Handle resume ingestion for annotation VCFs (#643)

TileDB-Inc · Dec 21, 2023 · eeb71ef · eeb71ef
1 parent 128d736
commit eeb71ef
Show file tree

Hide file tree

Showing 2 changed files with 11 additions and 5 deletions.
diff --git a/libtiledbvcf/src/vcf/vcf_utils.cc b/libtiledbvcf/src/vcf/vcf_utils.cc
@@ -59,7 +59,13 @@ bcf_hdr_t* VCFUtils::hdr_read_header(const std::string& path) {
 std::vector<std::string> VCFUtils::get_sample_name_from_vcf(
     const std::string& path) {
   SafeBCFHdr hdr(hdr_read_header(path), bcf_hdr_destroy);
-  return hdr_get_samples(hdr.get());
+  auto samples = hdr_get_samples(hdr.get());
+  // If there are no samples, add an empty string to the list
+  // to indicate this is a sampleless VCF.
+  if (samples.empty()) {
+    samples.push_back("");
+  }
+  return samples;
 }
 
 std::vector<std::string> VCFUtils::hdr_get_samples(bcf_hdr_t* hdr) {

diff --git a/libtiledbvcf/src/write/writer.cc b/libtiledbvcf/src/write/writer.cc
@@ -774,9 +774,9 @@ std::pair<uint64_t, uint64_t> Writer::ingest_samples_v4(
   if (params.resume_sample_partial_ingestion &&
       !existing_sample_contig_fragments.empty()) {
     const std::string first_sample_name =
-        VCFUtils::get_sample_name_from_vcf(samples.front().sample_uri)[0];
+        VCFUtils::get_sample_name_from_vcf(samples.front().sample_uri).at(0);
     const std::string last_sample_name =
-        VCFUtils::get_sample_name_from_vcf(samples.back().sample_uri)[0];
+        VCFUtils::get_sample_name_from_vcf(samples.back().sample_uri).at(0);
 
     LOG_INFO("Resume: checking for regions to skip");
     LOG_DEBUG("Resume: regions before resume check = {}", regions_v4.size());
@@ -787,7 +787,7 @@ std::pair<uint64_t, uint64_t> Writer::ingest_samples_v4(
       bool skip = false;
 
       LOG_DEBUG(
-          "Resume: Checking sample_range=({}, {}) contig={}",
+          "Resume: Checking sample_range=('{}', '{}') contig={}",
           first_sample_name,
           last_sample_name,
           contig);
@@ -803,7 +803,7 @@ std::pair<uint64_t, uint64_t> Writer::ingest_samples_v4(
             {first_sample_name, last_sample_name});
 
         LOG_DEBUG(
-            "Resume:   found fragments with sample_range=({}, {})",
+            "Resume:   found fragments with sample_range=('{}', '{}')",
             first_sample_name,
             last_sample_name);
         // Loop over contigs for the sample range