For example, say I create a template

A template with $value1, $value2

When I substitute this template with values, say I provide only $value1, I want other $variables to be substituted with a default value. If my default value is HAHA. and I provide this map {'value1': 'first_value'}, the result is

A template with first_value, HAHA

Creating the class was very simple (in retrospect).

I just subclassed string.Template and provided my own version of safe_substitute.

Here's the class, complete with doctest code.

import string

class Template(string.Template):

    """

        Template with modified functions

    -- doctests ----

    >>> t = 'this is a $test with $result'

    >>> st = Template(t, 'haha')

    >>> d = {'test': 'a big test'}

    >>> result = st.safe_substitute(d)

    >>> expected_result = 'this is a a big test with haha'

    >>> test_result = True if expected_result == result else 'expected %s >>> but got <<< %s' % (expected_result,  result)

    >>> test_result

    True

    >>> st = Template(t)

    >>> d = {'test': 'a big test'}

    >>> result = st.safe_substitute(d)

    >>> expected_result = 'this is a a big test with $result'

    >>> test_result = True if expected_result == result else 'expected %s >>> but got <<< %s' % (expected_result,  result)

    >>> test_result

    True

    """

    def __init__(self, template, default_substitution_value = None):

        string.Template(template)

        self.default_substitution_value = default_substitution_value

        self.template = template

    def safe_substitute(self, *args, **kws):

        '''

            Returns template with placeholders substituted

              if no substitution value is specified, the

                  default_substitution_value

                  is used

        '''

        if len(args) > 1:

            raise TypeError('Too many positional arguments')

        if not args:

            mapping = kws

        elif kws:

            mapping = _multimap(kws, args[0])

        else:

            mapping = args[0]

        # Helper function for .sub()

        def convert(mo):

            named = mo.group('named')

            if named is not None:

                try:

                    # We use this idiom instead of str() because the latter

                    # will fail if val is a Unicode containing non-ASCII

                    return '%s' % (mapping[named],)

                except KeyError:

                    if self.default_substitution_value:

                        return self.default_substitution_value

                    return self.delimiter + named

            braced = mo.group('braced')

            if braced is not None:

                try:

                    return '%s' % (mapping[braced],)

                except KeyError:

                    return self.delimiter + '{' + braced + '}'

            if mo.group('escaped') is not None:

                return self.delimiter

            if mo.group('invalid') is not None:

                return self.delimiter

            raise ValueError('Unrecognized named group in pattern',

                             self.pattern)

        return self.pattern.sub(convert, self.template)

I built the above because I was building an interface to populate P2 Energy's Field Operations equipment readings. The application sends data via xml files and there are many attributes in the file that I did not care to populate. Rather than hand code a default value for each of them, I built the above python class.

Read more python articles

in which tests in functions using externally 

defined decorators are ignored by doctest.

Here's a simple workaround.

@aspect.processedby(aspect.tracing_processor)

def aa():

    ''' Returns None

    -- doctests ----

    >>> aa()

    True

    '''

    return False

def all_tests():

    ''' Returns None

    -- doctests ----

    >>> aa()

    True

    >>> aa()

    False

    '''

    return False

In the above snippet, my aa function is decorated and as expected, doctest ignores the tests within the docstring.

However, the all_tests function is not decorated and doctest picks it up and runs the tests.

**********************************************************************

File "C:\1\python\test_doctest.py", line 31, in __main__.all_tests

Failed example:

    aa()

Expected:

    True

Got:

    False

**********************************************************************

1 items had failures:

   1 of   2 in __main__.all_tests

***Test Failed*** 1 failures.

completed doctest

This is a simple workaround (and solution) and the obvious disadvantage is that the tests are further away from the tested function.

That is not altogether too bad, considering that if you switch to other unit testing methods like nose, py.test, you would have to place your tests separately anyway.

Read more python articles

What do I use pywin32 for ?

I use it for:

.COM programming - Excel automation. There are alternatives for building Excel files without .COM but if you need to invoke Excel macros within your python program, you need .COM.

This is yet another reason to stick to Windows 32 bit for now.

Read more guide to buying a dell,python articles

I tested it first on a desktop. The default upgrade installed to a C:\Python25 directory and a lot of things broke. Python scripts that worked before stopped working.

easy_install would not work because it could not find Python.exe.

This was due to my previous Python installation was on c:\1Programs\Python.

Fortunately I created a system restore point before the upgrade and I restored to the point.

At this time, Python was still not working properly - the restore did not repair everything. So much for Window's System Restore.

Then I installed Python 2.5.4 again, this time on c:\1Programs\Python.

Now, everything works!

Lessons learnt:

1. Test the upgrade on a test machine first

2. Install upgrades on the same path of the original installation

3. Don't count on Window's System Restore Points

Read more python,work,wow articles

The gvim.exe at vim.org does not have +python for 2.5.

You can get a python 2.5 build at Yong Wei's site.

Now omnicomplete works for me !!

Read more python,work articles

kl_ofm.py -g -v

When the scheduled task runs, it runs as kl_ofm.py without the arguments.

The above is something to take note of when scheduling Python scripts.

One option around this is to put kl_ofm.py -g -v into a batch file and schedule the batch file as the job to run.

Read more python,work articles

import win32com.client

xl = win32com.client.Dispatch('Excel.Application')

wb = xl.Workbooks('Book1')

ws = wb.Worksheets('Sheet1')

cell = ws.Range('A1')

cell.SetValue(arg1 = 'test entry in Excel')

Above is a very simple python script to write something in Excel.

To run the above, you need Mark Hammond's pywin32 to generate the PythonCOM package. Read O'Reilly's Python Programming on Win32 for more information on what PythonCOM packages are.

Once you have installed pythonwin, start it and generate the python code for Excel like this:

1. Start PythonWin

2. Select MakePy

3. Select the Excel library

4. The python file is generated.

However, the generated file will not yet work. If you now type in the above code snippet, you will get an COM error:

com_error: (-2147352567, 'Exception occurred.', (0, None, None, None, 0, -214735

2565), None)

The cause of the com_error is various, on one machine, it was due to some bug in the file.

Use your editor and open the generated python file, find all the ret = Dispatch lines ...

ret = Dispatch(ret, 'Item', '{00020857-0000-0000-C000-000000000046}', UnicodeToString=0)

and replace with

ret = Dispatch(ret, 'Item', '{00020857-0000-0000-C000-000000000046}')

On another, it was because xl was linked to an Excel instance which was visible. I did this to make it show itself

xl.Visible = 1

Read more python,work articles