[skip ci] Update .gitignore

.gitignore

@@ -6,5 +6,3 @@ build/
 hbuild/
 subprojects/*
 !subprojects/*.wrap
-ssp.cpp
-ssp.bin

include/ss/common.hpp

@@ -12,6 +12,7 @@ using string_range = std::pair<const char*, const char*>;
 using split_data = std::vector<string_range>;
 
 constexpr inline auto default_delimiter = ",";
+constexpr static auto get_line_initial_buffer_size = 128;
 
 template <bool StringError>
 inline void assert_string_error_defined() {
@@ -46,7 +47,7 @@ inline ssize_t get_line_file(char** lineptr, size_t* n, FILE* stream) {
     }
 
     if (*lineptr == nullptr) {
-        *lineptr = static_cast<char*>(malloc(128));
+        *lineptr = static_cast<char*>(malloc(get_line_initial_buffer_size));
         if (*lineptr == nullptr) {
             return -1;
         }
@@ -57,8 +58,8 @@ inline ssize_t get_line_file(char** lineptr, size_t* n, FILE* stream) {
     while (c != EOF) {
         if (pos + 1 >= *n) {
             size_t new_size = *n + (*n >> 2);
-            if (new_size < 128) {
-                new_size = 128;
+            if (new_size < get_line_initial_buffer_size) {
+                new_size = get_line_initial_buffer_size;
             }
             char* new_ptr = static_cast<char*>(
                 realloc(static_cast<void*>(*lineptr), new_size));

include/ss/parser.hpp

@@ -52,11 +52,16 @@ public:
            const std::string& delim = ss::default_delimiter)
         : file_name_{"buffer line"},
           reader_{csv_data_buffer, csv_data_size, delim} {
-        read_line();
-        if constexpr (ignore_header) {
-            ignore_next();
+        if (csv_data_buffer) {
+            read_line();
+            if constexpr (ignore_header) {
+                ignore_next();
+            } else {
+                raw_header_ = reader_.get_buffer();
+            }
         } else {
-            raw_header_ = reader_.get_buffer();
+            handle_error_null_buffer();
+            eof_ = true;
         }
     }
 
@@ -524,6 +529,19 @@ private:
         }
     }
 
+    void handle_error_null_buffer() {
+        constexpr static auto error_msg = " received null data buffer";
+
+        if constexpr (string_error) {
+            error_.clear();
+            error_.append(file_name_).append(error_msg);
+        } else if constexpr (throw_on_error) {
+            throw ss::exception{file_name_ + error_msg};
+        } else {
+            error_ = true;
+        }
+    }
+
     void handle_error_file_not_open() {
         constexpr static auto error_msg = " could not be opened";
 
@@ -728,19 +746,14 @@ private:
             size_t pos;
             int c;
 
-            // TODO remove check
-            if (lineptr == nullptr || buffer == nullptr || n == nullptr) {
-                return -1;
-            }
-
             if (curr_char >= csv_data_size) {
                 return -1;
             }
             c = buffer[curr_char++];
 
-            // TODO maybe remove this too
             if (*lineptr == nullptr) {
-                *lineptr = static_cast<char*>(malloc(128));
+                *lineptr =
+                    static_cast<char*>(malloc(get_line_initial_buffer_size));
                 if (*lineptr == nullptr) {
                     return -1;
                 }
@@ -751,13 +764,11 @@ private:
             while (curr_char <= csv_data_size) {
                 if (pos + 1 >= *n) {
                     size_t new_size = *n + (*n >> 2);
-                    // TODO maybe remove this too
-                    if (new_size < 128) {
-                        new_size = 128;
+                    if (new_size < get_line_initial_buffer_size) {
+                        new_size = get_line_initial_buffer_size;
                     }
                     char* new_ptr = static_cast<char*>(
                         realloc(static_cast<void*>(*lineptr), new_size));
-                    // TODO check for failed malloc in the callee
                     if (new_ptr == nullptr) {
                         return -1;
                     }
@@ -924,11 +935,11 @@ private:
         }
 
         void realloc_concat(char*& first, size_t& first_size,
-                            const char* const second, size_t second_size) {
-            // TODO make buffer_size an argument
-            next_line_buffer_size_ = first_size + second_size + 3;
+                            size_t& buffer_size, const char* const second,
+                            size_t second_size) {
+            buffer_size = first_size + second_size + 3;
             auto new_first = static_cast<char*>(
-                realloc(static_cast<void*>(first), next_line_buffer_size_));
+                realloc(static_cast<void*>(first), buffer_size));
             if (!first) {
                 throw std::bad_alloc{};
             }
@@ -958,7 +969,8 @@ private:
 
