Hi,
I did a patch using [text] objet that is able to search and replace words in a xml file by a random dictionary. It is working a bit, but with some errors and limitation...

1- Since [text search] is very limited, I use [text get] and loop every line and field to search a field that match my word... but I don't know the maximum field numbers per line, which is a bit annoying...
So I get red errors in the pd console when I am out of field.

2- Because of my [text get] field's looping system, every words need to be surrounded by a space. but the xml tag formatting is a bit limiting here... like "<noun ...>" string cannot match "noun".

Would be very happy to know how to improve it/get rid of some limitation here

Here is the source:
Search_and_replace.zip

Cheers

@60htz Regarding the first issue, i would get the whole line, determine the list length and iterate over the fields with [list store].

As for the second point or more like the whole project, i think it is an uphill battle to build something like an xml parser with [text]. It would only make sense for a very specific use case, where the structure of the xml is well known and won't change in the future. But then if the structure is known, you can simply search for "<noun" instead of "noun".

I was working on a similar idea but I didnt get very far because the purpose I needed for was more quickly satisfied with paper than a computer! But I would love to see how this develops.

There is an object [detox] from jasch_lib, that seems to dissect xml tags. Maybe this could help.

Thanks all for the clever clues!

I got rid of error message by checking the lines length with a simple [list length] like @ingox said... sometimes things are so simple you need to ask on a forum and then check back your code to see it's one or two object ahead!

here is the result (it's a particular case, but it might help others...)

Search_and_replace2.zip

As for the xml tags, I will investigate detox thanks... I remember that with some xml it freezes my pd so I try to be careful.

@60htz Here is a version that filters out the "<" at the beginning and any non-letter character at the end: search_and_replace.zip

Wow thanks @ingox it's working well

Last but not least, what would be the best way to know that the replace process is finished? I think about a logical check... if the max_lines equal the actual line AND if the max length equal to max_length then the replace process is done. But it's a bit painfull with the [until] loops, I need to use [tof/streamMinMax] in order to get the maximum...

@60htz The easiest way to determine the absolute order of operations is with trigger:

The bang at the end will be reliably triggered after all the until madness is finished.

In this example it will never happen as the untils never end, but you get the idea.

Ah, so simple, thanks for the explanation

@60htz Yes, the only things that break the order of operation are [delay] and [pipe].

In this example, [print 1] prints first, because [delay 0] already jumps out of the execution order and is processed after all the rest of the tree is finished and then executes with a zero delay: