How to add lines after a pattern using sed

Question

In shell script, how can I add lines after a certain pattern? Say I have the following file and I want to add two lines after block 1 and blk 2.

abc
def

[block 1]
    apples = 3
    grapes = 4

[blk 2]
    banana = 2
    apples = 3

[block 1] and [blk 2] will be present in the file.

The output I am expecting is below.

abc
def

[block 1]
    oranges = 5
    pears = 2
    apples = 3
    grapes = 4

[blk 2]
    oranges = 5
    pears = 2
    banana = 2
    apples = 3

I thought of doing this with sed. I tried the below command but it does not work on my Mac. I checked these posts but I couldn't find what I am doing wrong.

$sed -i '/\[block 1\]/a\n\toranges = 3\n\tpears = 2' sample2.txt
sed: 1: "sample2.txt": unterminated substitute pattern

How can I fix this? Thanks for your help!

[Edit] I tried the below and these didn't work on my Mac.

$sed -E '/\[block 1\]|\[blk 2\]/r\\n\\toranges = 3\\n\\tpears = 2' sample2.txt
abc
def

    [block 1]
        apples = 3
        grapes = 4

    [blk 2]
        banana = 2
        apples = 3

$sed -E '/\[block 1\]|\[blk 2\]/r\n\toranges = 3\n\tpears = 2' sample2.txt
abc
def

    [block 1]
        apples = 3
        grapes = 4

    [blk 2]
        banana = 2
        apples = 3

Awk attempt:

$awk -v RS= '/\[block 1\]/{$0 = $0 ORS "\toranges = 3" ORS "\tpears = 2" ORS}
      /\[blk 2\]/{$0 = $0 ORS "\toranges = 5" ORS "\tpears = 2" ORS} 1' sample2.txt
abc
def

    [block 1]
        apples = 3
        grapes = 4

    [blk 2]
        banana = 2
        apples = 3
    oranges = 3
    pears = 2

    oranges = 5
    pears = 2

I still get an error. sed '/\[block 1\]/a\n\toranges = 3\n\tpears = 2' sample2.txt sed: 1: "/\[block 1\]/a\n\torang ...": extra characters after \ at the end of a command — SyncMaster, Nov 02 '17 at 21:25

randomir · Accepted Answer · 2017-11-02T21:33:04.813

4

Note that the text provided to the a command has to be on a separate line:

sed '/\[block 1\]/ {a\
\toranges = 3\n\tpears = 2
}' file

and all embedded newlines have to be escaped. Another way to write it (probably more readable):

sed '/\[block 1\]/ {a\
    oranges = 3\
    pears = 2
}' file

Also, consider the r command as an alternative to the a command when larger amounts of text have to be inserted (e.g. more than one line). It will read data from a text file provided:

sed '/\[block 1\]/r /path/to/text' file

To handle multiple sections with one sed program, you can use the alternation operator (available in ERE, notice the -E flag):

sed -E '/\[block 1\]|\[blk 2\]/r /path/to/text' file

edited Nov 02 '17 at 21:33

answered Nov 02 '17 at 21:25

randomir

17,989
1
40
55

Or you can use `sed '/\[bl$k\|ock$ [12]\]/ {a\ ...` (and 4-spaces instead of a `\t`) – David C. Rankin Nov 02 '17 at 21:41
I tried both these and the file didn't get updated. 1. sed -E '/\[block 1\]|\[blk 2\]/r\\n\\toranges = 3\\n\\tpears = 2' sample2.txt 2. sed -E '/\[block 1\]|\[blk 2\]/r\n\toranges = 3\n\tpears = 2' sample2.txt Am I missing something? – SyncMaster Nov 02 '17 at 21:48
@DavidC.Rankin, Actually, (escaped) alternation operator (`\|`) in BRE is a [GNU `sed` extension](https://www.gnu.org/software/sed/manual/sed.html#index-GNU-extensions_002c-to-basic-regular-expressions-4). In [POSIX](http://pubs.opengroup.org/onlinepubs/009695399/basedefs/xbd_chap09.html), only ERE allows alternation. – randomir Nov 02 '17 at 21:48
@SyncMaster, if you want an inplace update, add the `-i` flag. – randomir Nov 02 '17 at 21:49
@SyncMaster, if you want to use `r` command, store the text to be added into a separate file, then reference it with `r file` (see my answer, `/path/to/text` is path to your text file, the contents of which will be inserted). – randomir Nov 02 '17 at 21:55
1

@SyncMaster, or just try the first (or second) example. You can simply copy & paste it, without creating an intermediate file you need for the last two examples. – randomir Nov 02 '17 at 22:38
ANSI-C quoting can be used to insert newlines: $'\n' – Joe Bowbeer Aug 07 '19 at 10:15

score 1 · Answer 2 · answered Nov 02 '17 at 21:25

1

This awk should work with empty RS. This breaks each block into a single record.

awk -v RS= '/\[block 1\]/{$0 = $0 ORS "\toranges = 3" ORS "\tpears = 2" ORS} 
      /\[blk 2\]/{$0 = $0 ORS "\toranges = 5" ORS "\tpears = 2" ORS} 1' file

abc
def
[block 1]
    apples = 3
    grapes = 4
    oranges = 3
    pears = 2

[blk 2]
    banana = 2
    apples = 3
    oranges = 5
    pears = 2

answered Nov 02 '17 at 21:25

anubhava

761,203
64
569
643

I didn't get the expected output. The new lines did not get inserted in [block 1]. This is what I got when I ran this. $awk -v RS= '/\[block 1\]/{$0 = $0 ORS "\toranges = 3" ORS "\tpears = 2" ORS} > /\[blk 2\]/{$0 = $0 ORS "\toranges = 5" ORS "\tpears = 2" ORS} 1' sample2.txt abc def [block 1] apples = 3 grapes = 4 [blk 2] banana = 2 apples = 3 oranges = 3 pears = 2 oranges = 5 pears = 2 – SyncMaster Nov 02 '17 at 21:42
1

Strange... OSX should use `'\n'` as it's end of line -- the same as Linux (unlike pre-OSX that used `'\r'` instead) – David C. Rankin Nov 02 '17 at 21:45
Same issue on my Linux machine too – SyncMaster Nov 02 '17 at 22:04
1

@SyncMaster: This has been tested on OSX awk. [Here is a working demo](https://ideone.com/E6lqAH) – anubhava Nov 03 '17 at 04:32
Also make sure your input file does not have DOS line ending (carriage returns). Check with `cat -A sample2.txt` command. If it shows `^M` at the end of line then it means a DOS file that needs to be converted to Unix file – anubhava Nov 03 '17 at 04:39

score 1 · Answer 3 · answered Nov 03 '17 at 08:02

This might work for you (GNU sed):

sed '/^\[\(block 1\|blk 2\)\]\s*$/{n;h;s/\S.*/oranges = 5/p;s//pears = 2/p;x}' file

Locate the required match, print it and then store the next line in the hold space. Replace the first non-space character to the end of the line with the first required line, repeat for the second required string and then revert to the original line.

How to add lines after a pattern using sed

3 Answers3