Because the question can be misleading, here is a little example. I have this kind of file:
some text
some text ##some-text-KEY-some-other-text##
text again ##some-text-KEY-some-other-text## ##some-text-KEY-some-other-text##
again ##some-text-KEY-some-other-text-KEY-text##
some text with KEY ##KEY-some-text##
blabla ##KEY##
In this example, I want to replace each occurrence of KEY- inside a pair of ## by VALUE-. I started with this sed command:
sed -i 's/\(##[^#]*\)KEY-\([^#]*##\)/\1VALUE-\2/g'
Here is how it works:
\(##[^#]*\): create a first group composed of two # and any characters except # ...
KEY-: ... until the last occurrence of KEY- on that line
\([^#]*##\): and create a second group with all the characters except # until the next pair of #.
The problem is my command can't handle correctly the following line because there are multiple KEY- inside my pair of ##:
again ##some-text-KEY-some-other-text-KEY-text##
Indeed, I get this result:
again ##some-text-KEY-some-other-text-VALUE-text##
If I want to replace all the occurrences of KEY- in that line, I have to run my command multiple times and I prefer to avoid that. I also tried with lazy operators but the problem is the same.
How can I create a regex and a sed command who can handle correctly all my file?
The problem is rather complex: you need to replace all occurrences of some multicharacter text inside blocks of text between identical multicharacter delimiters.
The easiest and safest way to solve the task is using Perl:
perl -i -pe 's/(##)(.*?)(##)/$end_delim=$3; "$1" . $2=~s|KEY-|VALUE-|gr . "$end_delim"/ge' file
See the online demo.
The (##)(.*?)(##) pattern will match strings between two adjacent ## substrings capturing the start delimiter into Group 1, end delimiter in Group 3, and all text in between into Group 2. Since the regex substitution re-sets all placeholders, the temporary variable is used to keep the value of the end delimiter ($end_delim=$3), then, "$1" . $2=~s|KEY-|VALUE-|gr . "$end_delim" replaces the match with the value in the Group 1 of the first match (the first ##), then the Group 2 value with all KEY- replaced with VALUE-, and then the end delimiter.
If there are no KEY-s in between matches on the same line you may use a branch with sed by enclosing your command with :A and tA:
sed -i ':A; s/\(##[^#]*\)KEY-\([^#]*##\)/\1VALUE-\2/g; tA' file
Note you missed the first placeholder in \VALUE-\2, it should be \1VALUE-\2.
See the online demo:
s="some KEY- text
some text ##some-text-KEY-some-other-text##
text again ##some-text-KEY-some-other-text## ##some-text-KEY-some-other-text##
again ##some-text-KEY-some-other-text-KEY-text##
some text with KEY ##KEY-some-text##
blabla ##KEY##"
sed ':A; s/\(##[^#]*\)KEY-\([^#]*##\)/\1VALUE-\2/g; tA' <<< "$s"
Output:
some KEY- text
some text ##some-text-VALUE-some-other-text##
text again ##some-text-VALUE-some-other-text## ##some-text-VALUE-some-other-text##
again ##some-text-VALUE-some-other-text-VALUE-text##
some text with KEY ##VALUE-some-text##
blabla ##KEY##
More details:
sed allows the usage of loops and branches. The :A in the code above is a label, a special location marker that can be "jumped at" using the appropriate operator. t is used to create a branch, this "command jumps to the label only if the previous substitute command was successful". So, once the pattern matched and the replacement occurred, sed goes back to where it was and re-tries a match. If it is not successful, sed goes on to search for the matches further in the string. So, tA means go back to the location marked with A if there was a successful search-and-replace operation.
This might work for you (GNU sed):
sed -E 's/##/\n/g;:a;s/^([^\n]*(\n[^\n]*\n[^\n]*)*\n[^\n]*)KEY-/\1VALUE-/;ta;s/\n/##/g' file
Convert ##'s to newlines. Using a loop, replace VAL- between matched newlines to VALUE-. When all done replace newlines by ##'s.
Related
I have a file foo.properties with contents like
foo=bar
# another property
test=true
allNames=alpha:.02,beta:0.25,ph:0.03,delta:1.0,gamma:.5
In my script, I need to replace whatever value is against ph (The current value is unknown to the bash script) and change it to 0.5. So the the file should look like
foo=bar
# another property
test=true
allNames=alpha:.02,beta:0.25,ph:0.5,delta:1.0,gamma:.5
I know it can be easily done if the current value is known by using
sed "s/\,ph\:0.03\,/\,ph\:0.5\,/" foo.properties
But in my case, I have to actually read the contents against allNames and search for the value and then replace within a for loop. Rest all is taken care of but I can't figure out the sed/perl command for this.
