sed regex with multiple matches and condition - regex

I would like to convert strings like:
abc=123.24|127.9|2891;xyz;hgy
to:
abc=123.24,127.9,2891;xyz;hgy
This is close:
echo "abc=123.24|127.9|2891;xyz;hgy" | sed -r 's/(=)([0-9.]+)\|/\1\2,/g'
but returns:
abc=123.24,127.9|2891;xyz;hgy
How can I do the rest of the numbers in a similar fashion if the number of bar-separated numbers is variable?
Clarification:
I hate it when people do not give me the whole picture in questions, but my original description above did just that. The small example is embedded in a much larger line that includes "|" separated text. I want to replace only the "|" with "," between numbers that follow an equal sign. Here is an entire line as an example:
chr1 69511 rs75062661 A G . QSS_ref ASP;BaseCounts=375,3,118,4;CAF=[0.348,0.652];COMMON=1;EFF=NON_SYNONYMOUS_CODING(MODERATE|MISSENSE|Aca/Gca|T141A|305|OR4F5|protein_coding|CODING|ENST00000335137|1|1);GNO;HRun=0;HaplotypeScore=0.0000;KGPROD;KGPhase1;LowMQ=0.0280,0.0580,500;MQ=49.32;MQ0=14;MSigDb=ACEVEDO_METHYLATED_IN_LIVER_CANCER_DN,KEGG_OLFACTORY_TRANSDUCTION,REACTOME_GPCR_DOWNSTREAM_SIGNALING,REACTOME_OLFACTORY_SIGNALING_PATHWAY,REACTOME_SIGNALING_BY_GPCR,chr1p36;NORMALT=86;NORMREF=228;NSM;NT=het;OTHERKG;QSS=8;QSS_NT=6;REF;RS=75062661;RSPOS=69511;S3D;SAO=0;SGT=AG->AG;SOMATIC;SSR=0;TQSS=1;TQSS_NT=2;TUMALT=15;TUMREF=227;TUMVAF=0.06198347107438017;TUMVARFRACTION=0.1485148514851485;VC=SNV;VLD;VP=0x050200000a05140116000100;WGT=1;dbNSFP_1000Gp1_AC=1424;dbNSFP_1000Gp1_AF=0.652014652014652;dbNSFP_1000Gp1_AFR_AC=162;dbNSFP_1000Gp1_AFR_AF=0.32926829268292684;dbNSFP_1000Gp1_AMR_AC=235;dbNSFP_1000Gp1_AMR_AF=0.649171270718232;dbNSFP_1000Gp1_ASN_AC=500;dbNSFP_1000Gp1_ASN_AF=0.8741258741258742;dbNSFP_1000Gp1_EUR_AC=527;dbNSFP_1000Gp1_EUR_AF=0.6952506596306068;dbNSFP_29way_logOdds=4.1978;dbNSFP_29way_pi=0.1516:0.0:0.6258:0.2226;dbNSFP_ESP6500_AA_AF=0.544101;dbNSFP_ESP6500_EA_AF=0.887429;dbNSFP_Ensembl_geneid=ENSG00000186092;dbNSFP_Ensembl_transcriptid=ENST00000534990|ENST00000335137;dbNSFP_FATHMM_score=0.51;dbNSFP_GERP++_NR=2.31;dbNSFP_GERP++_RS=1.15;dbNSFP_Interpro_domain=GPCR|_rhodopsin-like_superfamily_(1)|;dbNSFP_LRT_Omega=0.000000;dbNSFP_LRT_pred=N;dbNSFP_LRT_score=0.000427;dbNSFP_MutationAssessor_pred=neutral;dbNSFP_MutationAssessor_score=-1.295;dbNSFP_MutationTaster_pred=N;dbNSFP_MutationTaster_score=0.000162;dbNSFP_Polyphen2_HDIV_pred=B;dbNSFP_Polyphen2_HVAR_pred=B;dbNSFP_SIFT_score=0.950000;dbNSFP_Uniprot_aapos=141;dbNSFP_Uniprot_acc=Q8NH21;dbNSFP_Uniprot_id=OR4F5_HUMAN;dbNSFP_aaalt=A;dbNSFP_aapos=189|141;dbNSFP_aaref=T;dbNSFP_cds_strand=+;dbNSFP_codonpos=1;dbNSFP_fold-degenerate=0;dbNSFP_phyloP=0.267000;dbNSFP_refcodon=ACA;dbSNPBuildID=131 AU:CU:DP:FDP:GU:SDP:SUBDP:TU 228,232:3,3:322:4:86,109:0:0:1,2 227,228:0,0:244:1:15,16:0:0:1,2
The replacement in this line is of the string:
dbNSFP_aapos=189|141
with:
dbNSFP_aapos=189,141

why not:
sed 's/|/,/g'
kent$ echo "abc=123.24|127.9|2891;xyz;hgy"|sed 's/|/,/g'
abc=123.24,127.9,2891;xyz;hgy

