Take words from a sentence (help)

[cerberus_gr]

Hello,

which command do we use, if we have a sentence like "word1 word2 word3 word4 word5 word6" and we want
to save to a variable only "word2 word3 word4"? We don't know the number of words, but only which are these words.

Example
sentence 1 -> My name is George and I am 27 years old.
sentence 2 -> My first name is George and I am 27 years old and I live...
need -> George and I am 27

If I want to save to a variable the sentence from George until the end?

What should I do if the sentence has characters such as ", [. { etc.
Should I use join and split command?

thx

[Dereckson]

Instead join/split, you need something convert CORRECTLY your string into a list.

This is the solution I use :

Code: Select all

proc splitwords {string} { 
	set return ""
	foreach word [regexp -inline -all -- {\S+} $string] { 
		lappend return $word
	} 
	return $return
}

[caesar]

Or see/use the "lrange" here..
eg: set bla [lrange $text 4 end]

[ppslim]

The regexp is not needed, join and split are fine.

First you split the string into a list, so you can operate on it in a array like fassion.

One you have finished with it as a list, and need a string again, you use the join command.

In this, treat the incoming text as $text

Code: Select all

#Convert strin to list, and place it back in the same variable
set text [split $text]
#extract using lrange
set part [lrange $text 4 end]
#Now convert $part to a string so the characters do not look garbled.
set part [join $part]

However, this doesn't cover extracting the details you wanted.

Because of the way the 2 sentaces are structured (differently), ity would take some form of AI system, to know what part of the text is what.

If you know of a script, be it in another language, that can do this, paste it here, we can try and convert it to Tcl.

[strikelight]

Your original post is a bit confusing as to what you want exactly..

From your example, if you only wanted 'George and I am 27',
I would recommend using:

Code: Select all

regexp -nocase {name is (.*) years} garbage sentence

And the contents of $sentence would then be 'George and I am 27' from your given example data.

However, you went on to say 'from George until the end'.. If this is the case, I would recommend simply using:

Code: Select all

regexp -nocase {name is (.*)} garbage sentence

And $sentence would then contain from the name until the end of the sentence.

For more information, consult the re_syntax part of TCL Manpages.