             ++line_number_;
             size_t next_size = remove_eol(helper_buffer_, next_ssize);
-            realloc_concat(buffer, size, helper_buffer_, next_size);
+            realloc_concat(buffer, size, next_line_buffer_size_, helper_buffer_,
+                           next_size);
             return true;
         }
 

include/ss/setup.hpp

@@ -109,7 +109,7 @@ struct get_matcher<Matcher, T, Ts...> {
     struct is_matcher : is_instance_of_matcher<U, Matcher> {};
 
     static_assert(count_v<is_matcher, T, Ts...> <= 1,
-                  "the same matcher cannot"
+                  "the same matcher cannot "
                   "be defined multiple times");
     using type = std::conditional_t<is_matcher<T>::value, T,
                                     typename get_matcher<Matcher, Ts...>::type>;

ssp.hpp

@@ -625,6 +625,7 @@ using string_range = std::pair<const char*, const char*>;
 using split_data = std::vector<string_range>;
 
 constexpr inline auto default_delimiter = ",";
+constexpr static auto get_line_initial_buffer_size = 128;
 
 template <bool StringError>
 inline void assert_string_error_defined() {
@@ -659,7 +660,7 @@ inline ssize_t get_line_file(char** lineptr, size_t* n, FILE* stream) {
     }
 
     if (*lineptr == nullptr) {
-        *lineptr = static_cast<char*>(malloc(128));
+        *lineptr = static_cast<char*>(malloc(get_line_initial_buffer_size));
         if (*lineptr == nullptr) {
             return -1;
         }
@@ -670,8 +671,8 @@ inline ssize_t get_line_file(char** lineptr, size_t* n, FILE* stream) {
     while (c != EOF) {
         if (pos + 1 >= *n) {
             size_t new_size = *n + (*n >> 2);
-            if (new_size < 128) {
-                new_size = 128;
+            if (new_size < get_line_initial_buffer_size) {
+                new_size = get_line_initial_buffer_size;
             }
             char* new_ptr = static_cast<char*>(
                 realloc(static_cast<void*>(*lineptr), new_size));
@@ -803,7 +804,7 @@ struct get_matcher<Matcher, T, Ts...> {
     struct is_matcher : is_instance_of_matcher<U, Matcher> {};
 
     static_assert(count_v<is_matcher, T, Ts...> <= 1,
-                  "the same matcher cannot"
+                  "the same matcher cannot "
                   "be defined multiple times");
     using type = std::conditional_t<is_matcher<T>::value, T,
                                     typename get_matcher<Matcher, Ts...>::type>;
@@ -2178,11 +2179,16 @@ public:
            const std::string& delim = ss::default_delimiter)
         : file_name_{"buffer line"},
           reader_{csv_data_buffer, csv_data_size, delim} {
-        read_line();
-        if constexpr (ignore_header) {
-            ignore_next();
+        if (csv_data_buffer) {
+            read_line();
+            if constexpr (ignore_header) {
+                ignore_next();
+            } else {
+                raw_header_ = reader_.get_buffer();
+            }
         } else {
-            raw_header_ = reader_.get_buffer();
+            handle_error_null_buffer();
+            eof_ = true;
         }
     }
 
@@ -2650,6 +2656,19 @@ private:
         }
     }
 
+    void handle_error_null_buffer() {
+        constexpr static auto error_msg = " received null data buffer";
+
+        if constexpr (string_error) {
+            error_.clear();
+            error_.append(file_name_).append(error_msg);
+        } else if constexpr (throw_on_error) {
+            throw ss::exception{file_name_ + error_msg};
+        } else {
+            error_ = true;
+        }
+    }
+
     void handle_error_file_not_open() {
         constexpr static auto error_msg = " could not be opened";
 
@@ -2854,19 +2873,14 @@ private:
             size_t pos;
             int c;
 
-            // TODO remove check
-            if (lineptr == nullptr || buffer == nullptr || n == nullptr) {
-                return -1;
-            }
-
             if (curr_char >= csv_data_size) {
                 return -1;
             }
             c = buffer[curr_char++];
 
-            // TODO maybe remove this too
             if (*lineptr == nullptr) {
-                *lineptr = static_cast<char*>(malloc(128));
+                *lineptr =
+                    static_cast<char*>(malloc(get_line_initial_buffer_size));
                 if (*lineptr == nullptr) {
                     return -1;
                 }
@@ -2877,13 +2891,11 @@ private:
             while (curr_char <= csv_data_size) {
                 if (pos + 1 >= *n) {
                     size_t new_size = *n + (*n >> 2);
-                    // TODO maybe remove this too
-                    if (new_size < 128) {
-                        new_size = 128;
+                    if (new_size < get_line_initial_buffer_size) {
+                        new_size = get_line_initial_buffer_size;
                     }
                     char* new_ptr = static_cast<char*>(
                         realloc(static_cast<void*>(*lineptr), new_size));
-                    // TODO check for failed malloc in the callee
                     if (new_ptr == nullptr) {
                         return -1;
                     }
@@ -3050,11 +3062,11 @@ private:
         }
 
         void realloc_concat(char*& first, size_t& first_size,
-                            const char* const second, size_t second_size) {
-            // TODO make buffer_size an argument
-            next_line_buffer_size_ = first_size + second_size + 3;
+                            size_t& buffer_size, const char* const second,
+                            size_t second_size) {
+            buffer_size = first_size + second_size + 3;
             auto new_first = static_cast<char*>(
-                realloc(static_cast<void*>(first), next_line_buffer_size_));
+                realloc(static_cast<void*>(first), buffer_size));
             if (!first) {
                 throw std::bad_alloc{};
             }
@@ -3084,7 +3096,8 @@ private:
 
             ++line_number_;
             size_t next_size = remove_eol(helper_buffer_, next_ssize);
-            realloc_concat(buffer, size, helper_buffer_, next_size);
+            realloc_concat(buffer, size, next_line_buffer_size_, helper_buffer_,
+                           next_size);
             return true;
         }
 

test/CMakeLists.txt

@@ -35,8 +35,8 @@ enable_testing()
 foreach(name IN ITEMS test_splitter test_parser1_1 test_parser1_2
                       test_parser1_3 test_parser1_4 test_converter
                       test_extractions test_parser2_1 test_parser2_2
-                      test_parser2_3 test_parser2_4
-                      test_extractions_without_fast_float)
+                      test_parser2_3 test_parser2_4 test_parser2_5
+                      test_parser2_6 test_extractions_without_fast_float)
     add_executable("${name}" "${name}.cpp")
     target_link_libraries("${name}" PRIVATE ssp::ssp fast_float
                                             doctest::doctest)

test/meson.build

@@ -13,6 +13,8 @@ tests = [
   'parser2_2',
   'parser2_3',
   'parser2_4',
+  'parser2_5',
+  'parser2_6',
   'extractions_without_fast_float',
 ]
 

test/test_helpers.hpp

@@ -12,6 +12,11 @@
 #include <doctest.h>
 #endif
 
+namespace ss {
+template <typename... Ts>
+class parser;
+} /* ss */
+
 namespace {
 struct buffer {
     std::string data_;
@@ -34,7 +39,7 @@ struct buffer {
 
 [[maybe_unused]] inline buffer buff;
 
-std::string time_now_rand() {
+[[maybe_unused]] std::string time_now_rand() {
     srand(time(nullptr));
     std::stringstream ss;
     auto t = std::time(nullptr);
@@ -115,8 +120,8 @@ struct unique_file_name {
     }
 
