Update parser error messages, fix parser tests

* Resolve clang-tidy warnings, update single_header_generator.py

* Update single header test, resolve additional clang-tidy warnings

.github/fuzz/makefile

@@ -0,0 +1,7 @@
+EXE=ssp_fuzz
+
+all:
+	clang++ ${CXXFLAGS} ssp_fuzz.cpp -fsanitize=fuzzer -std=c++17 -o ${EXE}
+
+run:
+	./${EXE} -max_total_time=900

.github/fuzz/ssp_fuzz.cpp

@@ -0,0 +1,81 @@
+#include "../../ssp.hpp"
+#include <filesystem>
+#include <iostream>
+#include <unistd.h>
+
+template <typename... Ts>
+void test_ssp_file_mode(const uint8_t* data, size_t size,
+                        std::string delim = ss::default_delimiter) {
+    std::string file_name = std::filesystem::temp_directory_path().append(
+        "ss_fuzzer" + std::to_string(getpid()) + ".csv");
+    FILE* file = std::fopen(file_name.c_str(), "wb");
+    if (!file) {
+        std::exit(1);
+    }
+    std::fwrite(data, size, 1, file);
+    std::fclose(file);
+
+    ss::parser<Ts...> p{file_name.c_str(), delim};
+    while (!p.eof()) {
+        try {
+            const auto& [s0, s1] =
+                p.template get_next<std::string, std::string>();
+            if (s0.size() == 10000) {
+                std::cout << s0.size() << std::endl;
+            }
+        } catch (ss::exception& e) {
+            continue;
+        }
+    }
+
+    std::remove(file_name.c_str());
+}
+
+template <typename... Ts>
+void test_ssp_buffer_mode(const uint8_t* data, size_t size,
+                          std::string delim = ss::default_delimiter) {
+    ss::parser<Ts...> p{(const char*)data, size, delim};
+    while (!p.eof()) {
+        try {
+            const auto& [s0, s1] =
+                p.template get_next<std::string, std::string>();
+            if (s0.size() == 10000) {
+                std::cout << s0.size() << std::endl;
+            }
+        } catch (ss::exception& e) {
+            continue;
+        }
+    }
+}
+
+template <typename... Ts>
+void test_ssp(const uint8_t* data, size_t size) {
+    test_ssp_file_mode<Ts...>(data, size);
+    test_ssp_file_mode<Ts..., ss::throw_on_error>(data, size);
+
+    test_ssp_file_mode<Ts...>(data, size, ":::");
+    test_ssp_file_mode<Ts..., ss::throw_on_error>(data, size, ":::");
+
+    test_ssp_buffer_mode<Ts...>(data, size);
+    test_ssp_buffer_mode<Ts..., ss::throw_on_error>(data, size);
+
+    test_ssp_buffer_mode<Ts...>(data, size, ":::");
+    test_ssp_buffer_mode<Ts..., ss::throw_on_error>(data, size, ":::");
+}
+
+extern "C" int LLVMFuzzerTestOneInput(const uint8_t* data, size_t size) {
+    using escape = ss::escape<'\\'>;
+    using quote = ss::quote<'"'>;
+    using trim = ss::trim<' ', '\t'>;
+    using multiline_r = ss::multiline_restricted<5>;
+
+    test_ssp<>(data, size);
+    test_ssp<escape>(data, size);
+    test_ssp<quote>(data, size);
+    test_ssp<trim>(data, size);
+    test_ssp<quote, escape>(data, size);
+    test_ssp<escape, quote, multiline_r, trim>(data, size);
+    test_ssp<escape, quote, multiline_r, trim, ss::ignore_empty>(data, size);
+
+    return 0;
+}

.github/workflows/fuzz.yml

@@ -0,0 +1,43 @@
+name: fuzz-ci
+
+on:
+  workflow_dispatch:
+
+  push:
+    branches:
+      - master
+      - feature/**
+      - improvement/**
+      - bugfix/**
+
+  pull_request:
+    branches:
+      - master
+      - feature/**
+      - improvement/**
+      - bugfix/**
+
+jobs:
+  clang_tests:
+    if: >-
+      ! contains(toJSON(github.event.commits.*.message), '[skip ci]') &&
+      ! contains(toJSON(github.event.commits.*.message), '[skip github]')
+
+    runs-on: ubuntu-latest
+
+    name: "Fuzzing"
+
+    container:
+      image: silkeh/clang:15
+      options: -v /usr/local:/host_usr_local
+
+    steps:
+      - uses: actions/checkout@v1
+
+      - name: Build
+        working-directory: .github/fuzz
+        run: make
+
+      - name: Run
+        working-directory: .github/fuzz
+        run: make run

.gitignore

@@ -1,6 +1,8 @@
 compile_commands.json
 .clang-format
-.ccls-cache/*
+.clang-tidy
+.ccls-cache/
+.cache/
 experiment/
 build/
 hbuild/

CMakeLists.txt

@@ -2,7 +2,7 @@ cmake_minimum_required(VERSION 3.14)
 
 project(
     ssp
-    VERSION 1.7.1
+    VERSION 1.7.2
     DESCRIPTION "csv parser"
     HOMEPAGE_URL "https://github.com/red0124/ssp"
     LANGUAGES CXX

README.md

@@ -9,6 +9,7 @@
 
 [![License](https://img.shields.io/badge/license-MIT-blue.svg)](https://opensource.org/licenses/MIT)
 [![coverage](https://coveralls.io/repos/github/red0124/ssp/badge.svg?branch=master)](https://coveralls.io/github/red0124/ssp?branch=master)
+[![fuzz](https://github.com/red0124/ssp/workflows/fuzz-ci/badge.svg)](https://github.com/red0124/ssp/actions/workflows/fuzz.yml)
 [![single-header](https://github.com/red0124/ssp/workflows/single-header-ci/badge.svg)](https://github.com/red0124/ssp/actions/workflows/single-header.yml)
 [![ubuntu-latest-gcc](https://github.com/red0124/ssp/workflows/ubuntu-latest-gcc-ci/badge.svg)](https://github.com/red0124/ssp/actions/workflows/ubuntu-latest-gcc.yml)
 [![ubuntu-latest-clang](https://github.com/red0124/ssp/workflows/ubuntu-latest-clang-ci/badge.svg)](https://github.com/red0124/ssp/actions/workflows/ubuntu-latest-clang.yml)

include/ss/common.hpp

@@ -1,11 +1,15 @@
 #pragma once
 #include <cerrno>
-#include <cstdint>
 #include <cstdio>
 #include <cstdlib>
 #include <cstring>
 #include <vector>
 
+#if !__unix__
+#include <array>
+#include <cstdint>
+#endif
+
 namespace ss {
 
 struct none {};
@@ -17,13 +21,13 @@ constexpr inline auto default_delimiter = ",";
 constexpr inline auto get_line_initial_buffer_size = 128;
 
 template <bool StringError>
-inline void assert_string_error_defined() {
+void assert_string_error_defined() {
     static_assert(StringError,
                   "'string_error' needs to be enabled to use 'error_msg'");
 }
 
 template <bool ThrowOnError>
-inline void assert_throw_on_error_not_defined() {
+void assert_throw_on_error_not_defined() {
     static_assert(!ThrowOnError, "cannot handle errors manually if "
                                  "'throw_on_error' is enabled");
 }
@@ -45,8 +49,8 @@ inline ssize_t get_line_file(char*& lineptr, size_t& n, FILE* file) {
 
 using ssize_t = intptr_t;
 
-ssize_t get_line_file(char*& lineptr, size_t& n, FILE* file) {
-    char buff[get_line_initial_buffer_size];
+inline ssize_t get_line_file(char*& lineptr, size_t& n, FILE* file) {
+    std::array<char, get_line_initial_buffer_size> buff;
 
     if (lineptr == nullptr || n < sizeof(buff)) {
         size_t new_n = sizeof(buff);
@@ -57,9 +61,9 @@ ssize_t get_line_file(char*& lineptr, size_t& n, FILE* file) {
     lineptr[0] = '\0';
 
     size_t line_used = 0;
-    while (std::fgets(buff, sizeof(buff), file) != nullptr) {
+    while (std::fgets(buff.data(), sizeof(buff), file) != nullptr) {
         line_used = std::strlen(lineptr);
-        size_t buff_used = std::strlen(buff);
+        size_t buff_used = std::strlen(buff.data());
 
         if (n <= buff_used + line_used) {
             size_t new_n = n * 2;
@@ -67,7 +71,7 @@ ssize_t get_line_file(char*& lineptr, size_t& n, FILE* file) {
             n = new_n;
         }
 
-        std::memcpy(lineptr + line_used, buff, buff_used);
+        std::memcpy(lineptr + line_used, buff.data(), buff_used);
         line_used += buff_used;
         lineptr[line_used] = '\0';
 
@@ -81,7 +85,7 @@ ssize_t get_line_file(char*& lineptr, size_t& n, FILE* file) {
 
 #endif
 
-ssize_t get_line_buffer(char*& lineptr, size_t& n,
+inline ssize_t get_line_buffer(char*& lineptr, size_t& n,
                         const char* const csv_data_buffer, size_t csv_data_size,
                         size_t& curr_char) {
     if (curr_char >= csv_data_size) {
@@ -89,7 +93,7 @@ ssize_t get_line_buffer(char*& lineptr, size_t& n,
     }
 
     if (lineptr == nullptr || n < get_line_initial_buffer_size) {
-        auto new_lineptr = static_cast<char*>(
+        auto* new_lineptr = static_cast<char*>(
             strict_realloc(lineptr, get_line_initial_buffer_size));
         lineptr = new_lineptr;
         n = get_line_initial_buffer_size;
@@ -114,22 +118,18 @@ ssize_t get_line_buffer(char*& lineptr, size_t& n,
         }
     }
 
