Fix header

Author: mwiewior

datafusion/vcf/examples/datafusion_integration.rs

@@ -7,9 +7,9 @@ async fn main() -> datafusion::error::Result<()> {
     env_logger::init();
     // let path = "gs://gcp-public-data--gnomad/release/4.1/vcf/exomes/gnomad.exomes.v4.1.sites.chr21.vcf.bgz".to_string();
     // let path = "gs://gcp-public-data--gnomad/release/4.1/genome_sv/gnomad.v4.1.sv.sites.vcf.gz".to_string();
-    let path = "gs://genomics-public-data/platinum-genomes/vcf/NA12878_S1.genome.vcf".to_string();
+    // let path = "gs://genomics-public-data/platinum-genomes/vcf/NA12878_S1.genome.vcf".to_string();
     // let path ="/tmp/gnomad.exomes.v4.1.sites.chr21.vcf.bgz".to_string();
-    // let path ="/tmp/gnomad.v4.1.sv.sites.vcf.gz".to_string();
+    let path ="/tmp/gnomad.v4.1.sv.sites.vcf.gz".to_string();
     // let infos  = Some(Vec::from(["AC".to_string(), "AF".to_string(), "AN".to_string(), "FS".to_string(), "AN_raw".to_string(), "variant_type".to_string(), "AS_culprit".to_string(), "only_het".to_string()]));
     // let infos  = Some(Vec::from(["SVTYPE".to_string()]));
     // let infos  = Some(Vec::from(["AF".to_string()]));

datafusion/vcf/src/storage.rs

@@ -225,6 +225,19 @@ pub async fn get_remote_vcf_header(file_path: String) -> Result<vcf::Header, Err
     Ok(header)
 }
 
+pub async fn get_header(file_path: String) -> Result<vcf::Header, Error> {
+    let storage_type = get_storage_type(file_path.clone());
+    let header = match storage_type {
+        StorageType::LOCAL => {
+            get_local_vcf_header(file_path, 1).await?
+        }
+        _ => {
+            get_remote_vcf_header(file_path).await?
+        }
+    };
+    Ok(header)
+}
+
 pub enum VcfRemoteReader {
     BGZF( vcf::r#async::io::Reader<AsyncReader<StreamReader<FuturesBytesStream, Bytes>>>),
     PLAIN( vcf::r#async::io::Reader<StreamReader<FuturesBytesStream, Bytes>>)

datafusion/vcf/src/table_provider.rs

@@ -17,16 +17,15 @@ use noodles::vcf::Header;
 use noodles::vcf::header::Infos;
 use noodles::vcf::header::record::value::map::info::{Number, Type};
 use crate::physical_exec::VcfExec;
-use crate::storage::{get_local_vcf_bgzf_reader, get_local_vcf_header, get_remote_vcf_bgzf_reader, get_remote_vcf_header, get_storage_type, StorageType};
+use crate::storage::{get_header, get_local_vcf_bgzf_reader, get_local_vcf_header, get_remote_vcf_bgzf_reader, get_remote_vcf_header, get_storage_type, StorageType};
 
 async fn determine_schema_from_header(
     file_path: &str,
     info_fields: &Option<Vec<String>>,
     format_fields: &Option<Vec<String>>,
 ) -> datafusion::common::Result<SchemaRef> {
 
-    let storage_type = get_storage_type(file_path.to_string());
-    let header = get_remote_vcf_header(file_path.to_string()).await?;
+    let header = get_header(file_path.to_string()).await?;
     let header_infos = header.infos();
 
     let mut fields = vec![