include/ss/parser.hpp

@@ -531,6 +531,10 @@ private:
 
     [[nodiscard]] 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_);
@@ -558,6 +562,11 @@ private:
 
         for (const auto& [begin, end] : splitter.get_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_duplicate_header_field(field);

ssp.hpp

@@ -2805,6 +2805,10 @@ private:
 
     [[nodiscard]] 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_);
@@ -2832,6 +2836,11 @@ private:
 
         for (const auto& [begin, end] : splitter.get_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_duplicate_header_field(field);

test/CMakeLists.txt

@@ -33,11 +33,10 @@ set(DOCTEST "${FETCHCONTENT_BASE_DIR}/doctest-src")
 enable_testing()
 
 foreach(name IN ITEMS test_splitter test_parser1_1 test_parser1_2
-                      test_parser1_3 test_parser1_4 test_parser1_5
-                      test_converter test_extractions test_parser2_1
-                      test_parser2_2 test_parser2_3 test_parser2_4
-                      test_parser2_5 test_parser2_6
-                      test_extractions_without_fast_float)
+                      test_parser1_3 test_parser1_4 test_converter
+                      test_extractions test_parser2_1 test_parser2_2
+                      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

@@ -6,7 +6,6 @@ tests = [
   'parser1_2',
   'parser1_3',
   'parser1_4',
-  'parser1_5',
   'splitter',
   'converter',
   'extractions',

test/test_parser1_4.cpp

@@ -178,8 +178,7 @@ void test_invalid_fields(const std::vector<std::string>& lines,
 
     auto check_header = [&lines](auto& p) {
         if (lines.empty()) {
-            CHECK_EQ(p.header().size(), 1);
-            CHECK_EQ(p.header().at(0), "");
+            CHECK(p.header().empty());
             CHECK_EQ(merge_header(p.header(), ","), p.raw_header());
         } else {
             CHECK_EQ(lines[0], merge_header(p.header()));
@@ -264,7 +263,7 @@ void test_invalid_fields(const std::vector<std::string>& lines,
     }
 }
 
-TEST_CASE_TEMPLATE("test invalid header fields usage", T,
+TEST_CASE_TEMPLATE("test invalid fheader fields usage", T,
                    ParserOptionCombinations) {
     test_invalid_fields<T>({}, {});
 
@@ -398,3 +397,178 @@ TEST_CASE_TEMPLATE("test invalid rows with header", T,
         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] { std::ignore = 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] { std::ignore = 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] { std::ignore = 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] { std::ignore = 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] { std::ignore = 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] { std::ignore = p.header(); };
+        expect_error_on_command(p, command);
+        CHECK_EQ(p.raw_header(), "Int\\");
+    }
+}
+
+template <typename T>
+void test_ignore_empty(const std::vector<X>& data) {
+    constexpr auto buffer_mode = T::BufferMode::value;
+    using ErrorMode = typename T::ErrorMode;
+
+    unique_file_name f{"ignore_empty"};
+    make_and_write(f.name, data);
+
+    std::vector<X> expected;
+    for (const auto& d : data) {
+        if (d.s != X::empty) {
+            expected.push_back(d);
+        }
+    }
+
+    {
+        auto [p, _] =
+            make_parser<buffer_mode, ErrorMode, ss::ignore_empty>(f.name, ",");
+
+        std::vector<X> i;
+        for (const auto& a : p.template iterate<X>()) {
+            i.push_back(a);
+        }
+
+        CHECK_EQ(i, expected);
+    }
+
+    {
+        auto [p, _] = make_parser<buffer_mode, ErrorMode>(f.name, ",");
+        std::vector<X> i;
+        size_t n = 0;
+        while (!p.eof()) {
+            try {
+                ++n;
+                const auto& a = p.template get_next<X>();
+                if (data.at(n - 1).s == X::empty) {
+                    CHECK_FALSE(p.valid());
+                    continue;
+                }
+                i.push_back(a);
+            } catch (...) {
+                CHECK_EQ(data.at(n - 1).s, X::empty);
+            }
+        }
+
+        CHECK_EQ(i, expected);
+    }
+}
+
+TEST_CASE_TEMPLATE("test various cases with empty lines", T,
+                   ParserOptionCombinations) {
+    test_ignore_empty<T>(
+        {{1, 2, "x"}, {3, 4, "y"}, {9, 10, "v"}, {11, 12, "w"}});
+
+    test_ignore_empty<T>(
+        {{1, 2, X::empty}, {3, 4, "y"}, {9, 10, "v"}, {11, 12, "w"}});
+
+    test_ignore_empty<T>(
+        {{1, 2, "x"}, {3, 4, "y"}, {9, 10, "v"}, {11, 12, X::empty}});
+
+    test_ignore_empty<T>(
+        {{1, 2, "x"}, {5, 6, X::empty}, {9, 10, "v"}, {11, 12, "w"}});
+
+    test_ignore_empty<T>(
+        {{1, 2, X::empty}, {5, 6, X::empty}, {9, 10, "v"}, {11, 12, "w"}});
+
+    test_ignore_empty<T>(
+        {{1, 2, X::empty}, {3, 4, "y"}, {9, 10, "v"}, {11, 12, X::empty}});
+
+    test_ignore_empty<T>(
+        {{1, 2, "x"}, {3, 4, "y"}, {9, 10, X::empty}, {11, 12, X::empty}});
+
+    test_ignore_empty<T>(
+        {{1, 2, X::empty}, {3, 4, "y"}, {9, 10, X::empty}, {11, 12, X::empty}});
+
+    test_ignore_empty<T>({{1, 2, X::empty},
+                          {3, 4, X::empty},
+                          {9, 10, X::empty},
+                          {11, 12, X::empty}});
+
+    test_ignore_empty<T>(
+        {{1, 2, "x"}, {3, 4, X::empty}, {9, 10, X::empty}, {11, 12, X::empty}});
+
+    test_ignore_empty<T>(
+        {{1, 2, X::empty}, {3, 4, X::empty}, {9, 10, X::empty}, {11, 12, "w"}});
+
+    test_ignore_empty<T>({{11, 12, X::empty}});
+
+    test_ignore_empty<T>({});
+}