I tried using sed "s/\,ph\:.*\,/\,ph\:0.5\,/" foo.properties and some variations but it didn't work.
A simpler sed solution:
sed -E 's/([=,]ph:)[0-9.]+/\10.5/g' file
foo=bar
# another property
test=true
allNames=alpha:.02,beta:0.25,ph:0.5,delta:1.0,gamma:.5
Here we match ([=,]ph:) (i.e. , or = followed by ph:) and capture in group #1. This should be followed by 1+ of [0-9.] character to natch any number. In replacement we put \1 back with 0.5
With your shown samples, please try following awk code.
awk -v new_val="0.5" '
match($0,/,ph:[0-9]+(\.[0-9]+)?/){
val=substr($0,RSTART+1,RLENGTH-1)
sub(/:.*/,":",val)
print substr($0,1,RSTART) val new_val substr($0,RSTART+RLENGTH)
next
}
1
' Input_file
Detailed Explanation: Creating awk's variable named new_val which contains new value which needs to put in. In main program of awk using match function of awk to match ,ph:[0-9]+(\.[0-9]+)? regex in each line, if a match of regex is found then storing that matched value into variable val. Then substituting everything from : to till end of value in val variable with : here. Then printing values as pre requirement of OP(values before matched regex value with val(edited matched value in regex) with new value and rest of line), using next will avoid going further and by mentioning 1 printing rest other lines which are NOT having a matched value in it.
2nd solution: Using sub function of awk.
awk -v newVal="0.5" '/^allNames=/{sub(/,ph:[^,]*/,",ph:"newVal)} 1' Input_file
Would you please try a perl solution:
perl -pe '
s/(?<=\bph:)[\d.]+(?=,|$)/0.5/;
' foo.properties
The -pe option makes perl to read the input line by line, perform
the operation, then print it as sed does.
The regex (?<=\bph:) is a zero-length lookbehind which matches
the string ph: preceded by a word boundary.
The regex [\d.]+ will match a decimal number.
The regex (?=,|$) is a zero-length lookahead which matches
a comma or the end of the string.
As the lookbehind and the lookahead has zero length, they are not
substituted by the s/../../ operator.
[Edit]
As Dave Cross comments, the lookahead (?=,|$) is unnecessary as long as the input file is correctly formatted.
Works with decimal place or not, or no value, anywhere in the line.
sed -E 's/(^|[^-_[:alnum:]])ph:[0-9]*(.[0-9]+)?/ph:0.5/g'
Or possibly:
sed -E 's/(^|[=,[:space:]])ph:[0-9]+(.[0-9]+)?/ph:0.5/g'
The top one uses "not other naming characters" to describe the character immediately before a name, the bottom one uses delimiter characters (you could add more characters to either). The purpose is to avoid clashing with other_ph or autograph.
Here you go
#!/usr/bin/perl
use strict;
use warnings;
print "\nPerl Starting ... \n\n";
while (my $recordLine =<DATA>)
{
chomp($recordLine);
if (index($recordLine, "ph:") != -1)
{
$recordLine =~ s/ph:.*?,/ph:0.5,/g;
print "recordLine: $recordLine ...\n";
}
}
print "\nPerl End ... \n\n";
__DATA__
foo=bar
# another property
test=true
allNames=alpha:.02,beta:0.25,ph:0.03,delta:1.0,gamma:.5
output:
Perl Starting ...
recordLine: allNames=alpha:.02,beta:0.25,ph:0.5,delta:1.0,gamma:.5 ...
Perl End ...
Using any sed in any shell on every Unix box (the other sed solutions posted that use sed -E require GNU or BSD seds):
a) if ph: is never the first tag in the allNames list (as shown in your sample input):
$ sed 's/\(,ph:\)[^,]*/\10.5/' foo.properties
foo=bar
# another property
test=true
allNames=alpha:.02,beta:0.25,ph:0.5,delta:1.0,gamma:.5
b) or if it can be first:
$ sed 's/\([,=]ph:\)[^,]*/\10.5/' foo.properties
foo=bar
# another property
test=true
allNames=alpha:.02,beta:0.25,ph:0.5,delta:1.0,gamma:.5
I have a test property file with this in it:
-config.test=false
config.test=false
I'm trying to, using sed, update the values of these properties whether they have the - in front of them or not. Originally I was using this, which worked:
sed -i -e "s/#*\(config.test\)\s*=\s*\(.*\)/\1=$(echo "true" | sed -e 's/[\/&]/\\&/g')/" $FILE_NAME
However, since I was basically ignoring all characters before the match, I found that when I had properties with keys that ended in the same value, it'd give me problems. Such as:
# The regex matches both of these
config.test=true
not.config.test=true
Is there a way to either ignore the first character for a match or ignore the initial - specifically?