Without using perl you can use
input="abc=123;def=hello123test;dbNSFP_aapos=189|141|142;dbNSFP_aaref=T;another=test|hello;"
sed -r 's/(.*=)([0-9.]+\|)+(.*)/\1'$(sed -r 's/(.*=)([0-9.]+\|)+(.*)/\2/' <<< $input | tr '|' ,)'\3/' <<< $input
Output:
abc=123;def=hello123test;dbNSFP_aapos=189,141,142;dbNSFP_aaref=T;another=test|hello;
Replace <<< $input with your file or whatever you actually have as input :)
Explanation:
We have three capturing groups in the regex (I restructured the groups from the OP), the second will contain only the string where the replacement of the | is to happen, while the first and third contain everything before and after the second group.
See the demo # regex101.
Within the second command ($(...)) we grab the second capturing group with sed and replace every | inside with a comma. This substitution is then used to be put in the place of the second group within the other sed-call.

As alternative, you can try with perl and its evalutation flag:
echo "..." | perl -pe 's{=([\d.|]+)}{"=" . (join ",", split /\|/, $1)}eg'
It searches for a string after an equal sign, splits it with | and join it with commas.

Using tr
echo "abc=123.24|127.9|2891;xyz;hgy" | tr \| ,
abc=123.24,127.9,2891;xyz;hgy

Assuming semicolon is your field separator, how about something like
tr `;\n' '\n;' | sed '/=[0-9.|]*$/s/|/,/g' | tr '\n;' ';\n'
This has a serious flaw; it fails in weird ways for the first and last fields on a line. If you can't live with that, maybe try this:
awk -F ';' '{ for(i=1; i<=NF; ++i) if ($i ~ /=([0-9.]+\|)+[0-9.]+$/) gsub(/\|/,",",$i); print }'

Related

find recurring pattern with `sed`

I am using GNU bash 4.3.48
I expected that
echo "23S62M1I19M2D" | sed 's/.*\([0-9]*M\).*/\1/g'
would output 62M19M... But it doesn't.
sed 's/\([0-9]*M\)//g' deletes ALL [0-9]*M and retrieves 23S1I2D. but the group \1 is not working as I thought it would.
sed 's/.*\([0-9]*M\).*/ \1 /g', retrieves M...
What am I doing wrong?
Thank you!
With your shown samples and with awk you could try following program.
echo "23S62M1I19M2D" |
awk '
{
val=""
while(match($0,/[0-9]+M/)){
val=val substr($0,RSTART,RLENGTH)
$0=substr($0,RSTART+RLENGTH)
}
print val
}
'
Explanation: Simple explanation would be, using echo to print values and sending it as a standard input to awk program. In awk program using its match function to match regex mentioned in it(/[0-9]+M) running loop to find all matches in each line and printing the collected matched values at last of each line.
This might work for you (GNU sed):
sed -nE '/[0-9]*M/{s//\n&\n/g;s/(^|\n)[^\n]*\n?//gp}' file
Surround the match by newlines and then remove non-matching parts.
Alternative, using grep and tr:
grep -o '[0-9]*M' file | tr -d '\n'
N.B. tr removes all newlines (including the last one) to restore the last newline, use:
grep -o '[0-9]*M' file | tr -d '\n' | paste
The alternate solution will concatenate all results into a single line. To achieve the same result with the first solution use:
sed -nE '/[0-9]*M/{s//\n&\n/g;s/(^|\n)[^\n]*\n?//g;H};${x;s/\n//gp}' file
The problem is that the .* is greedy. Since only M is obligatory, when the engine finds last M, it satisfies the regex, so all string is matched, M is captured and thus kept after replacing with \1 backreference.
That means, you can't easily do this with sed. You can do that with Perl much easier since it supports matching and skipping pattern:
#!/bin/bash
perl -pe 's/\d+M(*SKIP)(*F)|.//g' <<< "23S62M1I19M2D"
See the online demo. The pattern matches
\d+M(*SKIP)(*F) - one or more digits, M, and then the match is omitted and the next match is searched for from the failure position
|. - or matches any char other than a line break char.
Or simply match all occurrences and concatenate them:
perl -lane 'BEGIN{$a="";} while (/\d+M/g) {$a .= $&} END{print $a;}' <<< "23S62M1I19M2D"
All \d+M matches are appended to the $a variable which is printed at the end of processing the string.
Your substitution is probably working, but not substituting what you think it is.
In the substitution s/\(foo...\)/\1/, the \1 matches whatever \(...\) matches and captures, so your substitution is replacing foo... by foo...!
% echo "1234ABC" | sed 's/\([A-Z]\)/-\1-/'g
1234-A--B--C-
So you'll need to match more, but capture only a portion of the match. For example:
echo "23S62M1I19M2D" | sed 's/[0-9]*[A-LN-Z]*\([0-9]*M\)/\1/g'
62M19M2D
In the case of sed 's/.*\([0-9]*M\).*/\1/g' (did that appear in an edit to the question, or did I just miss it?), the .* matches ‘greedily’ – it matches as much as it possibly can, thus including the digits before the M. In the example above, the [A-LN-Z] is required to be at the end of the uncaptured part, so the digits are forced to be matched by the [0-9] inside the capture.
Getting a clear idea of what ‘greedy’ means is a really important idea when writing or interpreting regexps.
If you know you will only encounter the suffixes S, M, I and D, an alternative approach would be explicitly deleting the combinations you don't want:
echo "23S62M1I19M2D" | sed 's/[0-9]\+[SID]//g'
This gives the expected:
62M19M
Update: This variant produces the same output, but rejects all non-numeric, non-M suffixes:
echo "23S62M1I19M2D" | sed 's/[0-9]\+[^0-9M]//g'

Parsing Karma Coverage Output in Bash for a Jenkins Job (Scripting)

I'm working with the following output:
=============================== Coverage summary ===============================
Statements : 26.16% ( 1681/6425 )
Branches : 6.89% ( 119/1727 )
Functions : 23.82% ( 390/1637 )
Lines : 26.17% ( 1680/6420 )
================================================================================
I would like to parse the 4 coverage percentage numbers without the percent via REGEX, into a comma separated list.
Any suggestions for a good regex expression for this? Or another good option?
The sed command:
sed -n '/ .*% /{s/.* \(.*\)% .*/\1/;p;}' input.txt | sed ':a;N;$!ba;s/\n/,/g'
gives the output:
26.16,6.89,23.82,26.17
Edit: A better answer, with only a single sed, would be:
sed -n '/ .*% /{s/.* \(.*\)% .*/\1/;H;};${g;s/\n/,/g;s/,//;p;}' input.txt
Explanation:
/ .*% / search for lines with a percentage value (note spaces)
s/.* \(.*\)% .*/\1/ and delete everything except the percentage value
H and then append it to the hold space, prefixed with a newline
$ then for the last line
g get the hold space
s/\n/,/g replace all the newlines with commas
s/,// and delete the initial comma
p and then finally output the result
To harden the regex, you could replace the search for the percentage value .*% with for example [0-9.]*%.
I think this is a grep job. This should help:
$ grep -oE "[0-9]{1,2}\.[0-9]{2}" input.txt | xargs | tr " " ","
Output:
26.16,6.89,23.82,26.17
The input file just contains what you have shown above. Obviously, there are other ways like cat to feed the input to the command.
Explanation:
grep -oE: only show matches using extended regex
xargs: put all results onto a single line
tr " " ",": translate the spaces into commas:
This is actually a nice shell tool belt example, I would say.
Including the consideration of Joseph Quinsey, the regex can be made more robust with a lookahead to assert a % sign after then numeric value using a Perl-compatible RE pattern:
grep -oP "[0-9]{1,2}\.[0-9]{2}(?=%)" input.txt | xargs | tr " " ","
Would you consider to use awk? Here's the command you may try,
$ awk 'match($0,/[0-9.]*%/){s=(s=="")?"":s",";s=s substr($0,RSTART,RLENGTH-1)}END{print s}' file
26.16,6.89,23.82,26.17
Brief explanation,
match($0,/[0-9.]*%/): find the record matched with regex [0-9.]*%
s=(s=="")?"":s",": since comma separated is required, we just need print commas before each matched except the first one.
s=s substr($0,RSTART,RLENGTH-1): print the matched part appended to s
Assuming the item names (Statements, Branches, ...) do not contain whitespaces, how about:
#!/bin/bash
declare -a keys
declare -a vaues
while read -r line; do
if [[ "$line" =~ ^([^\ ]+)\ *:\ *([0-9.]+)% ]]; then
keys+=(${BASH_REMATCH[1]})
values+=(${BASH_REMATCH[2]})
fi
done < output.txt
ifsback=$IFS # backup IFS
IFS=,
echo "${keys[*]}"
echo "${values[*]}"
IFS=$ifsback # restore IFS
which yields:
Statements,Branches,Functions,Lines
26.16,6.89,23.82,26.17
Yet another option, with perl:
cat the_file | perl -e 'while(<>){/(\d+\.\d+)%/ and $x.="$1,"}chop $x; print $x;'
The code, unrolled and explained:
while(<>){ # Read line by line. Put lines into $_
/(\d+\.\d+)%/ and $x.="$1,"
# Equivalent to:
# if ($_ =~ /(\d+\.\d+)%/) {$x.="$1,"}
# The regex matches "numbers", "dot", "numbers" and "%",
# stores just numbers on $1 (first capturing group)
}
chop $x; # Remove extra ',' and print result
print $x;
Somewhat shorter with an extra sed
cat the_file | perl -ne '/(\d+\.\d+)%/ and print "$1,"'|sed 's/.$//'
Uses "n" parameter which implies while(<>){}. For removing the last ',' we use sed.