 template <typename T>
-std::vector<std::vector<T>> vector_combinations(const std::vector<T>& v,
-                                                size_t n) {
+[[maybe_unused]] std::vector<std::vector<T>> vector_combinations(
+    const std::vector<T>& v, size_t n) {
     std::vector<std::vector<T>> ret;
     if (n <= 1) {
         for (const auto& i : v) {
@@ -134,4 +139,54 @@ std::vector<std::vector<T>> vector_combinations(const std::vector<T>& v,
     }
     return ret;
 }
+
+[[maybe_unused]] std::string make_buffer(const std::string& file_name) {
+    std::ifstream in{file_name, std::ios::binary};
+    std::string tmp;
+    std::string out;
+
+    auto copy_if_whitespaces = [&] {
+        std::string matches = "\n\r\t ";
+        while (std::any_of(matches.begin(), matches.end(),
+                           [&](auto c) { return in.peek() == c; })) {
+            if (in.peek() == '\r') {
+                out += "\r\n";
+                in.ignore(2);
+            } else {
+                out += std::string{static_cast<char>(in.peek())};
+                in.ignore(1);
+            }
+        }
+    };
+
+    out.reserve(sizeof(out) + 1);
+
+    copy_if_whitespaces();
+    while (in >> tmp) {
+        out += tmp;
+        copy_if_whitespaces();
+    }
+    return out;
+}
+
+template <bool buffer_mode, typename... Ts>
+[[maybe_unused]] std::tuple<ss::parser<Ts...>, std::string> make_parser(
+    const std::string& file_name, const std::string& delim = "") {
+    if (buffer_mode) {
+        auto buffer = make_buffer(file_name);
+        if (delim.empty()) {
+            return {ss::parser<Ts...>{buffer.data(), buffer.size()},
+                    std::move(buffer)};
+        } else {
+            return {ss::parser<Ts...>{buffer.data(), buffer.size(), delim},
+                    std::move(buffer)};
+        }
+    } else {
+        if (delim.empty()) {
+            return {ss::parser<Ts...>{file_name}, std::string{}};
+        } else {
+            return {ss::parser<Ts...>{file_name, delim}, std::string{}};
+        }
+    }
+}
 } /* namespace */

test/test_parser1.hpp

@@ -105,47 +105,4 @@ static void make_and_write(const std::string& file_name,
     }
 }
 
-std::string make_buffer(const std::string& file_name) {
-    std::ifstream in{file_name, std::ios::binary};
-    std::string tmp;
-    std::string out;
-    out.reserve(sizeof(out) + 1);
-    while (in >> tmp) {
-        out += tmp;
-        std::string matches = "\n\r\t ";
-        while (std::any_of(matches.begin(), matches.end(),
-                           [&](auto c) { return in.peek() == c; })) {
-            if (in.peek() == '\r') {
-                out += "\r\n";
-                in.ignore(2);
-            } else {
-                out += std::string{static_cast<char>(in.peek())};
-                in.ignore(1);
-            }
-        }
-    }
-    return out;
-}
-
-template <bool buffer_mode, typename... Ts>
-std::tuple<ss::parser<Ts...>, std::string> make_parser(
-    const std::string& file_name, const std::string& delim = "") {
-    if (buffer_mode) {
-        auto buffer = make_buffer(file_name);
-        if (delim.empty()) {
-            return {ss::parser<Ts...>{buffer.data(), buffer.size()},
-                    std::move(buffer)};
-        } else {
-            return {ss::parser<Ts...>{buffer.data(), buffer.size(), delim},
-                    std::move(buffer)};
-        }
-    } else {
-        if (delim.empty()) {
-            return {ss::parser<Ts...>{file_name}, std::string{}};
-        } else {
-            return {ss::parser<Ts...>{file_name, delim}, std::string{}};
-        }
-    }
-}
-
 } /* namespace */

test/test_parser1_1.cpp

@@ -21,6 +21,25 @@ TEST_CASE("test file not found") {
     }
 }
 
+TEST_CASE("test null buffer") {
+    {
+        ss::parser p{nullptr, 10, ","};
+        CHECK_FALSE(p.valid());
+    }
+
+    {
+        ss::parser<ss::string_error> p{nullptr, 10, ","};
+        CHECK_FALSE(p.valid());
+    }
+
+    try {
+        ss::parser<ss::throw_on_error> p{nullptr, 10, ","};
+        FAIL("Expected exception...");
+    } catch (const std::exception& e) {
+        CHECK_FALSE(std::string{e.what()}.empty());
+    }
+}
+
 template <bool buffer_mode, typename... Ts>
 void test_various_cases() {
     unique_file_name f{"test_parser"};

test/test_parser1_4.cpp

@@ -7,17 +7,13 @@ template <typename T, typename... Us>
 struct has_type<T, std::tuple<Us...>>
     : std::disjunction<std::is_same<T, Us>...> {};
 
-static inline void check_size(size_t size1, size_t size2) {
-    CHECK_EQ(size1, size2);
-}
-
-template <typename Setup, typename... Ts>
+template <bool buffer_mode, typename Setup, typename... Ts>
 static void test_fields_impl(const std::string file_name,
                              const std::vector<X>& data,
                              const std::vector<std::string>& fields) {
     using CaseType = std::tuple<Ts...>;
 
