Awesome

cpp-dump

This is an animated GIF.

cpp-dump is a C++ library for printing variables of any type for debugging.

Python has print(), JavaScript has console.log(), and PHP has var_dump() — functions that print variables of any type and are useful for debugging when you want a quick way to inspect variables. But what about C++? Enter cpp_dump(...). cpp-dump is a library that automatically formats and prints variables of any type. With features like auto-indentation, colored output, string representations similar to JavaScript, Python, and C++, and over 20 manipulators, cpp-dump is equipped with everything you need to effortlessly and clearly print variables.

Key points:

cpp-dump can print a wide variety of types, including multi-D vectors, (multi)maps, (multi)sets, and tuples.
cpp-dump has an auto-indentation feature. The output fits into the maximum line width, and nested containers are formatted for readability.
The string representation of variables is similar to JavaScript, Python, and C++ syntax. The output is readable without being overloaded with information. (You can add more details using manipulators if you want.)
Customizable colored output is available. You can achieve syntax highlighting similar to that of IDEs.
With over 20 manipulators, you can easily change the format or add information to the output.
By using macros, cpp-dump supports user-defined types as well. There is no need to write new functions for printing.
cpp-dump is a header-only library. No build or dependencies are required!

日本語記事はこちら！

Table of Contents

Introduction
Features
Advanced Feature
- 20+ Manipulators to change the display style
Requirement
Installation
- Simplest solution
- With CMake
  - Run cmake --install to copy the headers to /usr/local/include/ or equivalent
  - Use FetchContent
Configuration (as needed)
- Configuration options
Detailed usage
Supported types
- Display example

Introduction

You can print variables of a wide variety of types by passing them to the cpp_dump(expressions...) macro.
See Full Example Code

std::vector<std::vector<int>> my_vector{{3, 5, 8, 9, 7}, {9, 3, 2, 3, 8}};
cpp_dump(my_vector);

Features

A wide variety of supported types

cpp-dump supports a wide variety of types, including multi-D vectors, (multi)maps, (multi)sets, and tuples.
Their string representation is similar to JavaScript, Python, and C++ syntax, which is easy to read.
See All Supported Types
See Full Example Code

bool my_bool = true; double my_double = 3.141592; int my_int = 65;
char my_char = 'a', LF_char = '\n'; std::string my_string = "This is a string.";
int *int_ptr = &my_int; void *void_ptr = &my_int;
std::vector<std::vector<int>> my_vector{{3, 5, 8, 9, 7}, {9, 3, 2, 3, 8}};
std::set<char> my_set{'A', 'p', 'p', 'l', 'e'};
std::map<int, int> my_map{{2, 6}, {4, 6}, {5, 3}};
std::multiset<char> my_multiset{'A', 'p', 'p', 'l', 'e'};
std::multimap<int, int> my_multimap{{2, 4}, {4, 6}, {5, 3}, {4, 7}};
std::pair<int, char> my_pair{8, 'a'};
std::tuple<int, double, std::string> my_tuple{7, 4.5, "This is a string."};
std::queue<int> my_queue;
std::priority_queue<int> my_priority_queue;
std::stack<int> my_stack;
for (auto v : {1, 2, 3, 4, 5}) my_queue.push(v), my_priority_queue.push(v), my_stack.push(v);
std::bitset<8> my_bitset(0x3a);
std::complex<double> my_complex{1.0, -1.0};
std::optional<int> my_optional{15};
std::variant<int, std::string> my_variant{"This is a string."};
std::vector<std::pair<int, std::string>> vector_of_pairs{{1, "apple"}, {3, "banana"}};

std::clog << "\n// Basic Type" << std::endl;
cpp_dump(my_bool, my_double, my_int); cpp_dump(my_string, my_char, LF_char);
cpp_dump(int_ptr, void_ptr, nullptr);

std::clog << "\n// Container" << std::endl;
cpp_dump(my_vector);

std::clog << "\n// Set/Map" << std::endl;
cpp_dump(my_set); cpp_dump(my_map);

std::clog << "\n// Multiset/Multimap" << std::endl;
cpp_dump(my_multiset); cpp_dump(my_multimap);

std::clog << "\n// Tuple" << std::endl;
cpp_dump(my_tuple); cpp_dump(my_pair);

std::clog << "\n// FIFO/LIFO" << std::endl;
cpp_dump(my_queue); cpp_dump(my_priority_queue); cpp_dump(my_stack);

std::clog << "\n// Other" << std::endl;
cpp_dump(my_bitset); cpp_dump(my_complex);
cpp_dump(my_optional, std::nullopt); cpp_dump(my_variant);

std::clog << "\n// Combination" << std::endl;
cpp_dump(vector_of_pairs);

Auto indent

cpp-dump automatically indents so that the output does not exceed the maximum width.
Additionally, nested containers are indented by default (This behavior can be changed through the settings).
See Full Example Code

cpp_dump(my_vector);
my_vector.push_back("This is a test string.");
cpp_dump(my_vector);

Auto indent

Filename and line can be printed instead of `[dump]`

If you want to print the filename and line instead of [dump], use the following code. The cpp_dump() macro will automatically detect and print the filename and the line. You can attach the function name, too. See Customize [dump] for details.
See Full Example Code

// Print the filename and line instead of [dump]
CPP_DUMP_SET_OPTION(log_label_func, cp::log_label::filename());
// Print along with the function name
CPP_DUMP_SET_OPTION(log_label_func, cp::log_label::filename(true));

