From 90247ce90cbd243c7a6ff64079326398ca49e9aa Mon Sep 17 00:00:00 2001
From: Brent Pedersen <bpederse@gmail.com>
Date: Fri, 22 Sep 2023 10:08:30 +0200
Subject: [PATCH] use xcf

---
 Cargo.toml        |  2 +-
 src/bedder_bed.rs |  2 ++
 src/bedder_vcf.rs | 15 ++++++++-----
 src/sniff.rs      | 54 +++++++++++++++++++++++------------------------
 4 files changed, 40 insertions(+), 33 deletions(-)

diff --git a/Cargo.toml b/Cargo.toml
index 2291d94..2df8250 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -26,7 +26,7 @@ env_logger = "0.10.0"
 log = "0.4.19"
 linear-map = "1.2.0"
 hashbrown = "0.14.0"
-xvcf = { version = "0.1.1", git = "https://github.com/brentp/xvcf-rs" }
+xvcf = { version = "0.1.2", git = "https://github.com/brentp/xvcf-rs" }
 
 [features]
 default = ["bed", "vcf", "bcf", "csi", "core", "bam", "sam", "bgzf"]
diff --git a/src/bedder_bed.rs b/src/bedder_bed.rs
index fd84c2a..de97584 100644
--- a/src/bedder_bed.rs
+++ b/src/bedder_bed.rs
@@ -1,3 +1,5 @@
+#![allow(clippy::useless_conversion)] // these are needed to support e.g. smartstring
+
 use crate::position::{Field, FieldError, Position, Positioned, Value};
 use crate::string::String;
 pub use bed::record::Record;
diff --git a/src/bedder_vcf.rs b/src/bedder_vcf.rs
index 5e84045..f799e0e 100644
--- a/src/bedder_vcf.rs
+++ b/src/bedder_vcf.rs
@@ -1,8 +1,8 @@
+#![allow(clippy::useless_conversion)] // these are needed to support e.g. smartstring
 use crate::position::{Field, FieldError, Position, Positioned, Value};
 use crate::string::String;
-use noodles::bcf;
 use noodles::vcf::{self, record::Chromosome};
-use std::io::{self, BufRead, Read};
+use std::io::{self, Read};
 use std::result;
 use vcf::record::info::field;
 use vcf::record::QualityScore;
@@ -12,6 +12,7 @@ pub use xvcf;
 pub struct BedderVCF<R> {
     reader: xvcf::Reader<R>,
     record_number: u64,
+    header: vcf::Header,
 }
 
 impl<R> BedderVCF<R>
@@ -19,24 +20,28 @@ where
     R: Read + 'static,
 {
     pub fn new(r: xvcf::Reader<R>) -> io::Result<BedderVCF<R>> {
+        let h = r.header().clone();
         let v = BedderVCF {
             reader: r,
             record_number: 0,
+            header: h,
         };
         Ok(v)
     }
 }
 
-pub fn from_reader<R: Read>(r: Box<R>) -> io::Result<BedderVCF<R>> {
+/*
+pub fn _from_reader<R: Read>(r: Box<R>) -> io::Result<BedderVCF<R>> {
     let reader = xvcf::Reader::from_reader(r, None)?;
     BedderVCF::new(reader)
 }
+*/
 
 fn match_info_value(info: &vcf::record::Info, name: &str) -> result::Result<Value, FieldError> {
     //let info = record.info();
     let key: vcf::record::info::field::Key = name
         .parse()
-        .map_err(|_| FieldError::InvalidFieldName(String::from(name.clone())))?;
+        .map_err(|_| FieldError::InvalidFieldName(String::from(name)))?;
 
     match info.get(&key) {
         Some(value) => match value {
@@ -142,7 +147,7 @@ where
     ) -> Option<std::result::Result<Position, std::io::Error>> {
         let mut v = vcf::Record::default();
 
-        match self.reader.next_record(&self.reader.header(), &mut v) {
+        match self.reader.next_record(&self.header, &mut v) {
             Ok(0) => None, // EOF
             Ok(_) => {
                 self.record_number += 1;
diff --git a/src/sniff.rs b/src/sniff.rs
index 9983eb8..bed0310 100644
--- a/src/sniff.rs
+++ b/src/sniff.rs
@@ -3,10 +3,9 @@ use std::io::{BufRead, Read};
 use std::path::Path;
 
 use crate::bedder_bed::BedderBed;
-use crate::bedder_vcf::{BedderVCF, VCF};
+use crate::bedder_vcf::BedderVCF;
 use crate::position::PositionedIterator;
 use noodles::bgzf;
-use noodles::vcf;
 
 /// File formats supported by this file detector.
 #[derive(Debug, PartialEq)]
@@ -52,35 +51,36 @@ where
         format,
         compression
     );
-    let br: Box<dyn BufRead> = match compression {
-        Compression::None => Box::new(reader),
-        Compression::GZ => Box::new(std::io::BufReader::new(GzDecoder::new(reader))),
-        Compression::BGZF => match format {
-            // BCF|BAM will appear as bgzf so we don't want to do this outside
-            FileFormat::BCF | FileFormat::BAM => Box::new(reader),
-            _ => Box::new(bgzf::Reader::new(reader)),
-        },
-        Compression::RAZF => unimplemented!(),
-    };
+    /*
+     */
     match format {
-        FileFormat::VCF => {
-            let mut vcf = vcf::reader::Builder.build_from_reader(br)?;
-            let hdr = vcf.read_header()?;
-            let bed_vcf = BedderVCF::new(VCF::VCF(vcf), hdr)?;
-            Ok(Box::new(bed_vcf))
-        }
-        FileFormat::BCF => {
-            let mut bcf = noodles::bcf::Reader::new(br);
-            let hdr = bcf.read_header()?;
-            let bed_vcf = BedderVCF::new(VCF::BCF(bcf), hdr)?;
+        FileFormat::VCF | FileFormat::BCF => {
+            // get &str from path
+            let path = path.as_ref().to_str().unwrap();
+            let x = xvcf::Reader::from_reader(Box::new(reader), Some(path))?;
+            let bed_vcf = BedderVCF::new(x)?;
             Ok(Box::new(bed_vcf))
         }
-
-        FileFormat::BED => {
-            let reader = BedderBed::new(br);
-            Ok(Box::new(reader))
+        _ => {
+            let br: Box<dyn BufRead> = match compression {
+                Compression::None => Box::new(reader),
+                Compression::GZ => Box::new(std::io::BufReader::new(GzDecoder::new(reader))),
+                Compression::BGZF => match format {
+                    // BCF|BAM will appear as bgzf so we don't want to do this outside
+                    FileFormat::BCF | FileFormat::BAM => Box::new(reader),
+                    _ => Box::new(bgzf::Reader::new(reader)),
+                },
+                Compression::RAZF => unimplemented!(),
+            };
+
+            match format {
+                FileFormat::BED => {
+                    let reader = BedderBed::new(br);
+                    Ok(Box::new(reader))
+                }
+                _ => unimplemented!("{format:?} not yet supported"),
+            }
         }
-        _ => unimplemented!("{format:?} not yet supported"),
     }
 }