Clean up CSV writing duplicated logic

Author: vincentlaucsb

docs/source/Doxy.md

@@ -87,7 +87,7 @@ column extraction, editing, and grouping.
    * `set_throw_on_missing_key()`: Control exception behavior for missing keys
 
 ### ETL Utilities
- * csv::csv_data_types(): Infer SQL-friendly column data types
+ * csv::csv_data_types(): Infer SQL-friendly column data types from a CSVReader or any supported CSVReader constructor input
 
 ### CSV Writing
 The [CSV Writing Guide](@ref csv_writing_guide) contains a 

include/csv.hpp

@@ -33,7 +33,6 @@ SOFTWARE.
 #include "internal/csv_reader.hpp"
 #include "internal/csv_utility.hpp"
 #include "internal/csv_writer.hpp"
-#include "internal/csv_writer_extensions.hpp"
 
 /** INSERT_CSV_SOURCES **/
 

include/internal/CMakeLists.txt

@@ -18,7 +18,6 @@ target_sources(csv
 		csv_utility.cpp
 		csv_utility.hpp
 		csv_writer.hpp
-		csv_writer_extensions.hpp
 		data_type.hpp
 		raw_csv_data.hpp
 		raw_csv_data.hpp

include/internal/common.hpp

@@ -5,6 +5,7 @@
 #pragma once
 #include <algorithm>
 #include <array>
+#include <cassert>
 #include <cmath>
 #include <cstdlib>
 #include <deque>
@@ -134,6 +135,12 @@ namespace csv {
 // Allows static assertions without specifying a message
 #define STATIC_ASSERT(x) static_assert(x, "Assertion failed")
 
+#ifdef NDEBUG
+    #define CSV_DEBUG_ASSERT(x) ((void)sizeof(x), (void)0)
+#else
+    #define CSV_DEBUG_ASSERT(x) assert(x)
+#endif
+
 #ifdef CSV_HAS_CXX17
      /** @typedef string_view
       *  The string_view class used by this library.

include/internal/csv_utility.cpp

@@ -2,9 +2,8 @@
 #include "data_frame.hpp"
 
 namespace csv {
-    CSV_INLINE std::unordered_map<std::string, DataType> csv_data_types(const std::string& filename) {
+    CSV_INLINE std::unordered_map<std::string, DataType> csv_data_types(CSVReader& reader) {
         std::unordered_map<std::string, DataType> csv_dtypes;
-        CSVReader reader(filename);
         const auto col_names = reader.get_col_names();
         std::vector<std::unordered_map<DataType, size_t>> type_counts(col_names.size());
         constexpr size_t TYPE_CHUNK_SIZE = 5000;

include/internal/csv_utility.hpp

@@ -11,6 +11,7 @@
 #include <string>
 #include <type_traits>
 #include <unordered_map>
+#include <utility>
 
 namespace csv {
     /** Returned by get_file_info() */
@@ -87,7 +88,37 @@ namespace csv {
 
     /** @name Utility Functions */
     ///@{
-    std::unordered_map<std::string, DataType> csv_data_types(const std::string&);
+    /** Infer SQL-friendly column data types from an existing CSVReader.
+     *
+     *  This consumes rows from `reader` using the chunked ETL path and returns
+     *  one inferred `DataType` per column name.
+     */
+    std::unordered_map<std::string, DataType> csv_data_types(CSVReader& reader);
+
+    /** Infer SQL-friendly column data types from any CSVReader constructor input.
+     *
+     *  This convenience overload forwards its arguments directly to
+     *  `CSVReader`, so it supports filenames, `std::istream` sources, owned
+     *  streams, and custom `CSVFormat` combinations without additional wrapper
+     *  code.
+     *
+     *  @par Example
+     *  @code
+     *  std::istringstream input("name,age\nAlice,30\nBob,41\n");
+     *  CSVFormat format;
+     *  format.delimiter(',').header_row(0);
+     *
+     *  auto dtypes = csv::csv_data_types(input, format);
+     *  @endcode
+     */
+    template<
+        typename... ReaderArgs,
+        csv::enable_if_t<std::is_constructible<CSVReader, ReaderArgs...>::value, int> = 0
+    >
+    inline std::unordered_map<std::string, DataType> csv_data_types(ReaderArgs&&... reader_args) {
+        CSVReader reader(std::forward<ReaderArgs>(reader_args)...);
+        return csv_data_types(reader);
+    }
 
     /** Apply a per-column batch function over a CSVReader using a reusable executor.
      *

include/internal/csv_writer.hpp

@@ -375,6 +375,16 @@ namespace csv {
             write_range_impl(container);
             return *this;
         }
+
+        /** Write a row-like object that exposes to_sv_range(). */
+        template<typename RowLike>
+        DelimWriter& operator<<(const RowLike& row)
+            requires internals::has_to_sv_range<RowLike>
+                && !internals::csv_string_field_range<RowLike> {
+            append_row_like(row);
+            finish_write_call();
+            return *this;
+        }
         #else
         /** Write a range of string-like fields as one delimited row.
          *
@@ -432,12 +442,13 @@ namespace csv {
         }
 
     private:
-        /** Helper to write a range of values, handling first element undelimited,
-         *  rest prefixed with delimiter. Inlines aggressively across both C++20 and
-         *  C++11 operator<< entry points.
+        /** Append delimited fields from a range without terminating the record.
+         *
+         *  Shared by single-row writes and bulk row appends so escaping and
+         *  delimiter handling stay on one code path.
          */
         template<typename Range>
-        inline void write_range_impl(const Range& record) {
+        inline void append_range_fields(Range&& record) {
             auto it = std::begin(record);
             auto end = std::end(record);
 
@@ -450,6 +461,14 @@ namespace csv {
                 batch_buffer_.push_back(Delim);
                 write_field(*it);
             }
+        }
+
+        /** Helper to write a complete range-backed row and apply the normal
+         *  end-of-record and flush policy for operator<< entry points.
+         */
+        template<typename Range>
+        inline void write_range_impl(const Range& record) {
+            append_range_fields(record);
 
             end_record();
             finish_write_call();
@@ -467,23 +486,6 @@ namespace csv {
 
             end_record();
         }
-
-        template<std::ranges::input_range Range>
-            requires std::convertible_to<std::ranges::range_reference_t<Range>, csv::string_view>
-        void append_range_fields(Range&& record) {
-            auto it = std::begin(record);
-            auto end = std::end(record);
-
-            if (it != end) {
-                write_field(*it);
-                ++it;
-            }
-
-            for (; it != end; ++it) {
-                batch_buffer_.push_back(Delim);
-                write_field(*it);
-            }
-        }
         #endif
 
         template<