Customizable output color

You can modify the escape sequences to change the colors of the output using the following code.
See Full Example Code

// Use more colors
CPP_DUMP_SET_OPTION(es_value.log, "\x1b[02m");                 // log: dark
CPP_DUMP_SET_OPTION(es_value.expression, "\x1b[34m");          // expression: blue
CPP_DUMP_SET_OPTION(es_value.reserved, "\x1b[38;5;39m");       // reserved: light blue
CPP_DUMP_SET_OPTION(es_value.number, "\x1b[38;5;150m");        // number: light green
CPP_DUMP_SET_OPTION(es_value.character, "\x1b[38;5;172m");     // character: orange
CPP_DUMP_SET_OPTION(es_value.escaped_char, "\x1b[38;5;220m");  // escaped_char: light orange
CPP_DUMP_SET_OPTION(es_value.op, "\x1b[02m");                  // op: dark
CPP_DUMP_SET_OPTION(es_value.identifier, "\x1b[32m");          // identifier:  green
CPP_DUMP_SET_OPTION(es_value.member, "\x1b[96m");              // member: light cyan
CPP_DUMP_SET_OPTION(es_value.unsupported, "\x1b[31m");         // unsupported: red
CPP_DUMP_SET_OPTION(es_value.bracket_by_depth, (std::vector<std::string>{
    "\x1b[33m",  // bracket_by_depth[0]: yellow
    "\x1b[35m",  // bracket_by_depth[1]: magenta
    "\x1b[36m",  // bracket_by_depth[2]: cyan
}));
CPP_DUMP_SET_OPTION(es_value.class_op, "\x1b[02m");   // class_op: dark
CPP_DUMP_SET_OPTION(es_value.member_op, "\x1b[02m");  // member_op: dark
CPP_DUMP_SET_OPTION(es_value.number_op, "");          // number_op: default

// Use the 'class_op'/'member_op'/'number_op' color for operators
// in class names, members, and numbers (::, <>, (), -, +, etc...).
CPP_DUMP_SET_OPTION(detailed_class_es, true);
CPP_DUMP_SET_OPTION(detailed_member_es, true);
CPP_DUMP_SET_OPTION(detailed_number_es, true);

// Use a color scheme closer to standard syntax highlighting.
// CPP_DUMP_SET_OPTION(es_style, cp::types::es_style_t::by_syntax);

For the light theme, the output will be like this.

customizable-colors-light-plus

To turn off output coloring, use the following code.
See Full Example Code

// Turn off output coloring
CPP_DUMP_SET_OPTION(es_style, cp::types::es_style_t::no_es);

Can print even user-defined types

If you want to print a user-defined type, you can enable the library to print it by using macros or defining an operator. The following is an example of the use of macros. See How to print a user-defined type with cpp-dump for details.
See Full Example Code

CPP_DUMP_DEFINE_EXPORT_OBJECT_GENERIC(i, str());

Advanced Feature

20+ Manipulators to change the display style

This library has over 20 manipulators to change the display style.
See Formatting with manipulators for details.

Manipulator to omit part of a container

Manipulator to show indexes of an array

Manipulator to change the numerical base

Manipulator to escape strings

Requirement

C++17 or higher.
No build or dependencies are required since cpp-dump is a header-only library.

Installation

Simplest solution

git clone https://github.com/philip82148/cpp-dump

Then

#include "path/to/cpp-dump/cpp-dump.hpp"

Copy this code to test the library

With CMake

Run `cmake --install` to copy the headers to `/usr/local/include/` or equivalent

git clone https://github.com/philip82148/cpp-dump
cd cpp-dump
cmake -S . -B build # No configuration is needed because the library is header-only.
sudo cmake --install build
# (The 'cpp-dump' folder can be removed after this.)

Then

#include <cpp-dump.hpp>

Use `FetchContent`

CMakeLists.txt

set(CMAKE_EXPORT_COMPILE_COMMANDS ON) # Generate compile_commands.json (optional)

include(FetchContent)

# Fetch cpp-dump
FetchContent_Declare(cpp-dump
    GIT_REPOSITORY https://github.com/philip82148/cpp-dump
    GIT_TAG main
)
FetchContent_MakeAvailable(cpp-dump)

# Link cpp-dump to your app
target_link_libraries(MyApp PRIVATE cpp-dump)

Then

#include <cpp-dump.hpp>

Configuration (as needed)

If you want to customize the library, you can write the configuration code as follows:

custom-cpp-dump.hpp

#ifdef DEBUGGING
#include "path/to/cpp-dump/cpp-dump.hpp"
namespace cp = cpp_dump;
// You can use this in both a header file and a source file,
// but make sure not to use it more than once for the same option.
CPP_DUMP_SET_OPTION_GLOBAL(max_line_width, 100);

// To ensure proper instantiation of templates,
// include these in at least one translation unit where cpp_dump(...) prints each type.
// One way is to write them in a header file and then include it wherever needed.
CPP_DUMP_DEFINE_EXPORT_OBJECT(my_class, member1, member2());
CPP_DUMP_DEFINE_EXPORT_OBJECT_GENERIC(member3, member4());
CPP_DUMP_DEFINE_EXPORT_ENUM(my_enum, my_enum::a, my_enum::b, my_enum::c);
CPP_DUMP_DEFINE_EXPORT_ENUM_GENERIC(member_a, member_b, member_c);
#else
#define cpp_dump(...)
#define CPP_DUMP_SET_OPTION(...)
#endif

main.cpp

#include "path/to/custom-cpp-dump.hpp"

int main() {
  cpp_dump(vec | cp::back());
}

If you want to configure the library within a function, use CPP_DUMP_SET_OPTION() instead.

