GitHub - kamchatka-volcano/figcone: Read JSON, YAML, TOML, XML or INI configuration by declaring a struct

figcone - is a C++17 / C++20 library, providing a convenient declarative interface for configuration parsers and built-in support for reading JSON, YAML, TOML, XML, INI and shoal config files. To use it, create a configuration schema by declaring a structure for each level of your config file and load it by calling a method, matching the preferred configuration format:

#include <figcone/figcone.h>
#include <filesystem>
#include <iostream>
#include <vector>

struct ThumbnailCfg {
    int maxWidth;
    int maxHeight;
};

struct PhotoViewerCfg {
    std::filesystem::path rootDir;
    std::vector<std::string> supportedFiles;
    ThumbnailCfg thumbnailSettings;
};

int main()
{
    auto cfgReader = figcone::ConfigReader{};
    auto cfg = cfgReader.readToml<PhotoViewerCfg>(R"(
        rootDir = "~/Photos"
        supportedFiles = [".jpg", ".png"]
        [thumbnailSettings]
          maxWidth = 256
          maxHeight = 256
    )");
    //At this point your config is ready to use
    std::cout << "Launching PhotoViewer in directory " << cfg.rootDir << std::endl;
}

This example uses the static reflection interface based on the pfr library. It requires C++20 and only works with aggregate initializable structures. On C++17, the runtime reflection solution originally developed with figcone is used to register configuration fields:

#include <figcone/figcone.h>
#include <filesystem>
#include <iostream>
#include <vector>

struct ThumbnailCfg : public figcone::Config
{
    int maxWidth = param<&ThumbnailCfg::maxWidth>();
    int maxHeight = param<&ThumbnailCfg::maxHeight>();
};

struct PhotoViewerCfg : public figcone::Config{
    //alternatively config fields can be created with macros:
    FIGCONE_PARAM(rootDir, std::filesystem::path);
    FIGCONE_PARAMLIST(supportedFiles, std::vector<std::string>);
    FIGCONE_NODE(thumbnailSettings, ThumbnailCfg);
};

int main()
{
    auto cfgReader = figcone::ConfigReader{};
    auto cfg = cfgReader.readToml<PhotoViewerCfg>(R"(
        rootDir = "~/Photos"
        supportedFiles = [".jpg", ".png"]
        [thumbnailSettings]
          maxWidth = 256
          maxHeight = 256
    )");
    //At this point your config is ready to use
    std::cout << "Launching PhotoViewer in directory " << cfg.rootDir << std::endl;
}

Usage

Config structure for runtime reflection (C++17)

To register configuration structure, subclass figcone::Config and declare fields using the following macros:

FIGCONE_PARAM(name, type) - creates a type name; config field and registers it in the parser.
FIGCONE_PARAMLIST(name, listType) - creates a listType name; config field and registers it in the parser. listType can be any sequence container that supports the emplace_back operation, such as vector, deque, or list from the STL.
FIGCONE_NODE(name, type) - creates a type name; config field for a nested configuration structure and registers it in the parser. The type of the name field must be a subclass of figcone::Config.
FIGCONE_NODELIST(name, listType) - creates a listType name; config field for a list of nested configuration structures and registers it in the parser. listType can be any sequence container that supports the emplace_back operation, such as vector, deque, or list from the STL. The type stored in the list (listType::value_type) must be a subclass of figcone::Config.
FIGCONE_COPY_NODELIST(name, listType) - creates a listType name; config field for a list of nested configuration structures and registers it in the parser. listType can be any sequence container that supports the emplace_back operation, such as vector, deque, or list from the STL. The type stored in the list (listType::value_type) must be a subclass of figcone::Config. The first element of this list acts as a template for the other elements, which means that all unspecified parameters of the second and following elements will be copied from the first element without raising a parsing error for missing parameters.
FIGCONE_DICT(name, mapType) - creates a mapType name; config field for a nested dictionary and registers it in the parser. mapType can be any associative container that supports the emplace operation, such as map or unordered_map from the STL. The key type of the map must be std::string
The preprocessor doesn't handle commas between template arguments in the correct way, so you need to create an alias for your map in order to use it with this macro:

   using StringMap = std::map<std::string, std::string>;
   FIGCONE_DICT(testDict, StringMap);

Notes:

All config entities listed above provide the parenthesis operator () which sets the default value and makes this config field optional. This means that the field can be omitted from the configuration file without raising an error. The empty operator () makes a field's value default initialized, otherwise the passed parameters are used for initialization. FIGCONE_NODE, FIGCONE_NODELIST, and FIGCONE_COPY_NODELIST only support default initialization.
It is also possible to make any config field optional by placing it in figcone::optional (a std::optional-like wrapper with a similar interface). If a value for this field is missing from the config file, the field remains uninitialized and no error occurs.
Types used for config parameters must be default constructible and copyable.