-    if (line_used != 0) {
     lineptr[line_used] = '\0';
     return line_used;
-    }
-
-    return -1;
 }
 
-std::tuple<ssize_t, bool> get_line(char*& buffer, size_t& buffer_size,
+inline std::tuple<ssize_t, bool> get_line(char*& buffer, size_t& buffer_size,
                                    FILE* file,
                                    const char* const csv_data_buffer,
                                    size_t csv_data_size, size_t& curr_char) {
-    ssize_t ssize;
+    ssize_t ssize = 0;
     if (file) {
         ssize = get_line_file(buffer, buffer_size, file);
-        curr_char = std::ftell(file);
+        curr_char += ssize;
     } else {
         ssize = get_line_buffer(buffer, buffer_size, csv_data_buffer,
                                 csv_data_size, curr_char);
@@ -145,4 +145,4 @@ std::tuple<ssize_t, bool> get_line(char*& buffer, size_t& buffer_size,
     return {ssize, false};
 }
 
-} /* ss */
+} /* namespace ss */

include/ss/converter.hpp

@@ -224,8 +224,9 @@ private:
     }
 
     std::string error_sufix(const string_range msg, size_t pos) const {
+        constexpr static auto reserve_size = 32;
         std::string error;
-        error.reserve(32);
+        error.reserve(reserve_size);
         error.append("at column ")
             .append(std::to_string(pos + 1))
             .append(": \'")
@@ -391,7 +392,7 @@ private:
     ////////////////
 
     bool columns_mapped() const {
-        return column_mappings_.size() != 0;
+        return !column_mappings_.empty();
     }
 
     size_t column_position(size_t tuple_position) const {
@@ -404,7 +405,7 @@ private:
     // assumes positions are valid and the vector is not empty
     void set_column_mapping(std::vector<size_t> positions,
                             size_t number_of_columns) {
-        column_mappings_ = positions;
+        column_mappings_ = std::move(positions);
         number_of_columns_ = number_of_columns;
     }
 
@@ -490,7 +491,7 @@ private:
     friend class parser;
 
     std::vector<size_t> column_mappings_;
-    size_t number_of_columns_;
+    size_t number_of_columns_{0};
 };
 
-} /* ss */
+} /* namespace ss */

include/ss/exception.hpp

@@ -12,12 +12,12 @@ class exception : public std::exception {
     std::string msg_;
 
 public:
-    exception(const std::string& msg): msg_{msg} {
+    exception(std::string msg): msg_{std::move(msg)} {
     }
 
-    virtual char const* what() const noexcept {
+    char const* what() const noexcept override {
         return msg_.c_str();
     }
 };
 
-} /* ss */
+} /* namespace ss */

include/ss/extract.hpp

@@ -2,8 +2,8 @@
 
 #include "type_traits.hpp"
 #include <charconv>
+#include <cstdint>
 #include <cstring>
-#include <functional>
 #include <optional>
 #include <string>
 #include <string_view>
@@ -14,6 +14,7 @@
 #else
 #include <algorithm>
 #include <cstdlib>
+#include <array>
 #endif
 
 namespace ss {
@@ -45,16 +46,17 @@ std::enable_if_t<std::is_floating_point_v<T>, std::optional<T>> to_num(
                   "Conversion to long double is disabled");
 
     constexpr static auto buff_max = 64;
-    char short_buff[buff_max];
+    std::array<char, buff_max> short_buff;
+
     size_t string_range = std::distance(begin, end);
     std::string long_buff;
 
-    char* buff;
+    char* buff = nullptr;
     if (string_range > buff_max) {
         long_buff = std::string{begin, end};
         buff = long_buff.data();
     } else {
-        buff = short_buff;
+        buff = short_buff.data();
         buff[string_range] = '\0';
         std::copy_n(begin, string_range, buff);
     }
@@ -92,6 +94,8 @@ struct numeric_wrapper {
     numeric_wrapper& operator=(numeric_wrapper&&) = default;
     numeric_wrapper& operator=(const numeric_wrapper&) = default;
 
+    ~numeric_wrapper() = default;
+
     numeric_wrapper(T other) : value{other} {
     }
 
@@ -142,7 +146,7 @@ template <typename T>
 struct unsupported_type {
     constexpr static bool value = false;
 };
-} /* namespace */
+} /* namespace errors */
 
 template <typename T>
 std::enable_if_t<!std::is_integral_v<T> && !std::is_floating_point_v<T> &&
@@ -215,10 +219,13 @@ inline bool extract(const char* begin, const char* end, bool& value) {
             return false;
         }
     } else {
+        constexpr static auto true_size = 4;
+        constexpr static auto false_size = 5;
         size_t size = end - begin;
-        if (size == 4 && std::strncmp(begin, "true", size) == 0) {
+        if (size == true_size && std::strncmp(begin, "true", size) == 0) {
             value = true;
-        } else if (size == 5 && std::strncmp(begin, "false", size) == 0) {
+        } else if (size == false_size &&
+                   std::strncmp(begin, "false", size) == 0) {
             value = false;
         } else {
             return false;
@@ -247,4 +254,4 @@ inline bool extract(const char* begin, const char* end,
     return true;
 }
 
-} /* ss */
+} /* namespace ss */

include/ss/function_traits.hpp

@@ -2,7 +2,6 @@
 
 #include <cstdlib>
 #include <functional>
-#include <tuple>
 
 namespace ss {
 
@@ -77,4 +76,4 @@ struct member_wrapper<R T::*> {
     template <typename T>                                                      \
     constexpr bool has_m_##method##_t = has_m_##method<T>::value;
 
-} /* trait */
+} /* namespace ss */

include/ss/parser.hpp

@@ -5,7 +5,6 @@
 #include "exception.hpp"
 #include "extract.hpp"
 #include "restrictions.hpp"
-#include <cerrno>
 #include <cstdlib>
 #include <cstring>
 #include <optional>
@@ -32,10 +31,12 @@ class parser {
 
     constexpr static bool ignore_empty = setup<Options...>::ignore_empty;
 
+    using header_splitter = ss::splitter<
+        ss::filter_not_t<ss::is_instance_of_multiline, Options...>>;
+
 public:
-    parser(const std::string& file_name,
-           const std::string& delim = ss::default_delimiter)
-        : file_name_{file_name}, reader_{file_name_, delim} {
+    parser(std::string file_name, std::string delim = ss::default_delimiter)
+        : file_name_{std::move(file_name)}, reader_{file_name_, delim} {
         if (reader_.file_) {
             read_line();
             if constexpr (ignore_header) {
@@ -51,7 +52,7 @@ public:
 
     parser(const char* const csv_data_buffer, size_t csv_data_size,
            const std::string& delim = ss::default_delimiter)
-        : file_name_{"buffer line"},
+        : file_name_{"CSV data buffer"},
           reader_{csv_data_buffer, csv_data_size, delim} {
         if (csv_data_buffer) {
             read_line();
@@ -68,6 +69,7 @@ public:
 
     parser(parser&& other) = default;
     parser& operator=(parser&& other) = default;
+    ~parser() = default;
 
     parser() = delete;
     parser(const parser& other) = delete;
@@ -162,20 +164,49 @@ public:
         return value;
     }
 
+    std::string raw_header() const {
+        assert_ignore_header_not_defined();
+        return raw_header_;
+    }
+
+    std::vector<std::string> header() {
+        assert_ignore_header_not_defined();
+        clear_error();
+
+        header_splitter splitter;
+        std::string raw_header_copy = raw_header_;
+
+        if (!strict_split(splitter, raw_header_copy)) {
+            return {};
+        }
+
+        std::vector<std::string> split_header;
+        for (const auto& [begin, end] : splitter.split_data_) {
+            split_header.emplace_back(begin, end);
+        }
+
+        return split_header;
+    }
+
     bool field_exists(const std::string& field) {
+        assert_ignore_header_not_defined();
+        clear_error();
+
         if (header_.empty()) {
             split_header_data();
         }
 
+        if (!valid()) {
+            return false;
+        }
+
         return header_index(field).has_value();
     }
 
     template <typename... Ts>
     void use_fields(const Ts&... fields_args) {
-        if constexpr (ignore_header) {
-            handle_error_header_ignored();
-            return;
-        }
+        assert_ignore_header_not_defined();
+        clear_error();
 
         if (header_.empty() && !eof()) {
             split_header_data();
@@ -188,7 +219,7 @@ public:
         auto fields = std::vector<std::string>{fields_args...};
 
         if (fields.empty()) {
-            handle_error_empty_mapping();
+            handle_error_invalid_use_fields_argument();
             return;
         }
 
@@ -237,6 +268,10 @@ public:
 
             iterator(const iterator& other) = default;
             iterator(iterator&& other) = default;
+            ~iterator() = default;
+
+            iterator& operator=(const iterator& other) = delete;
+            iterator& operator=(iterator&& other) = delete;
 
             value& operator*() {
                 return value_;
@@ -261,8 +296,10 @@ public:
                 return *this;
             }
 
