From 1850b96e0df361c4051a941bd38ed4b85e26c00e Mon Sep 17 00:00:00 2001
From: Evelin <evelin.aasna@fu-berlin.de>
Date: Fri, 2 Aug 2024 21:29:43 +0200
Subject: [PATCH] Check unique IDs

---
 include/valik/search/cart_query_io.hpp   |  5 ++---
 include/valik/search/iterate_queries.hpp | 21 ++++++---------------
 2 files changed, 8 insertions(+), 18 deletions(-)

diff --git a/include/valik/search/cart_query_io.hpp b/include/valik/search/cart_query_io.hpp
index 2b9e8be6..5a5bfd21 100644
--- a/include/valik/search/cart_query_io.hpp
+++ b/include/valik/search/cart_query_io.hpp
@@ -36,7 +36,7 @@ inline bool get_cart_queries(rec_vec_t const & records,
                              seq_vec_t & seq_vec)
 {
 
-    std::set<TId> uniqueIds; // set of short IDs (cut at first whitespace)
+    std::set<std::string> uniqueIds; // set of short IDs (cut at first whitespace)
     bool idsUnique = true;
 
     size_t seqCount{0};
@@ -45,8 +45,7 @@ inline bool get_cart_queries(rec_vec_t const & records,
     {
         seq_vec.emplace_back(record.sequence);
         seqCount++;
-
-        //!TODO: idsUnique?
+        idsUnique &= stellar::_checkUniqueId(uniqueIds, record.sequence_id);
     }
 
     /*
diff --git a/include/valik/search/iterate_queries.hpp b/include/valik/search/iterate_queries.hpp
index e63f0318..318a5bd2 100644
--- a/include/valik/search/iterate_queries.hpp
+++ b/include/valik/search/iterate_queries.hpp
@@ -53,7 +53,7 @@ void iterate_all_queries(size_t const ref_seg_count,
                          search_arguments const & arguments,
                          cart_queue<shared_query_record<seq_t>> & queue)
 {
-    using TId = seqan2::CharString;
+    using TId = std::string;
     constexpr uint64_t chunk_size = (1ULL << 20) * 10;
 
     seqan3::sequence_file_input<dna4_adaptor_traits> fin{arguments.query_file};
@@ -65,10 +65,7 @@ void iterate_all_queries(size_t const ref_seg_count,
     size_t seqCount{0};
     for (auto & record : fin)
     {
-        //!TODO: IDs unique??
-        // TId id{};
-        // idsUnique &= stellar::_checkUniqueId(uniqueIds, id);
-
+        idsUnique &= stellar::_checkUniqueId(uniqueIds, record.id());
         query_records.emplace_back(record.id(), std::make_shared<seq_t>(std::move(record.sequence())));
 
         if (query_records.size() > chunk_size)
@@ -99,7 +96,7 @@ void iterate_short_queries(search_arguments const & arguments,
                            raptor::threshold::threshold const & thresholder,
                            cart_queue<shared_query_record<seq_t>> & queue)
 {
-    using TId = seqan2::CharString;
+    using TId = std::string;
     constexpr uint64_t chunk_size = (1ULL << 20) * 10;
 
     seqan3::sequence_file_input<dna4_adaptor_traits> fin{arguments.query_file};
@@ -111,10 +108,7 @@ void iterate_short_queries(search_arguments const & arguments,
     size_t seqCount{0};
     for (auto & record : fin)
     {
-        //!TODO: IDs unique??
-        // TId id{};
-        // idsUnique &= stellar::_checkUniqueId(uniqueIds, id);
-
+        idsUnique &= stellar::_checkUniqueId(uniqueIds, record.id());
         query_records.emplace_back(record.id(), std::make_shared<seq_t>(std::move(record.sequence())));
 
         if (query_records.size() > chunk_size)
@@ -147,7 +141,7 @@ void iterate_split_queries(search_arguments const & arguments,
                            cart_queue<shared_query_record<seq_t>> & queue,
                            metadata & meta)
 {
-    using TId = seqan2::CharString;
+    using TId = std::string;
     constexpr uint64_t chunk_size = (1ULL << 20) * 10;
 
     seqan3::sequence_file_input<dna4_adaptor_traits> fin{arguments.query_file};
@@ -159,10 +153,7 @@ void iterate_split_queries(search_arguments const & arguments,
     size_t seqCount{0};
     for (auto & record : fin)
     {
-        //!TODO: IDs unique??
-        // TId id{};
-        // idsUnique &= stellar::_checkUniqueId(uniqueIds, id);
-
+        idsUnique &= stellar::_checkUniqueId(uniqueIds, record.id());
         auto query_ptr = std::make_shared<seq_t>(std::move(record.sequence()));
         for (auto const & seg : meta.segments_from_ind(seqCount))
         {