class RDoc::Parser::C

RDoc::Parser::C 尝试解析 C 扩展文件。它查找扩展中常见的标准模式：rb_define_class、rb_define_method 等。它尝试查找方法对应的 C 源代码并提取注释，但如果失败，我们也不会太担心。

与 Ruby 方法关联的注释是从与实现该方法的例程关联的 C 注释块中提取的，也就是说，其名称在 rb_define_method 调用中给出的方法。例如，你可以这样写

/*
 * Returns a new array that is a one-dimensional flattening of this
 * array (recursively). That is, for every element that is an array,
 * extract its elements into the new array.
 *
 *    s = [ 1, 2, 3 ]           #=> [1, 2, 3]
 *    t = [ 4, 5, 6, [7, 8] ]   #=> [4, 5, 6, [7, 8]]
 *    a = [ s, t, 9, 10 ]       #=> [[1, 2, 3], [4, 5, 6, [7, 8]], 9, 10]
 *    a.flatten                 #=> [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
 */
 static VALUE
 rb_ary_flatten(VALUE ary)
 {
     ary = rb_obj_dup(ary);
     rb_ary_flatten_bang(ary);
     return ary;
 }

 ...

 void
 Init_Array(void)
 {
   ...
   rb_define_method(rb_cArray, "flatten", rb_ary_flatten, 0);

在这里，RDoc 将从 rb_define_method 行确定 Array 类中有一个名为“flatten”的方法，并将在 rb_ary_flatten 方法中查找其实现。然后，它将在 HTML 输出中使用该方法的注释。此方法必须与 rb_define_method 在同一个源文件中。

注释块可能包含特殊的指令

Document-class: name

用于命名类的文档。

Document-module: name

用于命名模块的文档。

Document-const: name

用于命名 rb_define_const 的文档。

常量值可以在注释的第一行提供，如下所示

/* 300: The highest possible score in bowling */
rb_define_const(cFoo, "PERFECT", INT2FIX(300));

只要使用 \ 转义，该值就可以包含内部冒号

Document-global: name

用于命名 rb_define_global_const 的文档

Document-variable: name

用于命名 rb_define_variable 的文档

Document-method: method_name

用于命名方法的文档。当方法名称不明确时使用此指令。

Document-method: ClassName::method_name

用于给定类中的单例方法的文档。当方法名称本身不明确时使用此指令。

Document-method: ClassName#method_name

用于给定类中的实例方法的文档。当方法名称本身不明确时使用此指令。

Document-attr: name

用于命名属性的文档。

call-seq: 文本到空行

由于 C 源代码没有为 Ruby 级别的参数提供描述性名称，因此你需要明确地记录调用序列

此外，默认情况下，RDoc 假定实现 Ruby 函数的 C 方法与 rb_define_method 调用位于同一个源文件中。如果不是这种情况，请添加注释

rb_define_method(....);  // in filename

例如，我们可能有一个扩展，在其 Init_xxx 方法中定义了多个类。我们可以使用以下方法记录它们

/*
 * Document-class:  MyClass
 *
 * Encapsulate the writing and reading of the configuration
 * file. ...
 */

/*
 * Document-method: read_value
 *
 * call-seq:
 *   cfg.read_value(key)            -> value
 *   cfg.read_value(key} { |key| }  -> value
 *
 * Return the value corresponding to +key+ from the configuration.
 * In the second form, if the key isn't found, invoke the
 * block and return its value.
 */

属性

classes[R]

将 C 变量名映射到 Ruby 类或模块的名称

content[RW]

解析器正在解析的 C 文件

enclosure_dependencies[R]

从缺失的封闭类到 missing_dependencies 中依赖它的类的依赖关系。

known_classes[R]

将 C 变量名映射到 Ruby 类（和单例类）的名称

missing_dependencies[R]

在解析 C 文件时找到的类，由于缺少封闭类而尚未注册。这些类由 do_missing 处理

singleton_classes[R]

将 C 变量名映射到 Ruby 单例类的名称

top_level[R]

已解析文件中 TopLevel 项目所属的位置

公共类方法

new(top_level, file_name, content, options, stats) 点击切换源代码

准备解析 C 文件。有关参数的详细信息，请参阅 RDoc::Parser#initialize。

调用超类方法 RDoc::Parser::new

# File rdoc/parser/c.rb, line 171
def initialize top_level, file_name, content, options, stats
  super

  @known_classes = RDoc::KNOWN_CLASSES.dup
  @content = handle_tab_width handle_ifdefs_in @content
  @file_dir = File.dirname @file_name

  @classes           = load_variable_map :c_class_variables
  @singleton_classes = load_variable_map :c_singleton_class_variables

