Re: [Pyparsing] How do fix this?

[Paul M. <pt...@au...>]

First of all, it is not necessary (and probably not even helpful) to define
every element of your grammar as an instance variable, preceded by 'self.'.
In nearly all cases, when I define a grammar within a parsing class, I'll do
something like this:

class BobParser(object):
    def __init__(self):
        expr1 = Literal("Bob's your")
        expr2 = oneOf("uncle aunt brother sister father mother")
        self.parser = expr1 + expr2('relation')
    
    def who_is_it(self, string):
        return self.parser.parseString(string).relation


All the sub-expressions can just be local variables, and save the "self."
business for just the top-level parser.

This really looks like you just started at the beginning of your input text
and started writing pyparsing expressions to it. Many times there's nothing
wrong with that, but in your case, there is a lot of complexity and
structure to your input. And more importantly, many internal patterns that
are repeated - these can be defined once and then reused by name over and
over.

I often encourage people to write a BNF for complex data like this. Or at
*least* look at the input at an overall level for which bits are common and
can be reused. There are many small pieces that can be defined and reused in
larger parts, and these will help simplify your grammar.  For instance,
there are many places where you use this:

    Combine(self.plain_number + 'x' + self.plain_number,
adjacent=False).setResultsName('something')

which really is challenging to the eye to see what is the parser and what is
the meta-information (adjacency, results name). And you repeat this whenever
you need a "0 x 0" or "640 x 480", so it makes things very messy.  If you
define for yourself this reusable expression:

    number_x_number = Combine(plain_number + 'x' + plain_number, adjacent =
False)

Then you can use it in multiple places as:

    number_x_number("clean_aperature")
    number_x_number("Dimensions")
    etc.
    
I also strongly recommend using the short-cut version of
expr.setResultsName('xyz'), as expr('xyz').  This notation can really clean
up your grammar definition, and make it easier to see the overall parser
without being distracted by all the function calls. 

Finally, you make heavy use of "Combine(something + something + something,
adjacent=False)". Please consider using Group instead. It's clearer to
follow, it implicitly allows whitespace (so adjacent=False is not
necessary), and it allows you to define results names within the group,
making for a useful substructure (as in sample_time below):

    number_x_number = Group(plain_number + 'x' + plain_number)
    number_sl_number = Group(plain_number + '/' + plain_number)
    word_sl_word = Gruop(word + '/' + word)
    timestamp = Regex(r'\d\d:\d\d:\d\d\.\d\d\d')
    sample_time = Group(number_sl_number('sample') + timestamp('time'))

So just going through and stripping out all the "self." stuff and using some
of these repeated sub-expressions might make things easier to follow, and
your overall intent and structure will be clearer. Here is one area of your
parser with these changes:

    audio_dimensions = 'Dimensions:' + number_x_number('Dimensions')
    audio_track_matrix = 'Track Matrix:' + restOfLine('track_matrix')
    audio_track_dimensions = audio_dimensions + audio_track_matrix 
    subtitle_dimensions = 'Dimensions:' + number_x_number('Dimensions')
    subtitle_track_matrix = 'Track Matrix:' + restOfLine('track_matrix')
    subtitle_track_dimensions = subtitle_dimensions + subtitle_track_matrix 
    video_dimensions = 'Dimensions:' + number_x_number('Dimensions')
    video_clean_aperture = 'CleanAperture: ' +
number_x_number('CleanAperture')
    video_production_aperture = 'ProductionAperture:' +
number_x_number('ProductionAperture')
    video_encoded_pixels = 'EncodedPixels:' +
number_x_number('EncodedPixels')

Now you can answer some other questions for yourself, like "why do I repeat
the 'Dimensions:' expression?", and further simplify your grammar.

Good luck,
-- Paul


---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus