Handle legacy stats arrays (#761)

TileDB-Inc · Aug 20, 2024 · 867f130 · 867f130
1 parent 6282fc3
commit 867f130
Show file tree

Hide file tree

Showing 3 changed files with 51 additions and 7 deletions.
diff --git a/libtiledbvcf/src/stats/sample_stats.cc b/libtiledbvcf/src/stats/sample_stats.cc
@@ -34,6 +34,10 @@
 namespace tiledb::vcf {
 
 SampleStats::~SampleStats() {
+  if (!enabled_) {
+    return;
+  }
+
   // Flush any remaining stats
   flush(true);
 
@@ -174,6 +178,30 @@ void SampleStats::init(
     return;
   }
 
+  // Check array version
+  {
+    auto array = Array(*ctx, uri, TILEDB_READ);
+
+    auto get_version = [&]() -> std::optional<int> {
+      tiledb_datatype_t value_type;
+      uint32_t value_num;
+      const void* value;
+      array.get_metadata("version", &value_type, &value_num, &value);
+      if (value_type == TILEDB_INT32 && value_num == 1) {
+        return *static_cast<const int*>(value);
+      }
+      return std::nullopt;
+    };
+
+    auto version = get_version();
+    if (version == std::nullopt) {
+      LOG_WARN(
+          "[SampleStats] Sample stats are deprecated for this array version.");
+      enabled_ = false;
+      return;
+    }
+  }
+
   LOG_DEBUG("[SampleStats] Open array '{}'", uri);
 
   // Open array

diff --git a/libtiledbvcf/src/stats/variant_stats.cc b/libtiledbvcf/src/stats/variant_stats.cc
@@ -38,6 +38,9 @@ namespace tiledb::vcf {
 int32_t VariantStats::max_length_ = 0;
 
 uint32_t VariantStats::array_version_ = VariantStats::VARIANT_STATS_MIN_VERSION;
+
+bool VariantStats::an_present_ = false;
+
 //===================================================================
 //= public static functions
 //===================================================================
@@ -213,6 +216,10 @@ void VariantStats::init(std::shared_ptr<Context> ctx, const Group& group) {
     throw std::runtime_error(
         "encountered variant stats array version out of range while writing");
 
+  // Check for presence of "an" attribute
+  auto schema = fetch_version.schema();
+  an_present_ = schema.has_attribute("an");
+
   // Open array
   array_ = std::make_unique<Array>(*ctx, uri, TILEDB_WRITE);
   enabled_ = true;
@@ -421,7 +428,9 @@ void VariantStats::flush(bool finalize) {
     }
 
     query_->set_data_buffer("ac", ac_buffer_);
-    query_->set_data_buffer("an", an_buffer_);
+    if (an_present_) {
+      query_->set_data_buffer("an", an_buffer_);
+    }
     query_->set_data_buffer("n_hom", n_hom_buffer_);
     if (array_version_ >= 3) {
       query_->set_data_buffer("max_length", max_length_buffer_);
@@ -469,7 +478,9 @@ void VariantStats::flush(bool finalize) {
     }
 
     query_->set_data_buffer("ac", ac_buffer_);
-    query_->set_data_buffer("an", an_buffer_);
+    if (an_present_) {
+      query_->set_data_buffer("an", an_buffer_);
+    }
     query_->set_data_buffer("n_hom", n_hom_buffer_);
     if (array_version_ >= 3) {
       query_->set_data_buffer("max_length", max_length_buffer_);

diff --git a/libtiledbvcf/src/stats/variant_stats.h b/libtiledbvcf/src/stats/variant_stats.h
@@ -241,15 +241,20 @@ class VariantStats {
   // Sample names included in the fragment
   inline static std::set<std::string> fragment_sample_names_;
 
-  //===================================================================
-  //= private non-static
-  //===================================================================
-
-  // maximum allele length ecountered
+  // maximum allele length encountered
   static int32_t max_length_;
 
+  // Array version
   static uint32_t array_version_;
 
+  // Flag to indicate if AN is present in the schema, needed to handle
+  // the case where AN could be missing in a version 2 schema
+  static bool an_present_;
+
+  //===================================================================
+  //= private non-static
+  //===================================================================
+
   // Count delta is +1 in ingest mode, -1 in delete mode
   int count_delta_ = 1;