  @markup = @options.markup

  # class_variable => { function => [method, ...] }
  @methods = Hash.new { |h, f| h[f] = Hash.new { |i, m| i[m] = [] } }

  # missing variable => [handle_class_module arguments]
  @missing_dependencies = {}

  # missing enclosure variable => [dependent handle_class_module arguments]
  @enclosure_dependencies = Hash.new { |h, k| h[k] = [] }
  @enclosure_dependencies.instance_variable_set :@missing_dependencies,
                                                @missing_dependencies

  @enclosure_dependencies.extend TSort

  def @enclosure_dependencies.tsort_each_node &block
    each_key(&block)
  rescue TSort::Cyclic => e
    cycle_vars = e.message.scan(/"(.*?)"/).flatten

    cycle = cycle_vars.sort.map do |var_name|
      delete var_name

      var_name, type, mod_name, = @missing_dependencies[var_name]

      "#{type} #{mod_name} (#{var_name})"
    end.join ', '

    warn "Unable to create #{cycle} due to a cyclic class or module creation"

    retry
  end

  def @enclosure_dependencies.tsort_each_child node, &block
    fetch(node, []).each(&block)
  end
end

公共实例方法

add_alias(var_name, class_obj, old_name, new_name, comment) 点击切换源代码

添加别名，无论是直接的别名定义，还是引用同一函数的两个方法。

# File rdoc/parser/c.rb, line 250
def add_alias(var_name, class_obj, old_name, new_name, comment)
  al = RDoc::Alias.new '', old_name, new_name, ''
  al.singleton = @singleton_classes.key? var_name
  al.comment = comment
  al.record_location @top_level
  class_obj.add_alias al
  @stats.add_alias al
  al
end

do_aliases() 点击切换源代码

扫描 content 中是否有 rb_define_alias

# File rdoc/parser/c.rb, line 222
def do_aliases
  @content.scan(/rb_define_alias\s*\(
                 \s*(\w+),
                 \s*"(.+?)",
                 \s*"(.+?)"
                 \s*\)/xm) do |var_name, new_name, old_name|
    class_name = @known_classes[var_name]

    unless class_name then
      @options.warn "Enclosing class or module %p for alias %s %s is not known" % [
        var_name, new_name, old_name]
      next
    end

    class_obj = find_class var_name, class_name
    comment = find_alias_comment var_name, new_name, old_name
    comment.normalize
    if comment.to_s.empty? and existing_method = class_obj.method_list.find { |m| m.name == old_name}
      comment = existing_method.comment
    end
    add_alias(var_name, class_obj, old_name, new_name, comment)
  end
end

do_attrs() 点击切换源代码

扫描 content 中是否有 rb_attr 和 rb_define_attr

# File rdoc/parser/c.rb, line 263
def do_attrs
  @content.scan(/rb_attr\s*\(
                 \s*(\w+),
                 \s*([\w"()]+),
                 #{BOOL_ARG_PATTERN},
                 #{BOOL_ARG_PATTERN},
                 \s*\w+\);/xmo) do |var_name, attr_name, read, write|
    handle_attr var_name, attr_name, read, write
  end

  @content.scan(%r%rb_define_attr\(
                           \s*([\w\.]+),
                           \s*"([^"]+)",
                           #{BOOL_ARG_PATTERN},
                           #{BOOL_ARG_PATTERN}\);
              %xmo) do |var_name, attr_name, read, write|
    handle_attr var_name, attr_name, read, write
  end
end

do_boot_defclass() 点击切换源代码

扫描 content 中是否有 boot_defclass

# File rdoc/parser/c.rb, line 286
def do_boot_defclass
  @content.scan(/(\w+)\s*=\s*boot_defclass\s*\(\s*"(\w+?)",\s*(\w+?)\s*\)/) do
    |var_name, class_name, parent|
    parent = nil if parent == "0"
    handle_class_module(var_name, :class, class_name, parent, nil)
  end
end

do_classes_and_modules() 点击切换源代码

扫描 content 中是否有 rb_define_class、boot_defclass、rb_define_class_under 和 rb_singleton_class

# File rdoc/parser/c.rb, line 298
def do_classes_and_modules
  do_boot_defclass if @file_name == "class.c"

  @content.scan(
    %r(
      (?<open>\s*\(\s*) {0}
      (?<close>\s*\)\s*) {0}
      (?<name>\s*"(?<class_name>\w+)") {0}
      (?<parent>\s*(?:
        (?<parent_name>[\w\*\s\(\)\.\->]+) |
        rb_path2class\s*\(\s*"(?<path>[\w:]+)"\s*\)
      )) {0}
      (?<under>\w+) {0}