-    ss::parser<Setup> p{file_name, ","};
+    auto [p, _] = make_parser<buffer_mode, Setup>(file_name, ",");
     CHECK_FALSE(p.field_exists("Unknown"));
     p.use_fields(fields);
     std::vector<CaseType> i;
@@ -26,7 +22,7 @@ static void test_fields_impl(const std::string file_name,
         i.push_back(a);
     }
 
-    check_size(i.size(), data.size());
+    CHECK_EQ(i.size(), data.size());
     for (size_t j = 0; j < i.size(); ++j) {
         if constexpr (has_type<int, CaseType>::value) {
             CHECK_EQ(std::get<int>(i[j]), data[j].i);
@@ -43,11 +39,16 @@ static void test_fields_impl(const std::string file_name,
 template <typename... Ts>
 static void test_fields(const std::string file_name, const std::vector<X>& data,
                         const std::vector<std::string>& fields) {
-    test_fields_impl<ss::setup<>, Ts...>(file_name, data, fields);
-    test_fields_impl<ss::setup<ss::string_error>, Ts...>(file_name, data,
-                                                         fields);
-    test_fields_impl<ss::setup<ss::throw_on_error>, Ts...>(file_name, data,
-                                                           fields);
+    test_fields_impl<false, ss::setup<>, Ts...>(file_name, data, fields);
+    test_fields_impl<false, ss::setup<ss::string_error>, Ts...>(file_name, data,
+                                                                fields);
+    test_fields_impl<false, ss::setup<ss::throw_on_error>, Ts...>(file_name,
+                                                                  data, fields);
+    test_fields_impl<true, ss::setup<>, Ts...>(file_name, data, fields);
+    test_fields_impl<true, ss::setup<ss::string_error>, Ts...>(file_name, data,
+                                                               fields);
+    test_fields_impl<true, ss::setup<ss::throw_on_error>, Ts...>(file_name,
+                                                                 data, fields);
 }
 
 TEST_CASE("parser test various cases with header") {
@@ -196,34 +197,37 @@ TEST_CASE("parser test various cases with header") {
     test_fields<double, int, str>(o, d, {Dbl, Int, Str});
 }
 
-template <typename... Ts>
+template <bool buffer_mode, typename... Ts>
 void test_invalid_fields_impl(const std::vector<std::string>& lines,
                               const std::vector<std::string>& fields) {
     unique_file_name f{"test_parser"};
-    std::ofstream out{f.name};
-    for (const auto& line : lines) {
-        out << line << std::endl;
+    {
+        std::ofstream out{f.name};
+        for (const auto& line : lines) {
+            out << line << std::endl;
+        }
     }
-    out.close();
 
     {
         // No fields specified
-        ss::parser<Ts...> p{f.name, ","};
-        auto command = [&] { p.use_fields(); };
+        auto [p, _] = make_parser<buffer_mode, Ts...>(f.name, ",");
+        auto command = [&p = p] { p.use_fields(); };
         expect_error_on_command(p, command);
     }
 
     {
         // Unknown field
-        ss::parser<Ts...> p{f.name, ","};
-        auto command = [&] { p.use_fields("Unknown"); };
+        auto [p, _] = make_parser<buffer_mode, Ts...>(f.name, ",");
+        auto command = [&p = p] { p.use_fields("Unknown"); };
         expect_error_on_command(p, command);
     }
 
     {
         // Field used multiple times
-        ss::parser<Ts...> p{f.name, ","};
-        auto command = [&] { p.use_fields(fields.at(0), fields.at(0)); };
+        auto [p, _] = make_parser<buffer_mode, Ts...>(f.name, ",");
+        auto command = [&p = p, &fields = fields] {
+            p.use_fields(fields.at(0), fields.at(0));
+        };
         if (!fields.empty()) {
             expect_error_on_command(p, command);
         }
@@ -231,8 +235,8 @@ void test_invalid_fields_impl(const std::vector<std::string>& lines,
 
     {
         // Mapping out of range
-        ss::parser<Ts...> p{f.name, ","};
-        auto command = [&] {
+        auto [p, _] = make_parser<buffer_mode, Ts...>(f.name, ",");
+        auto command = [&p = p, &fields = fields] {
             p.use_fields(fields.at(0));
             p.template get_next<std::string, std::string>();
         };
@@ -243,8 +247,8 @@ void test_invalid_fields_impl(const std::vector<std::string>& lines,
 
