Parsing a string for nested patterns

Question

What would be the best way to do this.

The input string is

<133_3><135_3><116_2>The other system worked for about 1 month</116_2> got some good images <137_3>on it then it started doing the same thing as the first one</137_3> so then I quit using either camera now they are just sitting and collecting dust.</135_3></133_3>

the expected output is

{'The other system worked for about 1 month got some good images on it then it started doing the same thing as the first one so then I quit \
using either camera now they are just sitting and collecting dust.':[133, 135],

'The other system worked for about 1 month': [116],

'on it then it started doing the same thing as the first one':[137]

}

that seems like a recursive regexp search but I can't figure out how exactly.

I can think of a tedious recursive function as of now, but have a feeling that there should be a better way.

Related question: Can regular expressions be used to match nested patterns?

Answer 1

Use expat or another XML parser; it's more explicit than anything else, considering you're dealing with XML data anyway.

However, note that XML element names can't start with a number as your example has them.

Here's a parser that will do what you need, although you'll need to tweak it to combine duplicate elements into one dict key:

from xml.parsers.expat import ParserCreate

open_elements = {}
result_dict = {}

def start_element(name, attrs):
    open_elements[name] = True

def end_element(name):
    del open_elements[name]

def char_data(data):
    for element in open_elements:
        cur = result_dict.setdefault(element, '')
        result_dict[element] = cur + data

if __name__ == '__main__':
    p = ParserCreate()

    p.StartElementHandler = start_element
    p.EndElementHandler = end_element
    p.CharacterDataHandler = char_data

    p.Parse(u'<_133_3><_135_3><_116_2>The other system worked for about 1 month</_116_2> got some good images <_137_3>on it then it started doing the same thing as the first one</_137_3> so then I quit using either camera now they are just sitting and collecting dust.</_135_3></_133_3>', 1)

    print result_dict

Answer 2

Take an XML parser, make it generate a DOM (Document Object Model) and then build a recursive algorithm that traverses all the nodes, calls "text()" in each node (that should give you the text in the current node and all children) and puts that as a key in the dictionary.