      (?<var_name>[\w\.]+)\s* =
      \s*rb_(?:
        define_(?:
          class(?: # rb_define_class(name, parent_name)
            \(\s*
              \g<name>,
              \g<parent>
            \s*\)
          |
            _under\g<open> # rb_define_class_under(under, name, parent_name...)
              \g<under>,
              \g<name>,
              \g<parent>
            \g<close>
          )
        |
          (?<module>)
          module(?: # rb_define_module(name)
            \g<open>
              \g<name>
            \g<close>
          |
            _under\g<open> # rb_define_module_under(under, name)
              \g<under>,
              \g<name>
            \g<close>
          )
        )
    |
      (?<attributes>(?:\s*"\w+",)*\s*NULL\s*) {0}
      struct_define(?:
        \g<open> # rb_struct_define(name, ...)
          \g<name>,
      |
        _under\g<open> # rb_struct_define_under(under, name, ...)
          \g<under>,
          \g<name>,
      |
        _without_accessor(?:
          \g<open> # rb_struct_define_without_accessor(name, parent_name, ...)
        |
          _under\g<open> # rb_struct_define_without_accessor_under(under, name, parent_name, ...)
            \g<under>,
        )
          \g<name>,
          \g<parent>,
          \s*\w+,        # Allocation function
      )
        \g<attributes>
      \g<close>
    |
      singleton_class\g<open> # rb_singleton_class(target_class_name)
        (?<target_class_name>\w+)
      \g<close>
      )
    )mx
  ) do
    if target_class_name = $~[:target_class_name]
      # rb_singleton_class(target_class_name)
      handle_singleton $~[:var_name], target_class_name
      next
    end

    var_name = $~[:var_name]
    type = $~[:module] ? :module : :class
    class_name = $~[:class_name]
    parent_name = $~[:parent_name] || $~[:path]
    under = $~[:under]
    attributes = $~[:attributes]

    handle_class_module(var_name, type, class_name, parent_name, under)
    if attributes and !parent_name # rb_struct_define *not* without_accessor
      true_flag = 'Qtrue'
      attributes.scan(/"\K\w+(?=")/) do |attr_name|
        handle_attr var_name, attr_name, true_flag, true_flag
      end
    end
  end
end

do_constants() 点击切换源代码

扫描 content 中是否有 rb_define_variable、rb_define_readonly_variable、rb_define_const 和 rb_define_global_const

# File rdoc/parser/c.rb, line 396
def do_constants
  @content.scan(%r%\Wrb_define_
                 ( variable          |
                   readonly_variable |
                   const             |
                   global_const        )
             \s*\(
               (?:\s*(\w+),)?
               \s*"(\w+)",
               \s*(.*?)\s*\)\s*;
               %xm) do |type, var_name, const_name, definition|
    var_name = "rb_cObject" if !var_name or var_name == "rb_mKernel"
    type = "const" if type == "global_const"
    handle_constants type, var_name, const_name, definition
  end

  @content.scan(%r%
                \Wrb_curses_define_const
                \s*\(
                  \s*
                  (\w+)
                  \s*
                \)
                \s*;%xm) do |consts|
    const = consts.first

    handle_constants 'const', 'mCurses', const, "UINT2NUM(#{const})"
  end

  @content.scan(%r%
                \Wrb_file_const
                \s*\(
                  \s*
                  "([^"]+)",
                  \s*
                  (.*?)
                  \s*
                \)
                \s*;%xm) do |name, value|
    handle_constants 'const', 'rb_mFConst', name, value
  end
end

do_includes() 点击切换源代码

扫描 content 中是否有 rb_include_module

# File rdoc/parser/c.rb, line 443
def do_includes
  @content.scan(/rb_include_module\s*\(\s*(\w+?),\s*(\w+?)\s*\)/) do |c, m|
    next unless cls = @classes[c]
    m = @known_classes[m] || m

    comment = new_comment '', @top_level, :c
    incl = cls.add_include RDoc::Include.new(m, comment)
    incl.record_location @top_level
  end
end

do_methods() 点击切换源代码

扫描 content 中是否有 rb_define_method、rb_define_singleton_method、rb_define_module_function、rb_define_private_method、rb_define_global_function 和 define_filetest_function

# File rdoc/parser/c.rb, line 459
def do_methods
  @content.scan(%r%rb_define_
                 (
                    singleton_method |
                    method           |
                    module_function  |
                    private_method
                 )
                 \s*\(\s*([\w\.]+),
                   \s*"([^"]+)",
                   \s*(?:RUBY_METHOD_FUNC\(|VALUEFUNC\(|\(METHOD\))?(\w+)\)?,
                   \s*(-?\w+)\s*\)
                 (?:;\s*/[*/]\s+in\s+(\w+?\.(?:cpp|c|y)))?
               %xm) do |type, var_name, meth_name, function, param_count, source_file|

    # Ignore top-object and weird struct.c dynamic stuff
    next if var_name == "ruby_top_self"
    next if var_name == "nstr"

    var_name = "rb_cObject" if var_name == "rb_mKernel"
    handle_method(type, var_name, meth_name, function, param_count,
                  source_file)
  end

