SOLVE-IT CASE/UCO TTL Examiner. A single HTML file that renders a CASE/UCO Turtle export and lets you inspect and traverse the graph without losing your place in the source text.
A CASE/UCO export is readable as text but not navigable as a graph. Object identifiers are
opaque, so following an edge means copying an identifier such as
kb:bookmark-finding-a6114c5f-7ba3-5238-9315-794ff5a2a5a2 and searching the file for it,
and the target is rarely nearby because these files are grouped by kind rather than by
derivation. Facets add a second level of indirection: an object's properties do not live on
the object. Reading the SHA-256 of a source image means traversing a File, its
uco-core:hasFacet edge, a ContentDataFacet, its uco-observable:hash edge, and finally
a Hash node, across four separate blocks of the file.
SCUTTLE addresses both. Selecting an identifier opens a resolved view of that object with the properties of every facet it owns folded in and marked, together with the places the object is referenced from. Following an edge from inside that view replaces its contents and leaves the source pane where it is, so an arbitrarily long chain costs no position. The source pane moves only when you press the jump control.
The source text itself is never rewritten. Identifiers appear exactly as written and facets stay where the file puts them, because the primary use is checking that a generator produced correct CASE, and a pane that silently improved what it showed could not serve that use.
Class and property definitions come from an index bundled into the page, covering CASE
1.5.0, UCO 1.5.0 and the SOLVE-IT ontology.
Those hosts send no Access-Control-Allow-Origin header, so a page cannot fetch them at
runtime. SOLVE-IT technique records are fetched live from
data.solveit-df.org instead, because that host does permit
it and its knowledge base is regenerated daily.
The latest build is hosted at https://scuttle.hargs.co.uk/. Your file is
read in the browser and is not uploaded anywhere. To run it offline, download
scuttle.html from that page, or build it yourself as below.
npm run build:index # regenerate the bundled definitions (needs network; cached afterwards)
npm run build # produce dist/scuttle.html
npm test # run the test suitedist/scuttle.html is self-contained. Open it directly from the filesystem, with no server:
open dist/scuttle.htmlChoose a file with the picker, or drag one onto the page. The document is read through a
File object and never travels over the network, which is what allows the page to work
over file://.
The report opens with the file and lists what could not be resolved or verified. It is dismissible, and the control in the header reopens it and carries the count.
It reports identifiers that are referenced but never declared, classes and properties absent from the bundled index, techniques the SOLVE-IT knowledge base did not account for, and actions whose result disagrees with the classes their technique declares it produces. That last check is worth explaining: these exports type an action with a SOLVE-IT technique and state in a comment that the output is among the classes the technique produces. SCUTTLE compares the two rather than taking the comment's word for it.
The report also states which CASE/UCO version the bundled index was built from, and that
the file is assumed to target the same version. That assumption is necessary because these
exports carry their version only in a comment banner: there is no owl:Ontology node and
no owl:versionIRI anywhere in the graph. A reported disagreement can therefore mean
version skew inside the viewer rather than a defect in the export, and the report says so
where it matters.
- Blank nodes are not independently selectable. They are reached through the object that owns them. A facet written as a blank-node property list has its properties folded into its owner as normal, but you cannot click the blank node itself.
- Technique details need network access. Without it the technique stub is shown as the file declares it, with a link to SOLVE-IT, and the report records that the check was unavailable. Class and property definitions are unaffected because they are bundled.
- The bundled index is regenerated manually. Nothing detects that a newer CASE or UCO
release exists, so
npm run build:indexhas to be re-run deliberately. - The assumed version is not read from the data. See the load report section above.
Application modules under src/ are authored as ES modules so they can be imported by the
test suite, and tools/bundle.mjs concatenates them into one classic script at build time.
Two constraints follow from that transform and are enforced by tests in
test/bundle.test.mjs and test/smoke.test.mjs: no two modules may declare the same
top-level name, and imports may not be aliased.
The Turtle parser is N3.js, vendored as its UMD build in
vendor/ and loaded by a classic script, so the page carries no module semantics. Source
positions come from the N3 lexer, which reports line and column, rather than from its
parser, which does not put positions on quads. The two passes are joined by IRI string.
MIT. See LICENSE. N3.js is also MIT-licensed; its licence is in
vendor/n3-LICENSE.md.