Supporting non-aggregate config structures

Runtime reflection interface of figcone relies on aggregate initialization of user-provided structures. If your config object needs to contain private data or virtual functions, it becomes a non-aggregate type. In this case, you must use the following using declaration to inherit figcone::Config's constructors: using Config::Config;

struct PhotoViewerCfg : public figcone::Config
{
    using Config::Config;
    virtual ~PhotoViewerCfg() = default; //virtual destructor makes PhotoViewerCfg non-aggregate
    FIGCONE_PARAM(rootDir, std::filesystem::path);
    FIGCONE_PARAMLIST(supportedFiles, std::vector<std::string>);
    FIGCONE_NODE(thumbnailSettings, ThumbnailCfg);
};

Registration without macros

Runtime reflection interface of figcone can be used without macros, as every configuration entity described earlier can be registered with the similarly named figcone::Config's member templates:

    struct Cfg : public figcone::Config{
        int testParam                               = param<&Cfg::testParam>();
        int testParam2                              = param<&Cfg::testParam2>()(100);
        figcone::optional<int> testParam3           = param<&Cfg::testParam3>();
        std::vector<double> testParamList           = paramList<&Cfg::testParamList>();
        TestNode testNode                           = node<&Cfg::testNode>();
        figcone::optional<TestNode> testNode2       = node<&Cfg::testNode2>();
        std::vector<TestNode> testNodeList          = nodeList<&Cfg::testNodeList>();
        std::vector<TestNode> copyTestNodeList      = copyNodeList<&Cfg::copyTestNodeList>();
        std::map<std::string, std::string> testDict = dict<&Cfg::testDict>();
    };

Internally, these methods use the nameof library to get config fields' names as strings. Note that nameof relies on non-standard functionality of C++ compilers, so if you don't like it, you can use figcone without it by providing names for config fields yourself:

    struct Cfg : public figcone::Config{
        int testParam                          = param<&Cfg::testParam>("testParam");
        int testParam2                         = param<&Cfg::testParam2>("testParam2")(100);
        figcone::optional<int> testParam3      = param<&Cfg::testParam3>("testParam3");
        std::vector<double> testParamList      = paramList<&Cfg::testParamList>("testParamList");
        TestNode testNode                      = node<&Cfg::testNode>("testNode");
        figcone::optional<TestNode> testNode2  = node<&Cfg::testNode2>("testNode2");
        std::vector<TestNode> testNodeList     = nodeList<&Cfg::testNodeList>("testNodeList");
        std::vector<TestNode> copyTestNodeList = copyNodeList<&Cfg::copyTestNodeList>("copyTestNodeList");
        std::map<std::string, std::string> testDict = dict<&Cfg::testDict>("testDict");
    };

Please note that on the MSVC compiler, the nameof features used by figcone require the C++20 standard. This is handled automatically by CMake configuration if MSVC is your default compiler. Otherwise, you will need to enable the C++20 standard manually.

Config structures declared using the macro-free methods are fully compatible with all of figcone's functionality. Examples in the documentation use registration with macros, as it is the least verbose method.

Config structure for static reflection (C++20)

Static reflection provided by the pfr library allows us to register configuration by using aggregate structures without any base class and macros. Let's return to the example from the beginning of this document:

struct PhotoViewerCfg {
    std::filesystem::path rootDir;
    std::vector<std::string> supportedFiles;
    ThumbnailCfg thumbnailSettings;
};

All field types and names are registered during compile time. In this example, rootDir is registered as a parameter, supportedFiles as a parameter list, and thumbnailSettings as a node.
If a field can be converted to a string (by a conversion operator, std::stringstream, or figcone::StringConverter—which will be discussed later), it is registered as a parameter; otherwise, it is registered as a node. The same rule applies to the container elements.

To create node or parameter lists, use any sequence container that supports the emplace_back operation, such as vector, deque, or list from the STL.
To create dictionaries, use any associative container that supports the emplace operation, such as map or unordered_map from the STL. The key type of the map must be std::string.

Optional fields are created with std::optional; there's no need to use figcone::optional. If a field has a default value, it's not ergonomic to place it in std::optional just to specify that it can be omitted from the config file. An alternative way to make a field optional is using the figcone::OptionalField trait and placing it in the figcone::FieldTraits type list. Let's make supportedFiles from the previous example optional to see how it's done:

struct PhotoViewerCfg {
    std::filesystem::path rootDir;
    std::vector<std::string> supportedFiles;
    ThumbnailCfg thumbnailSettings;
    
    using traits=figcone::FieldTraits<
      figcone::OptionalField<&PhotoViewerCfg::supportedFiles>     
    >;
};

Another trait is figcone::CopyNodeListField. It enables the use of a copy node list functionality. The first element of such a list acts as a template for the other elements. This means that all unspecified parameters of the second and following elements will be copied from the first element without raising a parsing error for missing parameters (it's the same field type as FIGCONE_COPYNODELIST from the runtime reflection interface).

