I have a large file (50 GB) and I could like to count the number of occurrences of different lines in it. Normally I'd use
sort bigfile | uniq -c
but the file is large enough that sorting takes a prohibitive amount of time and memory. I could do
grep -cfx 'one possible line'
for each unique line in the file, but this would mean n passes over the file for each possible line, which (although much more memory friendly) takes even longer than the original.
Any ideas?
A related question asks about a way to find unique lines in a big file, but I'm looking for a way to count the number of instances of each -- I already know what the possible lines are.
Use awk
awk '{c[$0]++} END {for (line in c) print c[line], line}' bigfile.txt
This is O(n) in time, and O(unique lines) in space.
Here is a solution using jq 1.5. It is essentially the same as the awk solution, both in approach and performance characteristics, but the output is a JSON object representing the hash. (The program can be trivially modified to produce output in an alternative format.)
Invocation:
$ jq -nR 'reduce inputs as $line ({}; .[$line] += 1)' bigfile.txt
If bigfile.txt consisted of these lines:
a
a
b
a
c
then the output would be:
{
"a": 3,
"b": 1,
"c": 1
}
#!/bin/bash
# port this logic to awk or ksh93 to make it fast
declare -A counts=( )
while IFS= read -r line; do
counts[$line]=$(( counts[$line] + 1 )) # increment counter
done
# print results
for key in "${!counts[#]}"; do
count=${counts[$key]}
echo "Found $count instances of $key"
done
Related
I have a file with 2 columns, and i want to use the values from the second column to set the range in the cut command to select a range of characters from another file. The range i desire is the character in the position of the value in the second column plus the next 10 characters. I will give an example in a while.
My files are something like that:
File with 2 columns and no blank lines between lines (file1.txt):
NAME1 10
NAME2 25
NAME3 48
NAME4 66
File that i want to extract the variable range of characters(just one very long line with no spaces and no bold font) (file2.txt):
GATCGAGCGGGATTCTTTTTTTTTAGGCGAGTCAGCTAGCATCAGCTACGAGAGGCGAGGGCGGGCTATCACGACTACGACTACGACTACAGCATCAGCATCAGCGCACTAGAGCGAGGCTAGCTAGCTACGACTACGATCAGCATCGCACATCGACTACGATCAGCATCAGCTACGCATCGAAGAGAGAGC
...or, more literally (for copy/paste to test):
GATCGAGCGGGATTCTTTTTTTTTAGGCGAGTCAGCTAGCATCAGCTACGAGAGGCGAGGGCGGGCTATCACGACTACGACTACGACTACAGCATCAGCATCAGCGCACTAGAGCGAGGCTAGCTAGCTACGACTACGATCAGCATCGCACATCGACTACGATCAGCATCAGCTACGCATCGAAGAGAGAGC
Desired resulting file, one sequence per line (result.txt):
GATTCTTTTT
GGCGAGTCAG
CGAGAGGCGA
TATCACGACT
The resulting file would have the characters from 10-20, 25-35, 48-58 and 66-76, each range in a new line. So, it would always keep the range of 10, but in different start points and those start points are set by the values in the second column from the first file.
I tried the command:
for i in $(awk '{print $2}' file1.txt);
do
p1=$i;
p2=`expr "$1" + 10`
cut -c$p1-$2 file2.txt > result.txt;
done
I don't get any output or error message.
I also tried:
while read line; do
set $line
p2=`expr "$2" + 10`
cut -c$2-$p2 file2.txt > result.txt;
done <file1.txt
This last command gives me an error message:
cut: invalid range with no endpoint: -
Try 'cut --help' for more information.
expr: non-integer argument
There's no need for cut here; dd can do the job of indexing into a file, and reading only the number of bytes you want. (Note that status=none is a GNUism; you may need to leave it out on other platforms and redirect stderr otherwise if you want to suppress informational logging).
while read -r name index _; do
dd if=file2.txt bs=1 skip="$index" count=10 status=none
printf '\n'
done <file1.txt >result.txt
This approach avoids excessive memory requirements (as present when reading the whole of file2 -- assuming it's large), and has bounded performance requirements (overhead is equal to starting one copy of dd per sequence to extract).
