searching patterns

[Guest_imported]

The following script matches a pattern specified on the command line (e.g. gawk -f file.awk pattern="the" test.txt):

$0~pattern{
gsub(pattern, "|&|",)
print $0
}

it matches sentences like:

John is the greatest.
He's up there.

and turns this into:

John is |the| greatest.
He's up |the|re.

But in fact, I only wish it to match the first sentence (John is |the| greatest.) Can someone tell me what I should add to my script so that only the word "the" will be found and not a sequence of 't' 'h' 'e' in words like "there", "them", "ether", etc. ?

Thanks

 

[Guest_imported]

{
$0 = tolower($0)
for (i=0 ; i <= NF ; i++) {
if ($i ~ /the/ && length($i) == 3) {
print gensub(/the/,&quot;|&|&quot;,G,$x)
}
}
}

Output:
|the| red man ran up the back stair
|the| red man ran up the back stair
a man and his honey in absconded in |the| night
&quot;failed and failed again&quot;, said |the| reverend bitter.

You can see from the program output the problems with
any of these approaches.
#1: it doesn't get all matches of &quot;the&quot;.
#2: it doesn't perform the substitution (magic &quot;&&quot

comp.lang.awk is full of jerks, but they may be able to
help you out if no one here can come up with anything.

 

[aigles]

Hi,

You can adapt the following awk script for your definition of a word (see matchEre variable)

awk -v pattern=the -f SearchWord.awk test.txt

--- test.txt ---
there is two lines in the input file
three substitutions, one in the fist line and two in the last line
----------------

--- Output ---
there is two lines in |the| input file
three substitutions, one in |the| fist line and two in |the| last line
--------------


--- SearchWord.awk ---
BEGIN {
matchEre = &quot;(^|[ \t]+){1,1}&quot; pattern &quot;($|[ \t]+){1,1}&quot;
}
{
if (pattern != &quot;&quot {
while (1) {
len = length($0)
pos = match($0,matchEre)
if (pos == 0) break
wrk=substr($0,RSTART,RLENGTH)
gsub(pattern,&quot;|&|&quot;, wrk)
$0=substr($0,1,RSTART-1) wrk substr($0,RSTART+RLENGTH)
}
}
print $0
}
[COLOR=/blue]----------------------[/color]
Jean Pierre.

 

[CaKiwi]

Aigles' suggestion is about the best you can do in awk (IMHO). Another option would be to use ex which has the \< \> syntax to find a word. Also, Perl has \w RE pattern. CaKiwi

 

[jescat]

This should work as well:

BEGIN { pattern = &quot;the&quot; }
$0 ~ pattern {
gsub(&quot; &quot;pattern&quot; &quot;,&quot; |&quot;pattern&quot;| &quot
print $0
}

 

[aigles]

Sorry Jescat, with your solution, when the pattern is at the begining or the end of the line it is not substitued.
Jean Pierre.

 

[jescat]

oh, yes your right. I verified that the solution will work if you check for beginning of line and end of line. However the solution given above is cleaner.

Thanks for the correction

 

[aigles]

Hi,

CaKiwi' solution using ex works fine:

ex -s -c '%s/\<the\>/|the|/g' -c 'wq' test.txt Jean Pierre.

 

[Guest_imported]

So, it's not possible to use the \< \> syntax(cf. aigles' command line ex) in an AWK script for finding a particular word like 'the'?

Stold

 

[aigles]

You can use the \<\> syntax with Gawk (GNU Project's implementation of the AWK programming language.

BEGIN {
matchEre = &quot;\\<&quot; pattern &quot;\\>&quot;
}
{
if (pattern != &quot;&quot {
gsub(matchEre,&quot;|&|&quot;,$0)
print $0
}
}
Jean Pierre.

 

[Guest_imported]

Ok, thanks JP!