vcf: Add indexer

See #214.

noodles-vcf/CHANGELOG.md

@@ -4,9 +4,15 @@
 
 ### Added
 
+  * vcf: Add indexer (`vcf::index`) ([#214]).
+
+    This is a convenience function to index a bgzip-compressed VCF file.
+
   * vcf/header/record/value/map: Add mutable getter for other fields
     (`Map::<I>::other_fields_mut`).
 
+[#214]: https://github.com/zaeleus/noodles/issues/214
+
 ## 0.45.0 - 2023-11-02
 
 ### Changed

noodles-vcf/examples/vcf_index.rs

@@ -6,57 +6,19 @@
 
 use std::{
     env,
-    fs::File,
     io::{self, BufWriter},
 };
 
-use noodles_bgzf as bgzf;
-use noodles_core::Position;
-use noodles_csi::{self as csi, index::reference_sequence::bin::Chunk};
 use noodles_tabix as tabix;
 use noodles_vcf as vcf;
 
 fn main() -> io::Result<()> {
     let src = env::args().nth(1).expect("missing src");
 
-    let mut reader = File::open(src)
-        .map(bgzf::Reader::new)
-        .map(vcf::Reader::new)?;
-
-    let header = reader.read_header()?;
-
-    let mut record = vcf::Record::default();
-
-    let mut indexer = tabix::index::Indexer::default();
-    indexer.set_header(csi::index::header::Builder::vcf().build());
-
-    let mut start_position = reader.get_ref().virtual_position();
-
-    while reader.read_record(&header, &mut record)? != 0 {
-        let end_position = reader.get_ref().virtual_position();
-        let chunk = Chunk::new(start_position, end_position);
-
-        let reference_sequence_name = record.chromosome().to_string();
-        let start = Position::try_from(usize::from(record.position()))
-            .map_err(|e| io::Error::new(io::ErrorKind::InvalidData, e))?;
-        let end = record
-            .end()
-            .map_err(|e| io::Error::new(io::ErrorKind::InvalidData, e))
-            .and_then(|position| {
-                Position::try_from(usize::from(position))
-                    .map_err(|e| io::Error::new(io::ErrorKind::InvalidData, e))
-            })?;
-
-        indexer.add_record(&reference_sequence_name, start, end, chunk)?;
-
-        start_position = end_position;
-    }
-
-    let index = indexer.build();
+    let index = vcf::index(src)?;
 
     let stdout = io::stdout().lock();
     let mut writer = tabix::Writer::new(BufWriter::new(stdout));
-
     writer.write_index(&index)?;
 
     Ok(())

noodles-vcf/src/indexer.rs

@@ -0,0 +1,51 @@
+use std::{fs::File, io, path::Path};
+
+use noodles_bgzf as bgzf;
+use noodles_core::Position;
+use noodles_csi::{self as csi, index::reference_sequence::bin::Chunk};
+use noodles_tabix as tabix;
+
+use super::{Reader, Record};
+
+/// Indexes a bgzipped-compressed VCF file.
+///
+/// ```no_run
+/// use noodles_vcf as vcf;
+/// let index = vcf::index("sample.vcf.gz")?;
+/// # Ok::<_, std::io::Error>(())
+/// ```
+pub fn index<P>(src: P) -> io::Result<csi::Index>
+where
+    P: AsRef<Path>,
+{
+    let mut reader = File::open(src).map(bgzf::Reader::new).map(Reader::new)?;
+    let header = reader.read_header()?;
+
+    let mut indexer = tabix::index::Indexer::default();
+    indexer.set_header(csi::index::header::Builder::vcf().build());
+
+    let mut record = Record::default();
+    let mut start_position = reader.get_ref().virtual_position();
+
+    while reader.read_record(&header, &mut record)? != 0 {
+        let end_position = reader.get_ref().virtual_position();
+        let chunk = Chunk::new(start_position, end_position);
+
+        let reference_sequence_name = record.chromosome().to_string();
+        let start = Position::try_from(usize::from(record.position()))
+            .map_err(|e| io::Error::new(io::ErrorKind::InvalidData, e))?;
+        let end = record
+            .end()
+            .map_err(|e| io::Error::new(io::ErrorKind::InvalidData, e))
+            .and_then(|position| {
+                Position::try_from(usize::from(position))
+                    .map_err(|e| io::Error::new(io::ErrorKind::InvalidData, e))
+            })?;
+
+        indexer.add_record(&reference_sequence_name, start, end, chunk)?;
+
+        start_position = end_position;
+    }
+
+    Ok(indexer.build())
+}

noodles-vcf/src/lib.rs

@@ -24,6 +24,7 @@ mod r#async;
 
 pub mod header;
 pub mod indexed_reader;
+mod indexer;
 pub mod io;
 pub mod lazy;
 pub mod reader;
@@ -33,7 +34,7 @@ mod variant_writer;
 pub mod writer;
 
 pub use self::{
-    header::Header, indexed_reader::IndexedReader, reader::Reader, record::Record,
+    header::Header, indexed_reader::IndexedReader, indexer::index, reader::Reader, record::Record,
     variant_reader::VariantReader, variant_writer::VariantWriter, writer::Writer,
 };