test/test_parser1_5.cpp

@@ -1,301 +0,0 @@
-#include "test_parser1.hpp"
-
-TEST_CASE_TEMPLATE("test empty fields header", T, ParserOptionCombinations) {
-    constexpr auto buffer_mode = T::BufferMode::value;
-    using ErrorMode = typename T::ErrorMode;
-
-    unique_file_name f{"empty_fields_header"};
-
-    // Empty header
-    {
-        std::ofstream out{f.name};
-        out << "" << std::endl;
-        out << "1" << std::endl;
-    }
-
-    {
-        std::vector<std::string> expected_header = {""};
-        auto [p, _] = make_parser<buffer_mode, ErrorMode>(f.name);
-        CHECK_EQ_ARRAY(expected_header, p.header());
-        CHECK_EQ("", p.raw_header());
-        CHECK(p.valid());
-    }
-
-    // All empty header fields
-    {
-        std::ofstream out{f.name};
-        out << ",," << std::endl;
-        out << "1,2,3" << std::endl;
-    }
-
-    {
-        std::vector<std::string> expected_header = {"", "", ""};
-        auto [p, _] = make_parser<buffer_mode, ErrorMode>(f.name);
-        CHECK_EQ_ARRAY(expected_header, p.header());
-        CHECK_EQ(",,", p.raw_header());
-        CHECK(p.valid());
-
-        auto command1 = [&p = p] { std::ignore = p.field_exists("Int"); };
-        expect_error_on_command(p, command1);
-
-        auto command2 = [&p = p] { p.use_fields("Int"); };
-        expect_error_on_command(p, command2);
-    }
-
-    // One empty field
-    const std::vector<std::string> valid_fields = {"Int0", "Int1", ""};
-
-    using svec = std::vector<std::string>;
-    const std::vector<std::vector<std::string>> valid_field_combinations =
-        {svec{"Int0"},
-         svec{"Int1"},
-         svec{""},
-         svec{"", "Int0"},
-         svec{"Int0", "Int1"},
-         svec{"Int1", ""},
-         svec{"Int0", "", "Int1"},
-         svec{"", "Int1", "Int0"}};
-
-    // Last header field empty
-    {
-        std::ofstream out{f.name};
-        out << "Int0,Int1," << std::endl;
-        out << "1,2,3" << std::endl;
-    }
-
-    {
-        std::vector<std::string> expected_header = {"Int0", "Int1", ""};
-        auto [p, _] = make_parser<buffer_mode, ErrorMode>(f.name);
-        CHECK_EQ_ARRAY(expected_header, p.header());
-        CHECK_EQ("Int0,Int1,", p.raw_header());
-        CHECK(p.valid());
-
-        for (const auto& field : valid_fields) {
-            CHECK(p.field_exists(field));
-            CHECK(p.valid());
-        }
-
-        for (const auto& fields : valid_field_combinations) {
-            p.use_fields(fields);
-            CHECK(p.valid());
-        }
-    }
-
-    // First header field empty
-    {
-        std::ofstream out{f.name};
-        out << ",Int0,Int1" << std::endl;
-        out << "1,2,3" << std::endl;
-    }
-
-    {
-        std::vector<std::string> expected_header = {"", "Int0", "Int1"};
-        auto [p, _] = make_parser<buffer_mode, ErrorMode>(f.name);
-        CHECK_EQ_ARRAY(expected_header, p.header());
-        CHECK_EQ(",Int0,Int1", p.raw_header());
-        CHECK(p.valid());
-
-        for (const auto& field : valid_fields) {
-            CHECK(p.field_exists(field));
-            CHECK(p.valid());