  @content.scan(%r%rb_define_global_function\s*\(
                           \s*"([^"]+)",
                           \s*(?:RUBY_METHOD_FUNC\(|VALUEFUNC\()?(\w+)\)?,
                           \s*(-?\w+)\s*\)
              (?:;\s*/[*/]\s+in\s+(\w+?\.[cy]))?
              %xm) do |meth_name, function, param_count, source_file|
    handle_method("method", "rb_mKernel", meth_name, function, param_count,
                  source_file)
  end

  @content.scan(/define_filetest_function\s*\(
                   \s*"([^"]+)",
                   \s*(?:RUBY_METHOD_FUNC\(|VALUEFUNC\()?(\w+)\)?,
                   \s*(-?\w+)\s*\)/xm) do |meth_name, function, param_count|

    handle_method("method", "rb_mFileTest", meth_name, function, param_count)
    handle_method("singleton_method", "rb_cFile", meth_name, function,
                  param_count)
  end
end

do_missing() 点击切换源代码

创建由于文件顺序与声明顺序不同而定义的缺失的类和模块。

# File rdoc/parser/c.rb, line 508
def do_missing
  return if @missing_dependencies.empty?

  @enclosure_dependencies.tsort.each do |in_module|
    arguments = @missing_dependencies.delete in_module

    next unless arguments # dependency on existing class

    handle_class_module(*arguments)
  end
end

find_alias_comment(class_name, new_name, old_name) 点击切换源代码

查找从 new_name 到 old_name 在 class_name 上的别名的注释

# File rdoc/parser/c.rb, line 524
def find_alias_comment class_name, new_name, old_name
  content =~ %r%((?>/\*.*?\*/\s+))
                rb_define_alias\(\s*#{Regexp.escape class_name}\s*,
                                 \s*"#{Regexp.escape new_name}"\s*,
                                 \s*"#{Regexp.escape old_name}"\s*\);%xm

  new_comment($1 || '', @top_level, :c)
end

find_attr_comment(var_name, attr_name, read = nil, write = nil) 点击切换源代码

查找 rb_define_attr、rb_attr 或 Document-attr 的注释。

var_name 是在其上定义属性的 C 类变量。attr_name 是属性的名称。

read 和 write 是读/写标志（“1”或“0”）。必须同时提供或都不提供。

# File rdoc/parser/c.rb, line 542
def find_attr_comment var_name, attr_name, read = nil, write = nil
  attr_name = Regexp.escape attr_name

  rw = if read and write then
         /\s*#{read}\s*,\s*#{write}\s*/xm
       else
         /.*?/m
       end

  comment = if @content =~ %r%((?>/\*.*?\*/\s+))
                              rb_define_attr\((?:\s*#{var_name},)?\s*
                                              "#{attr_name}"\s*,
                                              #{rw}\)\s*;%xm then
              $1
            elsif @content =~ %r%((?>/\*.*?\*/\s+))
                                 rb_attr\(\s*#{var_name}\s*,
                                          \s*#{attr_name}\s*,
                                          #{rw},.*?\)\s*;%xm then
              $1
            elsif @content =~ %r%(/\*.*?(?:\s*\*\s*)?)
                                 Document-attr:\s#{attr_name}\s*?\n
                                 ((?>(.|\n)*?\*/))%x then
              "#{$1}\n#{$2}"
            else
              ''
            end

  new_comment comment, @top_level, :c
end

find_body(class_name, meth_name, meth_obj, file_content, quiet = false) 点击切换源代码

查找与 Ruby 方法对应的 C 代码

# File rdoc/parser/c.rb, line 599
def find_body class_name, meth_name, meth_obj, file_content, quiet = false
  if file_content
    @body_table ||= {}
    @body_table[file_content] ||= gen_body_table file_content
    type, *args = @body_table[file_content][meth_name]
  end

  case type
  when :func_def
    comment = new_comment args[0], @top_level, :c
    body = args[1]
    offset, = args[2]

    comment.remove_private if comment

    # try to find the whole body
    body = $& if /#{Regexp.escape body}[^(]*?\{.*?^\}/m =~ file_content

    # The comment block may have been overridden with a 'Document-method'
    # block. This happens in the interpreter when multiple methods are
    # vectored through to the same C method but those methods are logically
    # distinct (for example Kernel.hash and Kernel.object_id share the same
    # implementation

    override_comment = find_override_comment class_name, meth_obj
    comment = override_comment if override_comment

    comment.normalize
    find_modifiers comment, meth_obj if comment

    #meth_obj.params = params
    meth_obj.start_collecting_tokens
    tk = { :line_no => 1, :char_no => 1, :text => body }
    meth_obj.add_token tk
    meth_obj.comment = comment
    meth_obj.line    = file_content[0, offset].count("\n") + 1

    body
  when :macro_def
    comment = new_comment args[0], @top_level, :c
    body = args[1]
    offset, = args[2]

