../ but absolute only states /root/second-level that it returns ../ which should not be a valid solution? Shouldnt this result in an error?

You are correct, haven't thought about it. Will be fixed in the next commit by throwing an error !

lfbrehm · 2026-01-09T13:02:26Z

src/ro_crate/rdf/context.rs

+        || term.starts_with("../")
+        || term.starts_with('#')
+        || term.starts_with('/')
+        || term.contains('/')


Why is this a relative IRI instead of an absolute IRI? If it contains a / this can also be absolute, because you the (absolute) base is split in resolve_relative_iri.

This occurs, after absolute iris have already been filtered, out, so the functionality is correct, but I will rename the function and add a check for absolute iris to prevent misuse !

lfbrehm · 2026-01-09T13:47:06Z

src/ro_crate/rdf/context.rs

+                if best.is_none() || len > best.unwrap().2 {
+                    best = Some((prefix, namespace, len));
+                }


I see that this cannot fail, but wouldnt it be better to just rewrite this with match guards to avoid unwraps?

match best { Some((p, n, l)) if len > l => ... }

Yes, even better is map_or:

if best.map_or(true, |(_, _, best_len)| len > best_len) {...

lfbrehm · 2026-01-09T14:08:04Z

src/ro_crate/rdf/rdf_io.rs

+            if subject_str.contains("ro-crate-metadata.json") {
+                metadata_iri = Some(subject_str.to_string());
+                break;
+            }


If you have multiple subcrates that are possibly defined with ./prefix/ro-crate-metadata.json, this would also match these subcrates, but you only want to find the root crate in this function.

The ro-crate spec also states that the root metadata descriptor must have @id ro-crate-metadata.json or ro-crate-metadata.jsonld for legacy crates.

This behavior is directly from the spec:

https://www.researchobject.org/ro-crate/specification/1.2/appendix/relative-uris.html#finding-ro-crate-root-in-rdf-triple-stores

but yes it can match multiple results (and some results that are not the root crate). We should maybe check if the entity matched is a subcrate entity if yes skip it. Btw. contains("ro-crate-metadata.json") also matches: "ro-crate-metadata.jsonld"

lfbrehm · 2026-01-09T14:11:40Z

src/ro_crate/rdf/convert.rs

+    /// Resolve relative IRIs against the provided base IRI.
+    /// This takes precedence over @base defined in the context.
+    WithBase(String),


Dont you want also the opposite where @base takes precendence and only if this cannot not be resolved you resolve to the specified one?

You are correct, will be fixed.

lfbrehm · 2026-01-13T10:38:11Z

src/ro_crate/rdf/rdf_io.rs

+            properties.entry(property_name).or_default().push(value);
+        }
+    }
+
+    // Convert Vec<EntityValue> to EntityValue (single or array)
+    let mut result = HashMap::new();
+    for (key, values) in properties {
+        if values.len() == 1 {
+            result.insert(key, values.into_iter().next().unwrap());
+        } else if values.len() > 1 {
+            result.insert(key, EntityValue::EntityVec(values));
+        }
+    }


Same as above: Why not do this when using the entry api of the HashMap to skip looping again over your results?

The entry API will not really work here ?! I mean I want to iterate all elements from the list. But we could make it more neat using an iterator / filter_map statement.

lfbrehm · 2026-01-13T10:39:44Z

src/ro_crate/rdf/rdf_io.rs

+    let compacted_id = if metadata_iri.contains("ro-crate-metadata.json") {
+        "ro-crate-metadata.json".to_string()
+    } else {
+        context.compact_iri(metadata_iri)
+    };


This will probably also match subcrates right?

Not a problem, this only occurs if the correct metadata descriptor has already been found, so this will not target subcrates !

lfbrehm · 2026-01-13T10:44:05Z

src/ro_crate/rdf/rdf_io.rs

+/// Internal implementation: Convert RDF triples to RoCrate using a ResolvedContext.
+fn rdf_to_rocrate_with_context(
+    triples: Vec<Triple>,
+    context: ResolvedContext,
+) -> Result<RoCrate, RdfError> {


Why not merge this and the rdf_graph_to_ro_crate fns together, if you use the wrapper fn only anyway?