-            iterator& operator++(int) {
-                return ++*this;
+            iterator operator++(int) {
+                auto result = *this;
+                ++*this;
+                return result;
             }
 
             friend bool operator==(const iterator& lhs, const iterator& rhs) {
@@ -326,7 +363,7 @@ public:
             Fun&& fun = none{}) {
             using Value = no_void_validator_tup_t<Us...>;
             std::optional<Value> value;
-            try_convert_and_invoke<Value, Us...>(value, fun);
+            try_convert_and_invoke<Value, Us...>(value, std::forward<Fun>(fun));
             return composite_with(std::move(value));
         }
 
@@ -335,7 +372,7 @@ public:
         template <typename U, typename... Us, typename Fun = none>
         composite<Ts..., std::optional<U>> or_object(Fun&& fun = none{}) {
             std::optional<U> value;
-            try_convert_and_invoke<U, Us...>(value, fun);
+            try_convert_and_invoke<U, Us...>(value, std::forward<Fun>(fun));
             return composite_with(std::move(value));
         }
 
@@ -443,7 +480,8 @@ private:
             using Ret = decltype(try_invoke_impl(arg, std::forward<Fun>(fun)));
             constexpr bool returns_void = std::is_same_v<Ret, void>;
             if constexpr (!returns_void) {
-                if (!try_invoke_impl(arg, std::forward<Fun>(fun))) {
+                if (!try_invoke_impl(std::forward<Arg>(arg),
+                                     std::forward<Fun>(fun))) {
                     handle_error_failed_check();
                 }
             } else {
@@ -478,22 +516,58 @@ private:
         if (valid()) {
             try_invoke(*value, std::forward<Fun>(fun));
         }
-        return {valid() ? std::move(value) : std::nullopt, *this};
+        return {valid() ? std::forward<T>(value) : std::nullopt, *this};
     }
 
     ////////////////
     // header
     ////////////////
 
+    void assert_ignore_header_not_defined() const {
+        static_assert(!ignore_header,
+                      "cannot use this method when 'ignore_header' is defined");
+    }
+
+    bool strict_split(header_splitter& splitter, std::string& header) {
+        if (header.empty()) {
+            return false;
+        }
+
+        if constexpr (throw_on_error) {
+            try {
+                splitter.split(header.data(), reader_.delim_);
+            } catch (const ss::exception& e) {
+                decorate_rethrow_invalid_header_split(e);
+            }
+        } else {
+            splitter.split(header.data(), reader_.delim_);
+            if (!splitter.valid()) {
+                handle_error_invalid_header_split(splitter);
+                return false;
+            }
+        }
+
+        return true;
+    }
+
     void split_header_data() {
-        ss::splitter<Options...> splitter;
+        header_splitter splitter;
         std::string raw_header_copy = raw_header_;
-        splitter.split(raw_header_copy.data(), reader_.delim_);
+
+        if (!strict_split(splitter, raw_header_copy)) {
+            return;
+        }
+
         for (const auto& [begin, end] : splitter.split_data_) {
             std::string field{begin, end};
+            if (field.empty()) {
+                handle_error_duplicate_header_field(field);
+                header_.clear();
+                return;
+            }
             if (std::find(header_.begin(), header_.end(), field) !=
                 header_.end()) {
-                handle_error_invalid_header(field);
+                handle_error_duplicate_header_field(field);
                 header_.clear();
                 return;
             }
@@ -524,7 +598,7 @@ private:
     }
 
     void handle_error_failed_check() {
-        constexpr static auto error_msg = " failed check";
+        constexpr static auto error_msg = ": failed check";
 
         if constexpr (string_error) {
             error_.clear();
@@ -537,7 +611,7 @@ private:
     }
 
     void handle_error_null_buffer() {
-        constexpr static auto error_msg = " received null data buffer";
+        constexpr static auto error_msg = ": received null data buffer";
 
         if constexpr (string_error) {
             error_.clear();
@@ -550,7 +624,7 @@ private:
     }
 
     void handle_error_file_not_open() {
-        constexpr static auto error_msg = " could not be opened";
+        constexpr static auto error_msg = ": could not be opened";
 
         if constexpr (string_error) {
             error_.clear();
@@ -563,7 +637,7 @@ private:
     }
 
     void handle_error_eof_reached() {
-        constexpr static auto error_msg = " read on end of file";
+        constexpr static auto error_msg = ": read on end of file";
 
         if constexpr (string_error) {
             error_.clear();
@@ -588,20 +662,6 @@ private:
         }
     }
 
-    void handle_error_header_ignored() {
-        constexpr static auto error_msg =
-            ": the header row is ignored within the setup it cannot be used";
-
-        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_invalid_field(const std::string& field) {
         constexpr static auto error_msg =
             ": header does not contain given field: ";
@@ -629,8 +689,9 @@ private:
         }
     }
 
-    void handle_error_empty_mapping() {
-        constexpr static auto error_msg = "received empty mapping";
+    void handle_error_invalid_use_fields_argument() {
+        constexpr static auto error_msg =
+            "received invalid argument for 'use_fields'";
 
         if constexpr (string_error) {
             error_.clear();
@@ -642,19 +703,53 @@ private:
         }
     }
 
-    void handle_error_invalid_header(const std::string& field) {
-        constexpr static auto error_msg = "header contains duplicates: ";
+    void handle_error_invalid_header_field() {
+        constexpr static auto error_msg = ": header contains empty field";
 
         if constexpr (string_error) {
             error_.clear();
-            error_.append(error_msg).append(error_msg);
+            error_.append(file_name_).append(error_msg);
         } else if constexpr (throw_on_error) {
-            throw ss::exception{error_msg + field};
+            throw ss::exception{file_name_ + error_msg};
         } else {
             error_ = true;
         }
     }
 
+    void handle_error_duplicate_header_field(const std::string& field) {
+        constexpr static auto error_msg = ": header contains duplicate: ";
+
+        if constexpr (string_error) {
+            error_.clear();
+            error_.append(file_name_).append(error_msg).append(field);
+        } else if constexpr (throw_on_error) {
+            throw ss::exception{file_name_ + error_msg + field};
+        } else {
+            error_ = true;
+        }
+    }
+
+    void handle_error_invalid_header_split(const header_splitter& splitter) {
+        constexpr static auto error_msg = ": failed header parsing: ";
+
+        if constexpr (string_error) {
+            error_.clear();
+            error_.append(file_name_)
+                .append(error_msg)
+                .append(splitter.error_msg());
+        } else {
+            error_ = true;
+        }
+    }
+
+    void decorate_rethrow_invalid_header_split(const ss::exception& e) const {
+        static_assert(throw_on_error,
+                      "throw_on_error needs to be enabled to use this method");
+        throw ss::exception{std::string{file_name_}
+                                .append(": failed header parsing: ")
+                                .append(e.what())};
+    }
+
     void decorate_rethrow(const ss::exception& e) const {
         static_assert(throw_on_error,
                       "throw_on_error needs to be enabled to use this method");
@@ -674,17 +769,18 @@ private:
     }
 
     struct reader {
-        reader(const std::string& file_name_, const std::string& delim)
-            : delim_{delim}, file_{std::fopen(file_name_.c_str(), "rb")} {
+        reader(const std::string& file_name_, std::string delim)
+            : delim_{std::move(delim)},
+              file_{std::fopen(file_name_.c_str(), "rb")} {
         }
 
         reader(const char* const buffer, size_t csv_data_size,
-               const std::string& delim)
-            : delim_{delim}, csv_data_buffer_{buffer},
+               std::string delim)
+            : delim_{std::move(delim)}, csv_data_buffer_{buffer},
               csv_data_size_{csv_data_size} {
         }
 
-        reader(reader&& other)
+        reader(reader&& other) noexcept
             : buffer_{other.buffer_},
               next_line_buffer_{other.next_line_buffer_},
               helper_buffer_{other.helper_buffer_},
@@ -705,7 +801,7 @@ private:
             other.file_ = nullptr;
         }
 
-        reader& operator=(reader&& other) {
+        reader& operator=(reader&& other) noexcept {
             if (this != &other) {
                 buffer_ = other.buffer_;
                 next_line_buffer_ = other.next_line_buffer_;
@@ -853,7 +949,7 @@ private:
         }
 
         bool escaped_eol(size_t size) {
-            const char* curr;
+            const char* curr = nullptr;
             for (curr = next_line_buffer_ + size - 1;
                  curr >= next_line_buffer_ &&
                  setup<Options...>::escape::match(*curr);
@@ -899,7 +995,7 @@ private:
                             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*>(
+            auto* new_first = static_cast<char*>(
                 strict_realloc(static_cast<void*>(first), buffer_size));
 
             first = new_first;
@@ -928,7 +1024,7 @@ private:
         }
 
         std::string get_buffer() {
-            return std::string{next_line_buffer_, next_line_buffer_size_};
+            return std::string{next_line_buffer_, next_line_size_};
         }
 
         ////////////////
@@ -971,4 +1067,4 @@ private:
     bool eof_{false};
 };
 
-} /* ss */
+} /* namespace ss */

include/ss/restrictions.hpp

@@ -124,4 +124,4 @@ struct ne {
     }
 };
 
-} /* ss */
+} /* namespace ss */

include/ss/setup.hpp

@@ -293,4 +293,7 @@ private:
 template <typename... Options>
 struct setup<setup<Options...>> : setup<Options...> {};
 