    find_body class_name, args[3], meth_obj, file_content, true

    comment.normalize
    find_modifiers comment, meth_obj

    meth_obj.start_collecting_tokens
    tk = { :line_no => 1, :char_no => 1, :text => body }
    meth_obj.add_token tk
    meth_obj.comment = comment
    meth_obj.line    = file_content[0, offset].count("\n") + 1

    body
  when :macro_alias
    # with no comment we hope the aliased definition has it and use it's
    # definition

    body = find_body(class_name, args[0], meth_obj, file_content, true)

    return body if body

    @options.warn "No definition for #{meth_name}"
    false
  else # No body, but might still have an override comment
    comment = find_override_comment class_name, meth_obj

    if comment then
      comment.normalize
      find_modifiers comment, meth_obj
      meth_obj.comment = comment

      ''
    else
      @options.warn "No definition for #{meth_name}"
      false
    end
  end
end

find_class(raw_name, name, base_name = nil) 点击切换源代码

查找 raw_name 的 RDoc::NormalClass 或 RDoc::NormalModule

# File rdoc/parser/c.rb, line 683
def find_class(raw_name, name, base_name = nil)
  unless @classes[raw_name]
    if raw_name =~ /^rb_m/
      container = @top_level.add_module RDoc::NormalModule, name
    else
      container = @top_level.add_class RDoc::NormalClass, name
    end
    container.name = base_name if base_name

    container.record_location @top_level
    @classes[raw_name] = container
  end
  @classes[raw_name]
end

find_class_comment(class_name, class_mod) 点击切换源代码

查找 Init_class_name(void) 上方，Document-class class_name（或模块）注释中或 rb_define_class（或模块）上方的类或模块文档。如果匹配的 Init_ 上方和 rb_define_class 上方都提供了注释，则使用 Init_ 注释。

/*
 * This is a comment for Foo
 */
Init_Foo(void) {
    VALUE cFoo = rb_define_class("Foo", rb_cObject);
}

/*
 * Document-class: Foo
 * This is a comment for Foo
 */
Init_foo(void) {
    VALUE cFoo = rb_define_class("Foo", rb_cObject);
}

/*
 * This is a comment for Foo
 */
VALUE cFoo = rb_define_class("Foo", rb_cObject);

# File rdoc/parser/c.rb, line 724
def find_class_comment class_name, class_mod
  comment = nil

  if @content =~ %r%
      ((?>/\*.*?\*/\s+))
      (static\s+)?
      void\s+
      Init(?:VM)?_(?i:#{class_name})\s*(?:_\(\s*)?\(\s*(?:void\s*)?\)%xm then
    comment = $1.sub(%r%Document-(?:class|module):\s+#{class_name}%, '')
  elsif @content =~ %r%Document-(?:class|module):\s+#{class_name}\s*?
                       (?:<\s+[:,\w]+)?\n((?>.*?\*/))%xm then
    comment = "/*\n#{$1}"
  elsif @content =~ %r%((?>/\*.*?\*/\s+))
                       ([\w\.\s]+\s* = \s+)?rb_define_(class|module)[\t (]*?"(#{class_name})"%xm then
    comment = $1
  elsif @content =~ %r%((?>/\*.*?\*/\s+))
                       ([\w\. \t]+ = \s+)?rb_define_(class|module)_under[\t\w, (]*?"(#{class_name.split('::').last})"%xm then
    comment = $1
  else
    comment = ''
  end

  comment = new_comment comment, @top_level, :c
  comment.normalize

  look_for_directives_in class_mod, comment

  class_mod.add_comment comment, @top_level
end

find_const_comment(type, const_name, class_name = nil) 点击切换源代码

查找与 type 和 const_name 匹配的注释，无论是在注释上方还是在匹配的 Document- 部分中。

# File rdoc/parser/c.rb, line 794
def find_const_comment(type, const_name, class_name = nil)
  @const_table ||= {}
  @const_table[@content] ||= gen_const_table @content
  table = @const_table[@content]

  comment =
    table[[type, const_name]] ||
    (class_name && table[class_name + "::" + const_name]) ||
    table[const_name] ||
    ''

  new_comment comment, @top_level, :c
end

find_modifiers(comment, meth_obj) 点击切换源代码

处理 comment 中的修饰符并适当地更新 meth_obj。

# File rdoc/parser/c.rb, line 811
def find_modifiers comment, meth_obj
  comment.normalize
  comment.extract_call_seq meth_obj

  look_for_directives_in meth_obj, comment
end

find_override_comment(class_name, meth_obj) 点击切换源代码

查找 class_name 上 meth_obj 的 Document-method 覆盖

# File rdoc/parser/c.rb, line 821
def find_override_comment class_name, meth_obj
  name = Regexp.escape meth_obj.name
  prefix = Regexp.escape meth_obj.name_prefix

