CODEWITHSUNDEEP
pythonxmlpython-2.7installshield
Noam
Noam

Reputation: 1804

Python: xml.etree.ElementTree destroys xml format

I'm having an ISM file (InstallShield project) that is formatted as XML.

I need to change some attributes in the file, so I used xml.etree.ElementTree (Python Library).

I can find the values and change them, however, after saving the file with updated values, I can't open it in InstallShield (I get a general error that file cant be open).

When I compare the old file with the new one, I see that beside the values I changed, some lines are simply missing from new XML and in some line the tags name had changed.

Why does it happen? Is there anything to do to make the file stay exactly as it was except for the changes I've made? Should I use other tool to make the change?

For example, the following section appears in original XML:

    <?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<?xml-stylesheet type="text/xsl" href="is.xsl" ?>
<!DOCTYPE msi [
   <!ELEMENT msi   (summary,table*)>
   <!ATTLIST msi version    CDATA #REQUIRED>
   <!ATTLIST msi xmlns:dt   CDATA #IMPLIED
                 codepage   CDATA #IMPLIED
                 compression (MSZIP|LZX|none) "LZX">

   <!ELEMENT summary       (codepage?,title?,subject?,author?,keywords?,comments?,
                            template,lastauthor?,revnumber,lastprinted?,
                            createdtm?,lastsavedtm?,pagecount,wordcount,
                            charcount?,appname?,security?)>

   <!ELEMENT codepage      (#PCDATA)>
   <!ELEMENT title         (#PCDATA)>
   <!ELEMENT subject       (#PCDATA)>
   <!ELEMENT author        (#PCDATA)>
   <!ELEMENT keywords      (#PCDATA)>
   <!ELEMENT comments      (#PCDATA)>
   <!ELEMENT template      (#PCDATA)>
   <!ELEMENT lastauthor    (#PCDATA)>
   <!ELEMENT revnumber     (#PCDATA)>
   <!ELEMENT lastprinted   (#PCDATA)>
   <!ELEMENT createdtm     (#PCDATA)>
   <!ELEMENT lastsavedtm   (#PCDATA)>
   <!ELEMENT pagecount     (#PCDATA)>
   <!ELEMENT wordcount     (#PCDATA)>
   <!ELEMENT charcount     (#PCDATA)>
   <!ELEMENT appname       (#PCDATA)>
   <!ELEMENT security      (#PCDATA)>                            

   <!ELEMENT table         (col+,row*)>
   <!ATTLIST table
                name        CDATA #REQUIRED>

   <!ELEMENT col           (#PCDATA)>
   <!ATTLIST col
                 key       (yes|no) #IMPLIED
                 def       CDATA #IMPLIED>

   <!ELEMENT row            (td+)>

   <!ELEMENT td             (#PCDATA)>
   <!ATTLIST td
                 href       CDATA #IMPLIED
                 dt:dt     (string|bin.base64) #IMPLIED
                 md5        CDATA #IMPLIED>
]>
<msi version="2.0" xmlns:dt="urn:schemas-microsoft-com:datatypes" codepage="65001">

But in the new XML it's gone and instead there is only:

<msi xmlns:ns0="urn:schemas-microsoft-com:datatypes" codepage="65001" version="2.0">

There are more differences, this is just an example.

The python code I use to make the change is

   tree = Et.parse(ism_file_path)
    root = tree.getroot()

    for attributes_group in root:
        for attribute in attributes_group:

            if attribute.tag == "revnumber":

                new_package_code = increment_hex_number(attribute.text)

                attribute.text = new_package_code

tree.write(ism_file_path)

Thank you!

Upvotes: 2

Views: 798

Answers (2)

Christopher Painter
Christopher Painter

Reputation: 55581

First I'd like to point out that InstallShield has a default compiler setting of generating a new PackageCode every time you build your MSI. This is a best practice so I really don't see why you would need to edit the ISM at all for this scenario.

The second thing I would point out is that the RAW DTD XML formatting of an ISM is very difficult to work with. I tend to use the COM automation inferface for the majority of dynamic authoring and then save the project in Binary format so I can edit it as a relational database just like I would using ORCa. WiX C#/DTF custom actions have LINQ support that makes this much easier to do. You could even write a C# helper class and call it from python too. You can then open it in the automation interface and save it back in XML format if you want to.

Upvotes: 0

Noam
Noam

Reputation: 1804

Eventually I moved to a new library - lxml. This library, in opposed to xml.etree.ElementTree keeps the order of all tags, so I did exactly the same and it worked:

def modify_ism_file(ism_file_path):
    context = etree.iterparse(ism_file_path)
    for action, attributes_group in context:
        for attribute in attributes_group:
            if attribute.tag == "revnumber":
                print "Found package code. TAG = {0} TEXT = {1}".format(attribute.tag, attribute.text)
                new_package_code = increment_hex_number(attribute.text)
                print "New package code is {0}".format(new_package_code)
                attribute.text = new_package_code

obj_xml = etree.tostring(context.root, pretty_print=True, xml_declaration=True,   encoding="utf-8")

    with open(ism_file_path, "w") as f:
        f.write(obj_xml)

Upvotes: 2

Related Questions