Refactor db graph code

Cargo.toml

@@ -21,8 +21,8 @@ name = "gen"
 path = "src/main.rs"
 
 [workspace]
-members = [".", "gen-core", "gen-models", "gen-graph", "gen-diff", "gen-tui", "gen-sugiyama", "gen-annotations", "gen-capnp-schemas"]
-default-members = [".", "gen-core", "gen-models", "gen-graph", "gen-diff", "gen-capnp-schemas", "gen-annotations"]
+members = [".", "gen-core", "gen-models", "gen-models/graph-tests", "gen-graph", "gen-diff", "gen-tui", "gen-sugiyama", "gen-annotations", "gen-capnp-schemas"]
+default-members = [".", "gen-core", "gen-models", "gen-models/graph-tests", "gen-graph", "gen-diff", "gen-capnp-schemas", "gen-annotations"]
 exclude = ["gen-python", "gen-r/src/rust"]
 
 [features]
@@ -93,6 +93,7 @@ pprof = { version = "0.15.0", optional = true }
 [dev-dependencies]
 cargo-llvm-cov = "0.8.5"
 cargo-deny = "0.19.0"
+gen-models-graph-tests = { path = "gen-models/graph-tests" }
 more-asserts = "0.3.1"
 insta = "1.43.1"
 

gen-annotations/src/gff.rs

@@ -304,7 +304,8 @@ mod tests {
             preserve_edge: true,
         };
 
-        BlockGroup::insert_change(conn, &change).expect("should apply AA update to child sample");
+        gen_graph::models::insert_change(conn, &change)
+            .expect("should apply AA update to child sample");
 
         let edge_to_insert = Edge::query(
             conn,

gen-annotations/src/test_helpers.rs

@@ -190,5 +190,5 @@ pub fn setup_test_data(conn: &GraphConnection) {
         preserve_edge: false,
     };
 
-    BlockGroup::insert_change(conn, &change).expect("should apply variant change");
+    gen_graph::models::insert_change(conn, &change).expect("should apply variant change");
 }

gen-annotations/src/translate/bed.rs

@@ -77,7 +77,8 @@ where
                 Entry::Occupied(entry) => entry.into_mut(),
                 Entry::Vacant(entry) => {
                     let path = BlockGroup::get_current_path(conn, &bg.id, history_ref)?;
-                    let graph = BlockGroup::get_graph(conn, &bg.id, history_ref)?;
+                    let graph =
+                        gen_graph::models::load_block_group_graph(conn, &bg.id, history_ref)?;
                     let mut tree = IntervalTree::default();
                     let mut position: i64 = 0;
                     for (node, strand) in project_path(&graph, &path.blocks(conn, history_ref)?) {

gen-annotations/src/translate/gff.rs

@@ -72,7 +72,8 @@ where
                 Entry::Occupied(entry) => entry.into_mut(),
                 Entry::Vacant(entry) => {
                     let path = BlockGroup::get_current_path(conn, &bg.id, history_ref)?;
-                    let graph = BlockGroup::get_graph(conn, &bg.id, history_ref)?;
+                    let graph =
+                        gen_graph::models::load_block_group_graph(conn, &bg.id, history_ref)?;
                     let mut tree = IntervalTree::default();
                     let mut position: i64 = 0;
                     for (node, strand) in project_path(&graph, &path.blocks(conn, history_ref)?) {

gen-core/Cargo.toml

@@ -16,6 +16,7 @@ gen-capnp-schemas = { path = "../gen-capnp-schemas", version = "0.2.1" }
 itertools = "0.14.0"
 capnp = { version = "0.24" }
 noodles = { version = "0.101.0", features = ["gff"] }
+petgraph = "0.6.5"
 rusqlite = { package = "rusqdoltlite", version = "0.40.14", features = ["bundled", "fallible_uint"] }
 serde = {  version = "1.0.219", features = ["derive"] }
 thiserror = "1.0"