  comment = if @content =~ %r%Document-method:
                              \s+#{class_name}#{prefix}#{name}
                              \s*?\n((?>.*?\*/))%xm then
              "/*#{$1}"
            elsif @content =~ %r%Document-method:
                                 \s#{name}\s*?\n((?>.*?\*/))%xm then
              "/*#{$1}"
            end

  return unless comment

  new_comment comment, @top_level, :c
end

gen_body_table(file_content) 点击切换源代码

生成 Ruby 方法表

# File rdoc/parser/c.rb, line 575
def gen_body_table file_content
  table = {}
  file_content.scan(%r{
    ((?>/\*.*?\*/\s*)?)
    ((?:\w+\s+){0,2} VALUE\s+(\w+)
      \s*(?:\([^\)]*\))(?:[^\);]|$))
  | ((?>/\*.*?\*/\s*))^\s*(\#\s*define\s+(\w+)\s+(\w+))
  | ^\s*\#\s*define\s+(\w+)\s+(\w+)
  }xm) do
    case
    when name = $3
      table[name] = [:func_def, $1, $2, $~.offset(2)] if !(t = table[name]) || t[0] != :func_def
    when name = $6
      table[name] = [:macro_def, $4, $5, $~.offset(5), $7] if !(t = table[name]) || t[0] == :macro_alias
    when name = $8
      table[name] ||= [:macro_alias, $9]
    end
  end
  table
end

gen_const_table(file_content) 点击切换源代码

生成常量表

# File rdoc/parser/c.rb, line 757
def gen_const_table file_content
  table = {}
  @content.scan(%r{
    (?<doc>(?>^\s*/\*.*?\*/\s+))
      rb_define_(?<type>\w+)\(\s*(?:\w+),\s*
                         "(?<name>\w+)"\s*,
                         .*?\)\s*;
  | (?<doc>(?>^\s*/\*.*?\*/\s+))
      rb_define_global_(?<type>const)\(\s*
                         "(?<name>\w+)"\s*,
                         .*?\)\s*;
  |  (?<doc>(?>^\s*/\*.*?\*/\s+))
      rb_file_(?<type>const)\(\s*
                         "(?<name>\w+)"\s*,
                         .*?\)\s*;
  |  (?<doc>(?>^\s*/\*.*?\*/\s+))
      rb_curses_define_(?<type>const)\(\s*
                         (?<name>\w+)
                         \s*\)\s*;
  | Document-(?:const|global|variable):\s
      (?<name>(?:\w+::)*\w+)
      \s*?\n(?<doc>(?>.*?\*/))
  }mxi) do
    name, doc, type = $~.values_at(:name, :doc, :type)
    if type
      table[[type, name]] = doc
    else
      table[name] = "/*\n" + doc
    end
  end
  table
end

handle_attr(var_name, attr_name, read, write) 点击切换源代码

在类 var_name 上创建新的 RDoc::Attr attr_name，该属性要么是 read，要么是 write，或者两者都是

# File rdoc/parser/c.rb, line 843
def handle_attr(var_name, attr_name, read, write)
  rw = ''
  rw += 'R' if TRUE_VALUES.include?(read)
  rw += 'W' if TRUE_VALUES.include?(write)

  class_name = @known_classes[var_name]

  return unless class_name

  class_obj = find_class var_name, class_name

  return unless class_obj

  comment = find_attr_comment var_name, attr_name
  comment.normalize

  name = attr_name.gsub(/rb_intern(?:_const)?\("([^"]+)"\)/, '\1')

  attr = RDoc::Attr.new '', name, rw, comment

  attr.record_location @top_level
  class_obj.add_attribute attr
  @stats.add_attribute attr
end

handle_class_module(var_name, type, class_name, parent, in_module) 点击切换源代码

基于 type 在 parent 中创建名为 class_name 的新的 RDoc::NormalClass 或 RDoc::NormalModule，该类或模块已分配给 C var_name。

# File rdoc/parser/c.rb, line 872
def handle_class_module(var_name, type, class_name, parent, in_module)
  parent_name = @known_classes[parent] || parent

  if in_module then
    enclosure = @classes[in_module] || @store.find_c_enclosure(in_module)

    if enclosure.nil? and enclosure = @known_classes[in_module] then
      enc_type = /^rb_m/ =~ in_module ? :module : :class
      handle_class_module in_module, enc_type, enclosure, nil, nil
      enclosure = @classes[in_module]
    end

    unless enclosure then
      @enclosure_dependencies[in_module] << var_name
      @missing_dependencies[var_name] =
        [var_name, type, class_name, parent, in_module]

      return
    end
  else
    enclosure = @top_level
  end

  if type == :class then
    full_name = if RDoc::ClassModule === enclosure then
                  enclosure.full_name + "::#{class_name}"
                else
                  class_name
                end

    if @content =~ %r%Document-class:\s+#{full_name}\s*<\s+([:,\w]+)% then
      parent_name = $1
    end

    cm = enclosure.add_class RDoc::NormalClass, class_name, parent_name
  else
    cm = enclosure.add_module RDoc::NormalModule, class_name
  end

  cm.record_location enclosure.top_level

  find_class_comment cm.full_name, cm

  case cm
  when RDoc::NormalClass
    @stats.add_class cm
  when RDoc::NormalModule
    @stats.add_module cm
  end

