diff --git a/dataload/05_materialise/grebi_materialise/Cargo.toml b/dataload/05_link/grebi_link/Cargo.toml similarity index 92% rename from dataload/05_materialise/grebi_materialise/Cargo.toml rename to dataload/05_link/grebi_link/Cargo.toml index 18c7525..95a60f7 100644 --- a/dataload/05_materialise/grebi_materialise/Cargo.toml +++ b/dataload/05_link/grebi_link/Cargo.toml @@ -1,6 +1,6 @@ [package] -name = "grebi_materialise" +name = "grebi_link" version = "0.1.0" edition = "2021" diff --git a/dataload/05_materialise/grebi_materialise/src/main.rs b/dataload/05_link/grebi_link/src/main.rs similarity index 100% rename from dataload/05_materialise/grebi_materialise/src/main.rs rename to dataload/05_link/grebi_link/src/main.rs diff --git a/dataload/05_materialise/merge_summary_jsons.py b/dataload/05_link/merge_summary_jsons.py similarity index 100% rename from dataload/05_materialise/merge_summary_jsons.py rename to dataload/05_link/merge_summary_jsons.py diff --git a/dataload/nextflow/load_subgraph.nf b/dataload/nextflow/load_subgraph.nf index 39b3f74..c689f25 100644 --- a/dataload/nextflow/load_subgraph.nf +++ b/dataload/nextflow/load_subgraph.nf @@ -31,7 +31,7 @@ workflow { indexed = index(merged.collect()) link(merged.flatten(), indexed.metadata_jsonl, indexed.summary_json, Channel.value(config.exclude_edges + config.identifier_props), Channel.value(config.exclude_self_referential_edges + config.identifier_props), groups_txt) - merge_summary_jsons(indexed.summary_json.collect() + link.out.mat_summary.collect()) + merge_summary_jsons(indexed.summary_json.collect() + link.out.linked_summary.collect()) compressed_blobs = create_compressed_blobs(link.out.nodes.mix(link.out.edges)) sqlite = create_sqlite(compressed_blobs.collect()) @@ -250,19 +250,19 @@ process link { output: path("linked_nodes_${task.index}.jsonl"), emit: nodes path("linked_edges_${task.index}.jsonl"), emit: edges - path("mat_summary_${task.index}.json"), emit: mat_summary + path("linked_summary_${task.index}.json"), emit: linked_summary script: """ #!/usr/bin/env bash set -Eeuo pipefail cat ${merged_filename} \ - | ${params.home}/target/release/grebi_materialise \ + | ${params.home}/target/release/grebi_link \ --in-metadata-jsonl ${metadata_jsonl} \ --in-summary-json ${index_summary_json} \ --groups-txt ${groups_txt} \ --out-edges-jsonl linked_edges_${task.index}.jsonl \ - --out-summary-json mat_summary_${task.index}.json \ + --out-summary-json linked_summary_${task.index}.json \ --exclude ${exclude.iterator().join(",")} \ --exclude-self-referential ${exclude_self_referential.iterator().join(",")} \ > linked_nodes_${task.index}.jsonl