feat(sparql-anything): add SparqlAnythingConverter for chunked non-RDF to RDF conversion

AGENTS.md

@@ -99,7 +99,7 @@ The `@nx/js:library` generator’s output diverges from the conventions in this
    - `name` → `@lde/<new-name>`
    - `description` — write something useful
    - `repository.directory` → `packages/<new-name>`
-   - `version` → `0.0.0` from the get-go (do NOT keep the sibling’s version). nx release bumps from there via conventional commits, so the introducing `feat:` commit lands the first release at `0.1.0`; any higher starting version overshoots it. This must be in place before the PR merges — see [Releasing a new package](#releasing-a-new-package).
+   - `version` → `0.1.0` from the get-go (do NOT keep the sibling’s version). nx release bumps from there via conventional commits, so the introducing `feat:` commit lands the first release at `0.1.0`; any higher starting version overshoots it. This must be in place before the PR merges — see [Releasing a new package](#releasing-a-new-package).
    - `dependencies` and `peerDependencies` — replace with what the new package actually needs
 4. **Replace the source.** Empty out `src/` and `test/`, write the new code.
 5. **Update `tsconfig.lib.json` `references`** to match the new package’s actual `@lde/*` peers.
@@ -129,8 +129,6 @@ For releasing the new package’s first version, see [Releasing a new package](#
 
 `.github/workflows/release.yml` publishes existing packages on every push to main, but the CI workflow alone cannot bring up a brand-new `@lde/<name>` package: npm’s Trusted Publisher configuration can only be added to a package that already exists on the registry. The first version has to be published manually by a maintainer; CI takes over from the second version onwards.
 
-The package’s `version` must already be `0.0.0` before the PR merges (set in [Creating New Packages](#creating-new-packages) step 3) so the introducing `feat:` commit publishes `0.1.0`.
-
 One-time bootstrap for a new package (do this once it has been merged to main):
 
 1. **Publish the first version manually.** From a maintainer’s machine:

README.md

@@ -127,6 +127,11 @@ await pipeline.run();
   <td><a href="https://www.npmjs.com/package/@lde/sparql-importer"><img src="https://img.shields.io/npm/v/@lde/sparql-importer" alt="npm"></a></td>
   <td>Import data dumps to a local SPARQL endpoint for querying</td>
 </tr>
+<tr>
+  <td><a href="packages/sparql-anything">@lde/sparql-anything</a></td>
+  <td><a href="https://www.npmjs.com/package/@lde/sparql-anything"><img src="https://img.shields.io/npm/v/@lde/sparql-anything" alt="npm"></a></td>
+  <td>Convert tabular and other non-RDF sources to RDF with the SPARQL Anything CLI</td>
+</tr>
 <tr><th colspan="3" align="left">Publication – Serve and document your data</th></tr>
 <tr>
   <td><a href="packages/fastify-rdf">@lde/fastify-rdf</a></td>
@@ -222,6 +227,7 @@ graph TD
     distribution-probe --> dataset
     pipeline --> distribution-probe
     sparql-importer --> dataset
+    sparql-anything --> task-runner
     distribution-health --> distribution-probe
     distribution-health --> sparql-importer
   end

