I. Overview of the compilation process
When we execute the Python code, in the Python interpreter we use four processes to "disassemble" our code, which is eventually returned to the user by CPU execution.
First, when the user type code to the Python processing will be the first lexical analysis, such as the user typed the keyword or when the input keyword is wrong, will be triggered by lexical analysis, the incorrect code will not be executed.
Next Python will parse the syntax, for example, if the colon after test is written as a different symbol in the "for I in Test:", the code will still not be executed.
Following the most critical process, before executing Python, Python generates the. pyc file, which is the byte code, if we accidentally modified the bytecode, Python next recompile the program will be compared with its last generated bytecode file, if the mismatch will be modified bytecode file overwrite, to ensure the accuracy of the bytecode after each compilation.
So what is a byte code? The byte code in the Python virtual machine program corresponds to the Pycodeobject object. The. pyc file is the representation of the bytecode on disk. Simply put, in the process of compiling the code, the functions, classes and other objects in the code are processed first, and then the bytecode file is generated. With the bytecode file, the CPU can directly identify the bytecode file for processing, and then Python can execute it.
Second, the process diagram
Third, compile the byte code
Python has a built-in function compile (), you can compile the source file into CodeObject, first look at the description of this function:
Compile (...) compile (source, filename, mode[, flags[, Dont_inherit]), code object
Parameter 1: Content string for source file
Parameter 2: Source file name
Parameter 3:exec-compile module,single-compile a declaration, eval-compile an expression generally using the first three parameters is enough
Examples of Use:
#src_file. Py#some functiondef F (d=0): c=1 print "Hello" a=9b=8f () >>> a=open (' src_file.py ', ' R '). Read () #命令行模式中打开源文件进行编译 >>> co=compile (A, ' src_file ', ' exec ') >>> type (CO) <type ' code ' > #编译出了codeobject对象
Iv. Properties of CodeObject objects
What are the variables of codeobject, and then analyze the contents of the previous section?
>>> Print co.co_names #所有的符号名称 (' f ', ' a ', ' B ') >>> print co.co_name #模块名, function name, class name <module >>>> print co.co_consts #常量集合, function f, and two int constants A,b,d (0, <code object f at 0xb7273b18, file ' Src_file ', line 2>, 9, 8, None) >>> print co.co_consts[1].co_varnames #可以看到f函数也是一个codeobject, prints local variables in f (' C ',) >> > Print co.co_code #字节码指令dZdZdZedS >>> print Co.co_consts[1].co_firstlineno #代码块在文件中的起始行号2 > >> print co.co_stacksize #代码栈大小2 >>> print co.co_filename #文件名src_file #模块名, function name, class name
CodeObject's Co_code represents a byte code, what does this bytecode mean? We can use the DIS module for the anti-compilation of Python:
Import Disdis.dis (CO) >>> output 2 0 load_const 0 (0) 3 load_const 1 (<code object F at 0xb727 3b18, file "Src_file", line 2>) 6 make_function 1 9 store_name 0 (f) 5 Load_const 2 (9) c10/>15 store_name 1 (a) 6 load_const 3 (8) store_name 2 (b) 7 load_name 0 ( f) call_function 0 pop_top load_const 4 (None) Return_value
From the results of the anti-compilation, Python bytecode is actually a parody of the x86 assembly, the code is compiled into a single instruction to a virtual CPU to execute.
- First column: line number
- Second column: The offset of the instruction in the code block
- Column three: directives
- Fourth column: Operands
- Column Fifth: Description of operands
Python Internal Execution process