Using awk
$ awk 'FNR==NR{a=$0; next} {print substr(a,$2+1,10)}' file2 file1
GATTCTTTTT
GGCGAGTCAG
CGAGAGGCGA
TATCACGACT
If file2.txt is not too large, then you can read it in memory,
and use Bash sub-strings to extract the desired ranges:
data=$(<file2.txt)
while read -r name index _; do
echo "${data:$index:10}"
done <file1.txt >result.txt
This will be much more efficient than running cut or another process for every single range definition.
(Thanks to #CharlesDuffy for the tip to read data without a useless cat, and the while loop.)
One way to solve it:
#!/bin/bash
while read line; do
pos=$(echo "$line" | cut -f2 -d' ')
x=$(head -c $(( $pos + 10 )) file2.txt | tail -c 10)
echo "$x"
done < file1.txt > result.txt
It's not the solution an experienced bash hacker would use, but it is very good for someone who is new to bash. It uses tools that are very versatile, although somewhat bad if you need high performance. Shell scripting is commonly used by people who rarely shell scripts, but knows a few commands and just wants to get the job done. That's why I'm including this solution, even if the other answers are superior for more experienced people.
The first line is pretty easy. It just extracts the numbers from file1.txt. The second line uses the very nice tools head and tail. Usually, they are used with lines instead of characters. Nevertheless, I print the first pos + 10 characters with head. The result is piped into tail which prints the last 10 characters.
Thanks to #CharlesDuffy for improvements.
given this question Appending a line to a file only if it does not already exist
is there a faster way than the solution provided by #drAlberT?
grep -q -F 'string' foo.bar || echo 'string' >> foo.bar
I have implemented the above solution and I have to iterate it over a 500k lines file (i.e. check if a line is not already in a 500k lines set). Moreover, I've to run this process for a lot of times, maybe 10-50 million times. Needless to say it's kind of slow as it takes 25-30ms to run on my server (so 3-10+ days of runtime in total).
EDIT: the flow is the following: I have a file with 500k lines, each time I run, I get maybe 10-30 new lines and I check if they are already there or not. If not I add them, then I repeat many times. The order of my 500k lines files is important as I'm going through it with another process.
EDIT2: the 500k lines file is always containing unique lines, and I only care about "full lines", no substrings.
Thanks a lot!
Few suggested improvements:
Try using awk instead of grep so that you can both detect the string and write it in one action;
If you do use grep don't use a Bash loop to feed each potential match to grep and then append that one word to the file. Instead, read all the potential lines into grep as matches (using -f file_name) and print the matches. Then invert the matches and append the inverted match. See last pipeline here;
Exit as soon as you see the string (for a single string) rather than continuing to loop over a big file;
Don't call the script millions of times with one or just a few lines -- organize the glue script (in Bash I suppose) so that the core script is called once or a few times with all the lines instead;
Perhaps use multicores since the files are not dependent on each other. Maybe with GNU Parallel (or you could use Python or Ruby or Perl that has support for threads).
Consider this awk for a single line to add:
$ awk -v line=line_to_append 'FNR==NR && line==$0{f=1; exit}
END{if (!f) print line >> FILENAME}' file
Or for multiple lines:
$ awk 'FNR==NR {lines[$0]; next}
$0 in lines{delete lines[$0]}
END{for (e in lines) print e >> FILENAME}' lines file
Some timings using a copy of the Unix words file (235,886 lines) with a five line lines file that has two overlaps:
$ echo "frob
knob
kabbob
stew
big slob" > lines
$ time awk 'FNR==NR {lines[$0]; next}
$0 in lines{delete lines[$0]}
END{for (e in lines) print e >> FILENAME}' lines words
real 0m0.056s
user 0m0.051s
sys 0m0.003s
$ tail words
zythum
Zyzomys
Zyzzogeton
frob
kabbob
big slob
Edit 2
Try this as being the best of both:
$ time grep -x -f lines words |
awk 'FNR==NR{a[$0]; next} !($0 in a)' - lines >> words
real 0m0.012s
user 0m0.010s
sys 0m0.003s
Explanation:
grep -x -f lines words find the lines that ARE in words
awk 'FNR==NR{a[$0]; next} !($0 in a)' - lines invert those into lines that are NOT in words
>> words append those to the file
Turning the millions of passes over the file into a script with millions of actions will save you a lot of overhead. Searching for a single label at each pass over the file is incredibly inefficient; you can search for as many labels as you can comfortably fit into memory in a single pass over the file.
