Support for Semi-Structured XML

Semi-structured element should not be parsed and the content should be kept as is. For example:

<level1>
    <level2>
        <level3>First level3.</level3>
        text outside 1st level3 at the end
    </level2>
    <level2>
        text outside 2nd level3 at the beginning
        <level3>Second level3.</level3>
        text outside 2nd level3 at the end
    </level2>
    <level2>
        text outside 3rd level3 at the beginning
        <level3>Third leve3.</level3>
    </level2>
</level1>

will produce (at the corresponding levels):

level2: [
{ #text: "text outside 1st level3 at the end", level3: "First level3." },
{ #text: "text outside 2nd level3 at the beginningtext outside 2st level3 at the end", level3: "Second level3." },
{ #text: "text outside 3rd level3 at the end", level3: "Third level3." },
]

which is not only irreversible (not keeping order) but for 2nd level3 also meaningless (joining the texts). According to “spec” you claim you are adopting, at least the case of 2nd level3 should not be parsed.

It would be also great if a tag name(s) could be specified (as a parameter to parse function) whose content wouldn’t be parsed at all. It could also solve the described issue sometimes (as the user would specify that tag level2 shouldn’t be parsed and its content should be kept in #text property).

Thank you in advance for your comments.

[margru]

AFAIK, there are two options:

xml.sax.saxutils.unescape

which unescape &, <, and > in a string of data.

Or

in Python 2

HTMLParser.HTMLParser.unescape

in Python 3

html.parser.HTMLParser().unescape

which is not documented.

Both are a standard part of Python. The (current) resulting string could be passed into one of these functions to get rid of the entities. But I’m not familiar with the internal processes of your module so I don’t know if it can be really used as I suppose.

[javadev]

I agree. In case of external independent json parser we get different elements order.