-} /* ss */
+template <typename... Options>
+struct setup<std::tuple<Options...>> : setup<Options...> {};
+
+} /* namespace ss */

include/ss/splitter.hpp

@@ -2,11 +2,9 @@
 #include "common.hpp"
 #include "exception.hpp"
 #include "setup.hpp"
-#include "type_traits.hpp"
 #include <algorithm>
 #include <cstdlib>
 #include <cstring>
-#include <memory>
 #include <string>
 #include <vector>
 
@@ -322,8 +320,9 @@ private:
 
         trim_left_if_enabled(begin_);
 
-        for (done_ = false; !done_; read(delim))
-            ;
+        for (done_ = false; !done_;) {
+            read(delim);
+        }
 
         return split_data_;
     }
@@ -479,4 +478,4 @@ public:
     friend class converter;
 };
 
-} /* ss */
+} /* namespace ss */

include/ss/type_traits.hpp

@@ -34,7 +34,11 @@ struct left_of_impl;
 
 template <size_t N, typename T, typename... Ts>
 struct left_of_impl {
-    static_assert(N < 128, "recursion limit reached");
+private:
+    constexpr static auto recursion_limit = 128;
+
+public:
+    static_assert(N < recursion_limit, "recursion limit reached");
     static_assert(N != 0, "cannot take the whole tuple");
     using type = tup_cat_t<T, typename left_of_impl<N - 1, Ts...>::type>;
 };
@@ -378,4 +382,4 @@ T to_object(U&& data) {
     }
 }
 
-} /* trait */
+} /* namespace ss */

meson.build

@@ -6,7 +6,7 @@ project(
       'cpp_std=c++17', 
       'buildtype=debugoptimized',
       'wrap_mode=forcefallback'],
-  version: '1.7.1',
+  version: '1.7.2',
   meson_version:'>=0.54.0')
 
 fast_float_dep = dependency('fast_float')

script/single_header_generator.py

@@ -14,14 +14,21 @@ headers = ['type_traits.hpp',
 
 combined_file = []
 includes = []
+in_pp_block = False
 
 for header in headers:
     with open(headers_dir + header) as f:
         for line in f.read().splitlines():
+            if '#if ' in line:
+                in_pp_block = True
+
+            if '#endif' in line:
+                in_pp_block = False
+
             if '#include "' in line or '#include <fast_float' in line:
                 continue
 
-            if '#include <' in line:
+            if '#include <' in line and not in_pp_block:
                 includes.append(line)
                 continue
 

ssp.hpp

@@ -8,7 +8,6 @@
 #include <cstring>
 #include <exception>
 #include <functional>
-#include <memory>
 #include <optional>
 #include <string>
 #include <string_view>
@@ -50,7 +49,11 @@ struct left_of_impl;
 
 template <size_t N, typename T, typename... Ts>
 struct left_of_impl {
-    static_assert(N < 128, "recursion limit reached");
+private:
+    constexpr static auto recursion_limit = 128;
+
+public:
+    static_assert(N < recursion_limit, "recursion limit reached");
     static_assert(N != 0, "cannot take the whole tuple");
     using type = tup_cat_t<T, typename left_of_impl<N - 1, Ts...>::type>;
 };
@@ -394,7 +397,7 @@ T to_object(U&& data) {
     }
 }
 
-} /* trait */
+} /* namespace ss */
 
 namespace ss {
 
@@ -406,15 +409,15 @@ class exception : public std::exception {
     std::string msg_;
 
 public:
-    exception(const std::string& msg): msg_{msg} {
+    exception(std::string msg): msg_{std::move(msg)} {
     }
 
-    virtual char const* what() const noexcept {
+    char const* what() const noexcept override {
         return msg_.c_str();
     }
 };
 
-} /* ss */
+} /* namespace ss */
 
 
 namespace ss {
@@ -490,7 +493,7 @@ struct member_wrapper<R T::*> {
     template <typename T>                                                      \
     constexpr bool has_m_##method##_t = has_m_##method<T>::value;
 
-} /* trait */
+} /* namespace ss */
 
 namespace ss {
 
@@ -616,7 +619,12 @@ struct ne {
     }
 };
 
-} /* ss */
+} /* namespace ss */
+
+#if !__unix__
+#include <array>
+#include <cstdint>
+#endif
 
 namespace ss {
 
@@ -629,13 +637,13 @@ constexpr inline auto default_delimiter = ",";
 constexpr inline auto get_line_initial_buffer_size = 128;
 
 template <bool StringError>
-inline void assert_string_error_defined() {
+void assert_string_error_defined() {
     static_assert(StringError,
                   "'string_error' needs to be enabled to use 'error_msg'");
 }
 
 template <bool ThrowOnError>
-inline void assert_throw_on_error_not_defined() {
+void assert_throw_on_error_not_defined() {
     static_assert(!ThrowOnError, "cannot handle errors manually if "
                                  "'throw_on_error' is enabled");
 }
@@ -657,8 +665,8 @@ inline ssize_t get_line_file(char*& lineptr, size_t& n, FILE* file) {
 
 using ssize_t = intptr_t;
 
-ssize_t get_line_file(char*& lineptr, size_t& n, FILE* file) {
-    char buff[get_line_initial_buffer_size];
+inline ssize_t get_line_file(char*& lineptr, size_t& n, FILE* file) {
+    std::array<char, get_line_initial_buffer_size> buff;
 
     if (lineptr == nullptr || n < sizeof(buff)) {
         size_t new_n = sizeof(buff);
@@ -669,9 +677,9 @@ ssize_t get_line_file(char*& lineptr, size_t& n, FILE* file) {
     lineptr[0] = '\0';
 
     size_t line_used = 0;
-    while (std::fgets(buff, sizeof(buff), file) != nullptr) {
+    while (std::fgets(buff.data(), sizeof(buff), file) != nullptr) {
         line_used = std::strlen(lineptr);
-        size_t buff_used = std::strlen(buff);
+        size_t buff_used = std::strlen(buff.data());
 
         if (n <= buff_used + line_used) {
             size_t new_n = n * 2;
@@ -679,7 +687,7 @@ ssize_t get_line_file(char*& lineptr, size_t& n, FILE* file) {
             n = new_n;
         }
 
-        std::memcpy(lineptr + line_used, buff, buff_used);
+        std::memcpy(lineptr + line_used, buff.data(), buff_used);
         line_used += buff_used;
         lineptr[line_used] = '\0';
 
@@ -693,7 +701,7 @@ ssize_t get_line_file(char*& lineptr, size_t& n, FILE* file) {
 
 #endif
 
-ssize_t get_line_buffer(char*& lineptr, size_t& n,
+inline ssize_t get_line_buffer(char*& lineptr, size_t& n,
                         const char* const csv_data_buffer, size_t csv_data_size,
                         size_t& curr_char) {
     if (curr_char >= csv_data_size) {
@@ -701,7 +709,7 @@ ssize_t get_line_buffer(char*& lineptr, size_t& n,
     }
 
     if (lineptr == nullptr || n < get_line_initial_buffer_size) {
-        auto new_lineptr = static_cast<char*>(
+        auto* new_lineptr = static_cast<char*>(
             strict_realloc(lineptr, get_line_initial_buffer_size));
         lineptr = new_lineptr;
         n = get_line_initial_buffer_size;
@@ -726,22 +734,18 @@ ssize_t get_line_buffer(char*& lineptr, size_t& n,
         }
     }
 
-    if (line_used != 0) {
     lineptr[line_used] = '\0';
     return line_used;
-    }
-
-    return -1;
 }
 
-std::tuple<ssize_t, bool> get_line(char*& buffer, size_t& buffer_size,
+inline std::tuple<ssize_t, bool> get_line(char*& buffer, size_t& buffer_size,
                                    FILE* file,
                                    const char* const csv_data_buffer,
                                    size_t csv_data_size, size_t& curr_char) {
-    ssize_t ssize;
+    ssize_t ssize = 0;
     if (file) {
         ssize = get_line_file(buffer, buffer_size, file);
-        curr_char = std::ftell(file);
+        curr_char += ssize;
     } else {
         ssize = get_line_buffer(buffer, buffer_size, csv_data_buffer,
                                 csv_data_size, curr_char);
@@ -757,7 +761,7 @@ std::tuple<ssize_t, bool> get_line(char*& buffer, size_t& buffer_size,
     return {ssize, false};
 }
 
-} /* ss */
+} /* namespace ss */
 
 namespace ss {
 
@@ -1050,7 +1054,10 @@ private:
 template <typename... Options>
 struct setup<setup<Options...>> : setup<Options...> {};
 
-} /* ss */
+template <typename... Options>
+struct setup<std::tuple<Options...>> : setup<Options...> {};
+
+} /* namespace ss */
 
 namespace ss {
 
@@ -1364,8 +1371,9 @@ private:
 
         trim_left_if_enabled(begin_);
 
-        for (done_ = false; !done_; read(delim))
-            ;
+        for (done_ = false; !done_;) {
+            read(delim);
+        }
 
         return split_data_;
     }
@@ -1521,7 +1529,7 @@ public:
     friend class converter;
 };
 
-} /* ss */
+} /* namespace ss */
 
 
 #ifndef SSP_DISABLE_FAST_FLOAT
@@ -1557,16 +1565,17 @@ std::enable_if_t<std::is_floating_point_v<T>, std::optional<T>> to_num(
                   "Conversion to long double is disabled");
 
     constexpr static auto buff_max = 64;
-    char short_buff[buff_max];
+    std::array<char, buff_max> short_buff;
+
     size_t string_range = std::distance(begin, end);
     std::string long_buff;
 