There are two wrappers: rdf_graph_to_rocrate and rdf_to_rocrate both use this function (with different context)

lfbrehm · 2026-01-13T11:52:43Z

src/ro_crate/rdf/rdf_io.rs

+    let base_iri = base
+        .map(|b| b.to_string())
+        .or_else(|| infer_base_from_metadata(&metadata_iri))
+        .unwrap_or_else(|| "http://example.org/".to_string());


Is http://example.org really meant as a default fallback value or just a placeholder that never got replaced?

Ups, yes this is leftover from a testing session -> fixed

lfbrehm · 2026-01-13T12:09:16Z

src/ro_crate/rdf/resolver.rs

+        Self {
+            cache: HashMap::new(),
+            allow_remote: false,
+            client: reqwest::blocking::Client::builder()
+                .timeout(std::time::Duration::from_secs(30))
+                .build()
+                .expect("Failed to create HTTP client"),
+        }


Why initialize a client when allow_remote is false?

Fair point, will make it optional and initialize it only if allow_remote is set to true. Thinking about this allow_remote should omit the boolean so allow_remote should just set it to true !

microchange: Fix typo on "Research" !

…erplate

…xample

St4NNi · 2026-02-23T13:12:49Z

will merge in separate PR into upstream!

St4NNi requested review from das-Abroxas and lfbrehm December 23, 2025 12:28

St4NNi force-pushed the feat/rdf branch 2 times, most recently from 88b7876 to fecf2bf Compare January 8, 2026 09:31

lfbrehm reviewed Jan 13, 2026

View reviewed changes

St4NNi pushed a commit that referenced this pull request Jan 14, 2026

Merge pull request #3 from lubianat/patch-1

586ad9d

microchange: Fix typo on "Research" !

St4NNi force-pushed the feat/rdf branch from fdbc137 to 877cbee Compare January 14, 2026 14:12

St4NNi requested a review from lfbrehm January 15, 2026 13:47

St4NNi added 22 commits February 23, 2026 13:28

feat: Added boiler plate for rdf support, added context resolver boil…

f3749d9

…erplate

feat: Draft for context expansion

18ae037

feat: Added resolver for remote context resolving

7909a99

feat: Added draft for graph

c102021

feat: Initial draft for rocrate -> rdf conversion

7531c62

fix: Broken tests

9b08b65

feat: Update context handling to better expand @base and other iris

5373d47

feat: Updated conversion logic to use resolver to properly expand terms

b988dba

chore: Add oxrdfio and update the error with parsing errors

e1acc3e

feat: Add first draft for rdf_io module

883a58f

fix: Compile errors in benches and examples

d9373cf

fix: Fix context, make conversion options enum

0e5ef91

feat: Update rdf_io to properly serialize / deserialize from rdf

c8462f3

fix: fixture None is not a valid json keyword, remove warnings from e…

0195289

…xample

feat: Added rdf examples

f184f2c

feat: Added python (rdflib) compatibility testing

db1e71f

chore: Cargo fmt

f08ffa2

fix: Fixed tests wired allow_relative to oxrdf unchecked

8b8f1a1

fix: Fix triple serializing for empty values

2a0470f

fix: Fixed entity datatypes conversion

f1e1727

fix: Fragment compaction with / base

c32fb8a

feat: Added blanknode support

72b05f9

St4NNi added 8 commits February 23, 2026 13:31

fix: Path traversal above root

ce911fc

fix: Rename is_relative_iri to prevent misuse

f5269d0

chore: Fix root data resolving

4475de6

fix: @base takes precedence over WithBase

3e45e24

fix: Improve vec to hashmap conversions

1c3d935

fix: Wrong fallback to examples.org

62b9aba

fix: Lazy client initialization if allow_remote() is set

24a5fc5

chore: Upgrade dependencies

46c0b66

St4NNi force-pushed the feat/rdf branch from aa37715 to 46c0b66 Compare February 23, 2026 13:06

St4NNi closed this Feb 23, 2026

Conversation

St4NNi commented Dec 23, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

lfbrehm left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

St4NNi commented Feb 23, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

St4NNi commented Dec 23, 2025 •

edited

Loading