Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

xml data or other?

122 views
Skip to first unread message

Artie Ziff

unread,
Nov 9, 2012, 7:54:43 AM11/9/12
to pytho...@python.org
Hello,

I want to process XML-like data like this:

<testname=ltpacpi.sh>
<description>
ACPI (Advanced Control Power & Integration) testscript for 2.5 kernels.

<\description>
<test_location>
ltp/testcases/kernel/device-drivers/acpi/ltpacpi.sh
<\test_location>
<\testname>


After manually editing the data above, the python module
xml.etree.ElementTree parses it without failing due to error in the data
structure.

Edits were substituting '/' for '\' on the end tags, and adding the
following structure:

<?xml version="1.0"?>
<data>
<testname name=ltpacpi.sh>
...
<\testname>
</data>


Is there a name for the format above (perhaps xhtml)?
I'd like to find a python module that can translate it to proper xml.
Does one exist? etree?

Many thanks!
az

rusi

unread,
Nov 9, 2012, 8:50:15 AM11/9/12
to
On Nov 9, 5:54 pm, Artie Ziff <artie.z...@gmail.com> wrote:
> Hello,
>
> I want to process XML-like data like this:
<snipped>
> Edits were substituting '/' for '\' on the end tags, and adding the
> following structure:

If thats all you want, you can try the following:

# obviously this should come from a file
input= """<testname=ltpacpi.sh>
<description>
ACPI (Advanced Control Power & Integration) testscript
for 2.5 kernels.

<\description>
<test_location>
ltp/testcases/kernel/device-drivers/acpi/ltpacpi.sh
<\test_location>
<\testname>"""

prefix = """<?xml version="1.0"?>
<data>
"""

postfix = """</data>"""

correctedInput = prefix + input.replace("\\", "/") + postfix
# submit correctedinput to etree

shiver...@yahoo.co.uk

unread,
Nov 13, 2012, 9:05:49 AM11/13/12
to pytho...@python.org
maybe an xml tool would be better, a good list of xml tools here; http://www.xml-data.info

shiver...@yahoo.co.uk

unread,
Nov 13, 2012, 9:05:49 AM11/13/12
to comp.lan...@googlegroups.com, pytho...@python.org
On Friday, November 9, 2012 12:54:56 PM UTC, Artie Ziff wrote:

Dave Angel

unread,
Nov 13, 2012, 1:01:17 PM11/13/12
to Artie Ziff, pytho...@python.org
On 11/09/2012 07:54 AM, Artie Ziff wrote:
> Hello,
>
> I want to process XML-like data like this:
>
> <testname=ltpacpi.sh>
> <description>
> ACPI (Advanced Control Power & Integration) testscript for 2.5
> kernels.
>
> <\description>
> <test_location>
> ltp/testcases/kernel/device-drivers/acpi/ltpacpi.sh
> <\test_location>
> <\testname>
> <snip...>
>
>
> Is there a name for the format above (perhaps xhtml)?

The only word I can think of is "broken." xml and html and xhtml all
use forward slashes.

> I'd like to find a python module that can translate it to proper xml.
> Does one exist? etree?
>

I think you've already figured it out. Just take your description and
turn it into Python. in other words, replace all "<\" with "</" and
perhaps " \>" with " /", although your example doesn't happen to have
any of these. Tack a xml header on, and try to parse it with etree.
If you can't, then let someone manually fix it.

Or better, fix the program upstream that's creating this mess. There
isn't a reliable way to "fix" all the possible broken xml it might be
creating, without reverse engineering it.



--

DaveA

Artie Ziff

unread,
Nov 18, 2012, 8:32:26 AM11/18/12
to pytho...@python.org
On 11/9/12 5:50 AM, rusi wrote:
> On Nov 9, 5:54 pm, Artie Ziff <artie.z...@gmail.com> wrote:
> # submit correctedinput to etree
I was very grateful to get the "leg up" on getting started down that
right path with my coding. Many thanks to you, rusi. I took your
excellent advices and have this working.

class Converter():
PREFIX = """<?xml version="1.0"?>
<data>
"""
POSTFIX = "</data>"
def __init__(self, data):
self.data = data
self.writeXML()
def writeXML(self):
pattern = re.compile('<testname=(.*)>')
replaceStr = r'<testname name="\1">'
xmlData = re.sub(pattern, replaceStr, self.data)
self.dataXML = self.PREFIX + xmlData.replace("\\", "/") +
self.POSTFIX

### main
# input to script is directory:
# sanitize trailing slash
testPkgDir = sys.argv[1].rstrip('/')
# Within each test package directory is doc/testcase
tcDocDir = "doc/testcases"
# set input dir, containing broken files
tcTxtDir = os.path.join(testPkgDir, tcDocDir)
# set output dir, to write proper XML files
tcXmlDir = os.path.join(testPkgDir, tcDocDir + "_XML")
if not os.path.exists(tcXmlDir):
os.makedirs(tcXmlDir)
# iterate through files in input dir
for filename in os.listdir(tcTxtDir):
# set filepaths
filepathTXT = os.path.join(tcTxtDir, filename)
base = os.path.splitext(filename)[0]
fileXML = base + ".xml"
filepathXML = os.path.join(tcXmlDir, fileXML)
# read broken file, convert to proper XML
with open(filepathTXT) as f:
c = Converter(f.read())
xmlFO = open(filepathXML, 'w') # xmlFileObject
xmlFO.write(c.dataXML)
xmlFO.close()

###

Writing XML files so to see whats happening. My plan is to
keep xml data in memory and parse with xml.etree.ElementTree.