-    char* buff;
+    char* buff = nullptr;
     if (string_range > buff_max) {
         long_buff = std::string{begin, end};
         buff = long_buff.data();
     } else {
-        buff = short_buff;
+        buff = short_buff.data();
         buff[string_range] = '\0';
         std::copy_n(begin, string_range, buff);
     }
@@ -1604,6 +1613,8 @@ struct numeric_wrapper {
     numeric_wrapper& operator=(numeric_wrapper&&) = default;
     numeric_wrapper& operator=(const numeric_wrapper&) = default;
 
+    ~numeric_wrapper() = default;
+
     numeric_wrapper(T other) : value{other} {
     }
 
@@ -1654,7 +1665,7 @@ template <typename T>
 struct unsupported_type {
     constexpr static bool value = false;
 };
-} /* namespace */
+} /* namespace errors */
 
 template <typename T>
 std::enable_if_t<!std::is_integral_v<T> && !std::is_floating_point_v<T> &&
@@ -1727,10 +1738,13 @@ inline bool extract(const char* begin, const char* end, bool& value) {
             return false;
         }
     } else {
+        constexpr static auto true_size = 4;
+        constexpr static auto false_size = 5;
         size_t size = end - begin;
-        if (size == 4 && std::strncmp(begin, "true", size) == 0) {
+        if (size == true_size && std::strncmp(begin, "true", size) == 0) {
             value = true;
-        } else if (size == 5 && std::strncmp(begin, "false", size) == 0) {
+        } else if (size == false_size &&
+                   std::strncmp(begin, "false", size) == 0) {
             value = false;
         } else {
             return false;
@@ -1759,7 +1773,7 @@ inline bool extract(const char* begin, const char* end,
     return true;
 }
 
-} /* ss */
+} /* namespace ss */
 
 namespace ss {
 INIT_HAS_METHOD(tied)
@@ -1976,8 +1990,9 @@ private:
     }
 
     std::string error_sufix(const string_range msg, size_t pos) const {
+        constexpr static auto reserve_size = 32;
         std::string error;
-        error.reserve(32);
+        error.reserve(reserve_size);
         error.append("at column ")
             .append(std::to_string(pos + 1))
             .append(": \'")
@@ -2143,7 +2158,7 @@ private:
     ////////////////
 
     bool columns_mapped() const {
-        return column_mappings_.size() != 0;
+        return !column_mappings_.empty();
     }
 
     size_t column_position(size_t tuple_position) const {
@@ -2156,7 +2171,7 @@ private:
     // assumes positions are valid and the vector is not empty
     void set_column_mapping(std::vector<size_t> positions,
                             size_t number_of_columns) {
-        column_mappings_ = positions;
+        column_mappings_ = std::move(positions);
         number_of_columns_ = number_of_columns;
     }
 
@@ -2242,10 +2257,10 @@ private:
     friend class parser;
 
     std::vector<size_t> column_mappings_;
-    size_t number_of_columns_;
+    size_t number_of_columns_{0};
 };
 
-} /* ss */
+} /* namespace ss */
 
 
 namespace ss {
@@ -2268,10 +2283,12 @@ class parser {
 
     constexpr static bool ignore_empty = setup<Options...>::ignore_empty;
 
+    using header_splitter = ss::splitter<
+        ss::filter_not_t<ss::is_instance_of_multiline, Options...>>;
+
 public:
-    parser(const std::string& file_name,
-           const std::string& delim = ss::default_delimiter)
-        : file_name_{file_name}, reader_{file_name_, delim} {
+    parser(std::string file_name, std::string delim = ss::default_delimiter)
+        : file_name_{std::move(file_name)}, reader_{file_name_, delim} {
         if (reader_.file_) {
             read_line();
             if constexpr (ignore_header) {
@@ -2287,7 +2304,7 @@ public:
 
     parser(const char* const csv_data_buffer, size_t csv_data_size,
            const std::string& delim = ss::default_delimiter)
-        : file_name_{"buffer line"},
+        : file_name_{"CSV data buffer"},
           reader_{csv_data_buffer, csv_data_size, delim} {
         if (csv_data_buffer) {
             read_line();
@@ -2304,6 +2321,7 @@ public:
 
     parser(parser&& other) = default;
     parser& operator=(parser&& other) = default;
+    ~parser() = default;
 
     parser() = delete;
     parser(const parser& other) = delete;
@@ -2398,20 +2416,49 @@ public:
         return value;
     }
 
+    std::string raw_header() const {
+        assert_ignore_header_not_defined();
+        return raw_header_;
+    }
+
+    std::vector<std::string> header() {
+        assert_ignore_header_not_defined();
+        clear_error();
+
+        header_splitter splitter;
+        std::string raw_header_copy = raw_header_;
+
+        if (!strict_split(splitter, raw_header_copy)) {
+            return {};
+        }
+
+        std::vector<std::string> split_header;
+        for (const auto& [begin, end] : splitter.split_data_) {
+            split_header.emplace_back(begin, end);
+        }
+
+        return split_header;
+    }
+
     bool field_exists(const std::string& field) {
+        assert_ignore_header_not_defined();
+        clear_error();
+
         if (header_.empty()) {
             split_header_data();
         }
 
+        if (!valid()) {
+            return false;
+        }
+
         return header_index(field).has_value();
     }
 
     template <typename... Ts>
     void use_fields(const Ts&... fields_args) {
-        if constexpr (ignore_header) {
-            handle_error_header_ignored();
-            return;
-        }
+        assert_ignore_header_not_defined();
+        clear_error();
 
         if (header_.empty() && !eof()) {
             split_header_data();
@@ -2424,7 +2471,7 @@ public:
         auto fields = std::vector<std::string>{fields_args...};
 
         if (fields.empty()) {
-            handle_error_empty_mapping();
+            handle_error_invalid_use_fields_argument();
             return;
         }
 
@@ -2473,6 +2520,10 @@ public:
 
             iterator(const iterator& other) = default;
             iterator(iterator&& other) = default;
+            ~iterator() = default;
+
+            iterator& operator=(const iterator& other) = delete;
+            iterator& operator=(iterator&& other) = delete;
 
             value& operator*() {
                 return value_;
@@ -2497,8 +2548,10 @@ public:
                 return *this;
             }
 
-            iterator& operator++(int) {
-                return ++*this;
+            iterator operator++(int) {
+                auto result = *this;
+                ++*this;
+                return result;
             }
 
             friend bool operator==(const iterator& lhs, const iterator& rhs) {
@@ -2562,7 +2615,7 @@ public:
             Fun&& fun = none{}) {
             using Value = no_void_validator_tup_t<Us...>;
             std::optional<Value> value;
-            try_convert_and_invoke<Value, Us...>(value, fun);
+            try_convert_and_invoke<Value, Us...>(value, std::forward<Fun>(fun));
             return composite_with(std::move(value));
         }
 
@@ -2571,7 +2624,7 @@ public:
         template <typename U, typename... Us, typename Fun = none>
         composite<Ts..., std::optional<U>> or_object(Fun&& fun = none{}) {
             std::optional<U> value;
-            try_convert_and_invoke<U, Us...>(value, fun);
+            try_convert_and_invoke<U, Us...>(value, std::forward<Fun>(fun));
             return composite_with(std::move(value));
         }
 
@@ -2679,7 +2732,8 @@ private:
             using Ret = decltype(try_invoke_impl(arg, std::forward<Fun>(fun)));
             constexpr bool returns_void = std::is_same_v<Ret, void>;
             if constexpr (!returns_void) {
-                if (!try_invoke_impl(arg, std::forward<Fun>(fun))) {
+                if (!try_invoke_impl(std::forward<Arg>(arg),
+                                     std::forward<Fun>(fun))) {
                     handle_error_failed_check();
                 }
             } else {
@@ -2714,22 +2768,58 @@ private:
         if (valid()) {
             try_invoke(*value, std::forward<Fun>(fun));
         }
-        return {valid() ? std::move(value) : std::nullopt, *this};
+        return {valid() ? std::forward<T>(value) : std::nullopt, *this};
     }
 
