Agfdhyk

I'm an awk newbie. I have a file that looks like:

beans and celery  

beans and oatmeal  

beans and beans  

quinoa

<fo:external-graphic width="auto" height="auto" content-width="36pt" src="url(file:/C:/Users/xxx/images/tip.svg)"/>

<fo:external-graphic src="url(images/image1.png)" width="6.3in" height="auto" content-width="246px" content-height="322px"/>

I'm trying to perform a search and replace in-place for the "fo" tag. I want to capture the beginning of the tag, as well as the "src" parameter. Please note that the position of the src tag varies from line to line!

I've been able to get the fields I want using the following:

awk '/<fo:external-graphic.*/ {for (i=1; i<=NF; ++i) {if ($i ~ "src") print $1 " " $i}}' inventory.txt

How can I do an in-place substitution of this?
I also want to append a string to new contents of the line. I've tried:

awk '/<fo:external-graphic.*/ {for (i=1; i<=NF; ++i) {if ($i ~ "src") print $1 " " $i "misc stuff here"}}' inventory.txt

But it completely messes up the order of the resulting string, which I want to be of the form:

<fo:external-graphic src="url(images/image1.png)" misc stuff here

PS1:
Further clarification about what result I want:
The file contains strings like:

<fo:external-graphic width="auto" height="auto" content-width="36pt" src="url(file:/C:/Users/xxx/images/tip.svg)"/>

<fo:external-graphic src="url(images/image1.png)" width="6.3in" height="auto" content-width="246px" content-height="322px"/>

I want to process these in and get an output like:

<fo:external-graphic src="url(images/image1.png)" _completely new stuff here, till end of string_ />

for example:

<fo:external-graphic src="url(images/image1.png)" age="25" sex="M" />

I want the result to ALWAYS begin with:

<fo:external-graphic src="url(images/image1.png)"

then the extra stuff eg:

age="25" sex="M" />

No other part of the original string is needed in the final output.

PS2: Can I pack all this into a gsub? To the best of my knowledge, gsub only take two arguments. I've tried to make a complex expression, for the replace argument, but it keeps failing eg:

gawk '/<fo:external-graphic.*/ {for (i=1; i<=NF; ++i) {if ($i ~ "src") gsub($0, "boy band"); {print}}}' inventory.txt > testres

PS3: This is just a newbie observation, maybe I'm wrong. Consider a file with the following contents:

Donald Trump

Donald Duck

George Bush

Steve Austin

The regexp to search for all lines that begin with Donald is:

/^Donald/

If I want to replace all occurrences of "Donald" with "Barrack", I could do the following:

gawk -i inplace '{ gsub(/^Donald/, "Barrack"); { print } }' FILENAME

If I want to completely change all lines that contain "Donald" I would do:

gawk -i inplace '{ gsub(/^Donald.*/, "Barrack"); { print } }' FILENAME

gawk and gsub appear to only replace the span or whatever part of the string matches the given regexp. Thus if I want to completely change a whole line, my regexp should span the whole of that line.

PS4: Just to clear any ambiguities about the solution I expect. Given the following file:

<fo:external-graphic width="auto" height="auto" content-width="36pt" src="url(file:/C:/Users/xxx/images/tip.svg)"/>

<fo:external-graphic width="6.3in" height="auto" src="url(images/image1.png)" content-width="246px" content-height="322px"/>

<fo:external-graphic src="url(images/image1.png)" width="6.3in" content-width="246px" content-height="322px"/>

I'm looking for an awk/gawk solution that will replace this file with:

<fo:external-graphic src="url(file:/C:/Users/xxx/images/tip.svg)" age="25" sex="M" />

<fo:external-graphic src="url(images/image1.png)" age="25" sex="M"/>

<fo:external-graphic src="url(images/image1.png)" age="25" sex="M"/>

The target file must be changed.

Your attempt is right, but assuming your intention is add only on the word starting with src i.e. on $i, apply the action only for that field, and keep the other fields as-is

awk '/<fo:external-graphic.*/ {for (i=1; i<=NF; ++i) {if ($i ~ "src") $i = $i " misc stuff here"}}1' inventory.txt

The part $i = $i " misc stuff here" does action of appending the string only on the fields matching your regex condition. Notice the removal of print and appending of {..}1 at the end. What that basically means is re-construct the whole line based on the modifications done inside {..}. Since we are modifying only certain fields, the other ones are kept intact.

If you want to re-write the entire field starting with src and append some string, use a proper regex match with gsub() and append string after the matched text denoted by &

awk '/<fo:external-graphic.*/ {for (i=1; i<=NF; ++i) { if ($i ~ "src") gsub(/src="url([^"]*)"/, "& new string", $i ) }}1' inventory.txt

From OP's most recent edit, it seems OP just wants the src field modified with new string to be appended at the end. The rest of the fields can be ignored it seems. Using match() on GNU awk has an added advantage of adding a third parameter to store the captured groups as

awk -v newstr="age="25" sex="M"" 'match($0, /^(<fo:external-graphic).*(src="url([^"]*)").*(/>)$/, arr){ print arr[1]" "arr[2]" "newstr""arr[4]  }' file

Note that GNU awk did not have in-place modifications up until 4.1.2, on which you can just do

gawk -i inplace '{...}' inventory.txt

For versions earlier to that, use a temporary file

awk '{...}' inventory.txt > tmpfile && mv tmpfile inventory.txt

Or if you moreutils installed, use sponge to slurp the output from the first command and re-create the file back with the latest.

awk '{...}' inventory.txt | sponge inventory.txt

From your sample text on the question, one can't recognize if its some markup language of sorts (XML, HTML). If its a proper syntax aware language, then you should use a parser that knows that grammar.

EDIT3: Adding one more code here as per OP's new edit.

awk  '

/ width.*content-width.*src/{

  sub(/ width.*content-width.*src/," src")

  sub(//>$/," age="25" sex="M"&")

}

/src.*width/{

  match($0,/src[^)]*/)

  val=substr($0,RSTART,RLENGTH+2)

  sub(/src.*/,"")

  $0=$0 OFS val OFS "age="25" sex="M"/>"

}

1

'  Input_file

EDIT2: For changing complete line with respect to OP's PS3 could you please try following.

awk '/^Donald/{$0="new_line_value"} 1'  Input_file

new_line_value

new_line_value

George Bush

Steve Austin

EDIT: Since OP has changed expected output so adding solution as per that output now too.

awk '/^<fo:external-graphic src=.*/ && match($0,/src=.*)"/){$0=substr($0,1,RSTART+RLENGTH) " new_value_bla_bla_here.. />"} 1' Input_file

Could you please try following(haven't tested it thoroughly since your expected output is not clear).

awk '

/^<fo:/ && match($0,/src=.*>/){

  $0=substr($0,1,RSTART-1) OFS "new_value_here.." OFS substr($0,RSTART+RLENGTH+1)

}

1

'  Input_file

In this code checking a line which starts from <fo: string and then trying to catch string from src= till / by match and replacing that captured text with new string here.

In case you want to save output into Input_file itself then append > temp_file && mv temp_file Input_file in above code too.