Let's make thumbnailSettings a copy node list:

struct PhotoViewerCfg {
    std::filesystem::path rootDir;
    std::vector<std::string> supportedFiles;
    std::vector<ThumbnailCfg> thumbnailSettings;
    
    using traits=figcone::FieldTraits<
      figcone::OptionalField<&PhotoViewerCfg::supportedFiles>,
      figcone::CopyNodeListField<&PhotoViewerCfg::thumbnailSettings>
    >;
};

Sadly, using such trait lists is currently the only possible way to add additional information about config fields. If you have complex configuration that needs to have many traits, it might make sense to prefer the runtime reflection interface to keep the code more concise.

Name format

You do not need to change your code style when declaring config fields. camelCase, snake_case, and PascalCase names are supported, and can be converted to the format used by parameter names in the config file. To do this, specify the configuration names format with the figcone::NameFormat enum by passing its value to the figcone::ConfigReader template argument.

To demonstrate it, let's change our PhotoViewer example to use snake_case names in the configuration:

///examples_static_refl/ex02_static_refl.cpp
///
#include <figcone/figcone.h>
#include <filesystem>
#include <iostream>
#include <vector>

struct ThumbnailCfg {
    int maxWidth;
    int maxHeight;
};

struct PhotoViewerCfg {
    std::filesystem::path rootDir;
    std::vector<std::string> supportedFiles;
    ThumbnailCfg thumbnailSettings;
};


int main()
{
    auto cfgReader = figcone::ConfigReader<figcone::NameFormat::SnakeCase>{};
    auto cfg = cfgReader.readToml<PhotoViewerCfg>(R"(
        root_dir = "/home/kamchatka-volcano/photos"
        supported_files = [".jpg", ".png"]
        [thumbnail_settings]
          max_width = 256
          max_height = 256
    )");
    //At this point your config is ready to use
    std::cout << "Launching PhotoViewer in directory " << cfg.rootDir << std::endl;
    return 0;
}

Supported formats

Internally, the figcone library works on a tree-like structure provided by the figcone_tree library, and it is not aware of different configuration formats. The user needs to provide a parser implementing the figcone_tree::IParser interface to convert a configuration file to a tree structure based on the figcone_tree::TreeNode class. It is also possible to create a figcone compatible parser adapter that transforms the parsing result of some 3rd party configuration parsing library to a tree using figcone_tree::TreeNode. Five such adapters for popular configuration formats are included in figcone, and are fetched and built into a static library called figcone_formats which is automatically configured and linked by figcone's CMake configuration. An obscure configuration format called shoal, which was designed by the author of figcone, is also available and can be used as an example of an original parser implementation that is compatible with figcone.

Let's increase the complexity of our example config to demonstrate how configuration elements work with each format:

demo.h

///examples/demo.h
///
#pragma once
#include <figcone/config.h>
#include <string>
#include <vector>
#include <map>

struct ThumbnailCfg {
    bool enabled = true;
    int maxWidth;
    int maxHeight;

    using traits = figcone::FieldTraits<
            figcone::OptionalField<&ThumbnailCfg::enabled>>;
};
struct HostCfg {
    std::string ip;
    int port;
};
struct SharedAlbumCfg {
    std::filesystem::path dir;
    std::string name;
    std::vector<HostCfg> hosts;

    using traits = figcone::FieldTraits<
            figcone::OptionalField<&SharedAlbumCfg::hosts>>;
};
struct PhotoViewerCfg {
    std::filesystem::path rootDir;
    std::vector<std::string> supportedFiles;
    ThumbnailCfg thumbnails;
    std::vector<SharedAlbumCfg> sharedAlbums;
    std::map<std::string, std::string> envVars;

    using traits = figcone::FieldTraits<
            figcone::OptionalField<&PhotoViewerCfg::envVars>,
            figcone::OptionalField<&PhotoViewerCfg::sharedAlbums>,
            figcone::CopyNodeListField<&PhotoViewerCfg::sharedAlbums>>;
};

JSON

JSON support is provided by nlohmann/json library which is fetched and adapted to the figcone interface by the figcone_json library.

A JSON config that matches the configuration listed in demo.h earlier, looks like this:

demo.json

{
  "rootDir": "~/Photos",
  "supportedFiles": [
    ".jpg",
    "png"
  ],
  "thumbnails": {
    "enabled": "1",
    "maxWidth": "128",
    "maxHeight": "128"
  },
  "sharedAlbums": [
    {
      "dir": "summer_2019",
      "name": "Summer (2019)",
      "hosts" : [{"ip" : "127.0.0.1", "port" : "8080" }]
    },
    {
      "dir": "misc",
      "name": "Misc"
    }
  ],
  "envVars": {
    "DISPLAY" : "0.1"
  }
}

///examples/demo_json.cpp
///
#include "demo.h"
#include <figcone/configreader.h>
#include <iostream>

int main()
{
    auto cfgReader = figcone::ConfigReader{};
    auto cfg = cfgReader.readJsonFile<PhotoViewerCfg>(std::filesystem::canonical("../../examples/demo.json"));
    std::cout << "Launching PhotoViewer in directory" << cfg.rootDir << std::endl;
    return 0;
}