     ////////////////
     // header
     ////////////////
 
+    void assert_ignore_header_not_defined() const {
+        static_assert(!ignore_header,
+                      "cannot use this method when 'ignore_header' is defined");
+    }
+
+    bool strict_split(header_splitter& splitter, std::string& header) {
+        if (header.empty()) {
+            return false;
+        }
+
+        if constexpr (throw_on_error) {
+            try {
+                splitter.split(header.data(), reader_.delim_);
+            } catch (const ss::exception& e) {
+                decorate_rethrow_invalid_header_split(e);
+            }
+        } else {
+            splitter.split(header.data(), reader_.delim_);
+            if (!splitter.valid()) {
+                handle_error_invalid_header_split(splitter);
+                return false;
+            }
+        }
+
+        return true;
+    }
+
     void split_header_data() {
-        ss::splitter<Options...> splitter;
+        header_splitter splitter;
         std::string raw_header_copy = raw_header_;
-        splitter.split(raw_header_copy.data(), reader_.delim_);
+
+        if (!strict_split(splitter, raw_header_copy)) {
+            return;
+        }
+
         for (const auto& [begin, end] : splitter.split_data_) {
             std::string field{begin, end};
+            if (field.empty()) {
+                handle_error_duplicate_header_field(field);
+                header_.clear();
+                return;
+            }
             if (std::find(header_.begin(), header_.end(), field) !=
                 header_.end()) {
-                handle_error_invalid_header(field);
+                handle_error_duplicate_header_field(field);
                 header_.clear();
                 return;
             }
@@ -2760,7 +2850,7 @@ private:
     }
 
     void handle_error_failed_check() {
-        constexpr static auto error_msg = " failed check";
+        constexpr static auto error_msg = ": failed check";
 
         if constexpr (string_error) {
             error_.clear();
@@ -2773,7 +2863,7 @@ private:
     }
 
     void handle_error_null_buffer() {
-        constexpr static auto error_msg = " received null data buffer";
+        constexpr static auto error_msg = ": received null data buffer";
 
         if constexpr (string_error) {
             error_.clear();
@@ -2786,7 +2876,7 @@ private:
     }
 
     void handle_error_file_not_open() {
-        constexpr static auto error_msg = " could not be opened";
+        constexpr static auto error_msg = ": could not be opened";
 
         if constexpr (string_error) {
             error_.clear();
@@ -2799,7 +2889,7 @@ private:
     }
 
     void handle_error_eof_reached() {
-        constexpr static auto error_msg = " read on end of file";
+        constexpr static auto error_msg = ": read on end of file";
 
         if constexpr (string_error) {
             error_.clear();
@@ -2824,20 +2914,6 @@ private:
         }
     }
 
-    void handle_error_header_ignored() {
-        constexpr static auto error_msg =
-            ": the header row is ignored within the setup it cannot be used";
-
-        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_invalid_field(const std::string& field) {
         constexpr static auto error_msg =
             ": header does not contain given field: ";
@@ -2865,8 +2941,9 @@ private:
         }
     }
 
-    void handle_error_empty_mapping() {
-        constexpr static auto error_msg = "received empty mapping";
+    void handle_error_invalid_use_fields_argument() {
+        constexpr static auto error_msg =
+            "received invalid argument for 'use_fields'";
 
         if constexpr (string_error) {
             error_.clear();
@@ -2878,19 +2955,53 @@ private:
         }
     }
 
-    void handle_error_invalid_header(const std::string& field) {
-        constexpr static auto error_msg = "header contains duplicates: ";
+    void handle_error_invalid_header_field() {
+        constexpr static auto error_msg = ": header contains empty field";
 
         if constexpr (string_error) {
             error_.clear();
-            error_.append(error_msg).append(error_msg);
+            error_.append(file_name_).append(error_msg);
         } else if constexpr (throw_on_error) {
-            throw ss::exception{error_msg + field};
+            throw ss::exception{file_name_ + error_msg};
         } else {
             error_ = true;
         }
     }
 
+    void handle_error_duplicate_header_field(const std::string& field) {
+        constexpr static auto error_msg = ": header contains duplicate: ";
+
+        if constexpr (string_error) {
+            error_.clear();
+            error_.append(file_name_).append(error_msg).append(field);
+        } else if constexpr (throw_on_error) {
+            throw ss::exception{file_name_ + error_msg + field};
+        } else {
+            error_ = true;
+        }
+    }
+
+    void handle_error_invalid_header_split(const header_splitter& splitter) {
+        constexpr static auto error_msg = ": failed header parsing: ";
+
+        if constexpr (string_error) {
+            error_.clear();
+            error_.append(file_name_)
+                .append(error_msg)
+                .append(splitter.error_msg());
+        } else {
+            error_ = true;
+        }
+    }
+
+    void decorate_rethrow_invalid_header_split(const ss::exception& e) const {
+        static_assert(throw_on_error,
+                      "throw_on_error needs to be enabled to use this method");
+        throw ss::exception{std::string{file_name_}
+                                .append(": failed header parsing: ")
+                                .append(e.what())};
+    }
+
     void decorate_rethrow(const ss::exception& e) const {
         static_assert(throw_on_error,
                       "throw_on_error needs to be enabled to use this method");
@@ -2910,17 +3021,18 @@ private:
     }
 
     struct reader {
-        reader(const std::string& file_name_, const std::string& delim)
-            : delim_{delim}, file_{std::fopen(file_name_.c_str(), "rb")} {
+        reader(const std::string& file_name_, std::string delim)
+            : delim_{std::move(delim)},
+              file_{std::fopen(file_name_.c_str(), "rb")} {
         }
 
         reader(const char* const buffer, size_t csv_data_size,
-               const std::string& delim)
-            : delim_{delim}, csv_data_buffer_{buffer},
+               std::string delim)
+            : delim_{std::move(delim)}, csv_data_buffer_{buffer},
               csv_data_size_{csv_data_size} {
         }
 
-        reader(reader&& other)
+        reader(reader&& other) noexcept
             : buffer_{other.buffer_},
               next_line_buffer_{other.next_line_buffer_},
               helper_buffer_{other.helper_buffer_},
@@ -2941,7 +3053,7 @@ private:
             other.file_ = nullptr;
         }
 
-        reader& operator=(reader&& other) {
+        reader& operator=(reader&& other) noexcept {
             if (this != &other) {
                 buffer_ = other.buffer_;
                 next_line_buffer_ = other.next_line_buffer_;
@@ -3089,7 +3201,7 @@ private:
         }
 
         bool escaped_eol(size_t size) {
-            const char* curr;
+            const char* curr = nullptr;
             for (curr = next_line_buffer_ + size - 1;
                  curr >= next_line_buffer_ &&
                  setup<Options...>::escape::match(*curr);
@@ -3135,7 +3247,7 @@ private:
                             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*>(
+            auto* new_first = static_cast<char*>(
                 strict_realloc(static_cast<void*>(first), buffer_size));
 
             first = new_first;
@@ -3164,7 +3276,7 @@ private:
         }
 
         std::string get_buffer() {
-            return std::string{next_line_buffer_, next_line_buffer_size_};
+            return std::string{next_line_buffer_, next_line_size_};
         }
 
         ////////////////
@@ -3207,4 +3319,4 @@ private:
     bool eof_{false};
 };
 
-} /* ss */
+} /* namespace ss */

test/test_extractions.cpp

@@ -23,7 +23,7 @@ struct is_unsigned : public std::is_unsigned<T> {};
 template <>
 struct is_unsigned<ss::uint8> : public std::true_type {};
 
-} /* namespace */
+} /* anonymous namespace */
 
 static_assert(is_signed<ss::int8>::value);
 static_assert(is_unsigned<ss::uint8>::value);

test/test_helpers.hpp

@@ -19,7 +19,7 @@
 namespace ss {
 template <typename... Ts>
 class parser;
-} /* ss */
+} /* namespace ss */
 
 namespace {
 
@@ -145,6 +145,17 @@ struct unique_file_name {
         CHECK_FALSE(std::string{e.what()}.empty());                            \
     }
 
+#define CHECK_EQ_ARRAY(first, second)                                          \
+    {                                                                          \
+        const auto& first_ = (first);                                          \
+        const auto& second_ = (second);                                        \
+        CHECK_EQ(first_.size(), second_.size());                               \
+        for (size_t i_ = 0; i_ < std::min(first_.size(), second_.size());      \
+             ++i_) {                                                           \
+            CHECK_EQ(first_[i_], second_[i_]);                                 \
+        }                                                                      \
+    }
+
 template <typename T>
 [[maybe_unused]] std::vector<std::vector<T>> vector_combinations(
     const std::vector<T>& v, size_t n) {
@@ -166,6 +177,22 @@ template <typename T>
     return ret;
 }
 
+[[maybe_unused]] std::string merge_header(
+    const std::vector<std::string>& header,
+    const std::string& delimiter = ss::default_delimiter) {
+    std::string s;
+    if (!header.empty()) {
+        for (const auto& i : header) {
+            s.append(i);
+            s.append(delimiter);
+        }
+        for (size_t i = 0; i < delimiter.size(); ++i) {
+            s.pop_back();
+        }
+    }
+    return s;
+};
+
 [[maybe_unused]] std::string make_buffer(const std::string& file_name) {
     std::ifstream in{file_name, std::ios::binary};
     std::string tmp;
@@ -224,4 +251,4 @@ make_parser(const std::string& file_name,
     return make_parser_impl<buffer_mode, Ts...>(file_name, delim);
 }
 
-} /* namespace */
+} /* anonymous namespace */

test/test_parser1.hpp

@@ -28,6 +28,7 @@ void expect_error_on_command(ss::parser<Ts...>& p,
     if (ss::setup<Ts...>::throw_on_error) {
         try {
             command();
+            FAIL("expected exception");
         } catch (const std::exception& e) {
             CHECK_FALSE(std::string{e.what()}.empty());
         }
@@ -109,4 +110,4 @@ static void make_and_write(const std::string& file_name,
     }
 }
 