-        }
-
-        for (const auto& fields : valid_field_combinations) {
-            p.use_fields(fields);
-            CHECK(p.valid());
-        }
-    }
-
-    // Middle header field empty
-    {
-        std::ofstream out{f.name};
-        out << "Int0,,Int1" << std::endl;
-        out << "1,2,3" << std::endl;
-    }
-
-    {
-        std::vector<std::string> expected_header = {"Int0", "", "Int1"};
-        auto [p, _] = make_parser<buffer_mode, ErrorMode>(f.name);
-        CHECK_EQ_ARRAY(expected_header, p.header());
-        CHECK_EQ("Int0,,Int1", p.raw_header());
-        CHECK(p.valid());
-
-        for (const auto& field : valid_fields) {
-            CHECK(p.field_exists(field));
-            CHECK(p.valid());
-        }
-
-        for (const auto& fields : valid_field_combinations) {
-            p.use_fields(fields);
-            CHECK(p.valid());
-        }
-    }
-}
-
-template <typename T, typename... Ts>
-void test_unterminated_quote_header() {
-    constexpr auto buffer_mode = T::BufferMode::value;
-    using ErrorMode = typename T::ErrorMode;
-
-    unique_file_name f{"unterminated_quote_header"};
-
-    {
-        std::ofstream out{f.name};
-        out << "\"Int" << std::endl;
-        out << "1" << std::endl;
-    }
-
-    {
-        auto [p, _] = make_parser<buffer_mode, ErrorMode, Ts...>(f.name);
-
-        auto command0 = [&p = p] { std::ignore = p.header(); };
-        expect_error_on_command(p, command0);
-        CHECK_EQ(p.raw_header(), "\"Int");
-
-        auto command1 = [&p = p] { std::ignore = p.field_exists("Int"); };
-        expect_error_on_command(p, command1);
-
-        auto command2 = [&p = p] { p.use_fields("Int"); };
-        expect_error_on_command(p, command2);
-    }
-}
-
-TEST_CASE_TEMPLATE("test unterminated quote header", T,
-                   ParserOptionCombinations) {
-    using quote = ss::quote<'"'>;
-    using escape = ss::escape<'\\'>;
-    test_unterminated_quote_header<T, quote>();
-    test_unterminated_quote_header<T, quote, ss::multiline>();
-    test_unterminated_quote_header<T, quote, escape>();
-    test_unterminated_quote_header<T, quote, escape, ss::multiline>();
-}
-
-template <typename T, typename... Ts>
-void test_unterminated_escape_header() {
-    constexpr auto buffer_mode = T::BufferMode::value;
-    using ErrorMode = typename T::ErrorMode;
-
-    unique_file_name f{"unterminated_escape_header"};
-
-    // Unterminated escape in header
-    {
-        std::ofstream out{f.name};
-        out << "Int\\" << std::endl;
-        out << "1" << std::endl;
-    }
-
-    {
-        auto [p, _] = make_parser<buffer_mode, ErrorMode, Ts...>(f.name);
-
-        auto command0 = [&p = p] { std::ignore = p.header(); };
-        expect_error_on_command(p, command0);
-        CHECK_EQ(p.raw_header(), "Int\\");
-
-        auto command1 = [&p = p] { std::ignore = p.field_exists("Int"); };
-        expect_error_on_command(p, command1);
-
-        auto command2 = [&p = p] { p.use_fields("Int"); };
-        expect_error_on_command(p, command2);
-    }
-}
-
-TEST_CASE_TEMPLATE("test unterminated escape header", T,
