Magpie 简单分析(1)

一. target 目录是最顶层,通过/magpidl4.py -p platform(default:generic) -i interface

    (v4nicta_n2)   -g generator() TARGET_MAP 查表 
   获得['idl4/generic_l4v4'] target  目录下最终使用那个target, 比如-i v4, -p generic

    =>idl4/generic_l4v4_target.py
    TARGET_MAP 中某一项:
         generator ,platform , interface, mapping : target  
        ('idl4', 'generic', 'v4', 'c'): ['idl4/generic_l4v4'],

    通过命令行参数或者default 参数,将使targets/shared/options.py 下所有option 都设置

     完成以待后用:

def set_options(self): # set target/shared/options !!! liunote
  options['output_filename'] = self.options.dest
  options['target'] = self.get_target() !!!
  options['c_preprocessor'] = self.options.cpp
  self['c_preprocessor'] = "armcc"
  self['cpp_options'] = ['-E -D__arm__ ' +
                                               '-DARM_LITTLE_ENDIAN -D__ARMEL__']
  options['c_include_dirs'] = self.options.include_dirs
  options['hardware_arch'] = 'Generic 32'
  options['hardware_arch'] = 'MIG'
  options['cache_enabled'] = self.options.cache_enabled
  options['cache_dir'] = self.options.cache_dir
  options['output_type'] = self.options.output_type
  options['cpp_options'] = self.options.cpp_flags.split(' ')
  options['generator'] = self.options.generator ===>
  options.add_idl_files(cmdlineargs) =>idl文件

 

二. 开始targets/shared/target.py  class Target的run method 开始运行
     其中 Class Target init 时设置下面数据:

 

target_data = self.get_target_data(options['target']) # liunote options['target'] => traget/idl4/generic_l4v4_target.py
  # Create the type registry.
  # self.type_registry = new_registry(options['hardware_arch'], options['generator'])
  self.hardware_arch = options['hardware_arch']
  self.generator_type = options['generator'] ===> generator type = "idl4"
  #print target_data['Templates']
  self.templates = target_data['Templates']()
  self.output = GenericOutput()
  self.generator = target_data['Generator']
  self.dataStorage = target_data['DataStorage']
  self.inputparser = inputparser_construct() =>from inputparser import construct as

inputparser_construct
                    => inputparser.py 的 construct

target_data: 通过get_target_data('nicta/generic_n2')

       去获得default interface =>   targets/nicta/generic_n2_target.py
       或者 get_target_data('idl4/generic_l4v4') 去获得 -i v4 (interface=v4)  

       targets/nicta/generic_l4v4_target.py

       然后执行这个xxxxx_xxxx_target.py 去获得target_opts  =>

                                     execfile('generic_n2_target.py',target_opts)

 target_opts = {'Templates': None, 'Generator': None,'DataStorage':None}
  对于执行generic_l4v4_target.py , target_opt 输出结果如下:

{'Templates': , 'ServiceTemplates': ,  

'Generator': ,  'ClientTemplates': , 'DataStorage': , 

'V4Generator': ,

'SharedTemplates': ,

......

在target_opt 中预设的 'Templates'  'Generator'  'DataStorage' 都从generic_l4v4_target.py

的执行中获得了对应key的内容,

其中 Templates =>class Templates(base.Templates, SharedTemplates, ClientTemplates, ServiceTemplates, ServiceTemplateTemplates):
                  Generator =>from generator.v4generator import V4Generator
                              class Generator(V4Generator):
到此target_data已经很清楚了,就是target_opt ,其中包括了字典名和对应class Templates Generator DataStorage !!!

 

三. inputparser.py => 上面target.py中 self.inputparser = inputparser_construct  (from   inputparser import construct as inputparser_construct )
 该文件包含下面类层次:
  GenericASTGenerator, ASTGeneratorCImportMixIn(importer = CASTImporter(import_filenam))
             \               /
              \             /
             ImportingASTGenerator

         GenericASTGenerator ===> idl ast 生成
  ASTGeneratorCImportMixIn  ===> c preprocess 和c ast 生成

 

 

四. target.py  target 类实例化(__init__)的时候通过get_target_data中获得了target_data ,就是generic_l4v4_target.py (假定-i v4)
      generic_l4v4_target.py 的类层次如下:
base.Templates, SharedTemplates, ClientTemplates, ServiceTemplates, ServiceTemplateTemplates
     \     |                 |               |              /
      \    |                 |               |             /
                         Templates