JSON configurations that have a node list in its root are supported. Pass figcone::RootType::NodeList as the second template argument of the used read function:

///examples/demo_json_root_list.cpp
///
#include "demo.h"
#include "print_demo.h"
#include <figcone/configreader.h>
#include <iostream>

int main()
{
    auto cfgReader = figcone::ConfigReader{};
    auto cfgList = cfgReader.readJsonFile<PhotoViewerCfg, figcone::RootType::NodeList>(
            std::filesystem::canonical("../../examples/demo_root_list.json"));
    for (const auto& cfg : cfgList) {
        std::cout << "Launching PhotoViewer in directory " << cfg.rootDir << std::endl;
        printDemoConfig(cfg);
        std::cout << "---" << std::endl;
    }
}

YAML

YAML support is provided by the rapidyaml library which is fetched and adapted to the figcone interface by the figcone_yaml library.

A YAML config that matches the configuration listed in demo.h earlier, looks like this:
demo.yaml

  rootDir: ~/Photos
  supportedFiles: [ ".jpg", "png"]
  thumbnails:
    enabled: 1
    maxWidth: 128
    maxHeight: 128
  
  sharedAlbums:
    -
      dir: "summer_2019"
      name: "Summer (2019)"
      hosts: 
        -
          ip: "127.0.0.1"
          port: 8080
    -    
      dir: "misc"
      name: "Misc"      
  envVars: 
    DISPLAY : "0.1"

///examples/demo_yaml.cpp
///
#include "demo.h"
#include <figcone/configreader.h>
#include <iostream>

int main()
{
    auto cfgReader = figcone::ConfigReader{};
    auto cfg = cfgReader.readYamlFile<PhotoViewerCfg>(std::filesystem::canonical("../../examples/demo.yaml"));
    std::cout << "Launching PhotoViewer in directory" << cfg.rootDir << std::endl;
    return 0;
}

YAML configurations that have a node list in its root are supported. Pass figcone::RootType::NodeList as the second template argument of the used read function:

///examples/demo_yaml_root_list.cpp
///
#include "demo.h"
#include "print_demo.h"
#include <figcone/configreader.h>
#include <iostream>

int main()
{
    auto cfgReader = figcone::ConfigReader{};
    auto cfgList = cfgReader.readYamlFile<PhotoViewerCfg, figcone::RootType::NodeList>(
            std::filesystem::canonical("../../examples/demo_root_list.yaml"));
    for (const auto& cfg : cfgList) {
        std::cout << "Launching PhotoViewer in directory " << cfg.rootDir << std::endl;
        printDemoConfig(cfg);
        std::cout << "---" << std::endl;
    }
}

TOML

TOML support is provided by the toml11 library which is fetched and adapted to the figcone interface by the figcone_toml library.

A TOML config that matches the configuration listed in demo.h earlier, looks like this:
demo.toml

  rootDir = "~/Photos"
supportedFiles = [".jpg", "png"]
  [thumbnails]
    enabled = 1
    maxWidth = 128
    maxHeight = 128
  
  [[sharedAlbums]]
    dir  = "summer_2019"
    name = "Summer (2019)"
    [[sharedAlbums.hosts]]
      ip = "127.0.0.1"
      port = 8080 
  [[sharedAlbums]]
      dir  = "misc"
      name = "Misc"      
  [envVars] 
    DISPLAY = "0.1"

///examples/demo_toml.cpp
///
#include "demo.h"
#include <figcone/configreader.h>
#include <iostream>

int main()
{
    auto cfgReader = figcone::ConfigReader{};
    auto cfg = cfgReader.readTomlFile<PhotoViewerCfg>(std::filesystem::canonical("../../examples/demo.toml"));
    std::cout << "Launching PhotoViewer in directory" << cfg.rootDir << std::endl;
    return 0;
}

Notes:

TOML nested arrays aren't supported, as they can't be represented within figcone configuration tree.
TOML datetime types can be used in the config by including <figcone/format/toml/datetime.h> header:

#include <figcone/config.h>
#include <figcone/format/toml/datetime.h>

struct Cfg: figcone::Config
{   
    FICONE_PARAM(date, figcone::toml::DateTimePoint);
    FICONE_PARAM(time, figcone::toml::TimeDuration);
};

figcone::toml::DateTimePoint is used to read TOML local_date (2018-12-23), local_datetime (2018-12-23T12:30:00) and offset_datetime (2018-12-23T12:30:00+09:30) values.
figcone::toml::TimeDuration is used to read TOML time (12:30:00) values.

XML

XML support is provided by the rapidxml library which is fetched and adapted to the figcone interface by the figcone_xml library.

