Modify header parsing for empty headers, update old and add new tests for header parsing

This commit is contained in:
ado 2024-03-14 11:55:14 +01:00
parent 50de5b3a5a
commit 8b1cf56549
3 changed files with 79 additions and 50 deletions

View File

@ -531,10 +531,6 @@ private:
[[nodiscard]] bool strict_split(header_splitter& splitter, [[nodiscard]] bool strict_split(header_splitter& splitter,
std::string& header) { std::string& header) {
if (header.empty()) {
return false;
}
if constexpr (throw_on_error) { if constexpr (throw_on_error) {
try { try {
splitter.split(header.data(), reader_.delim_); splitter.split(header.data(), reader_.delim_);

View File

@ -2805,10 +2805,6 @@ private:
[[nodiscard]] bool strict_split(header_splitter& splitter, [[nodiscard]] bool strict_split(header_splitter& splitter,
std::string& header) { std::string& header) {
if (header.empty()) {
return false;
}
if constexpr (throw_on_error) { if constexpr (throw_on_error) {
try { try {
splitter.split(header.data(), reader_.delim_); splitter.split(header.data(), reader_.delim_);

View File

@ -178,7 +178,8 @@ void test_invalid_fields(const std::vector<std::string>& lines,
auto check_header = [&lines](auto& p) { auto check_header = [&lines](auto& p) {
if (lines.empty()) { if (lines.empty()) {
CHECK(p.header().empty()); CHECK_EQ(p.header().size(), 1);
CHECK_EQ(p.header().at(0), "");
CHECK_EQ(merge_header(p.header(), ","), p.raw_header()); CHECK_EQ(merge_header(p.header(), ","), p.raw_header());
} else { } else {
CHECK_EQ(lines[0], merge_header(p.header())); CHECK_EQ(lines[0], merge_header(p.header()));
@ -263,7 +264,7 @@ void test_invalid_fields(const std::vector<std::string>& lines,
} }
} }
TEST_CASE_TEMPLATE("test invalid fheader fields usage", T, TEST_CASE_TEMPLATE("test invalid header fields usage", T,
ParserOptionCombinations) { ParserOptionCombinations) {
test_invalid_fields<T>({}, {}); test_invalid_fields<T>({}, {});
@ -412,13 +413,42 @@ TEST_CASE_TEMPLATE("test invalid header", T, ParserOptionCombinations) {
} }
{ {
std::vector<std::string> expected_header = {""};
auto [p, _] = make_parser<buffer_mode, ErrorMode>(f.name); auto [p, _] = make_parser<buffer_mode, ErrorMode>(f.name);
CHECK(p.header().empty()); CHECK_EQ_ARRAY(expected_header, p.header());
CHECK_EQ(merge_header(p.header()), p.raw_header()); CHECK_EQ("", p.raw_header());
CHECK(p.valid()); CHECK(p.valid());
} }
// Unterminated quote in header // 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);
}
}
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}; std::ofstream out{f.name};
out << "\"Int" << std::endl; out << "\"Int" << std::endl;
@ -426,29 +456,36 @@ TEST_CASE_TEMPLATE("test invalid header", T, ParserOptionCombinations) {
} }
{ {
auto [p, _] = auto [p, _] = make_parser<buffer_mode, ErrorMode, Ts...>(f.name);
make_parser<buffer_mode, ErrorMode, ss::quote<'"'>>(f.name);
auto command = [&p = p] { std::ignore = p.header(); }; auto command0 = [&p = p] { std::ignore = p.header(); };
expect_error_on_command(p, command); expect_error_on_command(p, command0);
CHECK_EQ(p.raw_header(), "\"Int"); 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,
auto [p, _] = ParserOptionCombinations) {
make_parser<buffer_mode, ErrorMode, ss::quote<'"'>, ss::multiline>( using quote = ss::quote<'"'>;
f.name); using escape = ss::escape<'\\'>;
auto command = [&p = p] { std::ignore = p.header(); }; test_unterminated_quote_header<T, quote>();
expect_error_on_command(p, command); test_unterminated_quote_header<T, quote, ss::multiline>();
CHECK_EQ(p.raw_header(), "\"Int"); test_unterminated_quote_header<T, quote, escape>();
test_unterminated_quote_header<T, quote, escape, ss::multiline>();
} }
{ template <typename T, typename... Ts>
auto [p, _] = make_parser<buffer_mode, ErrorMode, ss::quote<'"'>, void test_unterminated_escape_header() {
ss::escape<'\\'>, ss::multiline>(f.name); constexpr auto buffer_mode = T::BufferMode::value;
auto command = [&p = p] { std::ignore = p.header(); }; using ErrorMode = typename T::ErrorMode;
expect_error_on_command(p, command);
CHECK_EQ(p.raw_header(), "\"Int"); unique_file_name f{"unterminated_escape_header"};
}
// Unterminated escape in header // Unterminated escape in header
{ {
@ -458,28 +495,28 @@ TEST_CASE_TEMPLATE("test invalid header", T, ParserOptionCombinations) {
} }
{ {
auto [p, _] = auto [p, _] = make_parser<buffer_mode, ErrorMode, Ts...>(f.name);
make_parser<buffer_mode, ErrorMode, ss::escape<'\\'>>(f.name);
auto command = [&p = p] { std::ignore = p.header(); }; auto command0 = [&p = p] { std::ignore = p.header(); };
expect_error_on_command(p, command); expect_error_on_command(p, command0);
CHECK_EQ(p.raw_header(), "Int\\"); 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,
auto [p, _] = make_parser<buffer_mode, ErrorMode, ss::escape<'\\'>, ParserOptionCombinations) {
ss::multiline>(f.name); using quote = ss::quote<'"'>;
auto command = [&p = p] { std::ignore = p.header(); }; using escape = ss::escape<'\\'>;
expect_error_on_command(p, command); test_unterminated_escape_header<T, escape>();
CHECK_EQ(p.raw_header(), "Int\\"); test_unterminated_escape_header<T, escape, ss::multiline>();
} test_unterminated_escape_header<T, escape, quote>();
test_unterminated_escape_header<T, escape, quote, ss::multiline>();
{
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> template <typename T>