[Tutor] deriving class from file to handle input line numbers?

[Kent Johnson]

Duncan Gibson wrote:
> I was sure that this must be a frequently asked [homework?] question,
> but trying to search for 'file open line number' didn't throw up the
> sort of answers I was looking for, so here goes...
> 
> I regularly write parsers for simple input files, and I need to give
> helpful error messages if the input is invalid. The three most helpful
> pieces on information are file name, line number and line content. Of
> course I can use a global variable to track how many times f.readline()
> is called, but I was wondering whether there is a more OO or Python
> way of encapsulating this within a class derived from file.

If you just want to keep track of line numbers as you read the file by lines, you could use enumerate():

f = open('myfile.txt')
for line_number, line in enumerate(f):
  ...

For any iterable object, enumerate() returns a sequence of (index, value) for each value in the object.

Your class has the advantage of maintaining the line number internally and customizing readline() to your needs.

More comments inline.

Kent

> What I have below is the minimal interface that I could come up with,
> but it is a file adaptor rather than a derived class, and it doesn't
> seem quite clean to me, because I have to open the file externally
> and then pass the file object into the constructor.

Why? You should be able to open the file in the constuctor and provide close() and __del__() methods to close the file. What problem did you have when deriving from file? 

> class LineCountedInputFile(object):
>     """
>     add input line count to minimal input File interface
> 
>     The file must be opened externally, and then passed into the
>     constructor.  All access should occur through the readLine method,
>     and not using normal File methods on the external file variable,
>     otherwise things will get out of sync and strange things could
>     happen, including incorrect line number.
>     """
>     
>     __slots__ = (
>             '_inputFile',
>             '_lineNumber')

__slots__ is overkill, it is intended as a way to save memory in classes that will have many instances, and discouraged otherwise.

>     
>     def __init__(self, inputFile):
>         """
>         create a LineCountedInputFile adaptor object
>         
>         :param inputFile: existing File object already open for reading only
>         :type  inputFile: `File`
>         """
>         
>         assert isinstance(inputFile, file)
>         assert not inputFile.closed and inputFile.mode == 'r'
>         
>         self._inputFile = inputFile
>         self._lineNumber = 0
> 
>     #----------------------------------------------------------------------
>         
>     def readLine(self):

I would call this method readline() to match what file does.

>         """
>         call file.readline(), strip excess whitespace, increment line number
>         
>         :return: next line of text from file minus excess whitespace, or
>                 None at end-of-file
>         :rtype:  str
>         """
>         
>         line = self._inputFile.readline()
>         if len(line) == 0:
>             return None
>         self._lineNumber += 1
>         return line.strip()
> 
>     #----------------------------------------------------------------------
>     
>     def _get_fileName(self):
>         return self._inputFile.name
>         
>     fileName = property(_get_fileName, None, None, """(read-only)""")

Maybe just call this property 'name' to match what file does.
 
>     #----------------------------------------------------------------------
>     
>     def _get_lineNumber(self):
>         return self._lineNumber
>         
>     lineNumber = property(_get_lineNumber, None, None, """(read-only)""")

The use of properties to make read-only attributes might be overkill also. I would just have a lineNumber attribute but that is more of a judgement call.

> 
> _______________________________________________
> Tutor maillist  -  Tutor at python.org
> http://mail.python.org/mailman/listinfo/tutor
>