pipe sed command to create multiple files

I need to get X to Y in the file with multiple occurrences, each time it matches an occurrence it will save to a file.
Here is an example file (demo.txt):
\x00START how are you? END\x00
\x00START good thanks END\x00
sometimes random things\x00\x00 inbetween it (ignore this text)
\x00START thats nice END\x00
And now after running a command each file (/folder/demo1.txt, /folder/demo2.txt, etc) should have the contents between \x00START and END\x00 (\x00 is null) in addition to 'START' but not 'END'.
/folder/demo1.txt should say "START how are you? ", /folder/demo2.txt should say "START good thanks".
So basicly it should pipe "how are you?" and using 'echo' I can prepend the 'START'.
It's worth keeping in mind that I am dealing with a very large binary file.
I am currently using
sed -n -e '/\x00START/,/END\x00/ p' demo.txt > demo1.txt
but that's not working as expected (it's getting lines before the '\x00START' and doesn't stop at the first 'END\x00').
If you have GNU awk, try:
awk -v RS='\0START|END\0' '
length($0) {printf "START%s\n", $0 > ("folder/demo"++i".txt")}
' demo.txt
RS='\0START|END\0' defines a regular expression acting as the [input] Record Separator which breaks the input file into records by strings (byte sequences) between \0START and END\0 (\0 represents NUL (null char.) here).
Using a multi-character, regex-based record separate is NOT POSIX-compliant; GNU awk supports it (as does mawk in general, but seemingly not with NUL chars.).
Pattern length($0) ensures that the associated action ({...}) is only executed if the records is nonempty.
{printf "START%s\n", $0 > ("folder/demo"++i)} outputs each nonempty record preceded by "START", into file folder/demo{n}.txt", where {n} represent a sequence number starting with 1.
You can use grep for that:
grep -Po "START\s+\K.*?(?=END)" file
how are you?
good thanks
thats nice
Explanation:
-P To allow Perl regex
-o To extract only matched pattern
-K Positive lookbehind
(?=something) Positive lookahead
EDIT: To match \00 as START and END may appear in between:
echo -e '\00START hi how are you END\00' | grep -aPo '\00START\K.*?(?=END\00)'
hi how are you
EDIT2: The solution using grep would only match single line, for multi-line it's better use perl instead. The syntax will be very similar:
echo -e '\00START hi \n how\n are\n you END\00' | perl -ne 'BEGIN{undef $/ } /\A.*?\00START\K((.|\n)*?)(?=END)/gm; print $1'
hi
how
are
you
What's new here:
undef $/ Undefine INPUT separator $/ which defaults to '\n'
(.|\n)* Dot matches almost any character, but it does not match
\n so we need to add it here.
/gm Modifiers, g for global m for multi-line
I would translate the nulls into newlines so that grep can find your wanted text on a clean line by itself:
tr '\000' '\n' < yourfile.bin | grep "^START"
from there you can take it into sed as before.

