I need to replace a numeric value, that occurs in a specific line of a series of config files in a pattern like this:
string number_1 number_to_replace number_2
I want to obtain something like this:
string number_1 number_replaced number_2
The difficulties I encountered are:
number_1 or number_2 can be equal to number_to_replace, so a simple replacement is not possible.
number_1 and number_2 vary between config files so I don't know them in advance.
The closest attempt I got until now is:
echo "field 4 4 4" | sed 's/\s4\s/3/'
Which ouputs:
field34 4
This is close, given that I want to replace the intermediate number I added another "\s" to try to use the known fact that the line starts with a character.
echo "field 4 4 4" | sed 's/\s\s4\s/3/'
Which gives:
field 4 4 4
So, nothing is replaced this time. How can I proceed? A somewhat detailed explanation would be ideal, because my knowledge of replacing expressions that involve patterns in nearly zero.
Thanks.
You can do something like below, which matches your exact sequence of digits as in the example. You could replace 3 with any digit of your choice.
sed 's/\([0-9]\{1,\}\)[[:space:]]\([0-9]\{1,\}\)[[:space:]]\([0-9]\{1,\}\)/\1 3 \3/'
Notice that I've used the POSIX bracket expression to match the whitespace character which should be supported in any variant of sed you are using. Note that \s is supported in only the GNU variants.
The literal meaning of the regex definition is to match a single digit followed by a space, then a digit and space and another digit. The captured groups are stored from \1. Since your intention is to remove the 2nd digit, you replace that with the word of your choice.
If the extra escapes causes it unreadable, use the -E flag for extended regex support. I've used the default BRE version
Related
I have a file in which I want to replace the "_" string with "-" in cases where it makes up a part of my gene name. Examples of the gene names and my intended output are:
aa1c1_123 -> aa1c1-123
aa1c2_456 -> aa1c1-456
aa1c10_789 -> aa1c1-789
In essence, the first four characters are fixed, followed by 1 or 2 characters depending on the chromosome, an underscore and then the remainder of the gene ID which could vary in length and character. Important is that there are other strings in this gene information column contains other strings with underscores (e.g. "gene_id", "transcript_id", "five_prime_utr") so using sed -i.bak s/_/-/g' file.gtf
can't be done.
Perhaps not the most elegant way, but this should work:
sed -i.bak 's/\([0-9a-z]\{4\}[0-9][0-9]\?\)_/\1-/g' file.gtf
i.e. capture a group (referenced by \1 in the substitution) of 4 characters consisting of lower case letters and digits followed by exactly one digit and perhaps another digit, which is followed by an underscore; if found, replace it by the group's content and a dash. This should exclude your other occurrences consisting of only characters and an underscore.
I've got an application that has no useful api implemented, and the only way to get certain information is to parse string output. This is proving to be very painful...
I'm trying to achieve this in bash on SLES12.
Given I have the following strings:
QMNAME(QMTKGW01) STATUS(Running)
QMNAME(QMTKGW01) STATUS(Ended normally)
I want to extract the STATUS value, ie "Ended normally" or "Running".
Note that the line structure can move around, so I can't count on the "STATUS" being the second field.
The closest I have managed to get so far is to extract a single word from inside STATUS like so
echo "QMNAME(QMTKGW01) STATUS(Running)" | sed "s/^.*STATUS(\(\S*\)).*/\1/"
This works for "Running" but not for "Ended normally"
I've tried switching the \S* for [\S\s]* in both "grep -o" and "sed" but it seems to corrupt the entire regex.
This is purely a regex issue, by doing \S you requested to match non-white space characters within (..) but the failing case has a space between which does not comply with the grammar defined. Make it simple by explicitly calling out the characters to match inside (..) as [a-zA-Z ]* i.e. zero or more upper & lower case characters and spaces.
sed 's/^.*STATUS(\([a-zA-Z ]*\)).*/\1/'
Or use character classes [:alnum:] if you want numbers too
sed 's/^.*STATUS(\([[:alnum:] ]*\)).*/\1/'
sed 's/.*STATUS(\([^)]*\)).*/\1/' file
Output:
Running
Ended normally
Extracting a substring matching a given pattern is a job for grep, not sed. We should use sed when we must edit the input string. (A lot of people use sed and even awk just to extract substrings, but that's wasteful in my opinion.)
So, here is a grep solution. We need to make some assumptions (in any solution) about your input - some are easy to relax, others are not. In your example the word STATUS is always capitalized, and it is immediately followed by the opening parenthesis (no space, no colon etc.). These assumptions can be relaxed easily. More importantly, and not easy to work around: there are no nested parentheses. You will want the longest substring of non-closing-parenthesis characters following the opening parenthesis, no mater what they are.
With these assumptions:
$ grep -oP '\bSTATUS\(\K[^)]*(?=\))' << EOF
> QMNAME(QMTKGW01) STATUS(Running)
> QMNAME(QMTKGW01) STATUS(Ended normally)
> EOF
Running
Ended normally
Explanation:
Command options: o to return only the matched substring; P to use Perl extensions (the \K marker and the lookahead). The regexp: we look for a word boundary (\b) - so the word STATUS is a complete word, not part of a longer word like SUBSTATUS; then the word STATUS and opening parenthesis. This is required for a match, but \K instructs that this part of the matched string will not be returned in the output. Then we seek zero or more non-closing-parenthesis characters ([^)]*) and we require that this be followed by a closing parenthesis - but the closing parenthesis is also not included in the returned string. That's a "lookahead" (the (?= ... ) construct).
I have files with many lines of the following form:
word -0.15636028 -0.2953045 0.29853472 ....
(one word preceding several hundreds floats delimited by blanks)
Due to some errors out of my control, the word sometimes has spaces in it.
a bbb c -0.15636028 -0.2953045 0.29853472 .... (several hundreds floats)
which I wish to substitute by underscores so to get:
a_bbb_c -0.15636028 -0.2953045 0.29853472 .... (several hundreds floats)
have tried for each line the following substitution code:
s/\s(?=(\s-?\d\.\d+)+)/_/g;
So lookarounds is apparently not the solution.
I'd be grateful for any clues.
Your idea for the lookahead is fine, but the question is how to replace only spaces in the part matched before the lookahead, when they are mixed with other things (the words, that is).
One way is to capture what precedes the first float (given by lookahead), and in the replacement part run another regex on what's been captured, to replace spaces
s{ (.*?) (?=\s+-?[0-9]+\.[0-9]) }{ $1 =~ s/\s+/_/gr }ex
Notes
Modifier /e makes the replacement part be evaluated as code; any valid Perl code goes
With s{}{} delimiters we can use s/// ones in the replacement part's regex
Regex in the replacement part, that changes spaces to _ in the captured text, has /r modifier so to return the modified string and leave the original unchanged. Thus we aren't attempting to change $1 (it's read only), and the modified string (being returned) is available as the replacement
Modifier /x allows use of spaces in patterns, for readability
Some assumptions must be made here. Most critical one is that the text to process is followed by a number in the given format, -?[0-9]+\.[0-9]+, and that there isn't such a number in the text itself. This follows the OP's sample and, more decidedly, the attempted solution
A couple of details with assumptions. (1) Leading digits are expected with [0-9]+\. -- if you can have numbers like .123 then use [0-9]*\. (2) The \s+ in the inner regex collapses multiple consecutive spaces into one _, so a b c becomes a_b_c (and not a__b_c)
In the lookahead I scoop up all spaces preceding the first float with \s+ -- and so they'll stay in front of the first float. This is as wanted with one space but with multiple ones it may be awkward
If they were included in the .*? capture (if the lookahead only has one space, \s) then we'd get an _ trailing the word(s). I thought that'd be more awkward. The ideal solution is to run another regex and clean that up, if such a case is possible and if it's a bother
An example
echo "a bbb c -0.15636028 -0.2953045" |
perl -wpe's{(.*?)(?=\s+-?[0-9]+\.[0-9])}{ $1 =~ s/\s+/_/gr }e'
prints
a_bbb_c -0.15636028 -0.2953045
Then to process all lines in a file you can do either
perl -wpe'...' file > new_file
and get a new_file with changes, or
perl -i.bak -wpe'...' file
to change the file in-place (that's -i), where .bak makes it save a backup.
Would something like this work for you:
s/\s+/_/g;
s/_(-?\d+\.)/ $1/g;
Use a negative lookahead to replace any spaces not followed by a float:
echo "a bbb cc -0.123232 -0.3232" | perl -wpe 's/ +(?! *-?\d+\.)/_/g'
Assuming from your comments your files look like that:
name float1 float2 float3
a bbb c -0.15636028 -0.2953045 0.29853472
abbb c -0.15636028 -0.2953045 0.29853472
a bbbc -0.15636028 -0.2953045 0.29853472
ab bbc -0.15636028 -0.2953045 0.29853472
abbbc -0.15636028 -0.2953045 0.29853472
Since you said in comments that the first field may contain digits, you can't use a lookahead that searches the first float to solve the problem. (you can nevertheless use a lookahead that counts the number of floats until the end of the line but it isn't very handy).
What I suggest is a solution based on fields number defined by the header first line.
You can use the header line to know the number of fields and replace spaces at the begining of other lines until the number of fields is the same.
You can use perl command line as awk like that:
perl -MEnglish -pae'$c=scalar #F if ($NR==1);for($i=0;$i<scalar(#F)-$c;$i++){s/\s+/_/}' file
The for loop counts the difference between the number of fields in the first row (stored in $c) and in the current line (given by scalar(#F) where #F is the fields array), and repeats the substitution.
The a switches the perl command line in autosplit mode and the -MEnglish makes available the number row variable as $NR (like the NR variable in awk).