EDIT:
Adding a little clarification in terms of what I'd want the regex to match:
config.test=false # Should match
-config.test=false # Should match
not.config.test=false # Should NOT match
sed -E 's/^(-?config\.test=).*/\1true/' file
? means zero or 1 repetitions of so it means the - can be present or not when matching the regexp.
I found some solution for a regex of a specific length instead of ignoring the first character with sed and awk. Sometimes the opposite does the same by an easier way.
If you only have the alternative to use sed I have two workaround depending on your file.
If your file looks like this
$ cat file
config.test=false
-config.test=false
not.config.test=false
you can use this one-liner
sed 's/^\(.\{11,12\}=\)\(.*$\)/\1true/' file
sed is looking at the beginning ^ of each line and is grouping \( ... \) for later back referencing every character . that occurs 11 or 12 times \{11,12\} followed by a =.
This first group will be replaced with the back reference \1.
The second group that match every character after the = to the end of line \(.*$\) will be dropped. Instead of the second group sed replaces with your desired string true.
This also means, that every character after the new string true will be chopped.
If you want to avoid this and your file looks like
$ cat file
config.test=true # Should match
-config.test=true # Should match
not.config.test=false # Should NOT match
you can use this one-liner
sed 's/^\(.\{11,12\}=\)\(false\)\(.*$\)/\1true\3/' file
This is like the example before but works with three groups for back referencing.
The content of the former group 2 is now in group 3. So no content after a change from false to true will be chopped.
The new second group \(false\) will be dropped and replaced by the string true.
If your file looks like in the example before and you are allowed to use awk, you can try this
awk -F'=' 'length($1)<=12 {sub(/false/,"true")};{print}'
For me this looks much more self-explanatory, but is up to your decision.
In both sed examples you invoke only one time the sed command which is always good.
The first sed command needs 39 and the second 50 character to type.
The awk command needs 52 character to type.
Please tell me if this works for you or if you need another solution.
I have a csv file and I am trying to delete all characters from the beginning of the line till it finds the first occurrence of "2015". I want to do this for each line in the csv file.
My csv file structure is as follows:
Field1 , Field2 , Field3 , Field4
sometext1 , 2015-07-15 , sometext2, sometext3
sometext1 , 2015-07-14 , sometext2, sometext3
sometext1 , 2015-07-13 , sometext2, sometext3
I cannot use the cut command or sed for the first occurrence of a comma because the text in the Field1 sometimes has commas in them too, which is making it complicated for parsing. I figured if I search for the first occurrence of the text 2015 for each line and replace all the preceding characters with nothing, then that should work.
FYI I only want to do this for the FIRST occurrence of 2015 only. There is another text field with 2015 in it within another column and I don't any text prior to that to be affected.
For example, if my original line is:
sometext1,#015,2015-07-10,sometext2,2015,sometext3
I want it to return:
2015-07-10,sometext2,2015,sometext3
Does anyone know the sed command to do this?
Any help will be appreciated!
Thanks
Here is a way to do it with sed assuming "#####" never occurs in a line:
sed -e 's/2015/#####&/'|sed -e 's/.*#####//'
For example:
> echo sometext1,#015,2015-07-10,sometext2,2015,sometext3\
|sed -e 's/2015/#####&/'|sed -e 's/.*#####//'
2015-07-10,sometext2,2015,sometext3
The first sed command prefixes "#####" to the first occurence of 2015 and the second sed command removes everything from the beginning to the end of the "#####" prefix.
The basic reason for using this two stage method is that sed's regular expression matcher has only greedy wildcards that always pick the longest match and does not support lazy matching which picks the shortest match.
If "#####" may occur in a line a more unlikely string could be substituted for it such as "7z#dNjm_wG8a3!esu#Rhv=".
