Skip to content

Latest commit

 

History

History
236 lines (171 loc) · 7.14 KB

PythonVsRepyV2.md

File metadata and controls

236 lines (171 loc) · 7.14 KB
Raw

All of the Python You Need to Forget to Use RepyV2

DISAMBIGUATION NOTE: This is the RepyV2 version of the page. There is a separate page for Repy version 1 in PythonVsRepy.md.

This tutorial is written for those who are familiar with the Python programming language, and want to start using RepyV2. The syntax and many of the properties of RepyV2 should be very familiar for Python users, but there are some noticeable differences.

Many Python built-ins are not allowed in RepyV2. This is to ensure maximum cross platform compatibility and security for end users. Python modules are not allowed to be imported in RepyV2 (see below), however porting Python code to RepyV2 takes minimal effort.

Global Variables

In RepyV2 global variables are not allowed. In place of this there is an implicit dictionary called mycontext which can be used to create globally visible variables.

mycontext['pagecount'] = mycontext['pagecount'] + 1

Import statements

In RepyV2 there are no import statements. Instead make use of the dynamic linking facility of seattlelib_v2/dylink.r2py.

If you have the file a.r2py:

def foo():
  log("foo")

You can use dylink to link the module code in:

a = dy_import_module("a.r2py")

a.foo()  # This calls into the foo function of a.r2py

This looks and works similar to importing a module in Python, and then accessing its definitions.

In order for your program to be able to use dy_import_module, provide dylink.r2py as the first argument to repy.py (when you start your sandbox on the command line):

python repy.y restrictionsfile dylink.r2py your_program_goes_here.r2py

or the run/start command (when using seash):

start dylink.r2py your_program_goes_here.r2py

Needless to say, you will need to upload dylink.r2py and all of the dependencies of your program before you can use this seash command.

Input

There are no RepyV2 mechanisms to read user input. Providing the python input and raw_input methods is impractical, since the intent of RepyV2 is to run as a background process on user machines. There is no terminal to accept input from.

However, your RepyV2 program can listen on network sockets, and you may use netcat or similar tools to send data to your RepyV2 program.

Python Built-Ins Not in RepyV2

Below is a list of the Python built-ins that are not allowed in RepyV2. See repy_v2/safe.py for details.

all, any, bin, callable, compile, complex, delattr, dir, enumerate,
eval, execfile, globals, hash, help, id, input, iter, lambda, locals,
next, property, raw_input, reload, reversed, sorted, staticmethod,
super, unichr, unicode, vars, yield, __import__

Common Python Constructs and Their Repy Counterparts

sys.argv

Programs written in Python use sys.argv``` to access arguments to the file. In RepyV2 the variable callargsbehaves the same assys.argv[1:]`

__name__

There is no __name__ variable. In Python it is common to use the following code to see if a file was invoked directly (on the other hand, importing a module invokes it indirectly):

if __name__ == "__main__":
 main()

In RepyV2 the corresponding implicit variable is callfunc. A similar RepyV2 program would include:

if callfunc == "initialize":
  main()

Python Modules

Python modules are not directly allowed in RepyV2. This is to ensure safety for Seattle users, so all code run in a VM can be strictly controlled. Many Python modules have RepyV2 equivalents. See the SeattleTestbed/seattlelib_v2 repository for a complete list.

Files

open()

RepyV2 has an openfile(filename, create) call that behaves similarly to the Python native open.

The file to be opened may only contain the characters a-z0-9.-_, and cannot start with a ..

The object returned by open() is similar to the Python file object,s and provides similar functionality: readat, writeat, close.

os.listdir()

RepyV2 provides the function listfiles() which is equivalent to a call to Python's os.listdir('.').

Since file access is only allowed within a VM, listfiles does not take a directory argument. The list returned will contain all the files in the current working directory.

os.remove()

Remove a file from the user program area. The call has the same filename character restrictions as openfile .

The RepyV2 call removefile(f) is equivalent to Python's os.remove(f).

Time

time.sleep()

The sleep(seconds) function in Repy behaves similarly to [time.sleep(seconds)] in Python, causing execution of the current program to halt for the specified number of seconds.

threading.Timer

In Python, to start a timer you typically have code such as:

import threading

def hello(str):
  print "hello,", str

t = threading.Timer(30.0, hello, ["world"])
t.start()
# after 30 seconds, "hello, world" will be printed

In RepyV2, combine the createthread and sleep functions. A Repy equivalent would be:

def hello(str):
  log("hello", str, "\n")

def sleep_then_call(pause_in_seconds, call_function, call_args):
  # This function returns a function that takes no arguments. Yet
  # the returned function "remembers" (via a closure) the variables
  # defined in the outside scope.
  # This construction is required as we cannot pass arguments to
  # the function that we supply to createthread.
  def configured_timer_function():
    sleep(pause_in_seconds)
    call_function(*call_args)
  return configured_timer_function

configured_function_for_new_thread = sleep_then_call(30, hello, ["world"])
createthread(configured_function_for_new_thread)

time.time()

In RepyV2 you make make use of the getruntime() function. This returns a float representing the runtime of the current program, which is equivalent to the Python:

time.time() - starttime #starttime is when the program started execution

For programs that actually want the current wallclock time (rather than using it to measure elapsed time as most programmers seem to use it), the library seattlelib_v2/time.r2py provides NTP time.

Networking

There are several native Repy functions provided to make common networking operations easier. Many don't have direct Python equivalents, but are convenience functions. See the RepyTutorial for examples of these functions.

socket.gethostbyname_ex(name)

In RepyV2, rather than use socket.gethostbyname(name), use gethostbyname(name). The behaviour is the same, so you can write RepyV2 which looks like

ip_address = gethostbyname("www.google.com")

Threading

threading.Lock()

In RepyV2 use createlock() to create new instances Mutex objects. The object's behaviour is the same in RepyV2 as it is in Python: it supports the acquire and release operations.