It's possible to shorten it like that:
perl -pae'$c=#F if $.<2;$i=#F-$c;s/\s+/_/ while $i--' file
To all,
I have spent alot of time searching for a solution to this but cannot find it.
Just for a background, I have a text database with thousands of records. Each record is delineated by :
"0 #nnnnnn# Xnnn" // no quotes
The records have many fields on a line of their own, but the field I am interested in to search and replace a substring (notice spaces) :
" 1 X94 User1.faculty.ventura.ca" // no quotes
I want to use sed to change the substring ".faculty.ventura.ca" to ".students.moorpark.ut", changing nothing else on the line, globally for ALL records.
I have tested many things with negative results.
How can this be done ?
Thank You for the assistance.
Bob Perez (robertperez1957#gmail.com)
If I understand you correctly, you want this:
sed 's/1 X94 \(.*\).faculty.ventura.ca/1 X94 \1.students.moorpark.ut/' mydatabase.file
This will replace all records of the form 1 X94 XXXXXX.faculty.ventura.ca with 1 X94 XXXXX.students.moorpark.ut.
Here's details on what it all does:
The '' let you have spaces and other messes in your script.
s/ means substitute
1 X94 \(.*\).faculty.ventura.ca is what you'll be substituting. The \(.*\) stores anything in that regular expression for use in the replacement
1 X94 \1.students.moorpark.ut is what to replace the thing you found with. \1 is filled in with the first thing that matched \(.*\). (You can have multiple of those in one line, and the next one would then be \2.)
The final / just tells sed that you're done. If your database doesn't have linefeeds to separate its records, you'll want to end with /g, to make this change multiple times per line.
mydatabase.file should be the filename of your database.
Note that this will output to standard out. You'll probably want to add
> mynewdatabasefile.name
to the end of your line, to save all the output in a file. (It won't do you much good on your terminal.)
Edit, per your comments
If you want to replace 1 F94 bperez.students.Napvil.NCC to 1 F94 bperez.JohnSmith.customer, you can use another set of \(.*\), as:
sed 's/1 X94 \(.*\).\(.*\).Napvil.NCC/1 X94 \1.JohnSmith.customer/' 251-2.txt
This is similar to the above, except that it matches two stored parameters. In this example, \1 evaluates to bperez and \2 evaluates to students. We match \2, but don't use it in the replace part of the expression.
You can do this with any number of stored parameters. (Sed probably has some limit, but I've never hit a sufficiently complicated string to hit it.) For example, we could make the sed script be '\(.\) \(...\) \(.*\).\(.*\).\(.*\).\(.*\)/\1 \2 \3.JohnSmith.customer/', and this would make \1 = 1, \2 = X94, \3 = bperez, \4 = Napvil and \5 = NCC, and we'd ignore \4 and \5. This is actually not the best answer though - just showing it can be done. It's not the best because it's uglier, and also because it's more accepting. It would then do a find and replace on a line like 2 Z12 bperez.a.b.c, which is presumably not what you want. The find query I put in the edit is as specific as possible while still being general enough to suit your tasks.
Another edit!
You know how I said "be as specific as possible"? Due to the . character being special, I wasn't. In fact, I was very generic. The . means "match any character at all," instead of "match a period". Regular expressions are "greedy", matching the most they could, so \(.*\).\(.*\) will always fill the first \(.*\) (which says, "take 0 to many of any character and save it as a match for later") as far as it can.
Try using:
sed 's/1 X94 \(.*\)\.\(.*\).Napvil.NCC/1 X94 \1.JohnSmith.customer/' 251-2.txt
That extra \ acts as an escape sequence, and changes the . from "any character" to "just the period". FYI, since I don't (but should) escape the other periods, technically sed would consider 1 X94 XXXX.StdntZNapvilQNCC as a valid match. Since . means any character, a Z or a Q there would be considered a fit.
The following tutorial helped me
sed - replace substring in file
try the same using a -i prefix to replace in the file directly
sed -i 's/unix/linux/' file.txt
I have a file like:
None44 DET20_22526;size=4; DET20_39906;size=2; DEX29.h_40767;size=4; DEX27.h_779;size=6757;
Goal:
None44 DET20_22526 DET20_39906 DEX29.h_40767 DEX27.h_779
Simply remove the ";size=**;
The digits after size= range from 1-6757 (at the most).
I have been trying:
sed 's/;size=*;//g'
My limited knowledge of sed and regX limited me to this.
Can someone point out how to either remove all between ;'s including the ;'s
or
How to make my sed realize what I can state in English... but can't code yet :(
You could try :
sed 's/;size=[0-9]*;//g'
What does this regex means ?
s/.../.../g stands for : replace every match of first expression with the second expression
first expression, in our case, is ;size=[0-9]*; that shoud be decomposed as:
the exact string ;size=, followed by
zero or more occurences of any digit in the range 0-9, followed by
;
second expression is empty, so the matched part is suppressed
the final g is an option that tells sed to match all matching parts, and not stop at the first one