An XML config that matches the configuration listed in demo.h earlier, looks like this:
demo.xml

<root rootDir="~/Photos"
      supportedFiles = "[ '.jpg', 'png']">
    <thumbnails enabled="1"
                maxWidth="128"
                maxHeight="128"/>
    <sharedAlbums>
        <album dir="summer_2019"
               name="Summer (2019)">
            <hosts>
                <host ip="127.0.0.1"
                      port="8080"/>
            </hosts>
        </album>
        <album dir="misc"
               name="Misc"/>
    </sharedAlbums>
    <envVars DISPLAY="0.1"/>
</root>

///examples/demo_xml.cpp
///
#include "demo.h"
#include <figcone/configreader.h>
#include <iostream>

int main()
{
    auto cfgReader = figcone::ConfigReader{};
    auto cfg = cfgReader.readXmlFile<PhotoViewerCfg>(std::filesystem::canonical("../../examples/demo.xml"));
    std::cout << "Launching PhotoViewer in directory" << cfg.rootDir << std::endl;
    return 0;
}

Notes:

Node list tags can't contain attributes and all its children tags representing node list elements must use the same name.
Parameter lists are stored in attributes and have the format: [value1, value2, ...]

INI

INI support is provided by the inifile-cpp library which is fetched and adapted to the figcone interface by the figcone_ini library.

An INI config that matches the configuration listed in demo.h earlier, looks like this:
demo.ini

rootDir = "~/Photos"
supportedFiles = [".jpg", "png"]
[thumbnails]
  enabled = 1
  maxWidth = 128
  maxHeight = 12
[sharedAlbums.0]
  dir  = "summer_2019"
  name = "Summer (2019)"
  [sharedAlbums.0.hosts.0]
    ip = "127.0.0.1"
    port = 8080
[sharedAlbums.1]
  dir  = "misc"
  name = "Misc"
[envVars]
DISPLAY = "0.1"

///examples/demo_ini.cpp
///
#include "demo.h"
#include <figcone/configreader.h>
#include <iostream>

int main()
{
    auto cfgReader = figcone::ConfigReader{};
    auto cfg = cfgReader.readIniFile<PhotoViewerCfg>(std::filesystem::canonical("../../examples/demo.ini"));
    std::cout << "Launching PhotoViewer in directory" << cfg.rootDir << std::endl;
    return 0;
}

Notes:

Nested structures can be created by using dotted ini section names listing the full hierarchy: [grandparent.parent.child]
Node lists can be created by using indices in dotted ini section names, starting with zero for the first element: [nodelist.0]
Parameter lists have the format: [value1, value2, ...]
Multiline values aren't supported.

shoal

shoal support is provided by the figcone_shoal library.

A shoal config that matches the configuration listed in demo.h earlier, looks like this:
demo.shoal

rootDir = "~/Photos"
supportedFiles = [ ".jpg", "png"]
#thumbnails:
  enabled = 1
  maxWidth = 128
  maxHeight = 128
---

#sharedAlbums:
###
  dir  = "summer_2019"
  name = "Summer (2019)"
  #hosts:
  ###
    ip = "127.0.0.1"
    port = 8080
  -
###
  dir  = "misc"
  name = "Misc"
---

#envVars:
  DISPLAY = "0.1"

///examples/demo_shoal.cpp
///
#include "demo.h"
#include "print_demo.h"
#include <figcone/configreader.h>
#include <iostream>

int main()
{
    auto cfgReader = figcone::ConfigReader{};
    auto cfg = cfgReader.readShoalFile<PhotoViewerCfg>(std::filesystem::canonical("../../examples/demo.shoal"));
    std::cout << "Launching PhotoViewer in directory" << cfg.rootDir << std::endl;
    printDemoConfig(cfg);
    return 0;
}

Creation of figcone-compatible parsers

To create a parser compatible with figcone, you will need to use the figcone_tree library, which provides all the necessary types and interfaces for this task. The parsing class should implement the figcone::IParser interface and return the result of the configuration parsing in the form of a tree-like structure, constructed using figcone::TreeNode and figcone::TreeParam objects. Let's demonstrate how to work with the figcone_tree library by creating a fake parser that provides a configuration tree for the demo structure listed in demo.h:

///examples/demo_parser.cpp
///
#include "demo.h"
#include <figcone_tree/iparser.h>
#include <figcone_tree/tree.h>
#include <figcone_tree/errors.h>
#include <figcone/configreader.h>

class DemoTreeProvider : public figcone::IParser
{
    figcone::Tree parse(std::istream& stream) final
    {
        auto tree = figcone::makeTreeRoot();
        tree->asItem().addParam("rootDir", "~/Photos");
        tree->asItem().addParamList("supportedFiles", {".jpg", ".png"});

        auto& thumbNode = tree->asItem().addNode("thumbnails");
        thumbNode.asItem().addParam("enabled", "1");
        thumbNode.asItem().addParam("maxWidth", "128");
        thumbNode.asItem().addParam("maxHeight", "128");

        auto& albumsNodeList = tree->asItem().addNodeList("sharedAlbums");
        auto& albumNode = albumsNodeList.asList().emplaceBack();
        albumNode.asItem().addParam("dir", "summer_2019");
        albumNode.asItem().addParam("name", "Summer (2019)");
        auto& hostsNodeList = albumNode.asItem().addNodeList("hosts");
        auto& hostNode = hostsNodeList.asList().emplaceBack();
        hostNode.asItem().addParam("ip", "127.0.0.1");
        hostNode.asItem().addParam("port", "80");