main.cpp

#include "path/to/custom-cpp-dump.hpp"

void func() {
  CPP_DUMP_SET_OPTION(print_expr, false);
  cpp_dump(vec | cp::back());
  CPP_DUMP_SET_OPTION(print_expr, true);
}

Configuration options

`max_line_width`

Type: std::size_t Default: 160
The maximum line width of the strings returned by cpp_dump() and cpp_dump::export_var(), which cpp_dump() internally uses to convert a variable into a string.

`max_depth`

Type: std::size_t Default: 4
The maximum number of times cpp_dump::export_var() is called recursively.

`max_iteration_count`

Type: std::size_t Default: 16
The maximum number of iterations of cpp_dump::export_var() over an iterator.
Note that in a single call, cpp_dump::export_var() calls itself at most (max_iteration_count^(max_depth+1)-1)/(max_iteration_count-1) times.

`cont_indent_style`

Type: enum class cpp_dump::types::cont_indent_style_t Default: cpp_dump::types::cont_indent_style_t::when_nested
The style of indents of the Container, Set and Map categories (See Supported types).

Name	Description
`minimal`	Don't indent unless the `max_line_width` is exceeded.
`when_nested`	Default. Always indent when the element/key/value type also falls into the Container/Set/Map/Tuple category.
`when_non_tuples_nested`	Always indent when the element/key/value type falls into the Container/Set/Map category, but don't when it falls into the Tuple category.
`always`	Always indent even if the Container/Set/Map is not nested.

`enable_asterisk`

Type: bool Default: false
Whether cpp_dump::export_var() prints types of the Asterisk category (See Supported types).

`print_expr`

Type: bool Default: true
Whether cpp_dump() prints the expressions.

`log_label_func`

Type: cpp_dump::types::log_label_func_t Default: cpp_dump::log_label::default_func
The function that returns the label that cpp_dump() prints at the beginning of the output.

`es_style`

Type: enum class cpp_dump::types::es_style_t Default cpp_dump::types::es_style_t::original
The style of the escape sequences (the output coloring).

Name	Description
`original`	Default.
`by_syntax`	Use a color scheme closer to standard syntax highlighting. Pointers, bitsets, complexes, and etc. are colored differently from `original`.
`no_es`	Turn off output coloring.

`es_value`

Type: cpp_dump::types::es_value_t Default: (Default constructor, see Types)
The values of the escape sequences.

`detailed_class_es`

Type: bool Default: false
If true, the es_value.class_op color is used for operators in class names (::, <>, etc...).

`detailed_member_es`

Type: bool Default: false
If true, the es_value.member_op color is used for operators in members ((), etc...).

`detailed_number_es`

Type: bool Default: false
If true, the es_value.number_op color is used for operators in numbers (-, +, etc...).

Detailed usage

Macros

/**
 * Print string representations of expressions and results to std::clog or other configurable outputs.
 * If you want to change the output, define an explicit specialization of cpp_dump::write_log().
 * This macro uses cpp_dump::export_var() internally.
 */
#define cpp_dump(expressions...)

/**
 * Make cpp_dump::export_var() support type T.
 * Member functions to be displayed must be const.
 */
#define CPP_DUMP_DEFINE_EXPORT_OBJECT(T, members...)

/**
 * Make cpp_dump::export_var() support every type that has the specified members.
 * Member functions to be displayed must be const.
 * Compile errors in this macro, such as ambiguous function calls, are never reported due to SFINAE.
 */
#define CPP_DUMP_DEFINE_EXPORT_OBJECT_GENERIC(members...)

/**
 * Make cpp_dump::export_var() support enum T.
 */
#define CPP_DUMP_DEFINE_EXPORT_ENUM(T, members...)

/**
 * Make cpp_dump::export_var() support every enum type that has the specified members.
 * Compile errors in this macro, such as ambiguous function calls, are never reported due to SFINAE.
 */
#define CPP_DUMP_DEFINE_EXPORT_ENUM_GENERIC(members...)

/**
 * Set a value to a variable in cpp_dump::options namespace.
 */
#define CPP_DUMP_SET_OPTION(variable, value)

/**
 * Set a value to a variable in cpp_dump::options namespace.
 * Use this if you want to run it in the global namespace, meaning before the main starts.
 */
#define CPP_DUMP_SET_OPTION_GLOBAL(variable, value)

Types

namespace cpp_dump::types {

/**
 * Type of cpp_dump::options::cont_indent_style.
 * cpp_dump::export_var() supports this type.
 */
enum class cont_indent_style_t { minimal, when_nested, when_non_tuples_nested, always };

/**
 * Type of cpp_dump::options::es_style.
 * cpp_dump::export_var() supports this type.
 */
enum class es_style_t { no_es, original, by_syntax };

