cdata value as integer generates TypeError

[pamiel]

Using xmltodict on RedHat runs like a charm. However, moving the exact same script to Alpine has generated an issue in the particular use case where the cdata value is an integer.

I reproduced the issue by using the yq tool (https://github.com/kislyuk/yq) that transforms a yaml input into a XML output, using xmltodict. My yaml input is quite simple:

test:
  element:
    '@id': 10
    '#': 10

Once given to yq (and thus xmltodict), I'm expecting an output similar to:

<test>
  <element id="10">10</element>
</test>

(I'm using # as cdata_key instead of the default #text, but this is not changing the game)

However, I have the following output and exception:

<test>
  <element id="10">TypeError: coercing to Unicode: need string or buffer, int found.

The attribute with an integer value is correctly handled and becomes the id="10" string, but the element value as an integer generates the error. More precisely, that's line 415 (https://github.com/martinblech/xmltodict/blob/master/xmltodict.py#L415) of xmltodict that generates the error... on Alpine (and for example, not on RedHat !... but don't ask me why!): the characters method of the Handler is expecting a string, and an integer is provided => TypeError!

The bugfix is quite simple and efficient: force the data to be a string with the str( ) so that the line 415 becomes: content_handler.characters(str(cdata))

I would be happy to provide a Pull Request with this change if you are ok with that update.

Thanks for your help, and great thanks for your work !

[javadev]

{
   "test": {
      "element": {
         "-id": "10",
         "#text": 10
      }
   },
   "#omit-xml-declaration": "yes"
}

may be converted to

<test>
  <element id="10" number="true">10</element>
</test>