        //For error notifications use figcone::ConfigError exceptions
        if (stream.bad())
            throw figcone::ConfigError{"stream error"};

        //Stream position information can be added to objects
        auto pos = figcone::StreamPosition{13, 1};
        auto& envVarsNode = tree->asItem().addNode("envVars", pos);
        envVarsNode.asItem().addParam("DISPLAY", "0.1");

        //and to ConfigError exceptions
        if (stream.fail())
            throw figcone::ConfigError{"stream error", pos};

        return tree;
    }
};

int main()
{
    auto cfgReader = figcone::ConfigReader{};
    auto parser = DemoTreeProvider{};
    auto cfg = cfgReader.read<PhotoViewerCfg>("", parser);
    std::cout << "Launching PhotoViewer in directory " << cfg.rootDir;
}

User defined types

To use user-defined types in your config, it's necessary to add a specialization of the struct figcone::StringConverter and implement its static method fromString.
Let's replace the HostCfg config structure with a parameter of type Host that is stored in the config as a string "ipAddress:port".

///examples_static_refl/ex03_static_refl.cpp
///
#include <figcone/figcone.h>
#include <filesystem>
#include <iostream>
#include <vector>
#include <map>

struct Host {
    std::string ip;
    int port;
};

namespace figcone {
template<>
struct StringConverter<Host> {
    static std::optional<Host> fromString(const std::string& data)
    {
        auto delimPos = data.find(':');
        if (delimPos == std::string::npos)
            return {};
        auto host = Host{};
        host.ip = data.substr(0, delimPos);
        host.port = std::stoi(data.substr(delimPos + 1, data.size() - delimPos - 1));
        return host;
    }
};
} //namespace figcone

struct SharedAlbumCfg {
    std::filesystem::path dir;
    std::string name;
    std::vector<Host> hosts;
};
struct PhotoViewerCfg {
    std::filesystem::path rootDir;
    std::vector<std::string> supportedFiles;
    std::vector<SharedAlbumCfg> sharedAlbums;
    std::map<std::string, std::string> envVars;

    using traits = figcone::FieldTraits<
            figcone::OptionalField<&PhotoViewerCfg::envVars>>;
};

int main()
{
    auto cfgReader = figcone::ConfigReader{};
    auto cfg = cfgReader.readYaml<PhotoViewerCfg>(R"(
      rootDir: ~/Photos
      supportedFiles: [ ".jpg", "png"]
      sharedAlbums:
        -
          dir: "summer_2019"
          name: "Summer 2019"
          hosts: ["127.0.0.1:8080"]
    )");

    std::cout << "Launching PhotoViewer in directory " << cfg.rootDir << std::endl;

    if (!cfg.supportedFiles.empty())
        std::cout << "Supported files:" << std::endl;
    for (const auto& file : cfg.supportedFiles)
        std::cout << "  " << file << std::endl;

    if (!cfg.sharedAlbums.empty())
        std::cout << "Shared albums:" << std::endl;
    for (const auto& album : cfg.sharedAlbums){
        std::cout << "  Album:" << album.name << std::endl;
        std::cout << "    Hosts:" << std::endl;
        for (const auto& host : album.hosts)
            std::cout << "      " << host.ip << ":" << host.port << std::endl;
    }

    return 0;
}

To provide additional information in the error message of the StringConverter, you can use the figcone::ValidationError exception:

namespace figcone{
template<>
struct StringConverter<Host>{
    static std::optional<Host> fromString(const std::string& data)
    {
        auto delimPos = data.find(':');
        if (delimPos == std::string::npos)
            throw ValidationError{"the host parameter must be in the format 'ipAddress:port'"};
        auto host = Host{};
        host.ip = data.substr(0, delimPos);
        host.port = std::stoi(data.substr(delimPos + 1, data.size() - delimPos - 1));
        return host;
    }
};
}

Unregistered fields handling

By default, figcone requires an exact match of the configuration file content with a registered configuration structure. Besides any missing or wrongly typed data, errors are raised when some config fields are present in the file but are not registered in figcone. This behavior is debatable; for example, it can make it impossible to store two different apps' configurations in the same file. For this reason, error signaling for unregistered fields can be suppressed by registering the necessary behavior with the UnregisteredFieldHandler template class.