  其中 base.Templates 中get 是用来获取 SharedTemplates, ClientTemplates, ServiceTemplates, ServiceTemplateTemplates 中各个模板名字对应的文件名!!!
(template_name = getattr(self, name, None) => 比如 client_function_body ==> 

                                                 'v4_generic/client_function_body.template.c')

 

 

五. 回到  target.py  => run
       run 之前 target_data 导出 targets/nicta/generic_l4v4_target.py 相关

              generator,datastorage,tmplate等等
       然后 output(file ,or stdout),output_type(client_stub),generator(from

             generator.v4generator import V4Generator)
             inputparser(from inputparser import construct as inputparser_construct)
             templates(target_data's templates)

 

       1. check output_type,必须3选1,default is client stub

       2. basicast = create_basictype_ast(self.hardware_arch, self.generator_type) =>

                                          from magpietypes.builtin import create_basictype_ast
          生成下面类型语法树:  Node + Type (parser_common.py)
MagpieAST  (0)
  type void  (1)
  +- meta_type = ['basic']
   info  (2)
   +- size = [0]
Type node 和Info node 的关系 (magpietypes/builtin.py !!! )
  newType = Type(ast, None)
  newType.leaf = name
  newType.add_attribute('meta_type', 'basic')
  info = Info(newType)
  newType.add_child(info)
  info.add_attribute('size', arch_info.size_in_bits(name))
  ast.add_child(newType)

  type short  (1)
  +- meta_type = ['basic']
   info  (2)
   +- size = [16]

  type long  (1)
  +- meta_type = ['basic']
   info  (2)
   +- size = [32]
  type long long  (1)
  +- meta_type = ['basic']
   info  (2)
   +- size = [64]
  type unsigned short  (1)
  +- meta_type = ['basic']
   info  (2)
   +- size = [16]
  type unsigned long  (1)
  +- meta_type = ['basic']
   info  (2)
   +- size = [32]
  type unsigned long long  (1)
  +- meta_type = ['basic']
   info  (2)
   +- size = [64]
  type float  (1)
  +- meta_type = ['basic']
   info  (2)
   +- size = [32]
  type double  (1)
  +- meta_type = ['basic']
   info  (2)
   +- size = [64]
  type long double  (1)
  +- meta_type = ['basic']
   info  (2)
   +- size = [80]
  type char  (1)
  +- meta_type = ['basic']
   info  (2)
   +- size = [8]
  type unsigned char  (1)
  +- meta_type = ['basic']
   info  (2)
   +- size = [8]
  type wchar_t  (1)
  +- meta_type = ['basic']
   info  (2)
   +- size = [16]
  type bool  (1)
  +- meta_type = ['basic']
   info  (2)
   +- size = [1]
  type octet  (1)
  +- meta_type = ['basic']
   info  (2)
   +- size = [8]
  type int  (1)
  +- meta_type = ['basic']
   info  (2)
   +- size = [32]
  type unsigned int  (1)
  +- meta_type = ['basic']
   info  (2)
   +- size = [32]
  type signed char  (1)
  +- meta_type = ['basic']
   info  (2)
   +- size = [8]
  type fpage  (1)
  +- meta_type = ['basic']
   info  (2)
   +- size = [64]
  type __int64  (1)
  +- meta_type = ['alias']
   info  (2)
   +- target_type = []
  type signed __int64  (1)
  +- meta_type = ['alias']
   info  (2)
   +- target_type = []
  type unsigned __int64  (1)
  +- meta_type = ['alias']
   info  (2)
   +- target_type = []
  type long int  (1)
  +- meta_type = ['alias']
   info  (2)
   +- target_type = []
  type unsigned long int  (1)
  +- meta_type = ['alias']
   info  (2)
   +- target_type = []
  type long unsigned int  (1)
  +- meta_type = ['alias']
   info  (2)
   +- target_type = []
  type long long int  (1)
  +- meta_type = ['alias']
   info  (2)
   +- target_type = []
  type unsigned long long int  (1)
  +- meta_type = ['alias']
   info  (2)
   +- target_type = []
  type unsigned  (1)
  +- meta_type = ['alias']
   info  (2)
   +- target_type = []
  type short int  (1)
  +- meta_type = ['alias']
   info  (2)
   +- target_type = []
  type boolean  (1)
  +- meta_type = ['alias']
   info  (2)
   +- target_type = []
  type signed short int  (1)
  +- meta_type = ['alias']
   info  (2)
   +- target_type = []
  type signed int  (1)
  +- meta_type = ['alias']
   info  (2)
   +- target_type = []
  type unsigned short int  (1)
  +- meta_type = ['alias']
   info  (2)
   +- target_type = []
  type Object  (1)
  +- meta_type = ['alias']
   info  (2)
   +- target_type = []
  type any  (1)
  +- meta_type = ['alias']
   info  (2)
   +- target_type = []
  type ValueBase  (1)
  +- meta_type = ['alias']
   info  (2)
   +- target_type = []
  type wchar  (1)
  +- meta_type = ['alias']
   info  (2)
   +- target_type = []
  type Word  (1)
  +- meta_type = ['alias']
   info  (2)
   +- target_type = []

