feat: port over assembly info from bioutils (#8)

.gitattributes

@@ -0,0 +1 @@
+src/static_data/**/*.json* filter=lfs diff=lfs merge=lfs -text

Cargo.toml

@@ -5,5 +5,10 @@ edition = "2021"
 
 [dependencies]
 anyhow = "1.0.69"
+enum-map = "2.4.2"
+flate2 = "1.0.25"
+lazy_static = "1.4.0"
 nom = "7.1.3"
 pretty_assertions = "1.3.0"
+serde = { version = "1.0.152", features = ["derive"] }
+serde_json = "1.0.93"

src/lib.rs

@@ -1,4 +1,5 @@
 pub mod parser;
+pub mod static_data;
 
 pub fn add(left: usize, right: usize) -> usize {
     left + right

src/static_data/_data/Makefile

@@ -0,0 +1,36 @@
+ASSEMBLIES := \
+	CHM1_1.0 \
+	CHM1_1.1 \
+	GRCh37 \
+	GRCh37.p10 \
+	GRCh37.p11 \
+	GRCh37.p12 \
+	GRCh37.p13 \
+	GRCh37.p2 \
+	GRCh37.p5 \
+	GRCh37.p9 \
+	GRCh38 \
+	GRCh38.p10 \
+	GRCh38.p11 \
+	GRCh38.p12 \
+	GRCh38.p1 \
+	GRCh38.p2 \
+	GRCh38.p3 \
+	GRCh38.p4 \
+	GRCh38.p5 \
+	GRCh38.p6 \
+	GRCh38.p7 \
+	GRCh38.p8 \
+	GRCh38.p9 \
+	NCBI33 \
+	NCBI34 \
+	NCBI35 \
+	NCBI36 \
+
+JSONS := $(addsuffix .json.gz,${ASSEMBLIES})
+
+.PHONY:
+all: $(JSONS)
+
+%.json.gz:
+	wget -O $@ https://github.com/biocommons/bioutils/raw/main/src/bioutils/_data/assemblies/$@

src/static_data/mod.rs

@@ -0,0 +1,74 @@
+//! Static data.
+
+use std::io::Read;
+
+use enum_map::{enum_map, Enum, EnumMap};
+use flate2::read::GzDecoder;
+use serde::Deserialize;
+
+const GRCH37_JSON_GZ: &[u8] = include_bytes!("_data/GRCh37.json.gz");
+const GRCH38_JSON_GZ: &[u8] = include_bytes!("_data/GRCh38.json.gz");
+
+#[derive(Debug, Deserialize, Enum)]
+pub enum Assembly {
+    Grch37,
+    Grch38,
+}
+
+impl Assembly {
+    /// Deserialize assembly info from embedded compressed JSON.
+    fn load_assembly_info(&self) -> AssemblyInfo {
+        let payload = match self {
+            Assembly::Grch37 => GRCH37_JSON_GZ,
+            Assembly::Grch38 => GRCH38_JSON_GZ,
+        };
+        let mut d = GzDecoder::new(payload);
+        let mut grch37_json = String::new();
+        d.read_to_string(&mut grch37_json).unwrap();
+        serde_json::from_str::<AssemblyInfo>(&grch37_json).unwrap()
+    }
+}
+
+#[derive(Debug, Deserialize)]
+pub struct Sequence {
+    pub aliases: Vec<String>,
+    pub assembly_unit: String,
+    pub genbank_ac: String,
+    pub length: usize,
+    pub name: String,
+    pub refseq_ac: String,
+    pub relationship: String,
+    pub sequence_role: String,
+}
+
+#[derive(Debug, Deserialize)]
+pub struct AssemblyInfo {
+    pub date: String,
+    pub description: String,
+    pub genbank_ac: String,
+    pub name: String,
+    pub refseq_ac: String,
+    pub sequences: Vec<Sequence>,
+    pub submitter: String,
+}
+
+lazy_static::lazy_static! {
+    /// Provide information about the assemblies.
+    pub static ref ASSEMBLY_INFOS: EnumMap<Assembly, AssemblyInfo> = enum_map! {
+        Grch37 => Assembly::Grch37.load_assembly_info(),
+        Grch38 => Assembly::Grch38.load_assembly_info(),
+    };
+}
+
+#[cfg(test)]
+mod test {
+    use pretty_assertions::assert_eq;
+
+    use crate::static_data::{Assembly, ASSEMBLY_INFOS};
+
+    #[test]
+    fn smoke() {
+        assert_eq!(ASSEMBLY_INFOS[Assembly::Grch37].sequences.len(), 92);
+        assert_eq!(ASSEMBLY_INFOS[Assembly::Grch38].sequences.len(), 92);
+    }
+}