sed substitutions within captured group

I think that I'm missing something obvious, but I've spent a lot of time Googling around and searching stackoverflow, and I can't find an answer. I'm looking for a way to perform an additional sed substitution on a captured group. So, for example, if I have the text:
MD5 (./x y.jpg) = 93845ad8b6fb2797d9f1d7e0622e9270
MD5 (./x y z.jpg) = 93845ad8b6fb2797d9f1d7e0622e9270
I'd like to replace the spaces within the parentheses with underscores and reformat the string to be the filenamemd5.
./x_y.jpg 93845ad8b6fb2797d9f1d7e0622e9270
./x_y_z.jpg 93845ad8b6fb2797d9f1d7e0622e9270
I'm able to capture the filename, but I don't know how to perform another substitution on the captured group.
echo 'MD5 (./x y.jpg) = 93845ad8b6fb2797d9f1d7e0622e9270' | sed 's/MD5 (\(.*\)) = \(.*\)/\1 \2/'
outputs:
./x y.jpg 93845ad8b6fb2797d9f1d7e0622e9270
Thoughts?
sed allows you to perform multiple substitutions (s/.../.../) in sequence, with each operating on the result of the previous:
echo 'MD5 (./x y.jpg) = 93845ad8b6fb2797d9f1d7e0622e9270' |
sed 's/MD5 (\(.*\)) = \(.*\)/\1|\2/; s/ /_/g; s/|/ /'
Here I've used a simple trick:
1st s command: I've used a | to temporarily separate the 2 backreferences.
2nd s: I then replace ALL spaces with _ chars.
3rd s: I replace the | with a space, so that a single space remains.
Of course, this trick only works if your input data never contains | chars.
You can use this sed,
sed 's/ /_/g; s/MD5_(\(.*\))_=_\(.*\)/\1 \2/'
Test:
sat:~$ echo 'MD5 (./x y z.jpg) = 93845ad8b6fb2797d9f1d7e0622e9270' |
sed 's/ /_/g; s/MD5_(\(.*\))_=_\(.*\)/\1 \2/'
./x_y_z.jpg 93845ad8b6fb2797d9f1d7e0622e9270

Extract multiple occurrences on the same line using sed/regex

I am trying to loop through each line in a file and find and extract letters that start with ${ and end with }. So as the final output I am expecting only SOLDIR and TEMP(from inputfile.sh).
I have tried using the following script but it seems it matches and extracts only the second occurrence of the pattern TEMP. I also tried adding g at the end but it doesn't help. Could anybody please let me know how to match and extract both/multiple occurrences on the same line ?
inputfile.sh:
.
.
SOLPORT=\`grep -A 4 '\[LocalDB\]' \${SOLDIR}/solidhac.ini | grep \${TEMP} | awk '{print $2}'\`
.
.
script.sh:
infile='inputfile.sh'
while read line ; do
echo $line | sed 's%.*${\([^}]*\)}.*%\1%g'
done < "$infile"
May I propose a grep solution?
grep -oP '(?<=\${).*?(?=})'
It uses Perl-style lookaround assertions and lazily matches anything between '${' and '}'.
Feeding your line to it, I get
$ echo "SOLPORT=\`grep -A 4 '[LocalDB]' \${SOLDIR}/solidhac.ini | grep \${TEMP} | awk '{print $2}'\`" | grep -oP '(?<=\${).*?(?=})'
SOLDIR
TEMP
This might work for you (but maybe only for your specific input line):
sed 's/[^$]*\(${[^}]\+}\)[^$]*/\1\t/g;s/$[^{$]\+//g'
Extracting multiple matches from a single line using sed isn't as bad as I thought it'd be, but it's still fairly esoteric and difficult to read:
$ echo 'Hello ${var1}, how is your ${var2}' | sed -En '
# Replace ${PREFIX}${TARGET}${SUFFIX} with ${PREFIX}\a${TARGET}\n${SUFFIX}
s#\$\{([^}]+)\}#\a\1\n#
# Continue to next line if no matches.
/\n/!b
# Remove the prefix.
s#.*\a##
# Print up to the first newline.
P
# Delete up to the first newline and reprocess what's left of the line.
D
'
var1
var2
And all on one line:
sed -En 's#\$\{([^}]+)\}#\a\1\n#;/\n/!b;s#.*\a##;P;D'
Since POSIX extended regexes don't support non-greedy quantifiers or putting a newline escape in a bracket expression I've used a BEL character (\a) as a sentinel at the end of the prefix instead of a newline. A newline could be used, but then the second substitution would have to be the questionable s#.*\n(.*\n.*)##, which might involve a pathological amount of backtracking by the regex engine.