 

 

       3. 继续target.py.run()

=> self.generate_asts(basicast)
   =>  self.inputparser.produce_ast(filename, 'CORBA', basicast)
     => ImportingASTGenerator.produce_ast ( self.inputparser = inputparser_construct()

         inputparser.py的 def construct():return ImportingASTGenerator())
           => GenericASTGenerator.perform_imports(baseast, data) data is idl file name
            预处理 + cparser
           => raw_ast = idlparser(idl_filename, data, baseast) 

             idlparse
           => add raw_ast to baseast's child ,and return raw_ast (see proc_ast.txt file) 

             到这里已经AST生成完毕
           =>  self.asts.append(ASTInfo(ast)) ,generator\generator.py 再封装到 ASTInfo类,

             其中一些方法用来方便处理ast
           => self.generate_output
                => generator_inst = self.generator(self.output, ast)
                   因为self.generator = target_data['Generator'] => targets\idl4

                   \generic_l4v4_target.py 的 class Generator(V4Generator):

class V4Generator(Generator, V4WordsizeMixIn): # generators\v4generator.py
  """ Code to generate V4 client stubs """
  def __init__(self, output, ast):

 

 target_data = self.get_target_data(options['target']) => self.get_target_data

     ("traget/idl4/generic_l4v4_target.py")

                 => execfile(traget/idl4/generic_l4v4_target.py,target_data)

                     get generic_l4v4_target.py object !!!!!!

获得v4generator(parent is generator) 的实例, ArchitectureInfo 等wordsize ,align 等信息
        => self.generator_go(generator_inst) 
           => 设置self.output.context 字典的一些内容(self.output = GenericOutput() 

                                                                   =>output/generic.py)

              self.output.context['templates'] = self.templates => 取templates的类实例
              self.output.context.file.version_string = helper.VERSION_STRING
              self.output.context['generator'] = generator_inst  => class generator  的类实例
              self.output.context.file.output_filename = options['output_filename'] 

                                                                                   输出文件名
              self.output.context.file.idl_filename = generator_inst.astinfo.get_filename()  

                                                                                   idl4 文件名
               base_template = self.templates.get(options['output_type'])

                     通过base.template的get 获得output_type(client or service)对应模板的

                      文件名路径
              

              self.output.go(base_template) ,开始output (output\generic.py => class 

                                         GenericOutput(BaseOutput): => GenericOutput.go )

                     接下部分到七.继续
           

 

六.  parser_common.py  的类层次:
         Node
          |
        ASTPart
          /\
       IDL elment Node (Declarator ,Interface ,Function ...)  
       比如:
       class Interface(ASTPart): ==> 在 idlparser\antlr\astgen.py 中def interf(self,ast,  pt): 使用该类生成实例!!!
         def __init__(self, parent,  children = None, name = None):
            ASTPart.__init__(self, parent,  'interface', children, name)

           
七. output\generic.py output
      self.fromtemplate(template_filename) v4_generic/main_client.template.h
         => markup(template_fullname, self.output, self.context)  ==> output\wsogmm.py (internal_run_fn 就是template 文件中的run !!!)
             markup = safe_markup_fromfile  (output/generic.py ===>  包含了 outputcontext !!!)
                   ===> return Markup.go(template, output, environment)
                   ===> inst._run_template(template) ( go = classmethod(go) )  output/wsogmm.py
                   ===> self._compile(template)