Something along the following lines, perhaps.
awk 'NR==FNR { a[$0]++; next }
$0 in a { delete a[$0] }
1
END { for (k in a) print k }' strings bigfile >bigfile.new
If you can't fit strings in memory all at once, splitting that into suitable chunks will obviously allow you to finish this in as many passes as you have chunks.
On the other hand, if you have already (effectively) divided the input set into sets of 10-30 labels, you can obviously only search for those 10-30 in one pass. Still, this should provide you with a speed improvement on the order of 10-30 times.
This assumes that a "line" is always a full line. If the label can be a substring of a line in the input file, or vice versa, this will need some refactoring.
If duplicates are not valid in the file, just append them all and filter out the duplicates:
cat myfile mynewlines | awk '!n[$0]++' > mynewfile
This will allow appending millions of lines in seconds.
If order additionally doesn't matter and your files are more than a few gigabytes, you can use sort -u instead.
Have the script read new lines from stdin after consuming the original file. All lines are stored in an associative array (without any compression such as md5sum).
Appending the suffix 'x' is targeted to handle inputs such as '-e'; better ways probably exist.
#!/bin/bash
declare -A aa
while read line; do aa["x$line"]=1;
done < file.txt
while read line; do
if [ x${aa[$line]} == x ]; then
aa[$line]=1;
echo "x$line" >> file.txt
fi
done
I have some output files (5000 files) of .log which are the results of QM computations. Inside each file there are two special lines indicate the number of electrons and orbitals, like this below as an example (with exact spaces as in output files):
Number of electrons = 9
Number of orbitals = 13
I thought about a script (bash or Fortran), as a solution to this problem, which grep these two lines (at same time) and get the corresponding integer values (9 and 13, for instance), compare them and finds the difference between two values, and finally, list them in a new text file with the corresponding filenames.
I would really appreciate any help given.
Am posting an attempt in GNU Awk, and have tested it in that only.
#!/bin/bash
for file in *.log
do
awk -F'=[[:blank:]]*' '/Number of/{printf "%s%s",$2,(NR%2?" ":RS)}' "$file" | awk 'function abs(v) {return v < 0 ? -v : v} {print abs($1-$2)}' >> output_"$file"
done
The reason I split the AWK logic to two was to reduce the complexity in doing it in single huge command. The first part is for extracting the numbers from your log file in a columnar format and second for getting their absolute value.
I will break-down the AWK logic:-
-F'=[[:blank:]]*' is a mult0 character delimiter logic including = and one or more instances of [[:blank:]] whitespace characters.
'/Number of/{printf "%s%s",$2,(NR%2?" ":RS)}' searches for lines starting with Number of and prints it in a columnar fashion, i.e. as 9 13 from your sample file.
The second part is self-explanatory. I have written a function to get the absolute value from the two returned values and print it.
Each output is saved in a file named output_, for you to process it further.
Run the script from your command line as bash script.sh, where script.sh is the name containing the above lines.
Update:-
In case if you are interested in negative values too i.e. without the absolute function, change the awk statement to
awk -F'=[[:blank:]]*' '/Number of/{printf "%s%s",$2,(NR%2?" ":RS)}' "$file" | awk '{print ($1-$2)}' >> output_"$file"
Bad way to do it (but it will work)-
while read file
do
first=$(awk -F= '/^Number/ {print $2}' "$file" | head -1)
second=$(awk -F= '/^Number/ {print $2}' "$file" | tail -1)
if [ "$first" -gt "$second" ]
then
echo $(("$first" - "$second"))
else
echo $(("$second" - "$first"))
fi > "$file"_answer ;
done < list_of_files
This method picks up the values (in the awk one liner and compares them.
It then subtracts them to give you one value which it saves in the file called "$file"_answer. i.e. the initial file name with '_answer' as a suffix to the name.
You may need to tweak this code to fit your purposes exactly.
I have a stupidly large text file (i.e. 40 gigabytes as of today) that I would like to filter for unique lines without sorting the file.
