[FIX] Remove whitespace from FASTA id (#100)

* [FIX] Remove flanking whitespace from fasta id

include/valik/split/metadata.hpp

@@ -13,6 +13,24 @@
 namespace valik
 {
 
+/**
+ * @brief Function that removes leading and trailing whitespace.
+*/
+template <typename id_t>
+void trim_fasta_id(id_t & id)
+{
+    std::string whitespace = " \t\n\r\v";
+    auto first_valid = id.find_first_not_of(whitespace);
+    if (first_valid == std::string::npos)
+        throw std::runtime_error{"Sequence name can not be empty."};
+    id.erase(0, first_valid);
+
+    auto last_valid = id.find_last_not_of(whitespace);
+    if (last_valid == std::string::npos)
+        throw std::runtime_error{"Sequence name can not be empty."};
+    id.erase(last_valid + 1);
+}
+
 /**
  * @brief Struct that stores the metadata for a split database.
  *  \param total_len    Total database length.
@@ -105,7 +123,6 @@ struct metadata
     size_t seg_count;
 
     private:
-
         std::vector<sequence_stats> sequences;
         size_t default_seg_len;
         std::vector<segment_stats> segments;
@@ -123,6 +140,7 @@ struct metadata
             size_t fasta_ind = 0;
             for (auto & record : fin)
             {
+                trim_fasta_id(record.id());
                 sequence_stats seq(record.id(), fasta_ind, record.sequence().size());
                 total_len += seq.len;
                 sequences.push_back(seq);

test/api/valik/split/CMakeLists.txt

@@ -5,3 +5,5 @@ target_use_datasources (write_seg_sequences_test FILES write_out_0_16_reference_
 target_use_datasources (write_seg_sequences_test FILES write_out_0_4_reference_metadata.txt)
 target_use_datasources (write_seg_sequences_test FILES write_out_20_16_reference_metadata.txt)
 target_use_datasources (write_seg_sequences_test FILES write_out_20_4_reference_metadata.txt)
+
+add_api_test (trim_fasta_id_test.cpp)

test/api/valik/split/trim_fasta_id_test.cpp

@@ -0,0 +1,38 @@
+#include <gtest/gtest.h>
+
+#include <valik/split/metadata.hpp>
+
+TEST(trim_fasta_id, no_whitespace)
+{
+    std::string id = "xy1";
+    std::string expected = id;
+
+    valik::trim_fasta_id(id);
+    EXPECT_EQ(id, expected);
+}
+
+TEST(trim_fasta_id, only_whitespace)
+{
+    EXPECT_THROW({
+        try
+        {
+            std::string id = "\n \t";
+            valik::trim_fasta_id(id);
+        }
+        catch( const std::runtime_error& e )
+        {
+            // and this tests that it has the correct message
+            EXPECT_STREQ( "Sequence name can not be empty.", e.what() );
+            throw;
+        }
+    }, std::runtime_error );
+}
+
+TEST(trim_fasta_id, remove_whitespace)
+{
+    std::string id = " \txy1 \n";
+    std::string expected = "xy1";
+
+    valik::trim_fasta_id(id);
+    EXPECT_EQ(id, expected);
+}