topic Re: Split String into Key / Value in Talend Studio

Split String into Key / Value

BooWseR — Fri, 05 Apr 2019 02:20:38 GMT

Hi guys,

I have a long string that I'd like to split into key values:

# Key1: Value1\n# Key2: Value2\n# Key3: Value3...

I split the string with a tNormalize and removed the \n with a tJavaRow, but how do I get the values written as Key / Value in a JSON? I tried before to convert the string into an array and use tMap to map part[0] and part[1] into a schema, but i can't use tMap with an array. In addition, there isn't always a key. Some points consist only of values:

# Key3: Value3\n# Value\n# Key4: Value4\n...

To use Javacode to write the key/values directly into a JSON file isnt realy an option, cause the data have to combine with other data later.

If anyone had an idea, I'd really appreciate it. I saw that there was already another topic here in the forum, but I failed a bit because of the complexity and I think it's more than I need.

Best regards

Re: Split String into Key / Value

TRF — Fri, 05 Apr 2019 07:06:40 GMT

Can share the expected result based on your input?

Re: Split String into Key / Value

BooWseR — Fri, 05 Apr 2019 07:27:37 GMT

Thanks for the quick response.

My final result should be like:

<product_description>
<key1>Value1</key1>
<key2>Value2</key2>
....
</product_description>

Like i said, these data are only a part of the job, so i have to write the JSON later. For the beginning, it would be great if I could handle the data as separated as possible. If Key and Value were in separate fields, I could assign the data in tMap as I want.

The problem is that the final structure is not completely finished yet, but I would like to handle the data as separated as possible.

I hope you know what I mean.

Re: Split String into Key / Value

TRF — Fri, 05 Apr 2019 08:09:32 GMT

Based on your sample, I used a tNormalize with "\n" for the separator:

Then the magic is with the tMap wich allow to separate keys and values using an array (yes, you can):

Here are the expressions to get the keys and values (when no key, a key name is generated).

Keys :

((String[])Var.var1).length > 1 ? ((String[])Var.var1)[0] : "newKey" + Numeric.sequence("newKey", 1, 1)

Values:

((String[])Var.var1).length > 1 ? ((String[])Var.var1)[1] : ((String[])Var.var1)[0]

And the result is:

[statistics] connecting to socket on port 3609
[statistics] connected
.-------+------.
|  tLogRow_10  |
|=------+-----=|
|key    |value |
|=------+-----=|
|Key1   |Value1|
|Key2   |Value2|
|Key3   |Value3|
|newKey1|Value |
|Key4   |Value4|
'-------+------'

[statistics] disconnected

(see the name "newKey1" when key is not in the input string).

Hope this helps.

Re: Split String into Key / Value

BooWseR — Fri, 05 Apr 2019 12:41:22 GMT

That's brilliant! Thank you very much for that.

But what doesn't work for me is the tNormalize with the "\n" separator. Here I have to split to # and then replace the \n in tMap. It works too.

Best regards

Re: Split String into Key / Value

TRF — Fri, 05 Apr 2019 12:55:17 GMT

Maybe you have a "\r\n" instead of a "\n" in your input.

If it comes from a text file, this is possible.

Re: Split String into Key / Value

BooWseR — Fri, 05 Apr 2019 13:46:21 GMT

Hm. The data comes from a JSON file. When I look in with Notepad++, I see "\\n". So yeah, maybe.

But another question: In a few places I have the case that I have a key but no value. This causes the key to be mistakenly recognized as a value. I thought about splitting the string after ": ", because there is always a space between colon and value. If the key has no value, there is no space after the colon.

(((String[])Var.var1).length > 1 || ((String[])Var.var1)[0].contains(":")) ? ...

So I get the key into the key field, but I don't get it out of the value field. Any idea?