The file has unix line endings, and all content matches [[:print:]]. I tried the following awk script to display only unique lines:
awk 'a[$0] {next} 1' stupid.txt > less_stupid.txt
The thought was that I'd populate an array by referencing its elements, using the contents of the file as keys, then skip lines that were already in the array. But this fails for two reasons -- firstly, because it inexplicably just doesn't work (even on small test files), and secondly because I know that my system will run out of memory before the entire set of unique lines is loaded into memory by awk.
After searching, I found this answer which recommended:
awk '!x[$0]++'
And while this works on small files, it also will run out of memory before reading my entire file.
What's a better (i.e. working) solution? I'm open to just about anything, though I'm more partial to solutions in languages I know (bash & awk, hence the tags). In trying to visualize the problem, the best I've come up with would be to store an array of line checksums or MD5s rather than the lines themselves, but that only saves a little space and runs the risk of checksum collisions.
Any tips would be very welcome. Telling me this is impossible would also be welcome, so that I stop trying to figure it out. :-P
The awk '!x[$0]++' trick is one of the most elegant solutions to de-duplicate a file or stream without sorting. However, it is inefficient in terms of memory and unsuitable for large files, since it saves all unique lines into memory.
However, a much more efficient implementation would be to save a constant-length hash representation of the lines in the array rather than the whole line. You can achieve this with Perl in one line and it is quite similar to the awk script.
perl -ne 'use Digest::MD5 qw(md5_base64); print unless $seen{md5_base64($_)}++' huge.txt
Here I used md5_base64 instead of md5_hex because the base64 encoding takes 22 bytes, while the hex representation 32.
However, since the Perl implementation of hashes still requires around 120bytes for each key, you may quickly run out of memory for your huge file.
The solution in this case is to process the file in chunks, splitting manually or using GNU Parallel with the --pipe, --keep-order and --block options (taking advantage of the fact that duplicate lines are not far apart, as you mentioned). Here is how you could do it with parallel:
cat huge.txt | pv |
parallel --pipe --keep-order --block 100M -j4 -q \
perl -ne 'use Digest::MD5 qw(md5_base64); print unless $seen{md5_base64($_)}++' > uniq.txt
The --block 100M option tells parallel to process the input in chunks of 100MB. -j4 means start 4 processes in parallel. An important argument here is --keep-order since you want the unique lines output to remain in the same order. I have included pv in the pipeline to get some nice statistics while the long running process is executing.
In a benchmark I performed with a random-data 1GB file, I reached a 130MB/sec throughput with the above settings, meaning you may de-duplicate your 40GB file in 4 minutes (if you have a sufficiently fast hard disk able to write at this rate).
Other options include:
Use an efficient trie structure to store keys and check for duplicates. For example a very efficient implementation is marisa-trie coded in C++ with wrappers in Python.
Sort your huge file with an external merge sort or distribution/bucket sort
Store your file in a database and use SELECT DISTINCT on an indexed column containing your lines or most efficiently md5_sums of your lines.
Or use bloom filters
Here is an example of using the Bloom::Faster module of Perl:
perl -e 'use Bloom::Faster; my $f = new Bloom::Faster({n => 100000000, e => 0.00001}); while(<>) { print unless $f->add($_); }' huge.txt > uniq.txt
You may install Bloom::Faster from cran (sudo cran and then install "Bloom::Faster")
Explanation:
You have to specify the probabilistic error rate e and the number of available buckets n. The memory required for each bucket is about 2.5 bytes. If your file has 100 million unique lines then you will need 100 million buckets and around 260MB of memory.
The $f->add($_) function adds the hash of a line to the filter and returns true if the key (i.e. the line here) is a duplicate.
You can get an estimation of the number of unique lines in your file, parsing a small section of your file with dd if=huge.txt bs=400M count=1 | awk '!a[$0]++' | wc -l (400MB) and multiplying that number by 100 (40GB). Then set the n option a little higher to be on the safe side.
In my benchmarks, this method achieved a 6MB/s processing rate. You may combine this approach with the GNU parallel suggestion above to utilize multiple cores and achieve a higher throughput.