To do this with sed without Perl-style non-greedy operators, you need to mark the first instance with something you know won't be in the line, as Tris describes. However, that solution requires knowledge of what won't be in the file. Fortunately, you can guarantee that a newline won't be in the line because that's what terminated the line. Thus you can do something like:
sed 's/2015/\n&/;s/.*\n//' input.txt > output.txt
NOTE: this won't modify the header row which you would have to treat specially.
Here are the contents of my text file named 'temp.txt'
---start of file ---
HEROKU_POSTGRESQL_AQUA_URL (DATABASE_URL) ----backup---> b687
Capturing... done
Storing... done
---end of file ----
I want to write a bash script in which I need to capture the string 'b687' in a variable. this is really a pattern (which is the letter 'b' followed by 'n' number of digits). I can do it the hard way by looping through the file and extracting the desired string (b687 in example above). Is there an easy way to do so? Perhaps by using awk or sed?
Try using grep
v=$(grep -oE '\bb[0-9]{3}\b' file)
This will seach for a word starting with b followed by '3' digits.
regex101 demo
Using sed
v=$(sed -nr 's/.*\b(b[0-9]{3})\b.*/\1/p' file)
varname=$(awk '/HEROKU_POSTGRESQL_AQUA_URL/{print $4}' filename)
what this does is reads the file when it matches the pattern HEROKU_POSTGRESQL_AQUA_URL print the 4th token in this case b687
your other option is to use sed
varname=$(sed -n 's/.* \(b[0-9][0-9]*\)/\1/p' filename)
In this case we are looking for the pattern you mentioned b####... and only print that pattern the -n tells sed not to print line that do not have that pattern. the rest of the sed command is a substitution .* is any string at the beginning. followed by a (...) which forms a group in which we put the regex that will match your b##### the second part says out of all that match only print the group 1 and the p at the end tells sed to print the result (since by default we told sed not to print with the -n)
I have a file that looks like this:
#"Afghanistan.png",
#"Albania.png",
#"Algeria.png",
#"American_Samoa.png",
I want it to look like this
#"Afghanistan.png",
#"Afghanistan",
#"Albania.png",
#"Albania",
#"Algeria.png",
#"Algeria",
#"American_Samoa.png",
#"American_Samoa",
I thought I could use sed to do this but I can't figure out how to store something in a buffer and then modify it.
Am I even using the right tool?
Thanks
You don't have to get tricky with regular expressions and replacement strings: use sed's p command to print the line intact, then modify the line and let it print implicitly
sed 'p; s/\.png//'
Glenn jackman's response is OK, but it also doubles the rows which do not match the expression.
This one, instead, doubles only the rows which matched the expression:
sed -n 'p; s/\.png//p'
Here, -n stands for "print nothing unless explicitely printed", and the p in s/\.png//p forces the print if substitution was done, but does not force it otherwise
That is pretty easy to do with sed and you not even need to use the hold space (the sed auxiliary buffer). Given the input file below:
$ cat input
#"Afghanistan.png",
#"Albania.png",
#"Algeria.png",
#"American_Samoa.png",
you should use this command:
sed 's/#"\([^.]*\)\.png",/&\
#"\1",/' input
The result:
$ sed 's/#"\([^.]*\)\.png",/&\
#"\1",/' input
#"Afghanistan.png",
#"Afghanistan",
#"Albania.png",
#"Albania",
#"Algeria.png",
#"Algeria",
#"American_Samoa.png",
#"American_Samoa",
This commands is just a replacement command (s///). It matches anything starting with #" followed by non-period chars ([^.]*) and then by .png",. Also, it matches all non-period chars before .png", using the group brackets \( and \), so we can get what was matched by this group. So, this is the to-be-replaced regular expression:
#"\([^.]*\)\.png",
So follows the replacement part of the command. The & command just inserts everything that was matched by #"\([^.]*\)\.png", in the changed content. If it was the only element of the replacement part, nothing would be changed in the output. However, following the & there is a newline character - represented by the backslash \ followed by an actual newline - and in the new line we add the #" string followed by the content of the first group (\1) and then the string ",.
This is just a brief explanation of the command. Hope this helps. Also, note that you can use the \n string to represent newlines in some versions of sed (such as GNU sed). It would render a more concise and readable command:
sed 's/#"\([^.]*\)\.png",/&\n#"\1",/' input
I prefer this over Carles Sala and Glenn Jackman's:
sed '/.png/p;s/.png//'
Could just say it's personal preference.
or one can combine both versions and apply the duplication only on lines matching the required pattern
sed -e '/^#".*\.png",/{p;s/\.png//;}' input