Update header parsing

include/ss/parser.hpp

@@ -45,7 +45,7 @@ public:
             if constexpr (ignore_header) {
                 ignore_next();
             } else {
-                header_ = reader_.get_header();
+                raw_header_ = reader_.get_buffer();
             }
         } else {
             handle_error_file_not_open();
@@ -123,6 +123,10 @@ public:
     }
 
     bool field_exists(const std::string& field) {
+        if (header_.empty()) {
+            split_header_data();
+        }
+
         return header_index(field).has_value();
     }
 
@@ -133,6 +137,10 @@ public:
             return;
         }
 
+        if (header_.empty()) {
+            split_header_data();
+        }
+
         if (!valid()) {
             return;
         }
@@ -432,6 +440,15 @@ private:
     // header
     ////////////////
 
+    void split_header_data() {
+        ss::splitter<Options...> splitter;
+        std::string raw_header_copy = raw_header_;
+        splitter.split(raw_header_copy.data(), reader_.delim_);
+        for (const auto& [begin, end] : splitter.split_data_) {
+            header_.emplace_back(begin, end);
+        }
+    }
+
     std::optional<size_t> header_index(const std::string& field) {
         auto it = std::find(header_.begin(), header_.end(), field);
 
@@ -508,8 +525,7 @@ private:
 
     void handle_error_header_ignored() {
         constexpr static auto error_msg =
-            ": \"the header row is ignored within the setup it cannot be "
-            "used\"";
+            ": the header row is ignored within the setup it cannot be used";
 
         if constexpr (string_error) {
             error_.clear();
@@ -796,15 +812,8 @@ private:
             return true;
         }
 
-        std::vector<std::string> get_header() {
-            std::vector<std::string> header;
-            std::string header_buffer = next_line_buffer_;
-            ss::splitter<Options...> splitter;
-            splitter.split(header_buffer.data(), delim_);
-            for (const auto& [begin, end] : splitter.split_data_) {
-                header.emplace_back(begin, end);
-            }
-            return header;
+        std::string get_buffer() {
+            return std::string{next_line_buffer_};
         }
 
         ////////////////
@@ -838,6 +847,7 @@ private:
     error_type error_{};
     reader reader_;
     std::vector<std::string> header_;
+    std::string raw_header_;
     bool eof_{false};
 };
 

test/test_parser.cpp

@@ -700,7 +700,8 @@ static inline std::string no_quote(const std::string& s) {
     return s;
 }
 
-TEST_CASE("parser test csv on multiple lines with quotes") {
+template <typename... Ts>
+void test_quote_multiline() {
     unique_file_name f{"test_parser"};
     std::vector<X> data = {{1, 2, "\"x\r\nx\nx\""},
                            {3, 4, "\"y\ny\r\ny\""},
@@ -720,11 +721,11 @@ TEST_CASE("parser test csv on multiple lines with quotes") {
         }
     }
 
-    ss::parser<ss::multiline, ss::quote<'"'>> p{f.name, ","};
+    ss::parser<ss::multiline, ss::quote<'"'>, Ts...> p{f.name, ","};
     std::vector<X> i;
 
     while (!p.eof()) {
-        auto a = p.get_next<int, double, std::string>();
+        auto a = p.template get_next<int, double, std::string>();
         i.emplace_back(ss::to_object<X>(a));
     }
 
@@ -733,13 +734,21 @@ TEST_CASE("parser test csv on multiple lines with quotes") {
     }
     CHECK_EQ(i, data);
 
-    ss::parser<ss::quote<'"'>> p_no_multiline{f.name, ","};
+    ss::parser<ss::quote<'"'>, Ts...> p_no_multiline{f.name, ","};
     while (!p.eof()) {
-        auto a = p_no_multiline.get_next<int, double, std::string>();
-        CHECK(!p.valid());
+        auto command = [&] {
+            p_no_multiline.template get_next<int, double, std::string>();
+        };
+        expect_error_on_command(p_no_multiline, command);
     }
 }
 
+TEST_CASE("parser test csv on multiple lines with quotes") {
+    test_quote_multiline();
+    test_quote_multiline<ss::string_error>();
+    test_quote_multiline<ss::throw_on_error>();
+}
+
 static inline std::string no_escape(std::string& s) {
     s.erase(std::remove(begin(s), end(s), '\\'), end(s));
     return s;