I don't have your data (or anything like it) handy, so I can't test this, but here's a proof of concept for you:
$ t='one\ntwo\nthree\none\nfour\nfive\n'
$ printf "$t" | nl -w14 -nrz -s, | sort -t, -k2 -u | sort -n | cut -d, -f2-
one
two
three
four
five
Our raw data includes one duplicated line. The pipes function as follows:
nl adds line numbers. It's a standard, low-impact unix tool.
sort the first time 'round sorts on the SECOND field -- what would have been the beginning of the line before nl. Adjust this as required for you data.
sort the second time puts things back in the order defined by the nl command.
cut merely strips off the line numbers. There are multiple ways to do this, but some of them depend on your OS. This one's portable, and works for my example.
Now... For obscenely large files, the sort command will need some additional options. In particular, --buffer-size and --temporary-directory. Read man sort for details about this.
I can't say I expect this to be fast, and I suspect you'll be using a ginormous amount of disk IO, but I don't see why it wouldn't at least work.
Assuming you can sort the file in the first place (i.e. that you can get sort file to work) then I think something like this might work (depends on whether a large awk script file is better then a large awk array in terms of memory usage/etc.).
sort file | uniq -dc | awk '{gsub("\"", "\\\"", $0); print "$0==\""substr($0, index($0, $1) + 2)"\"{x["NR"]++; if (x["NR"]>1){next}}"} END{print 7}' > dedupe.awk
awk -f dedupe.awk file
Which on a test input file like:
line 1
line 2
line 3
line 2
line 2
line 3
line 4
line 5
line 6
creates an awk script of:
$0=="line 2"{x[1]++; if (x[1]>1){next}}
$0=="line 3"{x[2]++; if (x[2]>1){next}}
7
and run as awk -f dedupe.awk file outputs:
line 1
line 2
line 3
line 4
line 5
line 6
If the size of the awk script itself is a problem (probably unlikely) you could cut that down by using another sentinel value something like:
sort file | uniq -dc | awk 'BEGIN{print "{f=1}"} {gsub("\"", "\\\"", $0); print "$0==\""substr($0, index($0, $1) + 2)"\"{x["NR"]++;f=(x["NR"]<=1)}"} END{print "f"}'
which cuts seven characters off each line (six if you remove the space from the original too) and generates:
{f=1}
$0=="line 2"{x[1]++;f=(x[1]<=1)}
$0=="line 3"{x[2]++;f=(x[2]<=1)}
f
This solution will probably run slower though because it doesn't short-circuit the script as matches are found.
If runtime of the awk script is too great it might even be possible to improve the time by sorting the duplicate lines based on match count (but whether that matters is going to be fairly data dependent).
I'd do it like this:
#! /bin/sh
usage ()
{
echo "Usage: ${0##*/} <file> [<lines>]" >&2
exit 1
}
if [ $# -lt 1 -o $# -gt 2 -o ! -f "$1" ]; then usage; fi
if [ "$2" ]; then
expr "$2" : '[1-9][0-9]*$' >/dev/null || usage
fi
LC_ALL=C
export LC_ALL
split -l ${2:-10000} -d -a 6 "$1"
for x in x*; do
awk '!x[$0]++' "$x" >"y${x}" && rm -f "$x"
done
cat $(sort -n yx*) | sort | uniq -d | \
while IFS= read -r line; do
fgrep -x -n "$line" /dev/null yx* | sort -n | sed 1d | \
while IFS=: read -r file nr rest; do
sed -i -d ${nr}d "$file"
done
done
cat $(sort -n yx*) >uniq_"$1" && rm -f yx*
(proof of concept; needs more polishing before being used in production).
What's going on here:
split splits the file in chunks of 10000 lines (configurable); the chunks are named x000000, x000001, ...
awk removes duplicates from each chunk, without messing with the line order; the resulting files are yx000000, yx000001, ... (since awk can't portably do changes in place)
cat $(sort -n yx*) | sort | uniq -d reassembles the chunks and finds a list of duplicates; because of the way the chunks were constructed, each duplicated line can appear at most once in each chunk
fgrep -x -n "$line" /dev/null yx* finds where each duplicated line lives; the result is a list of lines yx000005:23:some text
sort -n | sed 1d removes the first chunk from the list above (this is the first occurrence of the line, and it should be left alone)
IFS=: read -r file nr rest splits yx000005:23:some text into file=yx000005, nr=23, and the rest
sed -i -e ${nr}d "$file" removes line $nr from chunk $file
cat $(sort -n yx*) reassembles the chunks; they need to be sorted, to make sure they come in the right order.