     {
         // Invalid header
-        ss::parser<Ts...> p{f.name, ","};
-        auto command = [&] { p.use_fields(fields); };
+        auto [p, _] = make_parser<buffer_mode, Ts...>(f.name, ",");
+        auto command = [&p = p, &fields = fields] { p.use_fields(fields); };
 
         if (!fields.empty()) {
             // Pass if there are no duplicates, fail otherwise
@@ -267,9 +271,12 @@ void test_invalid_fields_impl(const std::vector<std::string>& lines,
 template <typename... Ts>
 void test_invalid_fields(const std::vector<std::string>& lines,
                          const std::vector<std::string>& fields) {
-    test_invalid_fields_impl(lines, fields);
-    test_invalid_fields_impl<ss::string_error>(lines, fields);
-    test_invalid_fields_impl<ss::throw_on_error>(lines, fields);
+    test_invalid_fields_impl<false>(lines, fields);
+    test_invalid_fields_impl<false, ss::string_error>(lines, fields);
+    test_invalid_fields_impl<false, ss::throw_on_error>(lines, fields);
+    test_invalid_fields_impl<true>(lines, fields);
+    test_invalid_fields_impl<true, ss::string_error>(lines, fields);
+    test_invalid_fields_impl<true, ss::throw_on_error>(lines, fields);
 }
 
 TEST_CASE("parser test invalid header fields usage") {
@@ -296,7 +303,7 @@ TEST_CASE("parser test invalid header fields usage") {
     test_invalid_fields({"Int,String,Int", "1,hi,3"}, {"Int", "String", "Int"});
 }
 
-template <typename... Ts>
+template <bool buffer_mode, typename... Ts>
 void test_invalid_rows_with_header() {
     unique_file_name f{"test_parser"};
     {
@@ -311,7 +318,7 @@ void test_invalid_rows_with_header() {
     }
 
     {
-        ss::parser<Ts...> p{f.name};
+        auto [p, _] = make_parser<buffer_mode, Ts...>(f.name);
 
         p.use_fields("Int", "String", "Double");
         using data = std::tuple<int, std::string, double>;
@@ -337,7 +344,7 @@ void test_invalid_rows_with_header() {
     }
 
     {
-        ss::parser<Ts...> p{f.name};
+        auto [p, _] = make_parser<buffer_mode, Ts...>(f.name);
 
         p.use_fields("Double", "Int");
         using data = std::tuple<double, int>;
@@ -361,7 +368,7 @@ void test_invalid_rows_with_header() {
     }
 
     {
-        ss::parser<Ts...> p{f.name};
+        auto [p, _] = make_parser<buffer_mode, Ts...>(f.name);
 
         p.use_fields("String", "Double");
         using data = std::tuple<std::string, double>;
@@ -389,12 +396,15 @@ void test_invalid_rows_with_header() {
 }
 
 TEST_CASE("parser test invalid rows with header") {
-    test_invalid_rows_with_header();
-    test_invalid_rows_with_header<ss::string_error>();
-    test_invalid_rows_with_header<ss::throw_on_error>();
+    test_invalid_rows_with_header<false>();
+    test_invalid_rows_with_header<false, ss::string_error>();
+    test_invalid_rows_with_header<false, ss::throw_on_error>();
+    test_invalid_rows_with_header<true>();
+    test_invalid_rows_with_header<true, ss::string_error>();
+    test_invalid_rows_with_header<true, ss::throw_on_error>();
 }
 
-template <typename... Ts>
+template <bool buffer_mode, typename... Ts>
 void test_ignore_empty_impl(const std::vector<X>& data) {
     unique_file_name f{"test_parser"};
     make_and_write(f.name, data);
@@ -407,7 +417,8 @@ void test_ignore_empty_impl(const std::vector<X>& data) {
     }
 
     {
-        ss::parser<ss::ignore_empty, Ts...> p{f.name, ","};
+        auto [p, _] =
+            make_parser<buffer_mode, ss::ignore_empty, Ts...>(f.name, ",");
 
         std::vector<X> i;
         for (const auto& a : p.template iterate<X>()) {
@@ -418,7 +429,7 @@ void test_ignore_empty_impl(const std::vector<X>& data) {
     }
 
