CSVRow::current_row_start(): track row start position of input stream

vincentlaucsb · Sep 13, 2024 · c05b6b3 · c05b6b3
1 parent 985966e
commit c05b6b3
Show file tree

Hide file tree

Showing 7 changed files with 1,108 additions and 1,026 deletions.
diff --git a/include/internal/basic_csv_parser.cpp b/include/internal/basic_csv_parser.cpp
@@ -235,6 +235,7 @@ namespace csv {
             this->field_start = UNINITIALIZED_FIELD;
             this->field_length = 0;
             this->reset_data_ptr();
+            this->data_ptr->_stream_pos = this->mmap_pos;
 
             // Create memory map
             size_t length = std::min(this->source_size - this->mmap_pos, bytes);

diff --git a/include/internal/basic_csv_parser.hpp b/include/internal/basic_csv_parser.hpp
@@ -321,6 +321,7 @@ namespace csv {
                 if (this->eof()) return;
 
                 this->reset_data_ptr();
+                this->data_ptr->_stream_pos = this->stream_pos;
                 this->data_ptr->_data = std::make_shared<std::string>();
 
                 if (source_size == 0) {

diff --git a/include/internal/csv_row.hpp b/include/internal/csv_row.hpp
@@ -130,6 +130,9 @@ namespace csv {
             internals::ColNamesPtr col_names = nullptr;
             internals::ParseFlagMap parse_flags;
             internals::WhitespaceMap ws_flags;
+
+            /** where in Stream we start */
+            uint64_t _stream_pos = {};
         };
 
         using RawCSVDataPtr = std::shared_ptr<RawCSVData>;
@@ -318,6 +321,9 @@ namespace csv {
         /** Return the number of fields in this row */
         CONSTEXPR size_t size() const noexcept { return row_length; }
 
+        /** Where in the Stream we start */
+        size_t current_row_start() const { return data->_stream_pos + data_start; }
+
         /** @name Value Retrieval */
         ///@{
         CSVField operator[](size_t n) const;