Accommodate '@' as a valid character in the quality strings.

This requires us to use a length-based approach to finding the end
of a quality string, given that '@' is not a reliable terminator.

CMakeLists.txt

@@ -1,7 +1,7 @@
 cmake_minimum_required(VERSION 3.24)
 
 project(kaori
-    VERSION 1.1.0
+    VERSION 1.1.1
     DESCRIPTION "Header-only C++ library for screen counting"
     LANGUAGES CXX)
 

include/kaori/FastqReader.hpp

@@ -29,13 +29,7 @@ class FastqReader {
     FastqReader(byteme::Reader* p) : pb(p) {
         sequence.reserve(200);
         name.reserve(200);
-
         okay = pb.valid();
-        if (okay) {
-            if (pb.get() != '@') {
-                throw std::runtime_error("first line containing FASTQ name should start with '@'");
-            }
-        }
     }
 
     /**
@@ -55,7 +49,12 @@ class FastqReader {
 
         // Processing the name. This should be on a single line, hopefully.
         name.clear();
-        char val = advance_and_check();
+        char val = pb.get();
+        if (val != '@') {
+            throw std::runtime_error("read name should start with '@' (starting line " + std::to_string(init_line + 1) + ")");
+        }
+
+        val = advance_and_check();
         while (!std::isspace(val)) {
             name.push_back(val);
             val = advance_and_check();
@@ -84,26 +83,29 @@ class FastqReader {
         } 
         ++line_count;
 
-        // Processing the qualities. Extraction is allowed to fail if we're at the
-        // end of the file.
-        size_t qual_length = 0;
+        // Processing the qualities. Extraction is allowed to fail if we're at
+        // the end of the file. Note that we can't check for '@' as a
+        // delimitor, as this can be a valid score, so instead we check at each
+        // newline whether we've reached the specified length, and quit if so.
+        size_t qual_length = 0, seq_length = sequence.size();
         okay = false;
+
         while (pb.advance()) {
             val = pb.get();
-            if (val == '@') {
-                okay = true;
-                break;
-            }
             if (val != '\n') {
                 ++qual_length;
+            } else if (qual_length >= seq_length) {
+                okay = pb.advance(); // sneak past the newline.
+                break;
             }
         }
-        ++line_count;
 
-        if (qual_length != sequence.size()) {
+        if (qual_length != seq_length) {
             throw std::runtime_error("non-equal lengths for quality and sequence strings (starting line " + std::to_string(init_line + 1) + ")");
         }
 
+        ++line_count;
+
         return true;
     }
 

tests/src/FastqReader.cpp

@@ -51,29 +51,42 @@ TEST(BasicTests, MultipleEntries) {
 }
 
 TEST(BasicTests, MultiLineEntries) {
-    std::string buffer = "@FOO\nA\nCG\nTGCA\n+\n!!\n!!\n!!!";
-    byteme::RawBufferReader reader(reinterpret_cast<const unsigned char*>(buffer.c_str()), buffer.size());
-    kaori::FastqReader fq(&reader);
+    for (size_t i = 0; i < 2; ++i) {
+        std::string buffer = "@FOO\nA\nCG\nTGCA\n+\n!!\n!!\n!!!\n@ARG\nACACGGT\nC\n+\n@@@\n@\n@@@@";
+        if (i == 2) {
+            buffer += '\n';
+        }
 
-    EXPECT_TRUE(fq());
-    const auto& name = fq.get_name();
-    EXPECT_EQ(std::string(name.begin(), name.end()), "FOO");
-    const auto& seq = fq.get_sequence();
-    EXPECT_EQ(std::string(seq.begin(), seq.end()), "ACGTGCA");
+        byteme::RawBufferReader reader(reinterpret_cast<const unsigned char*>(buffer.c_str()), buffer.size());
+        kaori::FastqReader fq(&reader);
 
-    EXPECT_FALSE(fq());
+        EXPECT_TRUE(fq());
+        const auto& name = fq.get_name();
+        EXPECT_EQ(std::string(name.begin(), name.end()), "FOO");
+        const auto& seq = fq.get_sequence();
+        EXPECT_EQ(std::string(seq.begin(), seq.end()), "ACGTGCA");
+
+        EXPECT_TRUE(fq());
+        const auto& name2 = fq.get_name();
+        EXPECT_EQ(std::string(name2.begin(), name2.end()), "ARG");
+        const auto& seq2 = fq.get_sequence();
+        EXPECT_EQ(std::string(seq2.begin(), seq2.end()), "ACACGGTC");
+
+        EXPECT_FALSE(fq());
+    }
 }
 
 TEST(BasicTests, Errors) {
     {
         std::string buffer = "FOO";
         byteme::RawBufferReader reader(reinterpret_cast<const unsigned char*>(buffer.c_str()), buffer.size());
+        kaori::FastqReader fq(&reader);
 
         EXPECT_ANY_THROW({
             try {
-                kaori::FastqReader fq(&reader);
+                fq();
             } catch (std::exception& e) {
-                EXPECT_TRUE(std::string(e.what()).find("first line") != std::string::npos);
+                EXPECT_TRUE(std::string(e.what()).find("read name should start") != std::string::npos);
                 throw e;
             }
         });
@@ -95,7 +108,25 @@ TEST(BasicTests, Errors) {
     }
 
     {
-        std::string buffer = "@FOO\nAC\n+\n!!\n@WHEE\nACGT\n+\n!!";
+        std::string buffer = "@FOO\nAC\n+\n!!\n@WHEE\nACGT\n+\n!!"; // too short.
+        byteme::RawBufferReader reader(reinterpret_cast<const unsigned char*>(buffer.c_str()), buffer.size());
+        kaori::FastqReader fq(&reader);
+        fq();
+
+        EXPECT_ANY_THROW({
+            try {
+                fq();
+            } catch (std::exception& e) {
+                std::string msg(e.what());
+                EXPECT_TRUE(msg.find("non-equal lengths") != std::string::npos);
+                EXPECT_TRUE(msg.find("line 5") != std::string::npos);
+                throw e;
+            }
+        });
+    }
+
+    {
+        std::string buffer = "@FOO\nAC\n+\n!!\n@WHEE\nACGT\n+\n!!!@@!!@\n"; // too long.
         byteme::RawBufferReader reader(reinterpret_cast<const unsigned char*>(buffer.c_str()), buffer.size());
         kaori::FastqReader fq(&reader);
         fq();
@@ -111,6 +142,24 @@ TEST(BasicTests, Errors) {
             }
         });
     }
+
+    {
+        std::string buffer = "@FOO\nAC\n+\n!!\nWHEE";
+        byteme::RawBufferReader reader(reinterpret_cast<const unsigned char*>(buffer.c_str()), buffer.size());
+        kaori::FastqReader fq(&reader);
+        fq();
+
+        EXPECT_ANY_THROW({
+            try {
+                fq();
+            } catch (std::exception& e) {
+                std::string msg(e.what());
+                EXPECT_TRUE(msg.find("should start") != std::string::npos);
+                EXPECT_TRUE(msg.find("line 5") != std::string::npos);
+                throw e;
+            }
+        });
+    }
 }
 
 class FastqReaderFileTest : public testing::TestWithParam<int> {};