fix incorrect sqlite equivalences and use btree instead of hash colle…

…ctions in id2group
EBISPOT · Jul 31, 2024 · 0fbd15c · 0fbd15c
1 parent 4a927ca
commit 0fbd15c
Show file tree

Hide file tree

Showing 2 changed files with 4 additions and 5 deletions.
diff --git a/01_ingest/grebi_ingest_sqlite/src/main.rs b/01_ingest/grebi_ingest_sqlite/src/main.rs
@@ -75,7 +75,6 @@ fn main() -> Result<()> {
 
                 if fk_info.is_some() {
                     json_obj[&col_name] = json!(format!("{}:{}:{}", prefix, fk_info.unwrap().0.to_singular(), v));
-                    ids.push(format!("{}:{}:{}", fk_info.unwrap().0.to_singular(), prefix, v));
                 } else {
                     json_obj[&col_name] = json!(v);
                 }

diff --git a/02_assign_ids/grebi_identifiers2groups/src/main.rs b/02_assign_ids/grebi_identifiers2groups/src/main.rs
@@ -1,5 +1,5 @@
 
-use std::collections::{HashSet, HashMap, BTreeMap};
+use std::collections::{BTreeSet, BTreeMap};
 use std::{env, io};
 use csv;
 use bloomfilter::Bloom;
@@ -21,15 +21,15 @@ struct Args {
 
 fn main() {
 
-	let mut group_to_entities:BTreeMap<u64, HashSet<Vec<u8>>> = BTreeMap::new();
+	let mut group_to_entities:BTreeMap<u64, BTreeSet<Vec<u8>>> = BTreeMap::new();
 	let mut entity_to_group:BTreeMap<Vec<u8>, u64> = BTreeMap::new();
 
 	let mut next_group_id:u64 = 1;
 
 	let args = Args::parse();
 	let add_group:Vec<String> = args.add_group;
 	for group in add_group {
-		let entries:HashSet<Vec<u8>> = group.split(",").map(|s| s.as_bytes().to_vec()).collect();
+		let entries:BTreeSet<Vec<u8>> = group.split(",").map(|s| s.as_bytes().to_vec()).collect();
 		let gid = next_group_id;
 		next_group_id = next_group_id + 1;
 		for id in &entries {
@@ -103,7 +103,7 @@ fn main() {
 			for id in &ids {
 				entity_to_group.insert(id.to_vec(), target_group);
 			} 
-			group_to_entities.insert(target_group, ids.iter().map(|id| id.to_vec()).collect::<HashSet<_>>());
+			group_to_entities.insert(target_group, ids.iter().map(|id| id.to_vec()).collect::<BTreeSet<_>>());
 		}
 	}