-                   ParserOptionCombinations) {
-    using quote = ss::quote<'"'>;
-    using escape = ss::escape<'\\'>;
-    test_unterminated_escape_header<T, escape>();
-    test_unterminated_escape_header<T, escape, ss::multiline>();
-    test_unterminated_escape_header<T, escape, quote>();
-    test_unterminated_escape_header<T, escape, quote, ss::multiline>();
-}
-
-template <typename T>
-void test_ignore_empty(const std::vector<X>& data) {
-    constexpr auto buffer_mode = T::BufferMode::value;
-    using ErrorMode = typename T::ErrorMode;
-
-    unique_file_name f{"ignore_empty"};
-    make_and_write(f.name, data);
-
-    std::vector<X> expected;
-    for (const auto& d : data) {
-        if (d.s != X::empty) {
-            expected.push_back(d);
-        }
-    }
-
-    {
-        auto [p, _] =
-            make_parser<buffer_mode, ErrorMode, ss::ignore_empty>(f.name, ",");
-
-        std::vector<X> i;
-        for (const auto& a : p.template iterate<X>()) {
-            i.push_back(a);
-        }
-
-        CHECK_EQ(i, expected);
-    }
-
-    {
-        auto [p, _] = make_parser<buffer_mode, ErrorMode>(f.name, ",");
-        std::vector<X> i;
-        size_t n = 0;
-        while (!p.eof()) {
-            try {
-                ++n;
-                const auto& a = p.template get_next<X>();
-                if (data.at(n - 1).s == X::empty) {
-                    CHECK_FALSE(p.valid());
-                    continue;
-                }
-                i.push_back(a);
-            } catch (...) {
-                CHECK_EQ(data.at(n - 1).s, X::empty);
-            }
-        }
-
-        CHECK_EQ(i, expected);
-    }
-}
-
-TEST_CASE_TEMPLATE("test various cases with empty lines", T,
-                   ParserOptionCombinations) {
-    test_ignore_empty<T>(
-        {{1, 2, "x"}, {3, 4, "y"}, {9, 10, "v"}, {11, 12, "w"}});
-
-    test_ignore_empty<T>(
-        {{1, 2, X::empty}, {3, 4, "y"}, {9, 10, "v"}, {11, 12, "w"}});
-
-    test_ignore_empty<T>(
-        {{1, 2, "x"}, {3, 4, "y"}, {9, 10, "v"}, {11, 12, X::empty}});
-
-    test_ignore_empty<T>(
-        {{1, 2, "x"}, {5, 6, X::empty}, {9, 10, "v"}, {11, 12, "w"}});
-
-    test_ignore_empty<T>(
-        {{1, 2, X::empty}, {5, 6, X::empty}, {9, 10, "v"}, {11, 12, "w"}});
-
-    test_ignore_empty<T>(
-        {{1, 2, X::empty}, {3, 4, "y"}, {9, 10, "v"}, {11, 12, X::empty}});
-
-    test_ignore_empty<T>(
-        {{1, 2, "x"}, {3, 4, "y"}, {9, 10, X::empty}, {11, 12, X::empty}});
-
-    test_ignore_empty<T>(
-        {{1, 2, X::empty}, {3, 4, "y"}, {9, 10, X::empty}, {11, 12, X::empty}});
-
-    test_ignore_empty<T>({{1, 2, X::empty},
-                          {3, 4, X::empty},
-                          {9, 10, X::empty},
-                          {11, 12, X::empty}});
-
-    test_ignore_empty<T>(
-        {{1, 2, "x"}, {3, 4, X::empty}, {9, 10, X::empty}, {11, 12, X::empty}});
-
-    test_ignore_empty<T>(
-        {{1, 2, X::empty}, {3, 4, X::empty}, {9, 10, X::empty}, {11, 12, "w"}});
-
-    test_ignore_empty<T>({{11, 12, X::empty}});
-
-    test_ignore_empty<T>({});
-}