Unfortunately, xml parsing fails due to angle brackets inside
description tags. In particular, xml.etree.ElementTree.parse()
aborts on '<' inside xml data such as the following:

<testname name="cron_test.sh">
<description>
This testcase tests if crontab <filename> installs the cronjob
and cron schedules the job correctly.
<\description>

##

What is right way to handle the extra angle brackets?
Substitute on line-by-line basis, if that works?
Or learn to write a simple stack-style parser, or
recursive descent, it may be called?

I am open to comments to improve my code more to be more readable,
pythonic, or better.

Many thanks
AZ

rusi

unread,
Nov 18, 2012, 10:54:05 AM11/18/12
to
On Nov 18, 6:32 pm, Artie Ziff <artie.z...@gmail.com> wrote:
> Unfortunately, xml parsing fails due to angle brackets inside
> description tags. In particular, xml.etree.ElementTree.parse()
> aborts on '<' inside xml data such as the following:
>
> <testname name="cron_test.sh">
>      <description>
>          This testcase tests if crontab <filename> installs the cronjob
>          and cron schedules the job correctly.
>      <\description>
>
> ##
>
> What is right way to handle the extra angle brackets?
> Substitute on line-by-line basis, if that works?
> Or learn to write a simple stack-style parser, or
> recursive descent, it may be called?
>
> I am open to comments to improve my code more to be more readable,
> pythonic, or better.
>
> Many thanks
> AZ

Start with cgi.escape perhaps?
http://docs.python.org/2/library/cgi.html

rusi

unread,
Nov 18, 2012, 10:58:12 AM11/18/12
to
On Nov 18, 8:54 pm, rusi <rustompm...@gmail.com> wrote:

> Start with cgi.escape perhaps?http://docs.python.org/2/library/cgi.html

This may be a better link for starters
http://wiki.python.org/moin/EscapingHtml
(Note the escaping xml at the bottom)

Prasad, Ramit

unread,
Nov 19, 2012, 4:42:00 PM11/19/12
to Artie Ziff, pytho...@python.org
Artie Ziff wrote:
>
> On 11/9/12 5:50 AM, rusi wrote:
> > On Nov 9, 5:54 pm, Artie Ziff <artie.z...@gmail.com> wrote:
> > # submit correctedinput to etree
> I was very grateful to get the "leg up" on getting started down that
> right path with my coding. Many thanks to you, rusi. I took your
> excellent advices and have this working.
>
> class Converter():
> PREFIX = """<?xml version="1.0"?>
> <data>
> """
> Unfortunately, xml parsing fails due to angle brackets inside
> description tags. In particular, xml.etree.ElementTree.parse()
> aborts on '<' inside xml data such as the following:
>
> <testname name="cron_test.sh">
> <description>
> This testcase tests if crontab <filename> installs the cronjob
> and cron schedules the job correctly.
> <\description>
>
> ##
>
> What is right way to handle the extra angle brackets?
> Substitute on line-by-line basis, if that works?
> Or learn to write a simple stack-style parser, or
> recursive descent, it may be called?

I think your description text should be in a CDATA section.
http://en.wikipedia.org/wiki/CDATA#CDATA_sections_in_XML

~Ramit


This email is confidential and subject to important disclaimers and
conditions including on offers for the purchase or sale of
securities, accuracy and completeness of information, viruses,
confidentiality, legal privilege, and legal entity disclaimers,
available at http://www.jpmorgan.com/pages/disclosures/email.

Stefan Behnel

unread,
Nov 20, 2012, 12:48:20 AM11/20/12
to pytho...@python.org
Prasad, Ramit, 19.11.2012 22:42:
> Artie Ziff wrote:
>> Writing XML files so to see whats happening. My plan is to
>> keep xml data in memory and parse with xml.etree.ElementTree.
>>
>> Unfortunately, xml parsing fails due to angle brackets inside
>> description tags. In particular, xml.etree.ElementTree.parse()
>> aborts on '<' inside xml data such as the following:
>>
>> <testname name="cron_test.sh">
>> <description>
>> This testcase tests if crontab <filename> installs the cronjob
>> and cron schedules the job correctly.
>> <\description>
>>
>> ##
>>
>> What is right way to handle the extra angle brackets?
>> Substitute on line-by-line basis, if that works?
>> Or learn to write a simple stack-style parser, or
>> recursive descent, it may be called?
>
> I think your description text should be in a CDATA section.
> http://en.wikipedia.org/wiki/CDATA#CDATA_sections_in_XML

Ah, don't bother with CDATA. Just make sure the data gets properly escaped,
any XML serialiser will do that for you. Just generate the XML using
ElementTree and you'll be fine. Generating XML as literal text is not a
good idea.

Stefan


0 new messages