-} /* namespace */
+} /* anonymous namespace */

test/test_parser1_4.cpp

@@ -9,6 +9,7 @@ struct has_type<T, std::tuple<Us...>>
 
 template <typename T, typename... Ts>
 static void test_fields(const std::string file_name, const std::vector<X>& data,
+                        const std::vector<std::string>& header,
                         const std::vector<std::string>& fields) {
     constexpr auto buffer_mode = T::BufferMode::value;
     using ErrorMode = typename T::ErrorMode;
@@ -17,9 +18,14 @@ static void test_fields(const std::string file_name, const std::vector<X>& data,
     auto [p, _] = make_parser<buffer_mode, ErrorMode>(file_name, ",");
     CHECK_FALSE(p.field_exists("Unknown"));
     p.use_fields(fields);
+
+    CHECK_EQ_ARRAY(header, p.header());
+    CHECK_EQ(merge_header(p.header(), ","), p.raw_header());
     std::vector<CaseType> i;
 
     for (const auto& a : p.template iterate<CaseType>()) {
+        CHECK_EQ_ARRAY(header, p.header());
+        CHECK_EQ(merge_header(p.header(), ","), p.raw_header());
         i.push_back(a);
     }
 
@@ -40,12 +46,12 @@ static void test_fields(const std::string file_name, const std::vector<X>& data,
 TEST_CASE_TEMPLATE("test various cases with header", T,
                    ParserOptionCombinations) {
     unique_file_name f{"various_cases_with_header"};
+    using str = std::string;
+
     constexpr static auto Int = "Int";
     constexpr static auto Dbl = "Double";
     constexpr static auto Str = "String";
-    using str = std::string;
-
-    std::vector<std::string> header{Int, Dbl, Str};
+    const std::vector<std::string> header{Int, Dbl, Str};
 
     std::vector<X> data = {{1, 2, "x"}, {3, 4, "y"},  {5, 6, "z"},
                            {7, 8, "u"}, {9, 10, "v"}, {11, 12, "w"}};
@@ -59,6 +65,8 @@ TEST_CASE_TEMPLATE("test various cases with header", T,
         std::vector<X> i;
 
         for (const auto& a : p.iterate<int, double, std::string>()) {
+            CHECK_EQ(header, p.header());
+            CHECK_EQ(merge_header(p.header(), ","), p.raw_header());
             i.emplace_back(ss::to_object<X>(a));
         }
 
@@ -71,46 +79,22 @@ TEST_CASE_TEMPLATE("test various cases with header", T,
 
         p.ignore_next();
         for (const auto& a : p.iterate<int, double, std::string>()) {
+            CHECK_EQ(header, p.header());
+            CHECK_EQ(merge_header(p.header(), ","), p.raw_header());
             i.emplace_back(ss::to_object<X>(a));
         }
 
         CHECK_EQ(i, data);
     }
 
-    {
-        ss::parser<ss::ignore_header> p{f.name, ","};
-        std::vector<X> i;
-
-        for (const auto& a : p.iterate<int, double, std::string>()) {
-            i.emplace_back(ss::to_object<X>(a));
-        }
-
-        CHECK_EQ(i, data);
-    }
-
-    {
-        ss::parser<ss::ignore_header, ss::string_error> p{f.name, ","};
-        p.use_fields(Int, Dbl, Str);
-        CHECK_FALSE(p.valid());
-    }
-
-    {
-        ss::parser<ss::ignore_header, ss::string_error> p{f.name, ","};
-        CHECK_FALSE(p.field_exists("Unknown"));
-
-        p.use_fields(Int, "Unknown");
-        CHECK_FALSE(p.valid());
-    }
-
-    {
-        ss::parser<ss::ignore_header, ss::string_error> p{f.name, ","};
-        p.use_fields(Int, Int);
-        CHECK_FALSE(p.valid());
-    }
-
     {
         ss::parser<ss::string_error> p{f.name, ","};
+        CHECK_EQ(header, p.header());
+        CHECK_EQ(merge_header(p.header(), ","), p.raw_header());
+
         p.use_fields(Int, Dbl);
+        CHECK_EQ(header, p.header());
+        CHECK_EQ(merge_header(p.header(), ","), p.raw_header());
 
         {
             auto [int_, double_] = p.get_next<int, double>();
@@ -119,6 +103,8 @@ TEST_CASE_TEMPLATE("test various cases with header", T,
         }
 
         p.use_fields(Dbl, Int);
+        CHECK_EQ(header, p.header());
+        CHECK_EQ(merge_header(p.header(), ","), p.raw_header());
 
         {
             auto [double_, int_] = p.get_next<double, int>();
@@ -163,25 +149,25 @@ TEST_CASE_TEMPLATE("test various cases with header", T,
                     template_params.append(type)
                     arg_params.append(header[type])
                 call = 'testFields<' + ', '.join(template_params) + \
-                    '>(o, d, {' + ', '.join(arg_params) + '});'
+                    '>(o, d, header, {' + ', '.join(arg_params) + '});'
                 print(call)
         */
 
-    test_fields<T, str>(o, d, {Str});
-    test_fields<T, int>(o, d, {Int});
-    test_fields<T, double>(o, d, {Dbl});
-    test_fields<T, str, int>(o, d, {Str, Int});
-    test_fields<T, str, double>(o, d, {Str, Dbl});
-    test_fields<T, int, str>(o, d, {Int, Str});
-    test_fields<T, int, double>(o, d, {Int, Dbl});
-    test_fields<T, double, str>(o, d, {Dbl, Str});
-    test_fields<T, double, int>(o, d, {Dbl, Int});
-    test_fields<T, str, int, double>(o, d, {Str, Int, Dbl});
-    test_fields<T, str, double, int>(o, d, {Str, Dbl, Int});
-    test_fields<T, int, str, double>(o, d, {Int, Str, Dbl});
-    test_fields<T, int, double, str>(o, d, {Int, Dbl, Str});
-    test_fields<T, double, str, int>(o, d, {Dbl, Str, Int});
-    test_fields<T, double, int, str>(o, d, {Dbl, Int, Str});
+    test_fields<T, str>(o, d, header, {Str});
+    test_fields<T, int>(o, d, header, {Int});
+    test_fields<T, double>(o, d, header, {Dbl});
+    test_fields<T, str, int>(o, d, header, {Str, Int});
+    test_fields<T, str, double>(o, d, header, {Str, Dbl});
+    test_fields<T, int, str>(o, d, header, {Int, Str});
+    test_fields<T, int, double>(o, d, header, {Int, Dbl});
+    test_fields<T, double, str>(o, d, header, {Dbl, Str});
+    test_fields<T, double, int>(o, d, header, {Dbl, Int});
+    test_fields<T, str, int, double>(o, d, header, {Str, Int, Dbl});
+    test_fields<T, str, double, int>(o, d, header, {Str, Dbl, Int});
+    test_fields<T, int, str, double>(o, d, header, {Int, Str, Dbl});
+    test_fields<T, int, double, str>(o, d, header, {Int, Dbl, Str});
+    test_fields<T, double, str, int>(o, d, header, {Dbl, Str, Int});
+    test_fields<T, double, int, str>(o, d, header, {Dbl, Int, Str});
 }
 
 template <typename T>
@@ -190,6 +176,17 @@ void test_invalid_fields(const std::vector<std::string>& lines,
     constexpr auto buffer_mode = T::BufferMode::value;
     using ErrorMode = typename T::ErrorMode;
 
+    auto check_header = [&lines](auto& p) {
+        if (lines.empty()) {
+            CHECK(p.header().empty());
+            CHECK_EQ(merge_header(p.header(), ","), p.raw_header());
+        } else {
+            CHECK_EQ(lines[0], merge_header(p.header()));
+            CHECK_EQ(merge_header(p.header(), ","), p.raw_header());
+        }
+        CHECK(p.valid());
+    };
+
     unique_file_name f{"invalid_fields"};
     {
         std::ofstream out{f.name};
@@ -203,6 +200,7 @@ void test_invalid_fields(const std::vector<std::string>& lines,
         auto [p, _] = make_parser<buffer_mode, ErrorMode>(f.name, ",");
         auto command = [&p = p] { p.use_fields(); };
         expect_error_on_command(p, command);
+        check_header(p);
     }
 
     {
@@ -210,6 +208,7 @@ void test_invalid_fields(const std::vector<std::string>& lines,
         auto [p, _] = make_parser<buffer_mode, ErrorMode>(f.name, ",");
         auto command = [&p = p] { p.use_fields("Unknown"); };
         expect_error_on_command(p, command);
+        check_header(p);
     }
 
     {
@@ -221,6 +220,7 @@ void test_invalid_fields(const std::vector<std::string>& lines,
         if (!fields.empty()) {
             expect_error_on_command(p, command);
         }
+        check_header(p);
     }
 
     {
@@ -230,15 +230,19 @@ void test_invalid_fields(const std::vector<std::string>& lines,
             p.use_fields(fields.at(0));
             p.template get_next<std::string, std::string>();
         };
+        check_header(p);
+
         if (!fields.empty()) {
             expect_error_on_command(p, command);
         }
+        check_header(p);
     }
 
     {
         // Invalid header
         auto [p, _] = make_parser<buffer_mode, ErrorMode>(f.name, ",");
         auto command = [&p = p, &fields = fields] { p.use_fields(fields); };
+        check_header(p);
 
         if (!fields.empty()) {
             // Pass if there are no duplicates, fail otherwise
@@ -255,6 +259,7 @@ void test_invalid_fields(const std::vector<std::string>& lines,
                 }
             }
         }
+        check_header(p);
     }
 }
 
@@ -289,7 +294,7 @@ TEST_CASE_TEMPLATE("test invalid rows with header", T,
     constexpr auto buffer_mode = T::BufferMode::value;
     using ErrorMode = typename T::ErrorMode;
 