     {
-        ss::parser<Ts...> p{f.name, ","};
+        auto [p, _] = make_parser<buffer_mode, Ts...>(f.name, ",");
         std::vector<X> i;
         size_t n = 0;
         while (!p.eof()) {
@@ -441,9 +452,12 @@ void test_ignore_empty_impl(const std::vector<X>& data) {
 
 template <typename... Ts>
 void test_ignore_empty(const std::vector<X>& data) {
-    test_ignore_empty_impl(data);
-    test_ignore_empty_impl<ss::string_error>(data);
-    test_ignore_empty_impl<ss::throw_on_error>(data);
+    test_ignore_empty_impl<false>(data);
+    test_ignore_empty_impl<false, ss::string_error>(data);
+    test_ignore_empty_impl<false, ss::throw_on_error>(data);
+    test_ignore_empty_impl<true>(data);
+    test_ignore_empty_impl<true, ss::string_error>(data);
+    test_ignore_empty_impl<true, ss::throw_on_error>(data);
 }
 
 TEST_CASE("parser test various cases with empty lines") {

test/test_parser2.hpp

@@ -121,7 +121,7 @@ column make_column(const std::string& input_header,
 }
 
 [[maybe_unused]] void replace_all2(std::string& s, const std::string& old_value,
-                  const std::string& new_value) {
+                                   const std::string& new_value) {
     for (size_t i = 0; i < 999; ++i) {
         size_t pos = s.find(old_value);
         if (pos == std::string::npos) {
@@ -257,7 +257,8 @@ std::vector<std::string> generate_csv_data(const std::vector<field>& data,
 }
 
 [[maybe_unused]] void write_to_file(const std::vector<std::string>& data,
-                   const std::string& delim, const std::string& file_name) {
+                                    const std::string& delim,
+                                    const std::string& file_name) {
     std::ofstream out{file_name, std::ios_base::app};
     std::string line;
     for (size_t i = 0; i < data.size(); ++i) {
@@ -299,7 +300,7 @@ std::vector<std::string> generate_csv_data(const std::vector<field>& data,
         CHECK(V1 == V2);                                                       \
     }
 
-template <typename... Ts>
+template <bool buffer_mode, typename... Ts>
 void test_data_combinations(const std::vector<column>& input_data,
                             const std::string& delim, bool include_header) {
     using setup = ss::setup<Ts...>;
@@ -388,7 +389,7 @@ void test_data_combinations(const std::vector<column>& input_data,
     }
 
     for (const auto& layout : unique_layout_combinations) {
-        ss::parser<setup> p{f.name, delim};
+        auto [p, _] = make_parser<buffer_mode, setup>(f.name, delim);
 
         if (include_header && !setup::ignore_header) {
             std::vector<std::string> fields;
@@ -409,7 +410,7 @@ void test_data_combinations(const std::vector<column>& input_data,
             REQUIRE(p.valid());
         }
 
-        auto check_error = [&p] {
+        auto check_error = [&p = p] {
             CHECK(p.valid());
             if (!p.valid()) {
                 if constexpr (setup::string_error) {
@@ -570,8 +571,14 @@ void test_option_combinations() {
                  {columns0, columns1, columns2, columns3, columns4, columns5,
                   columns6, columns7}) {
                 try {
-                    test_data_combinations<Ts...>(columns, delimiter, false);
-                    test_data_combinations<Ts...>(columns, delimiter, true);
+                    test_data_combinations<false, Ts...>(columns, delimiter,
+                                                         false);
+                    test_data_combinations<false, Ts...>(columns, delimiter,
+                                                         true);
+                    test_data_combinations<true, Ts...>(columns, delimiter,
+                                                        false);
+                    test_data_combinations<true, Ts...>(columns, delimiter,
+                                                        true);
                 } catch (std::exception& e) {
                     std::cout << typeid(ss::parser<Ts...>).name() << std::endl;
                     FAIL_CHECK(std::string{e.what()});
@@ -616,6 +623,7 @@ void test_option_combinations3() {
 
 } /* namespace */
 
+// Tests split into multiple compilation units
 #if 0
 
 TEST_CASE("parser test various cases version 2 segment 1") {
@@ -627,25 +635,33 @@ TEST_CASE("parser test various cases version 2 segment 1") {
     using multiline_r = ss::multiline_restricted<10>;
     using trimr = ss::trim_right<' '>;
     using triml = ss::trim_left<' '>;
+    using trim = ss::trim<' '>;
 
