Using Python on the Macintosh

(preliminary)

This document is an introduction to using Python on the Apple Macintosh. It does not introduce the language itself, for this you should refer to the Python Tutorial by Guido van Rossum. This guide more-or-less replaces chapter two of the tutorial, and provides some additional material.

There is currently no good tutorial for the mac-specific features of Python, but to whet your appetite: it has interfaces to many MacOS toolboxes (quickdraw, sound, quicktime, open scripting, etc) and various portable toolboxes are available too (Tk, stdwin, complex numbers, image manipulation, etc). Some annotated sample programs are available to give you an idea of Python's power.

The document refers to Python 1.3.3 or higher, some of the features (like setting applet options) will not work in earlier versions of Python.

Invoking the interpreter

The name of the interpreter may differ on different installations: it may be called Python, PythonPPC (for powerpc macs) or Python68K (indeed, for 68K macs). It will always be recognizable by the "16 ton" icon, though. You start the interpreter in interactive mode by double-clicking its icon:

This should give you a text window with an informative version string and a prompt, something like the following:

Python 1.3.3 (Apr  7 1996)  [CW PPC w/GUSI]
Copyright 1991-1995 Stichting Mathematisch Centrum, Amsterdam
>>>

The version string tells you the version of Python, whether it was built for PPC or 68K macs and possibly some options used to build the interpreter. If you find a bug or have a question about how the interpreter works it is a good idea to include the version information in your message.

At the prompt you can type interactive python commands. See the tutorial for more information. The interactive window works more-or-less like a Communication Toolbox or Telnet window: you type commands at the bottom and terminate them with the [return] or [enter] key. Interpreter feedback also appears at the bottom of the window, and the contents scroll as output is added. You can use copy and paste in the normal way, but be sure to paste only at the bottom of the document.

Creating Python scripts

The Python interpreter works in a way that is different from what you would expect of a macintosh program: the interpreter is just that: an interpreter. There is no builtin editor or other development support. Hence, to create a Python script you need an external text editor. For a first script you can use any editor that can create plain, unstyled text files, such as SimpleText.

For more serious scripts, though, it is advisable to use a programmers editor, such as BBEdit or Alpha. BBEdit is my favorite: it comes in a commercial version but also in a fully-functional free version BBEdit Lite. You can download it from the BareBones site. The free version will probably provide all the functionality you will ever need. Besides the standard edit facilities it has multi-file searches and many other goodies that can be very handy when editing programs.

After you have created your script in the editor of your choice you drop it on the interpreter. This will start the interpreter executing the script, again with a console window in which the output appears and in which you can type input if the script requires it. Normally the interpreter will close the window and quit as soon as the script is done executing, see below under startup options for a way to change this.

It is a good idea to have the names of all your scripts end in .py. While this is not necessary for standalone scripts it is needed for modules, and it is probably a good idea to start the habit now.

If you do not like to start the Python interpreter afresh for each edit-run cycle you can use the import statement and reload() function to speed things up in some cases. Here is Guido's original comment for how to do this, from the 1.1 release notes:

Make sure the program is a module file (filename must be a Python identifier followed by '.py'). You can then import it when you test it for the first time. There are now three possibilities: it contains a syntax error; it gets a runtime error (unhandled exception); or it runs OK but gives wrong results. (If it gives correct results, you are done testing and don't need to read the rest of this paragraph. :-) Note that the following is not Mac-specific -- it's just that on UNIX it's easier to restart the entire script so it's rarely useful.
Recovery from a syntax error is easy: edit the file and import it again.
Recovery from wrong output is almost as easy: edit the file and, instead of importing it, call the function reload() with the module name as argument (e.g., if your module is called foo, type reload(foo)).
Recovery from an exception is trickier. Once the syntax is correct, a 'module' entry is placed in an internal table, and following import statements will not re-read the file, even if the module's initialization terminated with an error (one reason why this is done is so that mutually recursive modules are initialized only once). You must therefore force re-reading the module with reload(), however, if this happens the first time you try to import the module, the import statement itself has not completed, and your workspace does not know the module name (even though the internal table of moduesl does!). The trick is to first import the module again, then reload it. For instance, import foo; reload(foo). Because the module object already exists internally, the import statement does not attempt to execute the module again -- it just places it in your workspace.

Clickable python scripts

If you create your script with the correct creator and type, creator 'Pyth' and type 'TEXT', you can double-click your script and it will automatically invoke the interpreter. If you use BBEdit you can tell it about the Python file type by adding it to the "file types" sections of the preferences. Then, if you save a file for the first time you can tell BBEdit to save the file as a Python script through the "options" choice of the save dialog.

The Scripts folder contains a script fixfiletypes that will recursively traverse a folder and set the correct creator and type for all files ending in .py.

Older releases of Python used the creator code 'PYTH' in stead of 'Pyth'. If you still have older Python sources on your system and named them with '.py' extension the fixfiletypes script will correct them.

Interaction with the user

Normally, the interpreter will check for user input (mouse clicks, keyboard input) every once in a while, so it is possible to switch to other applications while a script runs. It is also possible to interrupt the interpreter with the standard command-period keypress, this will raise the KeyboardInterrupt exception. Scripts may, however, turn off this behaviour to facilitate their own event handling. Such scripts can only be killed with the command-option-escape shortcut.

startup options

If the option key is depressed when Python starts executing the interpreter will bring up an options dialog thru which you can influence the way the interpreter behaves. Keep the option key depressed until the dialog comes up.

The options modify the interpreters behaviour in the following way:

the interpreter goes to interactive mode (in stead of exiting) after a script has terminated normally,
for every module imported a line is printed telling you where the module was loaded from,
do not print the values of expressions executed as statements in an interactive python,
do not buffer stdout and stderr,
print some debugging output during the parsing phase,
keep the output window open when a script terminates.