Provide a template definition with the void operator()(figcone::FieldType, const std::string& fieldName, const figcone::StreamPosition&) callable signature to apply the new behavior to all config levels:

namespace figcone {

template<typename T>
struct UnregisteredFieldHandler {
    void operator()(figcone::FieldType, const std::string&, const figcone::StreamPosition&) 
    {
        //Don't do anything here to suppress errors for all unregistered fields
    }
};

} //namespace figcone

Or create a specialization of UnregisteredFieldHandler with a specific node type:

namespace figcone {

template<>
struct UnregisteredFieldHandler<SharedAlbumCfg> {
    void operator()(figcone::FieldType, const std::string&, const figcone::StreamPosition&) 
    {
        //Don't do anything here to suppress errors for all unregistered fields inside the SharedAlbumCfg node.
    }
};

} //namespace figcone

Instead of suppressing all errors of this kind, it's possible to make more targeted adjustments. For example, let's allow SharedAlbumCfg to have unregistered parameters but forbid the unregistered subnodes:

namespace figcone {

template<>
struct UnregisteredFieldHandler<SharedAlbumCfg> {
    void operator()(figcone::FieldType fieldType, const std::string& fieldName, const figcone::StreamPosition& position) 
    {
        if (fieldType == figcone::FieldType::Node)
            throw ConfigError{"Unknown node '" + fieldName + "'", position};
            
        //Don't do anything for fieldType == figcone::FieldType::Param     
    }
};

} //namespace figcone

Validators

Processed config parameters and nodes can be validated by registering constraint-checking functions or callable objects. The signature must be compatible with void (const T&) where T is the type of validated config structure field T value or an optional field figcone::optional<T> (validators of optional fields aren't invoked if they are empty).
If the option's value is invalid, the validator must throw a figcone::ValidationError exception.

Runtime reflection validators

struct Cfg : figcone::Config<>{
    FIGCONE_PARAM(number, int).ensure( 
        [](int paramValue){
            if (paramValue < 0)
                throw figcone::ValidationError{"value can't be negative."};
        });
};

Let's improve the PhotoViewer program by checking that rootDir path exists and supportedFiles parameter list isn't empty:

///examples/ex04.cpp
///
#include <figcone/figcone.h>
#include <filesystem>
#include <iostream>
#include <vector>
#include<map>

struct NotEmpty{
    template<typename TList>
    void operator()(const TList& list)
    {
        if (!list.empty())
            throw figcone::ValidationError{"can't be empty."};
    }
};

struct PhotoViewerCfg : public figcone::Config{
    PARAM(rootDir, std::filesystem::path).ensure([](const auto& path) {
        if (!std::filesystem::exists(path))
            throw figcone::ValidationError{"a path must exist"};
    });
    PARAMLIST(supportedFiles, std::vector<std::string>).ensure<NotEmpty>();
    using StringMap = std::map<std::string, std::string>;
    DICT(envVars, StringMap)();
};

int main()
{
    try {
        auto cfgReader = figcone::ConfigReader{};
        auto cfg = cfgReader.readYaml<PhotoViewerCfg>(R"(
            rootDir: ~/Photos
            supportedFiles: []
        )");

        std::cout << "Launching PhotoViewer in directory " << cfg.rootDir << std::endl;

        if (!cfg.supportedFiles.empty())
            std::cout << "Supported files:" << std::endl;
        for (const auto& file : cfg.supportedFiles)
            std::cout << "  " << file << std::endl;

        return 0;
    }
    catch(const figcone::ConfigError& e){
        std::cout << "Config error:" << e.what();
        return 1;
    }
}

Now the read method will throw an exception if configuration provides invalid rootDir or supportedFiles parameters.

Static reflection validators

struct NotNegative{
    void operator()(int paramValue)
    {
        if (paramValue < 0)
                throw figcone::ValidationError{"value can't be negative."};
    }
};

struct Cfg {
    int number;
    
    using traits=figcone::FieldTraits<
      figcone::ValidatedField<&Cfg::number, NotNegative>
    >;  
};

Let's rewrite the PhotoViewer program again, this time using compile time reflection interface:

///examples_static_refl/ex04_static_refl.cpp
///
#include <figcone/figcone.h>
#include <filesystem>
#include <iostream>
#include <map>
#include <vector>

struct NotEmpty {
    template<typename TList>
    void operator()(const TList& list)
    {
        if (!list.empty())
            throw figcone::ValidationError{"can't be empty."};
    }
};

struct PathExists {
    void operator()(const std::filesystem::path& path)
    {
        if (!std::filesystem::exists(path))
            throw figcone::ValidationError{"a path must exist"};
    }
};

struct PhotoViewerCfg {
    std::filesystem::path rootDir;
    std::vector<std::string> supportedFiles;
    std::map<std::string, std::string> envVars;

    using traits = figcone::FieldTraits<
            figcone::ValidatedField<&PhotoViewerCfg::rootDir, PathExists>,
            figcone::ValidatedField<&PhotoViewerCfg::supportedFiles, NotEmpty>,
            figcone::OptionalField<&PhotoViewerCfg::envVars>>;
};

int main()
{
    try {
        auto cfgReader = figcone::ConfigReader{};
        auto cfg = cfgReader.readYaml<PhotoViewerCfg>(R"(
            rootDir: ~/Photos
            supportedFiles: []
        )");

        std::cout << "Launching PhotoViewer in directory " << cfg.rootDir << std::endl;

        if (!cfg.supportedFiles.empty())
            std::cout << "Supported files:" << std::endl;
        for (const auto& file : cfg.supportedFiles)
            std::cout << "  " << file << std::endl;

        return 0;
    }
    catch (const figcone::ConfigError& e) {
        std::cout << "Config error:" << e.what();
        return 1;
    }
}

