better handling of inconsistent assembly_summary file headers

muellan · Feb 29, 2024 · c563174 · c563174
1 parent 4c6db91
commit c563174
Show file tree

Hide file tree

Showing 3 changed files with 14 additions and 6 deletions.
diff --git a/src/building.cpp b/src/building.cpp
@@ -367,10 +367,16 @@ void add_targets_to_database(database& db,
                 }
 
                 try {
-                    const auto fileId = extract_accession_string(
+                    const auto fileAccession = extract_accession_string(
                                             filename, sequence_id_type::acc_ver);
 
-                    const taxon_id fileTaxId = find_taxon_id(sequ2taxid, fileId);
+                    const taxon_id fileTaxId = find_taxon_id(sequ2taxid, fileAccession);
+
+                    if(infoLvl == info_level::verbose) {
+                        std::lock_guard<std::mutex> lock(outputMtx);
+                        cout << "    accession '" << fileAccession
+                             << "' -> taxid " << fileTaxId << endl;
+                    }
 
                     sequence_reader reader{filename};
 

diff --git a/src/taxonomy_io.cpp b/src/taxonomy_io.cpp
@@ -237,9 +237,11 @@ void read_sequence_to_taxon_id_mapping(const string& mappingFile,
                 string header;
                 getline(is, header);
                 std::istringstream hs(header);
-                //get rid of comment chars
-                hs >> header;
                 while(hs >> header) {
+                    // handle comment chars
+                    if (header.size() == 1 && header[0] == '#') {
+                        hs >> header;
+                    }
                     if(header == "taxid") {
                         taxcol = col;
                     }

diff --git a/src/version.h b/src/version.h
@@ -24,11 +24,11 @@
 #define MC_VERSION_H_
 
 
-#define MC_VERSION 20230309
+#define MC_VERSION 20240229
 
 #define MC_DB_VERSION 20200820
 
-#define MC_VERSION_STRING "2.3.1"
+#define MC_VERSION_STRING "2.3.2"
 
 
 #endif