Graph Importer vertex and edgeSource only allows int type for id

[warframesam]

VertexConfig id() function should allow different id types other than int.

EdgeSourceConfig from() and to() functions should allow different types other than int.

Here is the problem I faced when loading graph vertex data with string id type:
JSONL file contained JSONs with id of type string that failed to parse. Here is a sample json, code, and error:

{"id": "W13696992", "title": "", "abstract": "", "authors": "", "publication_year": 2008}

.vertex("Document", new JsonlRowSource("vertices.jsonl"), v -> {
                            v.id("id");
                            v.property("id", "id");
                            v.property("title", "title");
                            v.property("abstract", "abstract");
                            v.property("authors", "authors");
                            v.intProperty("publication_year", "publication_year");
})

Caused by: com.arcadedb.serializer.json.JSONException: JSONObject[id] is not a int ("W13696992")
	at com.arcadedb.serializer.json.JSONObject.typeError(JSONObject.java:743)
	at com.arcadedb.serializer.json.JSONObject.getInt(JSONObject.java:262)
	at com.arcadedb.integration.importer.graph.JsonlRowSource$JsonlRecordReader.getInt(JsonlRowSource.java:92)
	at com.arcadedb.integration.importer.graph.GraphImporter.lambda$processVertexSource$3(GraphImporter.java:878)
	at com.arcadedb.integration.importer.graph.JsonlRowSource.forEach(JsonlRowSource.java:60)
	at com.arcadedb.integration.importer.graph.GraphImporter.processVertexSource(GraphImporter.java:850)
	at com.arcadedb.integration.importer.graph.GraphImporter.run(GraphImporter.java:768)

Here is the problem I faced when loading graph edge data with long id type:
CSV file contained long id values that failed to parse. Here is a sample CSV, code, and error:

from_id,to_id
2257721487,28031933
2186349252,2805857155
2186349252,1643034221

.edgeSource("Connection", new CsvRowSource("edges.csv"), e -> {
                            e.from("from_id", "Doc");
                            e.to("to_id", "Doc");
})

Caused by: java.lang.NumberFormatException: For input string: "2257721487"
	at java.base/java.lang.NumberFormatException.forInputString(NumberFormatException.java:67)
	at java.base/java.lang.Integer.parseInt(Integer.java:662)
	at java.base/java.lang.Integer.parseInt(Integer.java:778)
	at com.arcadedb.integration.importer.graph.CsvRowSource$CsvRecordReader.getInt(CsvRowSource.java:106)
	at com.arcadedb.integration.importer.graph.GraphImporter.lambda$processEdgeSource$8(GraphImporter.java:1037)
	at com.arcadedb.integration.importer.graph.CsvRowSource.forEach(CsvRowSource.java:84)
	at com.arcadedb.integration.importer.graph.GraphImporter.processEdgeSource(GraphImporter.java:1034)
	at com.arcadedb.integration.importer.graph.GraphImporter.run(GraphImporter.java:773)

[goutamadwant]

The implementation checked here has v.idByName("id") for string vertex IDs. Keep v.property("id", "id") to store the original value.

That doesn't solve the CSV edge case. processEdgeSource() reads both endpoints with getInt() and uses the integer ID map, so it cannot handle 2257721487 or string IDs. Switching the vertex to idByName() won't make edgeSource() resolve those IDs.

For the separate edge file, a workaround is to map external IDs to unique positive 32-bit integers in both files, use v.id() for those mapped IDs, and keep the original IDs as properties. Native string/long support needs changes to the importer.

Which ArcadeDB version are you on? The string vertex method above is from main.

[warframesam]

I'm using the latest stable ArcadeDB version -> 26.9.1
@goutamadwant Thank you for the response. I had also checked the source for GraphImporter and saw Interger.parseInt() calls on these id values and your response confirms my doubt.
Ideally, GraphImporter should have Object type for id and an additional parameter or maybe generics to resolve exact type.
void id(Object id, IdType idType)
<T> void id(T id)
Issue #7245 has already been opened for this discussion.

[goutamadwant]

good to know @warframesam thanks!

[lvca]

Fixed in #7245 (issue #7244), released in 26.10.1.

Vertex identities are now matched as text rather than as integers, so both of your cases work unchanged: v.id("id") accepts "W13696992", and e.from("from_id", "Doc") accepts 2257721487. No surrogate ids needed - values are compared exactly as your files spell them, so 007 and 7 stay two different vertices.

A few related problems were fixed along the way (a vertex id of 0 could not be pointed at, self-referencing edges could go the wrong way or miss forward references), and edges whose endpoint matches no vertex are now counted and logged instead of silently shrinking the graph. See the issue for details.

Thanks for the clear report - the samples made it easy to reproduce.

---

❤️ Thanks for helping make ArcadeDB better! ArcadeDB is free and open source, sustained by its community. If it's useful to you or your company, please consider becoming a sponsor to keep development going.