refactor(embl,gbk,record): deduplicate RangeValue; introduce GenericRecord and conversions; expose record module

microBioRust/src/embl.rs

@@ -747,24 +747,7 @@ where
      }
 }
 
-///stores a value for start or stop (end) which can be denoted as a < value or > value.
-#[derive(Debug, Hash, PartialEq, Eq, Clone)]
-pub enum RangeValue {
-    Exact(u32),
-    LessThan(u32),
-    GreaterThan(u32),
-}
-
-//trait for rangevalue
-impl RangeValue {
-    pub fn get_value(&self) -> u32 {
-        match self {
-            RangeValue::Exact(value) => *value,
-            RangeValue::LessThan(value) => *value,
-            RangeValue::GreaterThan(value) => *value,
-        }
-    }
-}
+pub use crate::record::RangeValue;
 
 ///stores the details of the source features in genbank (contigs)
 #[derive(Debug, Eq, PartialEq, Hash, Clone)]
@@ -1379,6 +1362,25 @@ impl Default for Record {
      }
 }
 
+// Provide a type alias and conversion to a generic record to aid interoperability
+pub type GenericRecordEmbl = crate::record::GenericRecord<SourceAttributeBuilder, FeatureAttributeBuilder, SequenceAttributeBuilder>;
+
+impl From<&Record> for GenericRecordEmbl {
+    fn from(r: &Record) -> Self {
+        Self {
+            id: r.id.clone(),
+            seq: r.sequence.clone(),
+            seqid: r.id.clone(),
+            start: r.start as u32,
+            end: r.end as u32,
+            strand: r.strand,
+            source: r.source_map.clone(),
+            cds: r.cds.clone(),
+            seq_features: r.seq_features.clone(),
+        }
+    }
+}
+
 #[allow(dead_code)]
 pub struct Config {
     filename: String,

microBioRust/src/gbk.rs

@@ -363,7 +363,7 @@ macro_rules! create_builder {
 	        self.$attributes
             }
 	    // function to iterate immutably through the BTreeMap as required
-	    pub fn iter_sorted(&self) -> std::collections::btree_map::Iter<String, HashSet<$enum_name>> {
+	    pub fn iter_sorted(&'_ self) -> std::collections::btree_map::Iter<String, HashSet<$enum_name>> {
 	        self.$attributes.iter()
 	    }
 	    //default function
@@ -816,24 +816,7 @@ where
      }
 }
 
-///stores a value for start or stop (end) which can be denoted as a < value or > value.
-#[derive(Debug, Hash, PartialEq, Eq, Clone)]
-pub enum RangeValue {
-    Exact(u32),
-    LessThan(u32),
-    GreaterThan(u32),
-}
-
-//trait for rangevalue
-impl RangeValue {
-    pub fn get_value(&self) -> u32 {
-        match self {
-            RangeValue::Exact(value) => *value,
-            RangeValue::LessThan(value) => *value,
-            RangeValue::GreaterThan(value) => *value,
-        }
-    }
-}
+pub use crate::record::RangeValue;
 
 //stores the details of the source features in genbank (contigs)
 #[derive(Debug, Eq, PartialEq, Hash, Clone)]
@@ -1549,6 +1532,25 @@ impl Default for Record {
      }
 }
 
+// Provide a type alias and conversion to a generic record to aid interoperability
+pub type GenericRecordGbk = crate::record::GenericRecord<SourceAttributeBuilder, FeatureAttributeBuilder, SequenceAttributeBuilder>;
+
+impl From<&Record> for GenericRecordGbk {
+    fn from(r: &Record) -> Self {
+        Self {
+            id: r.id.clone(),
+            seq: r.sequence.clone(),
+            seqid: r.id.clone(),
+            start: r.start as u32,
+            end: r.end as u32,
+            strand: r.strand,
+            source: r.source_map.clone(),
+            cds: r.cds.clone(),
+            seq_features: r.seq_features.clone(),
+        }
+    }
+}
+
 #[allow(dead_code)]
 pub struct Config {
     filename: String,

microBioRust/src/lib.rs

@@ -9,5 +9,6 @@
 //!  Additionally, you can create new features and records and save them either in genbank or gff3 format
 //!
 #![allow(non_snake_case)]
+pub mod record;
 pub mod embl;
 pub mod gbk;

microBioRust/src/record.rs

@@ -0,0 +1,74 @@
+///Shared generic record types to reduce duplication between gbk and embl
+///Minimal initial introduction: defines generic containers and builders that mirror the existing API where possible
+
+use std::collections::{HashMap, HashSet};
+
+#[derive(Clone, Debug, PartialEq, Eq, Hash)]
+pub enum RangeValue {
+    Exact(u32),
+    LessThan(u32),
+    GreaterThan(u32),
+}
+
+impl RangeValue {
+    pub fn get_value(&self) -> u32 {
+        match self {
+            RangeValue::Exact(v) => *v,
+            RangeValue::LessThan(v) => *v,
+            RangeValue::GreaterThan(v) => *v,
+        }
+    }
+}
+
+///Traits to unify attribute enums across formats. Existing enums can implement Into these trait views if needed
+pub trait HasStartStopStrand {
+    fn start(&self) -> Option<RangeValue> { None }
+    fn stop(&self) -> Option<RangeValue> { None }
+    fn strand(&self) -> Option<i8> { None }
+}
+
+///Generic attribute builders
+#[derive(Clone, Debug, Default)]
+pub struct AttributeBuilder<K, V> {
+    pub name: Option<String>,
+    pub attributes: HashMap<K, HashSet<V>>,
+}
+
+impl<K, V> AttributeBuilder<K, V>
+where
+    K: Eq + std::hash::Hash,
+    V: Eq + std::hash::Hash,
+{
+    pub fn set_name(&mut self, name: String) { self.name = Some(name); }
+    pub fn get_name(&self) -> Option<&String> { self.name.as_ref() }
+    pub fn add(&mut self, key: K, value: V) {
+        self.attributes.entry(key).or_insert_with(HashSet::new).insert(value);
+    }
+    pub fn get(&self, key: &K) -> Option<&HashSet<V>> { self.attributes.get(key) }
+}
+
+///Generic record and records container
+#[derive(Clone, Debug, Default)]
+pub struct GenericRecord<S, F, Q> {
+    pub id: String,
+    pub seq: String,
+    pub seqid: String,
+    pub start: u32,
+    pub end: u32,
+    pub strand: i32,
+    pub source: S,
+    pub cds: F,
+    pub seq_features: Q,
+}
+
+impl<S, F, Q> GenericRecord<S, F, Q> {
+    pub fn is_empty(&self) -> bool { self.id.is_empty() && self.seq.is_empty() }
+}
+
+pub struct GenericRecords<R> {
+    inner: R,
+}
+
+impl<R> GenericRecords<R> {
+    pub fn new(reader: R) -> Self { Self { inner: reader } }
+}