gen-core/src/graph.rs

@@ -0,0 +1,101 @@
+use core::fmt;
+
+use petgraph::graphmap::DiGraphMap;
+use serde::{Deserialize, Serialize};
+
+use crate::{HashId, Strand};
+
+pub type GenGraph = DiGraphMap<GraphNode, Vec<GraphEdge>>;
+
+#[derive(Clone, Copy, Debug, Eq, PartialEq, Hash, Ord, PartialOrd, Deserialize, Serialize)]
+pub struct GraphEdge {
+    pub edge_id: HashId,
+    pub source_strand: Strand,
+    pub target_strand: Strand,
+    pub chromosome_index: i64,
+    pub phased: i64,
+    pub created_on: i64,
+}
+
+#[derive(Clone, Copy, Debug, Eq, PartialEq)]
+pub struct GraphLoadBlock {
+    pub id: i64,
+    pub node_id: HashId,
+    pub start: i64,
+    pub end: i64,
+}
+
+#[derive(Clone, Copy, Debug, Eq, PartialEq)]
+pub struct GraphLoadEdge {
+    pub edge_id: HashId,
+    pub source_node_id: HashId,
+    pub source_coordinate: i64,
+    pub source_strand: Strand,
+    pub target_node_id: HashId,
+    pub target_coordinate: i64,
+    pub target_strand: Strand,
+    pub chromosome_index: i64,
+    pub phased: i64,
+    pub created_on: i64,
+}
+
+/// A contiguous slice of a stored sequence represented as a node in graph space.
+#[derive(Clone, Copy, Debug, Eq, PartialEq, Hash, Ord, PartialOrd, Deserialize, Serialize)]
+pub struct GraphNode {
+    pub node_id: HashId,
+    pub sequence_start: i64,
+    pub sequence_end: i64,
+}
+
+impl fmt::Display for GraphNode {
+    fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result {
+        write!(
+            formatter,
+            "{}[{}-{}]",
+            self.node_id, self.sequence_start, self.sequence_end
+        )
+    }
+}
+
+impl GraphNode {
+    pub const fn length(&self) -> i64 {
+        self.sequence_end - self.sequence_start
+    }
+}
+
+/// A local sequence slice within a graph node, including its traversal orientation.
+#[derive(Clone, Copy, Debug, PartialEq, Eq, Hash, Serialize, Deserialize)]
+pub struct GraphNodeSlice {
+    pub block: GraphNode,
+    /// Local start offset within the block's sequence slice (`0..block.length()`).
+    pub start: usize,
+    /// Local end offset, exclusive (`start..=block.length()`).
+    pub end: usize,
+    /// 5'→3' orientation of this slice.
+    pub strand: Strand,
+}
+
+impl GraphNodeSlice {
+    pub fn full(block: GraphNode, strand: Strand) -> Self {
+        Self {
+            block,
+            start: 0,
+            end: block.length() as usize,
+            strand,
+        }
+    }
+}
+
+/// A cursor within a graph node.
+#[derive(Clone, Copy, Debug, Eq, PartialEq, Hash, Ord, PartialOrd, Deserialize, Serialize)]
+pub struct GraphNodePosition {
+    pub graph_node: GraphNode,
+    /// Distance from `sequence_start` of `graph_node`.
+    pub offset: i64,
+}
+
+impl GraphNodePosition {
+    pub const fn coordinate(&self) -> i64 {
+        self.graph_node.sequence_start + self.offset
+    }
+}

gen-core/src/lib.rs

@@ -8,6 +8,7 @@ pub mod config;
 pub mod errors;
 #[allow(clippy::all)]
 pub mod generated;
+pub mod graph;
 pub mod path;
 pub mod range;
 pub mod region;
@@ -17,6 +18,10 @@ pub mod traits;
 pub use config::Workspace;
 use errors::HashError;
 pub use generated::gen_core_capnp;
+pub use graph::{
+    GenGraph, GraphEdge, GraphLoadBlock, GraphLoadEdge, GraphNode, GraphNodePosition,
+    GraphNodeSlice,
+};
 pub use path::PathBlock;
 #[cfg(feature = "python-bindings")]
 use pyo3::pyclass;

gen-graph/Cargo.toml

@@ -9,8 +9,10 @@ license = "Apache-2.0"
 
 [dependencies]
 gen-core = { path = "../gen-core", version = "0.2.1" }
+gen-models = { path = "../gen-models", version = "0.2.1" }
 interavl = "0.3.0"
 intervaltree = "0.2.7"
+itertools = "0.14.0"
 petgraph = "0.6.5"
 serde = { version = "1.0.219", features = ["derive"] }
 thiserror = "2.0.12"