/**
 * Type of cpp_dump::options::es_value.
 * cpp_dump::export_var() supports this type.
 */
struct es_value_t {
  std::string log = "\x1b[02m";                           // dark
  std::string expression = "\x1b[36m";                    // cyan
  std::string reserved{};                                 // default
  std::string number{};                                   // default
  std::string character{};                                // default
  std::string escaped_char = "\x1b[02m";                  // dark
  std::string op = "\x1b[02m";                            // dark
  std::string identifier = "\x1b[32m";                    // green
  std::string member = "\x1b[36m";                        // cyan
  std::string unsupported = "\x1b[31m";                   // red
  std::vector<std::string> bracket_by_depth{"\x1b[02m"};  // dark
  std::string class_op = "\x1b[02m";                      // dark
  std::string member_op = "\x1b[02m";                     // dark
  std::string number_op{};                                // default
};

using log_label_func_t = std::function<std::string(std::string_view, std::size_t, std::string_view)>;

}  // namespace cpp_dump::types

Variables

namespace cpp_dump::options {

/**
 * Maximum line width of the strings returned by cpp_dump() and cpp_dump::export_var().
 */
inline std::size_t max_line_width = 160;

/**
 * Maximum number of times cpp_dump::export_var() is called recursively.
 */
inline std::size_t max_depth = 4;

/**
 * Maximum number of iterations of cpp_dump::export_var() over an iterator.
 * Note that in a single call, cpp_dump::export_var() calls itself at most
 * (max_iteration_count^(max_depth+1)-1)/(max_iteration_count-1) times.
 */
inline std::size_t max_iteration_count = 16;

/**
 * Style of indents of the Container, Set and Map categories (See 'Supported types')
 */
inline types::cont_indent_style_t cont_indent_style = types::cont_indent_style_t::when_nested;

/**
 * Whether cpp_dump() prints types of the Asterisk category (See 'Supported types').
 */
inline bool enable_asterisk = false;

/**
 * Whether cpp_dump() prints the expressions.
 */
inline bool print_expr = true;

/**
 * Function that returns the label that cpp_dump() prints at the beginning of the output.
 */
inline types::log_label_func_t log_label_func = log_label::default_func;

/**
 * Style of the escape sequences (output coloring).
 */
inline types::es_style_t es_style = types::es_style_t::original;

/**
 * Values of the escape sequences (output coloring).
 */
inline types::es_value_t es_value;

/**
 * If true, the 'es_value.class_op' color is used for operators in class names (::, <>, etc...).
 */
inline bool detailed_class_es = false;

/**
 * If true, the 'es_value.member_op' color is used for operators in members ((), etc...).
 */
inline bool detailed_member_es = false;

/**
 * If true, the 'es_value.number_op' color is used for operators in numbers (-, +, etc...).
 */
inline bool detailed_number_es = false;


}  // namespace cpp_dump::options

Functions

namespace cpp_dump {

/**
 * Return a string representation of a variable.
 * cpp_dump() uses this function internally.
 */
template <typename T>
std::string export_var(const T &value);

/**
 * cpp_dump() uses this function to print logs.
 * Define an explicit specialization with 'void' to customize this function.
 */
template <typename = void>
void write_log(std::string_view output) {
  std::clog << output << std::endl;
}

// Manipulators (See 'Formatting with manipulators' for details.)
front(std::size_t iteration_count = options::max_iteration_count);
middle(std::size_t iteration_count = options::max_iteration_count);
back(std::size_t iteration_count = options::max_iteration_count);
both_ends(std::size_t half_iteration_count = options::max_iteration_count / 2);
index();
int_style(int base, int digits = -1, int chunk = 0,
    bool space_fill = false, bool make_unsigned_or_no_space_for_minus = false);
bin(int digits = -1, int chunk = 0, bool space_fill = false);
oct(int digits = -1, int chunk = 0, bool space_fill = false);
hex(int digits = -1, int chunk = 0, bool space_fill = false);
dec(int digits = -1, int chunk = 0, bool space_fill = true);
ubin(int digits = -1, int chunk = 0, bool space_fill = false);
uoct(int digits = -1, int chunk = 0, bool space_fill = false);
uhex(int digits = -1, int chunk = 0, bool space_fill = false);
udec(int digits = -1, int chunk = 0, bool space_fill = true);
map_k(return_value_of_manipulator);
map_v(return_value_of_manipulator);
map_kv(return_value_of_manipulator_for_key, return_value_of_manipulator_for_value);
format(const char *f);
bw(bool left = false);
boolnum();
stresc();
charhex();
addr(std::size_t depth = 0);

}  // namespace cpp_dump

// See 'Customize "[dump]"'.
namespace cpp_dump::log_label {

std::string default_func(std::string_view, std::size_t, std::string_view);
types::log_label_func_t line(bool show_func = false, int min_width = 0);
types::log_label_func_t basename(bool show_func = false, int min_width = 0);
types::log_label_func_t filename(bool show_func = false, int min_width = 0);
types::log_label_func_t fullpath(int substr_start, bool show_func = false, int min_width = 0);
types::log_label_func_t fixed_length(int min_width, int max_width,
    int substr_start, bool show_func = false);

}  // namespace cpp_dump::log_label

How to print a user-defined type with cpp-dump

There are three ways to enable the library to print a user type.