This is probably not very fast, but I'd say it should work. Increasing the number of lines in each chunk from 10000 can speed things up, at the expense of using more memory. The operation is O(N^2) in the number of duplicate lines across chunks; with luck, this wouldn't be too large.
The above assumes GNU sed (for -i). It also assumes there are no files named x* or yx* in the current directory (that's the part that could use some cleanup, perhaps by moving the junk into a directory created by mktemp -d).
Edit: Second version, after feedback from #EtanReisner:
#! /bin/sh
usage ()
{
echo "Usage: ${0##*/} <file> [<lines>]" >&2
exit 1
}
if [ $# -lt 1 -o $# -gt 2 -o ! -f "$1" ]; then usage; fi
if [ "$2" ]; then
expr "$2" : '[1-9][0-9]*$' >/dev/null || usage
fi
tdir=$(mktemp -d -p "${TEMP:-.}" "${0##*/}_$$_XXXXXXXX") || exit 1
dupes=$(mktemp -p "${TEMP:-.}" "${0##*/}_$$_XXXXXXXX") || exit 1
trap 'rm -rf "$tdir" "$dupes"' EXIT HUP INT QUIT TERM
LC_ALL=C
export LC_ALL
split -l ${2:-10000} -d -a 6 "$1" "${tdir}/x"
ls -1 "$tdir" | while IFS= read -r x; do
awk '!x[$0]++' "${tdir}/${x}" >"${tdir}/y${x}" && \
rm -f "${tdir}/$x" || exit 1
done
find "$tdir" -type f -name 'yx*' | \
xargs -n 1 cat | \
sort | \
uniq -d >"$dupes" || exit 1
find "$tdir" -type f -name 'yx*' -exec fgrep -x -n -f "$dupes" /dev/null {} + | \
sed 's!.*/!!' | \
sort -t: -n -k 1.3,1 -k 2,2 | \
perl '
while(<STDIN>) {
chomp;
m/^(yx\d+):(\d+):(.*)$/o;
if ($dupes{$3}++)
{ push #{$del{$1}}, int($2) }
else
{ $del{$1} = [] }
}
undef %dupes;
chdir $ARGV[0];
for $fn (sort <"yx*">) {
open $fh, "<", $fn
or die qq(open $fn: $!);
$line = $idx = 0;
while(<$fh>) {
$line++;
if ($idx < #{$del{$fn}} and $line == $del{$fn}->[$idx])
{ $idx++ }
else
{ print }
}
close $fh
or die qq(close $fn: $!);
unlink $fn
or die qq(remove $fn: $!);
}
' "$tdir" >uniq_"$1" || exit 1
If there's a lot of duplication, one possibility is to split the file using split(1) into manageable pieces and using something conventional like sort/uniq to make a summary of unique lines. This will be shorter than the actual piece itself. After this, you can compare the pieces to arrive at an actual summary.
Maybe not the answer you've been looking for but here goes: use a bloom filter.
https://en.wikipedia.org/wiki/Bloom_filter This sort of problem is one of the main reasons they exist.
I have a file with a long list of integers:
10
4
66
....
I want to find the maximum value using UNIX command line tools. I know I can use sort (and indeed there are solutions to this problem on SO that use sort), but that's inefficient, requiring O(N*log(N)) and plenty of memory. With a simple for loop, I should be able to find the maximum value in O(N) and a couple of bytes of memory.
It seems there must be some program out there (with a name like max) that does this out of the box---is that true?
You can use this if no negative number is expected:
awk '$0>x{x=$0};END{print x}' input.txt
Use this to support negative numbers:
awk 'BEGIN{x=-2147483648};$0>x{x=$0};END{print x}' input.txt
Initializing x allows the solution to properly handle integer lists with values <= 0. See the comments for more details.
max=1
while read i
do
if [[ "$i" > "$max" ]]; then
max="$i"
fi
done < a.txt
echo "$max" > b.txt
a.txt is the input file(with an integer on each line).
b.txt contains the maximum of the integers in a.txt.
awk '{if($1>a)a=$1;}END{print a}' temp3
sort -nr inputfile.txt | head -1
where inputfile.txt contains all numbers.