  @classes[var_name] = cm
  @known_classes[var_name] = cm.full_name
  @store.add_c_enclosure var_name, cm
end

handle_constants(type, var_name, const_name, definition) 点击切换源代码

添加常量。通过在注释的开头提供 some_value:，你可以覆盖注释的 C 值，以给出友好的定义。

/* 300: The perfect score in bowling */
rb_define_const(cFoo, "PERFECT", INT2FIX(300));

在 RDoc 输出中，将使用值 300 覆盖 INT2FIX(300)。值可能包括引号和转义的冒号 (:)。

# File rdoc/parser/c.rb, line 937
def handle_constants(type, var_name, const_name, definition)
  class_name = @known_classes[var_name]

  return unless class_name

  class_obj = find_class var_name, class_name, class_name[/::\K[^:]+\z/]

  unless class_obj then
    @options.warn 'Enclosing class or module %p is not known' % [const_name]
    return
  end

  comment = find_const_comment type, const_name, class_name
  comment.normalize

  # In the case of rb_define_const, the definition and comment are in
  # "/* definition: comment */" form.  The literal ':' and '\' characters
  # can be escaped with a backslash.
  if type.downcase == 'const' then
    if /\A(.+?)?:(?!\S)/ =~ comment.text
      new_definition, new_comment = $1, $'

      if !new_definition # Default to literal C definition
        new_definition = definition
      else
        new_definition = new_definition.gsub(/\\([\\:])/, '\1')
      end

      new_definition.sub!(/\A(\s+)/, '')

      new_comment = "#{$1}#{new_comment.lstrip}"

      new_comment = self.new_comment(new_comment, @top_level, :c)

      con = RDoc::Constant.new const_name, new_definition, new_comment
    else
      con = RDoc::Constant.new const_name, definition, comment
    end
  else
    con = RDoc::Constant.new const_name, definition, comment
  end

  con.record_location @top_level
  @stats.add_constant con
  class_obj.add_constant con
end

handle_ifdefs_in(body) 点击切换源代码

删除否则会使我们感到困惑的 ifdef

# File rdoc/parser/c.rb, line 987
def handle_ifdefs_in(body)
  body.gsub(/^#ifdef HAVE_PROTOTYPES.*?#else.*?\n(.*?)#endif.*?\n/m, '\1')
end

handle_method(type, var_name, meth_name, function, param_count, source_file = nil) 点击切换源代码

添加在分配给 var_name 的类或模块上定义的 RDoc::AnyMethod meth_name。type 是使用的方法定义函数的类型。singleton_method 和 module_function 创建一个单例方法。

# File rdoc/parser/c.rb, line 996
def handle_method(type, var_name, meth_name, function, param_count,
                  source_file = nil)
  class_name = @known_classes[var_name]
  singleton  = @singleton_classes.key? var_name

  @methods[var_name][function] << meth_name

  return unless class_name

  class_obj = find_class var_name, class_name

  if existing_method = class_obj.method_list.find { |m| m.c_function == function }
    add_alias(var_name, class_obj, existing_method.name, meth_name, existing_method.comment)
  end

  if class_obj then
    if meth_name == 'initialize' then
      meth_name = 'new'
      singleton = true
      type = 'method' # force public
    end

    meth_obj = RDoc::AnyMethod.new '', meth_name
    meth_obj.c_function = function
    meth_obj.singleton =
      singleton || %w[singleton_method module_function].include?(type)

    p_count = Integer(param_count) rescue -1

    if source_file then
      file_name = File.join @file_dir, source_file

      if File.exist? file_name then
        file_content = File.read file_name
      else
        @options.warn "unknown source #{source_file} for #{meth_name} in #{@file_name}"
      end
    else
      file_content = @content
    end

    body = find_body class_name, function, meth_obj, file_content

    if body and meth_obj.document_self then
      meth_obj.params = if p_count < -1 then # -2 is Array
                          '(*args)'
                        elsif p_count == -1 then # argc, argv
                          rb_scan_args body
                        else
                          args = (1..p_count).map { |i| "p#{i}" }
                          "(#{args.join ', '})"
                        end


      meth_obj.record_location @top_level

      if meth_obj.section_title
        class_obj.temporary_section = class_obj.add_section(meth_obj.section_title)
      end
      class_obj.add_method meth_obj