In addition, you can enter a unix-style command line which is passed to the script in sys.argv. Sys.argv[0] is always the name of the script being executed, additional values can be passed here. Quoting works as expected.

The default options are also settable on a system-wide basis, see the section on editing preferences.

Module search path

The module search path, sys.path, contains the folders python will search when you import a module. The path is settable on a system-wide basis (see the preferences section), and normally comprises the current folder (where the script lives), the Lib folder and some of its subfolders and possibly some more.

Working folder

The unix concept of a working directory does not translate directly to a similar concept on the Macintosh. To facilitate easy porting and the use of relative pathnames in scripts the interpreter simulates a working directory. When a script is started the initial working directory is the folder where the script lives. In case of an interactive interpreter the working directory is the folder where the interpreter lives.

By the way: the "standard file" folder, the folder that is presented to the user initially for an open or save dialog, does not follow the Python working directory. Which folder is initially shown to the user is usually one of (a) the application folder, (b) the "Documents" folder or (c) the folder most recently used for such a dialog (in any Python program). This is standard MacOS behaviour, so don't blame Python for it. The exact behaviour is settable through a control panel since System 7.5.

Interactive startup file

If the folder containing the interpreter contains a file named PythonStartup this file is executed when you start an interactive interpreter. In this file you could import modules you often use and other such things.

Compiled python scripts

Once a python module has been imported the interpreter creates a compiled version which is stored in a file with the ".py" extension replaced by ".pyc". These compiled files, with creator 'Pyth' and type 'PYC ' load faster when imported (because they do not have to be parsed). The Lib folder contains a script compileall.py, running this script will cause all modules along the python search path to be precompiled, which will speed up your programs. Compiled files are also double-clickable.

Python resources

MacPython has the ability to collect a number of compiled modules together in the resource fork of a single file. This feature is useful if you distribute a python program and want to minimize clutter: you can put all the needed modules in a single file (which could even be the interpreter itself).

If the module search path contains a filename as one of its entries (as opposed to a folder name, which is the normal case) this file will be searched for a resource with type 'PYC ' and a name matching the module being imported.

The scripts folder contains a script PackLibDir which will convert a number of modules (or possibly a complete subtree full of modules) into such a resource file.

Setting interpreter preferences

The python interpreter keeps a preferences file in the standard location in the system folder. In this preferences file it remembers the default module search path and the default settings for the runtime options. The preferences are settable via EditPythonPrefs. For PPC python this is a standalone program living in the main Python folder, for 68K python it is a script in the Scripts folder.

The interface to edit the preferences is rather clunky for the current release.

In the editable text field at the top you enter the initial module search path, using newline as a separator. There are two special values you can use here: an initial substring $(PYTHON) will expand to the Python home folder and a value of $(APPLICATION) will expand to the the python application itself. Note that the text field may extend "beyond the bottom" even though it does not have a scroll bar. Using the arrow keys works, though.

The Python home folder $(PYTHON) is initially, when you execute the interpreter for the first time, set to the folder where the interpreter lives. You can change it here.

Finally, you can set the default startup options here, through a sub-dialog.

Applets

An applet is a fullblown application written in Python, similar to an AppleScript applet (and completely different from a Java applet). Applets are currently only supported on PowerPC macintoshes, and are created using the mkapplet program. You create an applet by dropping the python source script onto mkapplet. The Demo folder contains an example of a more involved applet with its own resource file, etc.

Note that while an applet behaves as a fullblown Macintosh application it is not self-sufficient, so distributing it to a machine without an installed Python interpreter will not work: it needs the shared python execution engine PythonCore, and probably various modules from the Lib and PlugIns folders.

Customizing applets

Applets can have their own settings for the startup options and module search path. Dropping an applet on the EditPythonPrefs application allows you to set these, in the same way as double-clicking EditPythonPrefs allows you to set the system-wide defaults.

Actually, not only applets but also the interpreter itself can have non-default settings for path and options. If you make a copy of the interpreter and drop this copy onto EditPythonPrefs you will have an interpreter that has a different set of default settings.

Where to go from here

The previously mentioned Python Tutorial is an excellent place to start reading if you have never used Python before. Other documentation such as the library reference manual is indexed at the Python Documentation page.

There are some annotated sample programs available that show some mac-specific issues, like use of various toolboxes and creation of Python applets.

Finally, the Demo folder in the Macintosh distribution contains a number of other example programs. Most of these are only very lightly documented, but they may help you to understand some aspects of using Python.

The best way to contact fellow Macintosh Python programmers is to join the MacPython Special Interest Group mailing list. Send a message with "info" in the body to pythonmac-sig-request@python.org or view the Pythonmac SIG page on the www.python.org WWW server.

Troubleshooting

Python is a rather safe language, and hence it should be difficult to crash the interpreter of the system with a Python script. There is an exception to this rule, though: the modules that interface to the system toolboxes (windowing, quickdraw, etc) do very little error checking and therefore a misbehaving program using these modules may indeed crash the system. Such programs are unfortunately rather difficult to debug, since the crash does not generate the standard Python stack trace, obviously, and since debugging print statements will often interfere with the operation of the program. There is little to do about this currently.

Probably the most common cause of problems with modules ported from other systems is the Mac end-of-line convention. Where unix uses linefeed, 0x0a, to separate lines the mac uses carriage return, 0x0d. To complicate matters more a lot of mac programming editors like BBEdit and emacs will work happily with both conventions, so the file will appear to be correct in the editor but cause strange errors when imported. BBEdit has a popup menu which allows you to inspect (and set) the end-of-line convention used in a file.

Jack Jansen, jack@cwi.nl, 19-Apr-1996.