使用boost program_options处理帮助消息，删除默认值或重新格式化帮助消息

Question

我在写一个涉及boost :: program_options的C++程序，在这里我遇到了一些问题。我的一些代码在这里给出。

int main(int argc, char* argv[]) { 
    options_description desc("useage: filterfq", options_description::m_default_line_length * 2, options_description::m_default_line_length); 

    options_description generic("Gerneric options", options_description::m_default_line_length * 2, options_description::m_default_line_length); 
    generic.add_options() 
     ("help,h", "produce help message") 
    ; 

    options_description param("Parameters", options_description::m_default_line_length * 2, options_description::m_default_line_length); 
    param.add_options() 
     ("checkQualitySystem,c", bool_switch(), "only check quality system of the fastq file") 
     ("baseNrate,N", value<float>() -> default_value(0.05), "maximum rate of \'N\' base allowed along a read") 
     ("averageQuality,Q", value<float>() -> default_value(0), "minimum average quality allowed along a read") 
     ("perBaseQuality,q", value<int>() -> default_value(5), "minimum quality per base allowed along a read") 
     ("lowQualityRate,r", value<float>() -> default_value(0.5), "maximum low quality rate along a read") 
     ("rawQualitySystem,s", value<int>(), "specify quality system of raw fastq\n0: Sanger\n1: Solexa\n2: Illumina 1.3+\n3: Illumina 1.5+\n4: Illumina 1.8+") 
     ("preferSpecifiedRawQualitySystem,p", bool_switch(), "indicate that user prefers the given quality system to process") 
    ; 

    options_description input("Input", options_description::m_default_line_length * 2, options_description::m_default_line_length); 
    input.add_options() 
     ("rawFastq,f", value< vector<path> >() -> required() -> multitoken(), "raw fastq file(s) that need cleaned, required") 
    ; 

    options_description output("Output", options_description::m_default_line_length * 2, options_description::m_default_line_length); 
    output.add_options() 
     ("cleanQualitySystem,S", value<int>() -> default_value(4), "specify quality system of cleaned fastq, the same as rawQualitySystem") 
     ("outDir,O", value<path>() -> default_value(current_path()), "specify output directory, not used if cleanFastq is specified") 
     ("outBasename,o", value<string>(), "specify the basename for output file(s), required if outDir is specified") 
     ("cleanFastq,F", value< vector<path> >() -> multitoken(), "cleaned fastq file name(s), not used if outDir or outBasename is specified") 
     ("droppedFastq,D", value< vector<path> >() -> multitoken(), "fastq file(s) containing reads that are filtered out") 
    ; 

    desc.add(generic).add(param).add(input).add(output); 

    variables_map vm; 
    store(command_line_parser(argc, argv).options(desc).run(), vm); 
    if (vm.count("help")) { 
     cout << desc << "\n"; 
     return 0; 
    } 
    ... 
} 

#include using namespace parts are not given here。当我输入的命令来查看帮助信息，它表明我下面

useage: filterfq: 

Gerneric options: 
    -h [ --help ]   produce help message 

Parameters: 
    -c [ --checkQualitySystem ]    only check quality system of the fastq file 
    -N [ --baseNrate ] arg (=0.0500000007) maximum rate of 'N' base allowed along a read 
    -Q [ --averageQuality ] arg (=0)   minimum average quality allowed along a read 
    -q [ --perBaseQuality ] arg (=5)   minimum quality per base allowed along a read 
    -r [ --lowQualityRate ] arg (=0.5)  maximum low quality rate along a read 
    -s [ --rawQualitySystem ] arg    specify quality system of raw fastq 
              0: Sanger 
              1: Solexa 
              2: Illumina 1.3+ 
              3: Illumina 1.5+ 
              4: Illumina 1.8+ 
    -p [ --preferSpecifiedRawQualitySystem ] indicate that user prefers the given quality system to process 

Input: 
    -f [ --rawFastq ] arg raw fastq file(s) that need cleaned, required 

Output: 
    -S [ --cleanQualitySystem ] arg (=4)    specify quality system of cleaned fastq, the same as rawQualitySystem 
    -O [ --outDir ] arg (="/home/tanbowen/filterfq") specify output directory, not used if cleanFastq is specified 
    -o [ --outBasename ] arg       specify the basename for output file(s), required if outDir is specified 
    -F [ --cleanFastq ] arg       cleaned fastq file name(s), not used if outDir or outBasename is specified 
    -D [ --droppedFastq ] arg      fastq file(s) containing reads that are filtered out 

帮助消息看起来有点丑，尤其是“0.0500000007”，我想改善它。但我搜索了很长时间，我找不到解决方案。所以我在这里寻求帮助解决以下问题：

1. 有什么方法可以重新格式化帮助信息吗？
2. 如果1不可能，如何删除“arg”部分和默认值？
3. 是否有任何方法来对齐右侧的描述？

一个额外的问题：我如何才能防止被执行

filter -f <some file> -f <some file> 

即不允许被多次指定相同的选项下面的命令？

非常感谢！

Answer 1

1. 是的，见下文（示出了收集和显示格式的选项的一般形式）

2. 看的options_description构造。它允许你指定列的宽度。

以下是自定义选项值的（实际）示例。在我的情况下，我想收集以字节为单位的缓冲区大小，但也希望能够解析像4K或1M的东西。

struct bytesize_option 
{ 
    bytesize_option(std::size_t val = 0) : _size(val) {} 

    std::size_t value() const { return _size; } 
    void set(std::size_t val) { _size = val; } 

private: 
    std::size_t _size; 
}; 

std::ostream& operator<<(std::ostream& os, bytesize_option const& hs); 
std::istream& operator>>(std::istream& is, bytesize_option& hs); 

namespace { 
    static constexpr auto G = std::size_t(1024 * 1024 * 1024); 
    static constexpr auto M = std::size_t(1024 * 1024); 
    static constexpr auto K = std::size_t(1024); 
} 

std::ostream& operator<<(std::ostream& os, bytesize_option const& hs) 
{ 
    auto v = hs.value(); 
    if (v % G == 0) { return os << (v/G) << 'G'; } 
    if (v % M == 0) { return os << (v/M) << 'M'; } 
    if (v % K == 0) { return os << (v/K) << 'K'; } 
    return os << v; 
} 

std::istream& operator>>(std::istream& is, bytesize_option& hs) 
{ 
    std::string s; 
    is >> s; 
    static const std::regex re(R"regex((\d+)([GMKgmk]){0,1})regex"); 
    std::smatch match; 
    auto matched = std::regex_match(s, match, re); 
    if(!matched) { 
     throw po::validation_error(po::validation_error::invalid_option_value); 
    } 
    if (match[2].matched) 
    { 
     switch (match[2].str().at(0)) 
     { 
      case 'G': 
      case 'g': 
       hs.set(std::stoul(match[1].str()) * G); 
       break; 
      case 'M': 
      case 'm': 
       hs.set(std::stoul(match[1].str()) * M); 
       break; 
      case 'K': 
      case 'k': 
       hs.set(std::stoul(match[1].str()) * K); 
       break; 
     } 
    } 
    else { 
     hs.set(std::stoul(match[1].str())); 
    } 
    return is; 
} 

你会使用它，像这样：

return boost::shared_ptr<po::option_description> { 
     new po::option_description("server.max-header-size,x", 
            po::value(&_max_hdr_size) 
            ->default_value(_max_hdr_size), 
            "The maximum size (in bytes) of a HTTP header " 
            "that the server will accept") 
    }; 

凡在此情况下，_max_hdr_size定义：

bytesize_option _max_hdr_size;