      @stats.add_method meth_obj
      meth_obj.visibility = :private if 'private_method' == type
    end
  end
end

handle_singleton(sclass_var, class_var) 点击切换源代码

将单例类 sclass_var 注册为 class_var 的单例

# File rdoc/parser/c.rb, line 1066
def handle_singleton sclass_var, class_var
  class_name = @known_classes[class_var]

  @known_classes[sclass_var]     = class_name
  @singleton_classes[sclass_var] = class_name
end

load_variable_map(map_name) 点击切换源代码

如果存在，则从 RDoc::Store 中加载给定 name 的变量映射。

# File rdoc/parser/c.rb, line 1077
def load_variable_map map_name
  return {} unless files = @store.cache[map_name]
  return {} unless name_map = files[@file_name]

  class_map = {}

  name_map.each do |variable, name|
    next unless mod = @store.find_class_or_module(name)

    class_map[variable] = if map_name == :c_class_variables then
                            mod
                          else
                            name
                          end
    @known_classes[variable] = name
  end

  class_map
end

look_for_directives_in(context, comment) 点击切换源代码

在普通注释块中查找指令

/*
 * :title: My Awesome Project
 */

此方法修改 comment。:main: 和 :title: 指令均已弃用，并将在 RDoc 7 中移除。

# File rdoc/parser/c.rb, line 1107
  def look_for_directives_in context, comment
    @preprocess.handle comment, context do |directive, param|
      case directive
      when 'main' then
        @options.main_page = param

        warn <<~MSG
          The :main: directive is deprecated and will be removed in RDoc 7.

          You can use these options to specify the initial page displayed instead:
          - `--main=#{param}` via the command line
          - `rdoc.main = "#{param}"` if you use `RDoc::Task`
          - `main_page: #{param}` in your `.rdoc_options` file
        MSG
        ''
      when 'title' then
        @options.default_title = param if @options.respond_to? :default_title=

        warn <<~MSG
          The :title: directive is deprecated and will be removed in RDoc 7.

          You can use these options to specify the title displayed instead:
          - `--title=#{param}` via the command line
          - `rdoc.title = "#{param}"` if you use `RDoc::Task`
          - `title: #{param}` in your `.rdoc_options` file
        MSG
        ''
      end
    end

    comment
  end

new_comment(text = nil, location = nil, language = nil) 点击切换源代码

创建一个 RDoc::Comment 实例。

# File rdoc/parser/c.rb, line 1255
def new_comment text = nil, location = nil, language = nil
  RDoc::Comment.new(text, location, language).tap do |comment|
    comment.format = @markup
  end
end

rb_scan_args(method_body) 点击切换源代码

从 method_body 中提取参数，并返回方法参数字符串。遵循 1.9.3dev 的 scan-arg-spec，请参阅 README.EXT

# File rdoc/parser/c.rb, line 1144
def rb_scan_args method_body
  method_body =~ /rb_scan_args\((.*?)\)/m
  return '(*args)' unless $1

  $1.split(/,/)[2] =~ /"(.*?)"/ # format argument
  format = $1.split(//)

  lead = opt = trail = 0

  if format.first =~ /\d/ then
    lead = $&.to_i
    format.shift
    if format.first =~ /\d/ then
      opt = $&.to_i
      format.shift
      if format.first =~ /\d/ then
        trail = $&.to_i
        format.shift
        block_arg = true
      end
    end
  end

  if format.first == '*' and not block_arg then
    var = true
    format.shift
    if format.first =~ /\d/ then
      trail = $&.to_i
      format.shift
    end
  end

  if format.first == ':' then
    hash = true
    format.shift
  end

  if format.first == '&' then
    block = true
    format.shift
  end

  # if the format string is not empty there's a bug in the C code, ignore it

  args = []
  position = 1

  (1...(position + lead)).each do |index|
    args << "p#{index}"
  end

  position += lead

  (position...(position + opt)).each do |index|
    args << "p#{index} = v#{index}"
  end

  position += opt

  if var then
    args << '*args'
    position += 1
  end

  (position...(position + trail)).each do |index|
    args << "p#{index}"
  end

  position += trail

  if hash then
    args << "p#{position} = {}"
  end

  args << '&block' if block

  "(#{args.join ', '})"
end

remove_commented_out_lines() 点击切换源代码

移除被注释掉的行，这些行可能会在扫描类和方法时被错误地拾取。

# File rdoc/parser/c.rb, line 1227
def remove_commented_out_lines
  @content = @content.gsub(%r%//.*rb_define_%, '//')
end

scan() 点击切换源代码

从 C 文件中提取类、模块、方法、属性、常量和别名，并为此文件返回一个 RDoc::TopLevel。

# File rdoc/parser/c.rb, line 1235
def scan
  remove_commented_out_lines

  do_classes_and_modules
  do_missing

  do_constants
  do_methods
  do_includes
  do_aliases
  do_attrs

  @store.add_c_variables self

  @top_level
end