Method 1. Use `CPP_DUMP_DEFINE_EXPORT_OBJECT()` macro

This macro requires the user type to be accessible from the global scope, but it is the safest and easiest way to enable cpp_dump() to print a user type.
See Full Example Code

// Somewhere accessible from the global scope (not private or defined in a function)
struct class_A {
  int i;
  std::string str() const { return std::to_string(i); }
};

// In the global scope
// CPP_DUMP_DEFINE_EXPORT_OBJECT(type_name, members...)
CPP_DUMP_DEFINE_EXPORT_OBJECT(class_A, i, str());

// In a function
class_A my_class_A{10};
cpp_dump(my_class_A);

For enums, use CPP_DUMP_DEFINE_EXPORT_ENUM() macro.
See Full Example Code

// Somewhere accessible from the global scope (not private or defined in a function)
enum class enum_A { member_a, member_b, member_c };

// In the global scope
// CPP_DUMP_DEFINE_EXPORT_ENUM(enum_name, members...)
CPP_DUMP_DEFINE_EXPORT_ENUM(enum_A, enum_A::member_a, enum_A::member_b, enum_A::member_c);

// In a function
enum_A my_enum_A = enum_A::member_c;
cpp_dump(my_enum_A);

Method 2. Use `CPP_DUMP_DEFINE_EXPORT_OBJECT_GENERIC()` macro

This macro enables cpp_dump() to print any type with specified members.
This macro doesn't require the user type to be accessible from the global scope (or need even the type name).

If you use this macro two or more times, you need to be careful of ambiguous function call compile errors.
If such an error occurs, it won't be reported due to SFINAE, and the user-defined type will remain unsupported.
See Full Example Code

// In the global scope
// CPP_DUMP_DEFINE_EXPORT_OBJECT_GENERIC(members...)
CPP_DUMP_DEFINE_EXPORT_OBJECT_GENERIC(i, str());

// Anywhere
struct class_A {
  int i;
  std::string str() const { return std::to_string(i); }
};

// In a function
class_A my_class_A{10};
cpp_dump(my_class_A);

For enums, use CPP_DUMP_DEFINE_EXPORT_ENUM_GENERIC() macro.
See Full Example Code

// In the global scope
// CPP_DUMP_DEFINE_EXPORT_ENUM_GENERIC(members...)
CPP_DUMP_DEFINE_EXPORT_ENUM(member_a, member_b, member_c);

// Anywhere
enum class enum_A { member_a, member_b, member_c };

// In a function
enum_A my_enum_A = enum_A::member_c;
cpp_dump(my_enum_A);

Method 3. Define `std::ostream& operator<<(std::ostream&, const T &)` operator

The last way is to define the operator std::ostream& operator<<(std::ostream&, const T &).
See Full Example Code

// Somewhere accessible from the global scope (not private or defined in a function)
struct class_A {
  int i;
  std::string str() const { return std::to_string(i); }
};

// In the global scope
std::ostream &operator<<(std::ostream &os, const class_A &a) {
  os << "class_A{ i= " << a.i << ", str()= \"" << a.str() << "\" }";
  return os;
}

// In a function
class_A my_class_A{10};
cpp_dump(my_class_A);

Customize `[dump]`

Assigning a function to cpp_dump::options::log_label_func, you can customize [dump].
cpp-dump has some functions that create a function to assign to cpp_dump::options::log_label_func, so you don't have to make your own function.

namespace cpp_dump::types {

using log_label_func_t =
  std::function<std::string(std::string_view fullpath, std::size_t line, std::string_view func_name)>;

}  // namespace cpp_dump::types

namespace cpp_dump::log_label {

// Default function assigned to cpp_dump::options::log_label_func.
std::string default_func(std::string_view, std::size_t, std::string_view) {
  return "[dump] ";
}

// Functions that create a function that can be assigned to cpp_dump::options::log_label_func.
types::log_label_func_t line(bool show_func = false, int min_width = 0);
types::log_label_func_t basename(bool show_func = false, int min_width = 0);
types::log_label_func_t filename(bool show_func = false, int min_width = 0);
types::log_label_func_t fullpath(int substr_start, bool show_func = false, int min_width = 0);
types::log_label_func_t fixed_length(int min_width, int max_width,
    int substr_start, bool show_func = false);

}  // namespace cpp_dump::log_label

namespace cpp_dump::options {

inline types::log_label_func_t log_label_func = log_label::default_func;

}  // namespace cpp_dump::options

Formatting with manipulators

Using manipulators, you can easily change the format or add information to the output.
For example, you can select which elements and how many elements of an array, map, or set will be displayed using the front, middle, back, and both_ends manipulators.
See Full Example Code

// Show the last 10 elements for the 1st dimension, the first 5 and the last 5 for the 2nd dimension.
cpp_dump(some_huge_vector | cp::back(10) | cp::both_ends(5) | cp::dec(2));

And you can display the indexes of an array by using the index manipulator.
See Full Example Code

CPP_DUMP_SET_OPTION(max_iteration_count, 5);

// Show the indexes of the vector.
cpp_dump(some_huge_vector | cp::dec(2) | cp::index());

There are also many other manipulators, such as the int_style manipulators.
See Full Example Code

// Show integers in binary, minimum 16 digits, separated by every 4 characters.
cpp_dump(0x3e8u | cp::bin(16, 4));
// Show integers in octal, minimum 6 digits, separated by every 3 characters.
cpp_dump(0x3e8u | cp::oct(6, 3));
// Show integers in hexadecimal, minimum 4 digits, separated by every 2 characters.
cpp_dump(0x3e8u | cp::hex(4, 2));
// Show integers in minimum 4 digits.
cpp_dump(0x3e8u | cp::dec(4));

How to use manipulators

You can use the manipulators by applying the '|' operator or the '<<' operator.
The order of manipulators matters for some, but not for others.

cpp_dump(variable | manipulatorA() | manipulatorB());
cpp_dump(manipulatorA() << manipulatorB() << variable);

`front()`, `middle()`, `back()`, `both_ends()` manipulators

namespace cpp_dump {

front(std::size_t iteration_count = options::max_iteration_count);
middle(std::size_t iteration_count = options::max_iteration_count);
back(std::size_t iteration_count = options::max_iteration_count);
both_ends(std::size_t half_iteration_count = options::max_iteration_count / 2);

}  // namespace cpp_dump

These manipulators are order-sensitive.

