Hi! I'm looking for a solution for XML parsing. I found jasch_lib [detox] but it crashes pd in OSx and don't have RPi binary disponible in deken. I tried [jasch_lib/detox] in purr data and it still crash.

My goal is to recognize some identifiers and make it disponible in some matrix or vector for sonification purposes.

There are other external or vanilla way can it be done?

Cheers

@emviveros XML should be easy as it generally does not use reserved characters like backslash and curly brackets, that Pd will ignore.
You should be able to create a patch using [text] and the [list] objects in vanilla.
Post an example xml file and point us to the data you are looking to use..... if you need help.
David.

@emviveros You can also consider the option of converting your xml to json and then use this lib:
https://github.com/residuum/PuRestJson

Cheers~

Thanks for your suggestions!! Are very apreciated!!

@whale-av I'm advancing in this way you draw... Soon I will have some patch to share and to make more questions...

@EEight This option shine in my mind!!! o.O The project wich I'm a collaborator consist in sonify data of one week of public bus transport system of São Paulo city. With PuRest_Json lib I can retrieve data of SP bus system in real time!!! I'm correct?

Thanks again guys!!

I tried the two solutions.. and I had some problems...

1. @whale-av [text]:
   1.1 - pd crash when I tried to load one of the xml (16 Mb) (I have more 7 that)
   1.2 - text file truncate some values. 30003042 turn to 3.0003e+07
   1.3 - it don't output a line like in the xml file.. It counts line like a comma " , " here we have a sample of one of the files, the original have 2276 lines, and one of the biggest have 47046 lines.. Pd really can handle with these files?

2. @EEight PuRest_Json:
   2.1 I like it but I can't handle large files with [json-encode] and [json-decode] please correct me... I only see the way to parse json messages.

Thanks for any advance!!

@emviveros Then a better approach might be to parse the xml outside pd (choose your favorite programming language and send the values to pd). If I were you, I would use python (ElementTree) and use pyliblo (open sound control) to send the values to pd ([listen 5900< [netreceive -u -b] [parseosc]).

Cheers~

Yep!! It's time to learn python!!

Thanks again!

@emviveros Pd doesn't actually lose your value.... it just doesn't show or print it.
If you subtract 3e+07 from the value it shows as 3.0003e+07 then it will print 3042.
Yes, very large files, and yes, Pd sees a comma as "new list".
It might not have crashed, but the time necessary to read the file could well render Pd unusable.
@Eeight is most likely correct...!
David.

Thanks @whale-av for the explanations! Really usefull!

@emviveros I just made this out of old bits (so [pbank] does things you don't need).........
parse.zip (vanilla...... [parse] is the main patch).
It will lose the first and last values of each data set because they are stuck to the xml separators, but that is useful for separating the sets.
I don't know if that matters to your project?
They could be recovered I am sure, but more work would be required.
It would be interesting to know if it can deal with your huge files.

You could work through the data sets using a [metro] instead of [until] and setting [s $0-done] to fire a repeat count......... as you probably need to slow the output rate to make your patch useable........ bringing in the data at a set tempo..
As the data sets are in pairs (but part of the same sequence with radically different values) you are probably using them for different purposes. You could flip-flop them onwards to separate parts of your patch with a couple of spigots.
David.

Hum very interesting!! I'm going to dwell on your patch more calmly this weekend. Soon return with news and observations .. !! Gratitude @whale-av!

@whale-av I'm testing your patch, but with bigger files, it becomes really heavy. I don't think it will be usable for my project, but for reading some lines, like 1000 it works well... I tried until 50k reads, and after, hummm.. 20 to 30 seconds it retrieve some data (with some data changes, but ok.. thinks to fix latter). I don't think we can do it well like a xml parser algorithm can do out of pd.. I will test some python stuff and compare... I'll post results here.

if you want to test yourself, you have the data I'm using here: http://interscity.org/open_data/

Cheers

To register, I received a response in Pd-list and I return to try [pofxml], but it fail to parse bigger files, explicitly that: map_true.xml, wich freeze Pd...

on step
Esteban

Just an FYI-- We just removed a buffer overflow from jasch_lib/detox in Purr Data. Should be out in the next release.

Thanks @jancsika ! At this moment I'm parsing xml files in Python scripts. But I will try purrdata again to..