[CentOS] XML parsing in shell script

Sat Mar 20 00:26:36 UTC 2021
H <agents at meddatainc.com>

On 03/19/2021 12:40 PM, Fabian Arrotin wrote:
> On 18/03/2021 22:08, H wrote:
>> On 03/18/2021 04:30 PM, Paul Heinlein wrote:
>>> On Thu, 18 Mar 2021, H wrote:
>>>
>>>> I have a challenge I am interested in getting feedback on.
>>>>
>>>> I will on a regular basis download a series of data files from the web where the data is in XML-format. The format is known in advance but is different between the various data files. I then plan to extract the various data items ("elements?") from each data file, do some light formatting and then save desired parts of each original data file as a formatted CSV-file for later importing into a database.
>>>>
>>>> As the plan is to use a bash shell script using curl to get the files, I have begun looking at external XML parsers that I can call from my script, perhaps specify which elements I want, get the data back in some kind of bash data structure and finally format and save as CSV-files.
>>>>
>>>> There seems to be a number of XML parsers available but perhaps someone on the list has a recommendation for which one might suit my needs best? I should add that I am running CentOS 7.
>>> Will you be using an XSLT stylesheet to do the work? There's a somewhat steep learning curve, but in my experience it's the most reliable method for parsing XML except in the very simplest of cases.
>>>
>>> In that case, the libxslt stuff may be what you want:
>>>
>>>   http://xmlsoft.org/libxslt/
>>>
>>> The command-line tool is xsltproc.
>>>
>>> Again, it's not easy to use, but once you've built a toolchain, it will be reliable and fairly easy to modify if the source XML schema change.
>>>
>> I just checked and I cannot see that the organization publishing these data files offer any XSLT stylesheet. IOW, I am, perhaps incorrectly, assuming that the publisher of the data would be one with said stylesheet. (Although perhaps that is something an end-user could put together as well??)
>>
>> Although the data format of each data series is unique, it is simple and could conceivably be parsed using grep but I am looking for a more "forward-looking" solution for other applications in the future.
>>
>> If XSLT stylesheets are not available - would you suggest another tool? Or, would you suggest I design sheets, presumably one for for each data series?
>>
> I used in the past xmlstarlet (available in epel) for quick parsing from
> within bash scripts.
> For something more robust, maybe switch to python ? (ymmv)
>
I wanted to do this in bash and decided on calling xsltproc while investing in writing an XSLT stylesheet for each data file format.