The further left manipulator will act on the more outside dimensions of the array/map/set.
Caution:
These manipulators other than front() calculate the container's size. Containers whose size cannot be calculated with std::size() will cost O(N) in computation. In particular, passing an infinite sequence to these manipulators will result in an infinite loop.
See Full Example Code

// Show the last 10 elements for the 1st dimension, the first 5 and the last 5 for the 2nd dimension.
cpp_dump(some_huge_vector | cp::back(10) | cp::both_ends(5) | cp::dec(2));

`index()` manipulator

cpp_dump::index();

Unlike the front() and other manipulators, the index() manipulator acts on all sequence containers in the variable. (The order is irrelevant.)
It does not affect maps/sets.
See Full Example Code

cpp_dump(some_huge_vector | cp::dec(2) | cp::index());

`int_style()` and its alias manipulators

namespace cpp_dump {

int_style(int base, int digits = -1, int chunk = 0,
    bool space_fill = false, bool make_unsigned_or_no_space_for_minus = false);

bin(int digits = -1, int chunk = 0, bool space_fill = false) {
  return int_style(2, digits, chunk, space_fill, false);
}
oct(int digits = -1, int chunk = 0, bool space_fill = false) {
  return int_style(8, digits, chunk, space_fill, false);
}
hex(int digits = -1, int chunk = 0, bool space_fill = false) {
  return int_style(16, digits, chunk, space_fill, false);
}
dec(int digits = -1, int chunk = 0, bool space_fill = true) {
  return int_style(10, digits, chunk, space_fill, false);
}

ubin(int digits = -1, int chunk = 0, bool space_fill = false) {
  return int_style(2, digits, chunk, space_fill, true);
}
uoct(int digits = -1, int chunk = 0, bool space_fill = false) {
  return int_style(8, digits, chunk, space_fill, true);
}
uhex(int digits = -1, int chunk = 0, bool space_fill = false) {
  return int_style(16, digits, chunk, space_fill, true);
}
udec(int digits = -1, int chunk = 0, bool space_fill = true) {
  return int_style(10, digits, chunk, space_fill, true);
}

}  // namespace cpp_dump

The parameter base of int_style() supports values of 2, 8, 10, 16. For other values, this manipulator does nothing.
digits supports values of digits >= 0 and digits <= 'the maximum digits', where 'the maximum digits' is the maximum number of digits that can be represented by the integer type for the given base. For other values, it is treated as digits = 'the maximum digits'.
chunk supports values of chunk >= 0. For other values, it is treated as chunk = 0.

Like the index() manipulators, the int_style() manipulator acts on all integers in the variable. (The order is irrelevant.)
The bin(...), oct(...), hex(...), ubin(...), uoct(...), uhex(...), dec(...), udec(...), are aliases of int_style(...).
See Full Example Code

// Show integers in binary, minimum 16 digits, separated by every 4 characters.
cpp_dump(0x3e8u | cp::bin(16, 4));
// Show integers in octal, minimum 6 digits, separated by every 3 characters.
cpp_dump(0x3e8u | cp::oct(6, 3));
// Show integers in hexadecimal, minimum 4 digits, separated by every 2 characters.
cpp_dump(0x3e8u | cp::hex(4, 2));
// Show integers in minimum 4 digits.
cpp_dump(0x3e8u | cp::dec(4));

For signed integer types, the bin(), oct(), hex(), and dec() manipulators will add an extra space for positive values and a minus sign for negative values.
For unsigned integer types, these manipulators will not add any extra space or minus sign.
See Full Example Code

cpp_dump(signed_int_vector | cp::front(2) | cp::hex(2));
cpp_dump(unsigned_int_vector | cp::front(2) | cp::hex(2));
cpp_dump(signed_int_vector | cp::front(2) | cp::dec(2));
cpp_dump(unsigned_int_vector | cp::front(2) | cp::dec(2));

The ubin(), uoct(), and uhex() manipulators interpret all integer types as unsigned.
If the original type is not unsigned, the suffix u is shown.
However, the udec() manipulator acts differently from these.
The udec() manipulator interprets signed types as signed type, but it does not add an extra space for positive values.
This is suitable for showing a container of a signed integers when all values are positive.
See Full Example Code

cpp_dump(signed_int_vector | cp::front(2) | cp::uhex());
cpp_dump(unsigned_int_vector | cp::front(2) | cp::uhex(2));
cpp_dump(signed_int_vector | cp::front(2) | cp::udec(2));
cpp_dump(unsigned_int_vector | cp::front(2) | cp::udec(2));

`format()` manipulator

cpp_dump::format(const char *f);

