feat: automatically detect file endings for fasta files

include/silo/common/input_stream_wrapper.h

@@ -6,11 +6,13 @@
 #include <iostream>
 #include <memory>
 
+#include <boost/iostreams/filtering_stream.hpp>
+
 namespace silo {
 struct InputStreamWrapper {
   private:
    std::ifstream file;
-   std::unique_ptr<std::istream> input_stream;
+   std::unique_ptr<boost::iostreams::filtering_istream> input_stream;
 
   public:
    explicit InputStreamWrapper(const std::filesystem::path& filename);

include/silo/common/zstdfasta_reader.h

@@ -13,7 +13,7 @@
 namespace silo {
 class ZstdFastaReader {
   private:
-   silo::InputStreamWrapper in_file;
+   std::ifstream in_file;
    std::unique_ptr<silo::ZstdDecompressor> decompressor;
    std::string genome_buffer;
 

src/silo/common/input_stream_wrapper.cpp

@@ -2,26 +2,51 @@
 
 #include <utility>
 
+#include <spdlog/spdlog.h>
 #include <boost/iostreams/detail/error.hpp>
 #include <boost/iostreams/filter/lzma.hpp>
+#include <boost/iostreams/filter/zstd.hpp>
 #include <boost/iostreams/filtering_stream.hpp>
 #include <boost/iostreams/read.hpp>
 
+#include "silo/preprocessing/preprocessing_exception.h"
+
+namespace {
+
+std::filesystem::path appendXZ(const std::filesystem::path& filename) {
+   return {filename.string() + ".xz"};
+}
+
+std::filesystem::path appendZST(const std::filesystem::path& filename) {
+   return {filename.string() + ".zst"};
+}
+
+}  // namespace
+
 namespace silo {
-InputStreamWrapper::InputStreamWrapper(const std::filesystem::path& filename) {
-   if (filename.extension() == ".xz") {
+InputStreamWrapper::InputStreamWrapper(const std::filesystem::path& filename)
+    : input_stream(std::make_unique<boost::iostreams::filtering_istream>()) {
+   if (std::filesystem::exists(filename)) {
+      SPDLOG_INFO("Detected file without specialized ending, processing raw: " + filename.string());
       file = std::ifstream(filename, std::ios::binary);
-      std::unique_ptr<boost::iostreams::filtering_istream> archive =
-         std::make_unique<boost::iostreams::filtering_istream>();
-      archive->push(boost::iostreams::lzma_decompressor());
-      archive->push(file);
-      input_stream = std::move(archive);
+   } else if (std::filesystem::exists(appendXZ(filename))) {
+      SPDLOG_INFO("Detected file-ending .xz for input file " + filename.string());
+      file = std::ifstream(appendXZ(filename), std::ios::binary);
+      input_stream->push(boost::iostreams::lzma_decompressor());
+   } else if (std::filesystem::exists(appendZST(filename))) {
+      SPDLOG_INFO("Detected file-ending .zst for input file " + filename.string());
+      file = std::ifstream(appendZST(filename), std::ios::binary);
+      input_stream->push(boost::iostreams::zstd_decompressor());
    } else {
-      input_stream = make_unique<std::ifstream>(filename.string(), std::ios::binary);
+      throw silo::PreprocessingException(
+         "Cannot find file with name or associated endings (.xz, .zst): " + filename.string()
+      );
    }
+   input_stream->push(file);
 }
 
 std::istream& silo::InputStreamWrapper::getInputStream() const {
    return *input_stream;
 }
+
 }  // namespace silo

src/silo/common/zstdfasta_reader.cpp

@@ -18,7 +18,7 @@ silo::ZstdFastaReader::ZstdFastaReader(
 
 std::optional<std::string> silo::ZstdFastaReader::nextKey() {
    std::string key_with_prefix;
-   if (!getline(in_file.getInputStream(), key_with_prefix)) {
+   if (!getline(in_file, key_with_prefix)) {
       return std::nullopt;
    }
 
@@ -37,12 +37,12 @@ std::optional<std::string> silo::ZstdFastaReader::nextSkipGenome() {
    }
 
    std::string bytestream_length_str;
-   if (!getline(in_file.getInputStream(), bytestream_length_str)) {
+   if (!getline(in_file, bytestream_length_str)) {
       throw FastaFormatException("Missing bytestream length in line following key: " + *key);
    }
    const size_t bytestream_length = std::stoul(bytestream_length_str);
 
-   in_file.getInputStream().ignore(static_cast<std::streamsize>(bytestream_length));
+   in_file.ignore(static_cast<std::streamsize>(bytestream_length));
    return key;
 }
 
@@ -53,16 +53,14 @@ std::optional<std::string> silo::ZstdFastaReader::nextCompressed(std::string& co
    }
 
    std::string bytestream_length_str;
-   if (!getline(in_file.getInputStream(), bytestream_length_str)) {
+   if (!getline(in_file, bytestream_length_str)) {
       throw FastaFormatException("Missing bytestream length in line following key: " + *key);
    }
    const size_t bytestream_length = std::stoul(bytestream_length_str);
 
    compressed_genome.resize(bytestream_length);
-   in_file.getInputStream().read(
-      compressed_genome.data(), static_cast<std::streamsize>(compressed_genome.size())
-   );
-   in_file.getInputStream().ignore(1);
+   in_file.read(compressed_genome.data(), static_cast<std::streamsize>(compressed_genome.size()));
+   in_file.ignore(1);
    return key;
 }
 
@@ -79,6 +77,6 @@ std::optional<std::string> silo::ZstdFastaReader::next(std::string& genome) {
 }
 
 void silo::ZstdFastaReader::reset() {
-   in_file.getInputStream().clear();                  // clear fail and eof bits
-   in_file.getInputStream().seekg(0, std::ios::beg);  // g pointer back to the start
+   in_file.clear();                  // clear fail and eof bits
+   in_file.seekg(0, std::ios::beg);  // g pointer back to the start
 }