SED command to replace test with newline break

Question

I'm in need of some help, I need to replace text on a page using SED command with other text but it just will not work for me.

Need to replace this one:

            <key>disableMailRecentsSyncing</key>
            <true />
            <key>allowMailDrop</key>
            <false />
            <key>PreventMove</key>
            <true />

With this one:

            <key>disableMailRecentsSyncing</key>
            <false />
            <key>allowMailDrop</key>
            <true />
            <key>PreventMove</key>
            <false />

I've tried the following but it will not work:

sed -i 's/<key>disableMailRecentsSyncing</key>                 <true />                 <key>allowMailDrop</key>                 <false />                 <key>PreventMove</key>                 <true />/<key>disableMailRecentsSyncing</key>                 <false />                 <key>allowMailDrop</key>                 <true />                 <key>PreventMove</key>                 <false />/g' input.txt

Does this answer your question? [SED match pattern in two lines](https://stackoverflow.com/questions/38367648/sed-match-pattern-in-two-lines) — Barmar, Sep 23 '21 at 16:37

score 2 · Answer 1 · answered Sep 23 '21 at 16:45

For editing XML, use an XML-aware tool. For example, negating all the listed keys can be done in XSH (a wrapper around XML::LibXML) using the following command:

rename xsh:if(self::true, "false", "true")
       (//false[preceding-sibling::key[1]="allowMailDrop"]
       | //true[preceding-sibling::key[1]="PreventMove"
                or preceding-sibling::key[1]="disableMailRecentsSyncing"]) ;

Note that I'm the current maintainer of the tool.

score 0 · Answer 2 · answered Sep 23 '21 at 17:29

Assumptions:

data is nicely formatted as in the question (otherwise a proper XML/HTML-aware tool may be easier to use than re-inventing a parser)
objective is to toggle the current value (ie, true becomes false and false becomes true)
true/false values are all lowercase
true/false are always preceded by a <

With a view towards an awk solution I'd want the patterns (to search for) placed into a file as this allows for flexibility without having to hardcode the solution, eg:

$ cat key.list
disableMailRecentsSyncing
allowMailDrop
PreventMove

My input file with some additional data:

$ cat input.data
<key>disableMailRecentsSyncing</key>       # match but ...
<sometimes true />                         #    leave "true" alone

<key>disableMailRecentsSyncing</key>       # match so ...
<true />                                   #   switch to "false"
<key>allowMailDrop</key>                   # match so ...
<false />                                  #   switch to "true"
<key>PreventMove</key>                     # match so ...
<true />                                   #   switch to "false"

<key>allowMailDrop</key>                   # match but ...
<Tuesday />                                #   ignore

One awk idea:

awk '
FNR==NR    { keys[$1]; next }
           { split($0,arr,"[<>]") }
toggle     { if ( arr[2] ~ /^true/  ) gsub(/<true/, "<false")
             if ( arr[2] ~ /^false/ ) gsub(/<false/,"<true" )
             toggle=0
           }
           { if ( arr[3] in keys) toggle=1 }
1
' key.list input.data

This generates:

<key>disableMailRecentsSyncing</key>
<sometimes true />

<key>disableMailRecentsSyncing</key>
<false />
<key>allowMailDrop</key>
<true />
<key>PreventMove</key>
<false />

<key>allowMailDrop</key>
<Tuesday />

Due to the use of 2 input files OP will not be able to use (GNU) awk -i inplace so the output will need to be saved to a temp file and then copying/moving the temp file to replace the current file.

score 0 · Answer 3 · answered Sep 23 '21 at 17:40

This might work for you (GNU sed):

cat <<\! >matchFile
        <key>disableMailRecentsSyncing</key>
        <true />
        <key>allowMailDrop</key>
        <false />
        <key>PreventMove</key>
        <true />
!
cat <<\! >replaceFile
        <key>disableMailRecentsSyncing</key>
        <false />
        <key>allowMailDrop</key>
        <true />
        <key>PreventMove</key>
        <false />
!
cat file <(echo MATCH) matchFile <(echo REPLACE) replaceFile |
sed -Ez ':a;s/(.*)(.*MATCH\1REPLACE(.*))/\3\2/;ta;s/(.*)MATCH.*//'

To replace the original file, use:

sed -E -i 'H;1h;$!d;x;s/$/\n$(echo MATCH;cat matchFile;echo REPLACE;cat replaceFile)/
           s/.*/echo "&"/e;:a;s/(.*)(.*MATCH\1REPLACE(.*))/\3\2/;ta;s/(.*)MATCH.*/\1/' file

The solution appends a delimiter MATCH followed by the matchFile, followed by a second delimiter REPLACE followed by the replaceFile.

Then using a loop and pattern matching (involving back references) the matchFile is matched against places in the original file and replaced by the text in the replaceFile.

SED command to replace test with newline break

3 Answers3