This manipulator uses snprintf() to format numbers (integers and floating points).
Make sure that the types specified by format specifiers match the actual types.
See Full Example Code

cpp_dump(pi | cp::format("%.10f"));

`bw()`, `boolnum()` manipulators

cpp_dump::bw(bool left = false);
cpp_dump::boolnum();

These manipulators are for formatting bool.
The bw() manipulator adds a space when a bool value is true to match the width of false.
bw stands for "bool width".
The boolnum() manipulator shows bool values as 1 or 0.
See Full Example Code

cpp_dump(bool_vector | cp::bw());
cpp_dump(bool_vector | cp::bw(true));
cpp_dump(bool_vector | cp::boolnum());

`stresc()` manipulator

cpp_dump::stresc();

This manipulator escapes strings.
For escaped characters, the es_value.escaped_char color is used.
See Full Example Code

cpp_dump("\a\t\\\"\n\x7f need to be escaped.");
cpp_dump("\a\t\\\"\n\x7f need to be escaped." | cp::stresc());

`charhex()` manipulator

cpp_dump::charhex();

This manipulator shows chars with their hex.
The width of their string representation is fixed.
See Full Example Code

for (auto c : "\a\t\\\"\n\x7f ABC") cpp_dump(c | cp::charhex());

`addr()` manipulator

cpp_dump::addr(std::size_t depth = 0);

This manipulator shows pointers by their address.
Use the depth parameter to specify the depth of pointers for displaying addresses.
See Full Example Code

int my_int = 15;
int *int_ptr = &my_int;
int **int_ptr_ptr = &int_ptr;

cpp_dump(int_ptr_ptr);
cpp_dump(int_ptr_ptr | cp::addr());
cpp_dump(int_ptr_ptr | cp::addr(1));

`map_*()` manipulators

namespace cpp_dump {

map_k(return_value_of_manipulator);
map_v(return_value_of_manipulator);
map_kv(return_value_of_manipulator_for_key, return_value_of_manipulator_for_value);

}  // namespace cpp_dump

These manipulators are order-sensitive.

These manipulators act on (multi)maps.
In the following example, the keys are displayed in hexadecimal, and if the values are iterable, the front part of the values is omitted.

cpp_dump(cp::front() << cp::map_kv(cp::hex(), cp::back()) << map);
cpp_dump(map | cp::front() | cp::map_kv(cp::hex(), cp::back()));

Change the output destination from the standard error output

To change the output destination, define an explicit specialization of cpp_dump::write_log() with void.

// You can write this in a header file.
// If you write it in a source file, you can remove the inline keyword.
template <>
inline void cpp_dump::write_log(std::string_view output) {
  elsewhere << output << std::endl;
}

How to pass complex expressions to `cpp_dump(...)`

Expressions with commas

Enclose expressions that contain commas in parentheses.

cpp_dump(std::is_same_v<T, U>); // Compile error!
cpp_dump((std::is_same_v<T, U>)); // Correct

Variadic template arguments

When passing variadic template arguments, do not pass any additional arguments.

template <typename... Args>
void variadic_template_func(Args &&...args) {
  int i = 0;
  cpp_dump(args..., i); // Compile error!
  cpp_dump(args...), cpp_dump(i); // Correct
}

For competitive programming use

#ifdef DEFINED_ONLY_IN_LOCAL
#include "./cpp-dump/cpp-dump.hpp"
#define dump(...) cpp_dump(__VA_ARGS__)
namespace cp = cpp_dump;
CPP_DUMP_SET_OPTION_GLOBAL(max_line_width, 80);
CPP_DUMP_SET_OPTION_GLOBAL(log_label_func, cp::log_label::filename());
CPP_DUMP_SET_OPTION_GLOBAL(enable_asterisk, true);
#else
#define dump(...)
#define CPP_DUMP_SET_OPTION(...)
#define CPP_DUMP_DEFINE_EXPORT_OBJECT(...)
#define CPP_DUMP_DEFINE_EXPORT_OBJECT_GENERIC(...)
#define CPP_DUMP_DEFINE_EXPORT_ENUM(...)
#define CPP_DUMP_DEFINE_EXPORT_ENUM_GENERIC(...)
#endif

#include <bits/stdc++.h>

#define rep(i, n) for (int i = 0; i < (int)(n); ++i)

using namespace std;

int main() {
  int N;
  cin >> N;

  vector<int> X(N);
  rep(i, N) { cin >> X[i]; }
  dump(X);

  // To be continued...
}

Then

g++ ./main.cpp -D DEFINED_ONLY_IN_LOCAL

clang++ ./main.cpp -D DEFINED_ONLY_IN_LOCAL

Supported types

cpp_dump() and cpp_dump::export_var() print variables recursively, so they can dump nested variables of any combination of types in the table below.

