Custom changeset management #50
Conversation
| @@ -17,3 +17,6 @@ noodles = { version = "0.78.0", features = ["core", "vcf", "fasta", "async"] } | |||
| petgraph = "0.6.5" | |||
| chrono = "0.4.38" | |||
| tempdir = "0.3.7" | |||
| fallible-streaming-iterator = "0.1.9" | |||
| serde = { version = "1.0.210", features = ["derive"] } | |||
| serde_json = "1.0.128" | |||
There was a problem hiding this comment.
we can pick a more efficient format once we settle on this.
| @@ -49,7 +49,7 @@ pub fn parse_genotype(gt: &str) -> Vec<Option<Genotype>> { | |||
| true => Phasing::Unphased, | |||
| false => Phasing::Phased, | |||
| }; | |||
| for entry in gt.split_inclusive(|c| c == '|' || c == '/') { | |||
| for entry in gt.split_inclusive(['|', '/']) { | |||
There was a problem hiding this comment.
clippy made me do it
| #[derive(Clone, Debug, Eq, Hash, PartialEq)] | ||
| use crate::models::{edge::Edge, path_edge::PathEdge, sequence::Sequence, strand::Strand, Sample}; | ||
|
|
||
| #[derive(Clone, Debug, Eq, Hash, PartialEq, Deserialize, Serialize)] | ||
| pub struct Path { |
There was a problem hiding this comment.
reorganized so it matches the order in up.sql
| @@ -76,7 +81,33 @@ impl Path { | |||
| }) | |||
| }) | |||
| .unwrap(); | |||
| let path = rows.next().unwrap().unwrap(); | |||
| let path = match rows.next().unwrap() { | |||
There was a problem hiding this comment.
made creation get-or-create
src/models/block_group.rs
Outdated
| @@ -105,17 +106,29 @@ impl BlockGroup { | |||
| Err(rusqlite::Error::SqliteFailure(err, details)) => { | |||
| if err.code == rusqlite::ErrorCode::ConstraintViolation { | |||
| println!("{err:?} {details:?}"); | |||
| BlockGroup { | |||
| id: conn | |||
| let bg_id = match(sample_name) { | |||
There was a problem hiding this comment.
this was bugged behavior if the insert was conflicting it would always return the null sample
Chris7
force-pushed
the
changeset-v2
branch
from
45e509b
to
f16308d
Compare
| @@ -113,6 +114,7 @@ mod tests { | |||
| let mut fasta_path = PathBuf::from(env!("CARGO_MANIFEST_DIR")); | |||
| fasta_path.push("fixtures/simple.fa"); | |||
| let conn = get_connection(None); | |||
| let db_uuid = metadata::get_db_uuid(&conn); | |||
There was a problem hiding this comment.
There's a duplicate of this line a couple lines down, I think I added that to fix the test
| @@ -32,13 +34,28 @@ pub struct EdgeData { | |||
| pub phased: i32, | |||
| } | |||
|
|
|||
| impl From<&Edge> for EdgeData { | |||
| @@ -528,6 +545,79 @@ mod tests { | |||
| assert_eq!(edge_result3.target_coordinate, -1); | |||
| } | |||
|
|
|||
| #[test] | |||
| fn test_bulk_create_returns_edges_in_order() { | |||
There was a problem hiding this comment.
Oh, was bulk create not returning them in order? Oops, thanks!
src/models/path.rs
Outdated
| @@ -65,8 +68,10 @@ pub struct NewBlock { | |||
|
|
|||
| impl Path { | |||
| pub fn create(conn: &Connection, name: &str, block_group_id: i32, edge_ids: &[i32]) -> Path { | |||
| // TODO: Should we do something if edge_ids don't match here? | |||
There was a problem hiding this comment.
I don't understand what is meant by "if edge_ids don't match", can you add a bit more? I have a list of a few validations I'd like to add, I'll bring them up on slack
| @@ -30,6 +31,17 @@ pub struct NewSequence<'a> { | |||
| shallow: bool, | |||
| } | |||
|
|
|||
| impl<'a> From<&'a Sequence> for NewSequence<'a> { | |||
| .unwrap() | ||
| .to_string(); | ||
| created_edges.insert(edge_pk); | ||
| if !created_sequences.contains(&source_hash) { |
There was a problem hiding this comment.
How are we sure that this sequence doesn't show up later as a created sequence?
This is an extension to the changeset management to identify dependencies of a changeset and write those out. On imports, we ensure that the ids of dependent entities are created or retrieved from the current checked out database.
The workflow this handles is suppose we have branch A with operations 1->2->3, and branch B with operations 1->4->5. If we apply operation 3, things required from operation 2 may not be present. This ensures they will be. For the reversion, we create a new commit with the updates from this application (similar to a new commit id resulting from a cherry-pick).