packages/sparql-anything/README.md

@@ -0,0 +1,45 @@
+# SPARQL Anything
+
+Convert tabular and other non-RDF sources to RDF with the [SPARQL Anything](https://sparql-anything.cc) CLI.
+
+The `SparqlAnythingConverter` runs the SPARQL Anything jar **once per input chunk** to bound memory use, then concatenates the per-chunk N-Triples outputs into a single file. Processes are spawned through an [`@lde/task-runner`](../task-runner), so the same converter works on the host, in Docker, or anywhere else a `TaskRunner` is implemented.
+
+## Usage
+
+```typescript
+import { SparqlAnythingConverter } from '@lde/sparql-anything';
+import { NativeTaskRunner } from '@lde/task-runner-native';
+
+const converter = new SparqlAnythingConverter({
+  queryFile: 'config/places.rq', // CONSTRUCT query; `{SOURCE}` is replaced per chunk
+  jarPath: 'bin/sparql-anything.jar',
+  adminCodesFile: 'data/admin-codes.ttl', // loaded into the default graph via --load
+  taskRunner: new NativeTaskRunner(),
+});
+
+await converter.convert(
+  ['data/geonames_aa.csv', 'data/geonames_ab.csv'],
+  'output/geonames.nt',
+);
+```
+
+For each chunk, the converter:
+
+1. Replaces the literal `{SOURCE}` in the query file with the chunk's path and writes the result to a temporary `.rq` file.
+2. Runs `java -jar <jar> -q <query> --load <adminCodesFile> --format NT --output <chunk>.nt`.
+3. Waits for the process; a non-zero exit **aborts the whole conversion** so a crashed chunk can never be silently dropped from the output.
+
+Finally, the per-chunk `.nt` files are concatenated, in the order the chunks were given, into the output path. The concatenation streams, so multi-gigabyte outputs do not have to fit in memory. N-Triples has no prefixes or document structure, so concatenating per-chunk files always yields a single valid document.
+
+## Options
+
+| Option           | Type               | Description                                                                      |
+| ---------------- | ------------------ | -------------------------------------------------------------------------------- |
+| `queryFile`      | `string`           | Path to the SPARQL CONSTRUCT query. The literal `{SOURCE}` is replaced per chunk |
+| `jarPath`        | `string`           | Path to the SPARQL Anything CLI jar                                              |
+| `adminCodesFile` | `string`           | Path to the Turtle file loaded into the default graph (`--load`)                 |
+| `taskRunner`     | `TaskRunner<Task>` | Runs the SPARQL Anything process for each chunk                                  |
+
+## Chunking
+
+The converter consumes pre-split chunk files; it does not split sources itself. Producing the chunks (and the `adminCodesFile`) is the caller's responsibility.

packages/sparql-anything/eslint.config.mjs

@@ -0,0 +1,22 @@
+import baseConfig from '../../eslint.config.mjs';
+
+export default [
+  ...baseConfig,
+  {
+    files: ['**/*.json'],
+    rules: {
+      '@nx/dependency-checks': [
+        'error',
+        {
+          ignoredFiles: [
+            '{projectRoot}/eslint.config.{js,cjs,mjs}',
+            '{projectRoot}/vite.config.{js,ts,mjs,mts}',
+          ],
+        },
+      ],
+    },
+    languageOptions: {
+      parser: await import('jsonc-eslint-parser'),
+    },
+  },
+];

packages/sparql-anything/package.json

@@ -0,0 +1,31 @@
+{
+  "name": "@lde/sparql-anything",
+  "version": "0.1.0",
+  "description": "Convert tabular and other non-RDF sources to RDF with the SPARQL Anything CLI, running one process per input chunk via an @lde/task-runner.",
+  "repository": {
+    "url": "git+https://github.com/ldelements/lde.git",
+    "directory": "packages/sparql-anything"
+  },
+  "license": "MIT",
+  "type": "module",
+  "exports": {
+    "./package.json": "./package.json",
+    ".": {
+      "types": "./dist/index.d.ts",
+      "import": "./dist/index.js",
+      "development": "./src/index.ts",
+      "default": "./dist/index.js"
+    }
+  },
+  "main": "./dist/index.js",
+  "module": "./dist/index.js",
+  "types": "./dist/index.d.ts",
+  "files": [
+    "dist",
+    "!**/*.tsbuildinfo"
+  ],
+  "dependencies": {
+    "@lde/task-runner": "0.2.11",
+    "tslib": "^2.3.0"
+  }
+}

packages/sparql-anything/src/index.ts

@@ -0,0 +1,4 @@
+export {
+  SparqlAnythingConverter,
+  type SparqlAnythingConverterOptions,
+} from './sparql-anything-converter.js';