Category	Type T is supported if ...	Example
Arithmetic	`std::is_arithmetic_v<T> == true`	`bool`, `char`, `int`, `long`, `float`, `double`
String	T is convertible to `std::string_view`	`std::string`, `const char *`, `std::string_view`
Container	T is compatible with the range-based for loop	`std::vector`, `std::forward_list`, C-style arrays
Map	T is either `std::map`, `std::unordered_map`, `std::multimap`, or `std::unordered_multimap`
Set	T is either `std::set`, `std::unordered_set`, `std::multiset`, or `std::unordered_multiset`
Tuple	T is compatible with `std::tuple_size_v<T>`	`std::tuple`, `std::pair`, User-defined tuples
FIFO/LIFO	T is either `std::queue`, `std::priority_queue`, or `std::stack`
Pointer	T is a pointer or smart pointer	`int *`, `std::shared_ptr`, `std::unique_ptr`
Reference	T is `std::reference_wrapper`
Exception	T is convertible to `std::exception`
Other	T is either `std::bitset`, `std::complex`, `std::optional`, `std::variant`, `std::type_info`, `std::type_index` or `std::source_location`(C++20 or higher and g++ and MSVC only)
User-defined	`CPP_DUMP_DEFINE_EXPORT_OBJECT(T, members...);` is in the global scope and the member functions to be displayed is const.
Enum	`CPP_DUMP_DEFINE_EXPORT_ENUM(T, members...);` is in the global scope.
User-defined2	All of the above are not satisfied, T has all members specified by just one `CPP_DUMP_DEFINE_EXPORT_OBJECT_GENERIC(members...);` at top level, and the member functions to be displayed is const.
Enum2	All of the above are not satisfied, and T has all members specified by just one `CPP_DUMP_DEFINE_EXPORT_ENUM_GENERIC(members...);` at top level.
Ostream	All of the above are not satisfied, `std::is_function_v<T> == false && std::is_member_pointer_v<T> == false`, and the function `std::ostream& operator<<(std::ostream&, const T &)` is defined. The string representation of T must not be an empty string (This makes manipulators unsupported).
Asterisk	All of the above are not satisfied, `cpp_dump::options::enable_asterisk == true` and the function `TypeExceptT operator(const T &)` or the const member function `TypeExceptT T::operator() const` is defined.	Iterators

Display example

See also the image in the section A wide variety of supported types.

# Arithmetic
true, 'c', 1, 3.140000

# String
"A normal string"
`A string with '"' or newline(s)`

# Container
[ value1, value2, ... ]

# Map
{ key1: value1, key2: value2, ... },
{ key1 (multiplicity1): [ value1a, value1b, ... ], key2 (multiplicity2): [ ... ], ... }

# Set
{ value1, value2, ... },
{ value1 (multiplicity1), value2 (multiplicity2), ... }

# Tuple
( value1, value2, ... )

# FIFO/LIFO
std::queue{ size()= integer, front()= value, back()= value }

# Pointer
*value
nullptr
0x7fff2246c4d8
# (The address will be displayed when the pointer type is void *
#  or the type the pointer points to is not supported.)

# Reference
true, 'c', 1, 3.140000
# (No change)

# Exception
std::logic_error{ what()= "Error Message" }

# Asterisk
*value

For other categories, see the image(s) in the section...

Other -> A wide variety of supported types
User-defined, Enum -> How to print a user-defined type with cpp-dump

Awesome

cpp-dump

Table of Contents

Introduction

Features

A wide variety of supported types

Auto indent

Filename and line can be printed instead of [dump]

Customizable output color

Can print even user-defined types

Advanced Feature

20+ Manipulators to change the display style

Requirement

Installation

Simplest solution

With CMake

Run cmake --install to copy the headers to /usr/local/include/ or equivalent

Use FetchContent

Configuration (as needed)

Configuration options

max_line_width

max_depth

max_iteration_count

cont_indent_style

enable_asterisk

print_expr

log_label_func

es_style

es_value

detailed_class_es

detailed_member_es

detailed_number_es

Detailed usage

Macros

Types

Variables

Functions

How to print a user-defined type with cpp-dump

Method 1. Use CPP_DUMP_DEFINE_EXPORT_OBJECT() macro

Method 2. Use CPP_DUMP_DEFINE_EXPORT_OBJECT_GENERIC() macro

Method 3. Define std::ostream& operator<<(std::ostream&, const T &) operator

Customize [dump]

Formatting with manipulators

How to use manipulators

front(), middle(), back(), both_ends() manipulators

index() manipulator

int_style() and its alias manipulators

format() manipulator

bw(), boolnum() manipulators

stresc() manipulator

charhex() manipulator

addr() manipulator

map_*() manipulators

Change the output destination from the standard error output

How to pass complex expressions to cpp_dump(...)

Expressions with commas

Variadic template arguments

For competitive programming use

Supported types

Display example

Filename and line can be printed instead of `[dump]`

Run `cmake --install` to copy the headers to `/usr/local/include/` or equivalent

Use `FetchContent`

`max_line_width`

`max_depth`

`max_iteration_count`

`cont_indent_style`

`enable_asterisk`

`print_expr`

`log_label_func`

`es_style`

`es_value`

`detailed_class_es`

`detailed_member_es`

`detailed_number_es`

Method 1. Use `CPP_DUMP_DEFINE_EXPORT_OBJECT()` macro

Method 2. Use `CPP_DUMP_DEFINE_EXPORT_OBJECT_GENERIC()` macro

Method 3. Define `std::ostream& operator<<(std::ostream&, const T &)` operator

Customize `[dump]`

`front()`, `middle()`, `back()`, `both_ends()` manipulators

`index()` manipulator

`int_style()` and its alias manipulators

`format()` manipulator

`bw()`, `boolnum()` manipulators

`stresc()` manipulator

`charhex()` manipulator

`addr()` manipulator

`map_*()` manipulators

How to pass complex expressions to `cpp_dump(...)`