-    unique_file_name f{"invalid rows with header"};
+    unique_file_name f{"invalid_rows_with_header"};
     {
         std::ofstream out{f.name};
         out << "Int,String,Double" << std::endl;
@@ -301,8 +306,12 @@ TEST_CASE_TEMPLATE("test invalid rows with header", T,
         out << "six,line6,10.11" << std::endl;
     }
 
+    std::vector<std::string> header = {"Int", "String", "Double"};
+
     {
         auto [p, _] = make_parser<buffer_mode, ErrorMode>(f.name);
+        CHECK_EQ_ARRAY(header, p.header());
+        CHECK_EQ(merge_header(p.header()), p.raw_header());
 
         p.use_fields("Int", "String", "Double");
         using data = std::tuple<int, std::string, double>;
@@ -325,10 +334,14 @@ TEST_CASE_TEMPLATE("test invalid rows with header", T,
                                       {3, "line3", 67.8},
                                       {5, "line5", 9.10}};
         CHECK_EQ(i, expected);
+        CHECK_EQ_ARRAY(header, p.header());
+        CHECK_EQ(merge_header(p.header()), p.raw_header());
     }
 
     {
         auto [p, _] = make_parser<buffer_mode, ErrorMode>(f.name);
+        CHECK_EQ_ARRAY(header, p.header());
+        CHECK_EQ(merge_header(p.header()), p.raw_header());
 
         p.use_fields("Double", "Int");
         using data = std::tuple<double, int>;
@@ -349,10 +362,14 @@ TEST_CASE_TEMPLATE("test invalid rows with header", T,
 
         std::vector<data> expected = {{2.34, 1}, {67.8, 3}, {9.10, 5}};
         CHECK_EQ(i, expected);
+        CHECK_EQ_ARRAY(header, p.header());
+        CHECK_EQ(merge_header(p.header()), p.raw_header());
     }
 
     {
         auto [p, _] = make_parser<buffer_mode, ErrorMode>(f.name);
+        CHECK_EQ_ARRAY(header, p.header());
+        CHECK_EQ(merge_header(p.header()), p.raw_header());
 
         p.use_fields("String", "Double");
         using data = std::tuple<std::string, double>;
@@ -376,6 +393,92 @@ TEST_CASE_TEMPLATE("test invalid rows with header", T,
                                       {"line5", 9.10},
                                       {"line6", 10.11}};
         CHECK_EQ(i, expected);
+        CHECK_EQ_ARRAY(header, p.header());
+        CHECK_EQ(merge_header(p.header()), p.raw_header());
+    }
+}
+
+TEST_CASE_TEMPLATE("test invalid header", T, ParserOptionCombinations) {
+    constexpr auto buffer_mode = T::BufferMode::value;
+    using ErrorMode = typename T::ErrorMode;
+
+    unique_file_name f{"invalid_header"};
+
+    // Empty header
+    {
+        std::ofstream out{f.name};
+        out << "" << std::endl;
+        out << "1" << std::endl;
+    }
+
+    {
+        auto [p, _] = make_parser<buffer_mode, ErrorMode>(f.name);
+        CHECK(p.header().empty());
+        CHECK_EQ(merge_header(p.header()), p.raw_header());
+        CHECK(p.valid());
+    }
+
+    // Unterminated quote in header
+    {
+        std::ofstream out{f.name};
+        out << "\"Int" << std::endl;
+        out << "1" << std::endl;
+    }
+
+    {
+        auto [p, _] =
+            make_parser<buffer_mode, ErrorMode, ss::quote<'"'>>(f.name);
+        auto command = [&p = p] { p.header(); };
+        expect_error_on_command(p, command);
+        CHECK_EQ(p.raw_header(), "\"Int");
+    }
+
+    {
+        auto [p, _] =
+            make_parser<buffer_mode, ErrorMode, ss::quote<'"'>, ss::multiline>(
+                f.name);
+        auto command = [&p = p] { p.header(); };
+        expect_error_on_command(p, command);
+        CHECK_EQ(p.raw_header(), "\"Int");
+    }
+
+    {
+        auto [p, _] = make_parser<buffer_mode, ErrorMode, ss::quote<'"'>,
+                                  ss::escape<'\\'>, ss::multiline>(f.name);
+        auto command = [&p = p] { p.header(); };
+        expect_error_on_command(p, command);
+        CHECK_EQ(p.raw_header(), "\"Int");
+    }
+
+    // Unterminated escape in header
+    {
+        std::ofstream out{f.name};
+        out << "Int\\" << std::endl;
+        out << "1" << std::endl;
+    }
+
+    {
+        auto [p, _] =
+            make_parser<buffer_mode, ErrorMode, ss::escape<'\\'>>(f.name);
+        auto command = [&p = p] { p.header(); };
+        expect_error_on_command(p, command);
+        CHECK_EQ(p.raw_header(), "Int\\");
+    }
+
+    {
+        auto [p, _] = make_parser<buffer_mode, ErrorMode, ss::escape<'\\'>,
+                                  ss::multiline>(f.name);
+        auto command = [&p = p] { p.header(); };
+        expect_error_on_command(p, command);
+        CHECK_EQ(p.raw_header(), "Int\\");
+    }
+
+    {
+        auto [p, _] = make_parser<buffer_mode, ErrorMode, ss::escape<'\\'>,
+                                  ss::quote<'"'>, ss::multiline>(f.name);
+        auto command = [&p = p] { p.header(); };
+        expect_error_on_command(p, command);
+        CHECK_EQ(p.raw_header(), "Int\\");
     }
 }
 

test/test_parser2.hpp

@@ -1,13 +1,7 @@
 #include "test_helpers.hpp"
-#include <algorithm>
-#include <filesystem>
 #include <fstream>
-#include <iomanip>
 #include <iostream>
-#include <regex>
 #include <ss/parser.hpp>
-#include <sstream>
-#include <unordered_map>
 #include <unordered_set>
 
 #ifndef SEGMENT_NAME
@@ -333,8 +327,10 @@ void test_data_combinations(const std::vector<column>& input_data,
         field_header.push_back(field{el.header});
     }
 
+    std::string header_line;
     if (include_header) {
         auto header_data = generate_csv_data<Ts...>(field_header, delim);
+        header_line = merge_header(header_data, delim);
         if (input_data.size() == 0 && rand() % 10 == 0) {
             write_to_file(header_data, delim, f.name, false);
         } else {
@@ -403,7 +399,9 @@ void test_data_combinations(const std::vector<column>& input_data,
                 fields.push_back(header[index]);
             }
 
+            if constexpr (!setup::ignore_header) {
                 p.use_fields(fields);
+            }
 
             if (!p.valid()) {
                 if constexpr (setup::string_error) {
@@ -425,8 +423,19 @@ void test_data_combinations(const std::vector<column>& input_data,
             }
         };
 
+        auto check_header = [&p = p, &header = header, include_header,
+                             header_line] {
+            if (include_header) {
+                if constexpr (!setup::ignore_header) {
+                    CHECK_EQ_ARRAY(header, p.header());
+                    CHECK_EQ(header_line, p.raw_header());
+                }
+            }
+        };
+
         int num_columns = layout.size();
         for (size_t i = 0; i < n + 1; ++i) {
+            check_header();
             try {
                 switch (num_columns) {
                 case 1: {
@@ -616,7 +625,7 @@ void test_option_combinations3() {
     test_option_combinations2<Ts..., trim>();
 }
 
-} /* namespace */
+} /* anonymous namespace */
 
 // Tests split into multiple compilation units
 #if 0

test/test_single_header.sh

@@ -3,12 +3,14 @@
 set -x
 set -e
 
-python3 script/single_header_generator.py > ssp.cpp
+TMP_HDR=test_single_header.hpp
+TMP_SRC=test_single_header.cpp
+TMP_BIN=test_single_header
 
-echo 'int main(){ ss::parser p{""}; p.get_next<int, float>(); return 0; }' \
-    >> ssp.cpp
+python3 script/single_header_generator.py > ${TMP_HDR}
+cat ${TMP_HDR} test/test_single_header_main.txt > ${TMP_SRC}
 
-g++ -std=c++17 ssp.cpp -o ssp.bin -Wall -Wextra
-./ssp.bin
+g++ -std=c++17 ${TMP_SRC} -o ${TMP_BIN} -Wall -Wextra
+./${TMP_BIN}
 
-rm ssp.cpp ssp.bin
+rm ${TMP_HDR} ${TMP_SRC} ${TMP_BIN}

test/test_single_header_main.txt

@@ -0,0 +1,12 @@
+int main() {
+    using quote = ss::quote<'"'>;
+    using escape = ss::escape<'\\'>;
+    using trim = ss::trim<' '>;
+
+    std::string data = "1,string,2.34,c";
+
+    ss::parser<quote, escape, trim, ss::multiline> p{data.c_str(), data.size()};
+    auto tup = p.get_next<int, std::string, float, std::optional<char>>();
+
+    return 0;
+}

test/test_splitter.cpp

@@ -145,7 +145,7 @@ make_combinations(const std::vector<std::string>& input,
 
     return {std::move(lines), std::move(expectations)};
 }
-} /* namespace */
+} /* anonymous namespace */
 
 /* ********************************** */
 /* ********************************** */
@@ -548,7 +548,7 @@ public:
         return splitter.size_shifted();
     }
 };
-} /* ss */
+} /* namespace ss */
 
 TEST_CASE("splitter test resplit unterminated quote") {