     // segment 1
     test_option_combinations3<>();
     test_option_combinations3<escape>();
-    test_option_combinations3<quote>();
 
     // segment 2
+    test_option_combinations3<quote>();
     test_option_combinations3<escape, quote>();
+
+    // segment 3
     test_option_combinations3<escape, multiline>();
     test_option_combinations3<quote, multiline>();
 
-    // segment 3
+    // segment 4
     test_option_combinations3<escape, quote, multiline>();
     test_option_combinations3<escape, quote, multiline_r>();
 
-    // segment 4
+    // segment 5
     test_option_combinations<escape, quote, multiline, triml>();
     test_option_combinations<escape, quote, multiline, trimr>();
+
+    // segment 6
+    test_option_combinations3<escape, quote, multiline>();
+    test_option_combinations3<escape, quote, multiline, trim>();
 #else
+
     test_option_combinations3<escape, quote, multiline>();
 #endif
 }

test/test_parser2_1.cpp

@@ -3,12 +3,10 @@
 
 TEST_CASE("parser test various cases version 2 segment 1") {
 #ifdef CMAKE_GITHUB_CI
-    using quote = ss::quote<'"'>;
     using escape = ss::escape<'\\'>;
 
     test_option_combinations3<>();
     test_option_combinations3<escape>();
-    test_option_combinations3<quote>();
 #endif
 }
 

test/test_parser2_2.cpp

@@ -5,11 +5,9 @@ TEST_CASE("parser test various cases version 2 segment 2") {
 #ifdef CMAKE_GITHUB_CI
     using quote = ss::quote<'"'>;
     using escape = ss::escape<'\\'>;
-    using multiline = ss::multiline;
 
+    test_option_combinations3<quote>();
     test_option_combinations3<escape, quote>();
-    test_option_combinations3<escape, multiline>();
-    test_option_combinations3<quote, multiline>();
 #endif
 }
 

test/test_parser2_3.cpp

@@ -6,10 +6,9 @@ TEST_CASE("parser test various cases version 2 segment 3") {
     using quote = ss::quote<'"'>;
     using escape = ss::escape<'\\'>;
     using multiline = ss::multiline;
-    using multiline_r = ss::multiline_restricted<10>;
 
-    test_option_combinations3<escape, quote, multiline>();
-    test_option_combinations3<escape, quote, multiline_r>();
+    test_option_combinations3<escape, multiline>();
+    test_option_combinations3<quote, multiline>();
 #endif
 }
 

test/test_parser2_4.cpp

@@ -2,18 +2,14 @@
 #include "test_parser2.hpp"
 
 TEST_CASE("parser test various cases version 2 segment 4") {
+#ifdef CMAKE_GITHUB_CI
     using quote = ss::quote<'"'>;
     using escape = ss::escape<'\\'>;
     using multiline = ss::multiline;
+    using multiline_r = ss::multiline_restricted<10>;
 
-#ifdef CMAKE_GITHUB_CI
-    using trimr = ss::trim_right<' '>;
-    using triml = ss::trim_left<' '>;
-
-    test_option_combinations<escape, quote, multiline, triml>();
-    test_option_combinations<escape, quote, multiline, trimr>();
-#else
     test_option_combinations3<escape, quote, multiline>();
+    test_option_combinations3<escape, quote, multiline_r>();
 #endif
 }
 

test/test_parser2_5.cpp

@@ -0,0 +1,16 @@
+#define SEGMENT_NAME "segment5"
+#include "test_parser2.hpp"
+
+TEST_CASE("parser test various cases version 2 segment 5") {
+#ifdef CMAKE_GITHUB_CI
+    using quote = ss::quote<'"'>;
+    using escape = ss::escape<'\\'>;
+    using multiline = ss::multiline;
+    using trimr = ss::trim_right<' '>;
+    using triml = ss::trim_left<' '>;
+
+    test_option_combinations<escape, quote, multiline, triml>();
+    test_option_combinations<escape, quote, multiline, trimr>();
+#endif
+}
+

test/test_parser2_6.cpp

@@ -0,0 +1,11 @@
+#define SEGMENT_NAME "segment6"
+#include "test_parser2.hpp"
+
+TEST_CASE("parser test various cases version 2 segment 6") {
+    using quote = ss::quote<'"'>;
+    using escape = ss::escape<'\\'>;
+    using multiline = ss::multiline;
+
+    test_option_combinations3<escape, quote, multiline>();
+}
+