Post-processors

If you need to modify or validate the config object that is produced by figcone::ConfigReader, you can register the necessary action by creating a specialization of the figcone::PostProcessor class template:

///examples_static_refl/ex05_static_refl.cpp
///
struct ThumbnailCfg {
    int maxWidth;
    int maxHeight;
};

struct PhotoViewerCfg {
    std::filesystem::path rootDir;
    std::vector<std::string> supportedFiles;
    ThumbnailCfg thumbnailSettings;
};

namespace figcone {
    template<>
    void PostProcessor<PhotoViewerCfg>::operator()(PhotoViewerCfg& cfg)
    {
        auto supportPng = std::find(cfg.supportedFiles.begin(), cfg.supportedFiles.end(), ".png") != cfg.supportedFiles.end();
        if (supportPng && cfg.thumbnailSettings.maxWidth > 128)
            throw ValidationError{"thumbnail width can't be larger than 128px when png images are present"};

    }
}
}

Installation

Download and link the library from your project's CMakeLists.txt:

cmake_minimum_required(VERSION 3.14)

include(FetchContent)
FetchContent_Declare(figcone
    GIT_REPOSITORY "https://github.com/kamchatka-volcano/figcone.git"
    GIT_TAG "origin/master"
)
#uncomment if you need to install figcone with your target
#set(INSTALL_FIGCONE ON)
FetchContent_MakeAvailable(figcone)

add_executable(${PROJECT_NAME})
target_link_libraries(${PROJECT_NAME} PRIVATE figcone::figcone)

By default figcone fetches all supported configuration format libraries. You can control this with CMake option FIGCONE_USE_ALL. You can also enable support for a single configuration format or a combination of formats by setting the following options:

FIGCONE_USE_JSON - fetches and configures the figcone_json library;
FIGCONE_USE_YAML - fetches and configures the figcone_yaml library;
FIGCONE_USE_TOML - fetches and configures the figcone_toml library;
FIGCONE_USE_XML - fetches and configures the figcone_xml library;
FIGCONE_USE_INI - fetches and configures the figcone_ini library;
FIGCONE_USE_SHOAL - fetches and configures the figcone_shoal library;

To install the library system-wide, use the following commands:

git clone https://github.com/kamchatka-volcano/figcone.git
cd figcone
cmake -S . -B build
cmake --build build
cmake --install build

After installation, you can use the find_package() command to make the installed library available inside your project:

find_package(figcone 2.0.0 REQUIRED)
target_link_libraries(${PROJECT_NAME} PRIVATE figcone::figcone)

Running tests

cd figcone
cmake -S . -B build -DENABLE_TESTS=ON
cmake --build build
cd build/tests && ctest

Building examples

cd figcone
cmake -S . -B build -DENABLE_EXAMPLES=ON
cmake --build build
cd build/examples

License

figcone is licensed under the MS-PL license

Name		Name	Last commit message	Last commit date
Latest commit History 105 Commits
.github/workflows		.github/workflows
doc		doc
examples		examples
examples_static_refl		examples_static_refl
external		external
formats		formats
functional_tests		functional_tests
include/figcone		include/figcone
tests		tests
tests_cpp20		tests_cpp20
tests_static_refl		tests_static_refl
.clang-format		.clang-format
CMakeLists.txt		CMakeLists.txt
CMakePresets.json		CMakePresets.json
LICENSE.md		LICENSE.md
README.md		README.md

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Repository files navigation

Table of Contents

Usage

Config structure for runtime reflection (C++17)

Supporting non-aggregate config structures

Registration without macros

Config structure for static reflection (C++20)

Name format

Supported formats

demo.h

JSON

YAML

TOML

XML

INI

shoal

Creation of figcone-compatible parsers

User defined types

Unregistered fields handling

Validators

Runtime reflection validators

Static reflection validators

Post-processors

Installation

Running tests

Building examples

License

About

Releases 22

Packages

Languages

License

kamchatka-volcano/figcone

Folders and files

Latest commit

History

Repository files navigation

Table of Contents

Usage

Config structure for runtime reflection (C++17)

Supporting non-aggregate config structures

Registration without macros

Config structure for static reflection (C++20)

Name format

Supported formats

demo.h

JSON

YAML

TOML

XML

INI

shoal

Creation of figcone-compatible parsers

User defined types

Unregistered fields handling

Validators

Runtime reflection validators

Static reflection validators

Post-processors

Installation

Running tests

Building examples

License

About

Topics

Resources

License

Stars

Watchers

Forks

Releases 22

Packages 0

Languages

Packages