基于bash中特定列的排序错误

Question

Hi I am trying this thing but it doesn't work.

I know that it doesn't work because each line has different number of columns when words are separated by space but can we do the intended job any way.

Answer 1

NOTE: assuming the input file's columns are separated by spaces and not tabs, otherwise dan's comment - sort -nt $'\t' -k3,3 - should suffice

---

sort allows us to designate the field terminator as well as which fields (and optionally substrings of fields) to sort by.

If we set the field delimiter as a linefeed ( \n ) the entire line becomes a single field.

From here we can designate a substring of field #1 to sort by; -k1.x,1.y says to sort by field #1 from position x to position y (with the first character of the field/line having a position of 1 ).

Sample input:

$ cat animals.txt
         1         2         3         4         5         6
123456789012345678901234567890123456789012345678901234567890
alpaca   Intermediate Perl         2012   Schwatz, Randal
donkey   Cisco IOS in a Nutshell   2005   Boney, James
horse    Linux in a Nutshell       2009   Siever, Ellen

Where:

• the first 2 lines (the scale) do not exist in the file; the scale shows us ...
• the year part of the line runs from position 36 to 39

Pulling all of this into a sort call:

# sort numerically by year (ascending)

$ sort -t$'\n' -k1.36,1.39 -n animals.txt
donkey   Cisco IOS in a Nutshell   2005   Boney, James
horse    Linux in a Nutshell       2009   Siever, Ellen
alpaca   Intermediate Perl         2012   Schwatz, Randal

# sort numerically by year (descending)

$ sort -t$'\n' -k1.36,1.39 -rn animals.txt
alpaca   Intermediate Perl         2012   Schwatz, Randal
horse    Linux in a Nutshell       2009   Siever, Ellen
donkey   Cisco IOS in a Nutshell   2005   Boney, James

NOTE: assumes all lines have the year in the same position (ie, the contents of the file are formatted per a fixed-width scheme)

Obviously this approach requires we know the position of the year substring in advance; there are a few ways to determine this position ... one idea, assuming the year column will always be the 1st occurrence of a 4-digit substring ... use bash regex matching and the BASH_REMATCH[] array to determine the length of the line up to the 4-digit year , eg:

$ regex="^([^0-9]*)([0-9]{4}).*"
$ [[ $(head -1 animals.txt) =~ $regex ]] && typeset -p BASH_REMATCH
declare -ar BASH_REMATCH=([0]="alpaca   Intermediate Perl         2012   Schwatz, Randal" [1]="alpaca   Intermediate Perl         " [2]="2012")

From this we see that the BASH_REMATCH[1] contains the contents of the line up to the year ( 2012 for the alpaca line); now we grab the length of BASH_REMATCH[1] and add +1/+3 to get our x and y values:

$ (( x = ${#BASH_REMATCH[1]} + 1 ))
$ (( y = x + 3 ))
$ typeset -p x y
declare -- x="36"
declare -- y="39"

Plugging these variables into our previous sort call:

# sort numerically by year (ascending)

$ sort -t$'\n' -k1.${x},1.${y} -n animals.txt
donkey   Cisco IOS in a Nutshell   2005   Boney, James
horse    Linux in a Nutshell       2009   Siever, Ellen
alpaca   Intermediate Perl         2012   Schwatz, Randal

# sort numerically by year (descending)

$ sort -t$'\n' -k1.${x},1.${y} -rn animals.txt
alpaca   Intermediate Perl         2012   Schwatz, Randal
horse    Linux in a Nutshell       2009   Siever, Ellen
donkey   Cisco IOS in a Nutshell   2005   Boney, James

NOTE: OP hasn't defined a secondary sort requirement in the case of multiple lines having the same date but it shouldn't be too hard to extend this answer to include a secondary (and tertiary?) sort requirement

Answer 2

Try adding a seperator like a comma, as from there you will be able to use the sort command with the -t argument and specify the given field separator.

To find and replace a character with a seperator I would use cat animals.txt | sed {insert the pattern} cat animals.txt | sed {insert the pattern} .

Based on the file you've shared, you could attempt addding the seperator after the first word, and before and after the numerical values.

Answer 3

NOTE: assuming the input file's columns are separated by spaces and not tabs, otherwise dan's comment - sort -nt $'\t' -k3,3 - should suffice

---

If GNU awk is available we can have awk find the index for the year substring and then sort the output for us.

Sample input:

$ cat animals.txt
         1         2         3         4         5         6
123456789012345678901234567890123456789012345678901234567890
alpaca   Intermediate Perl         2012   Schwatz, Randal
donkey   Cisco IOS in a Nutshell   2005   Boney, James
horse    Linux in a Nutshell       2009   Siever, Ellen

Where:

• the first 2 lines (the scale) do not exist in the file; the scale shows us ...
• the year part of the line runs from position 36 to 39

One GNU awk idea:

awk '
FNR==1 { x=match($0, /[0-9]{4}/) }                # find index of the "year" substring in the 1st line of input; assumes the "year" is the 1st occurrence of a 4-digit substring
       { arr[substr($0,x,4)][FNR]=$0 }            # populate 2-dimensional array using "year" and row number (FNR) as indexes
END    { PROCINFO["sorted_in"]="@ind_num_asc"     # sort indexes as numbers in "asc"ending order
         for (i in arr)
             for (j in arr[i])
                 print arr[i][j]
       }
' animals.txt

This generates:

donkey   Cisco IOS in a Nutshell   2005   Boney, James
horse    Linux in a Nutshell       2009   Siever, Ellen
alpaca   Intermediate Perl         2012   Schwatz, Randal

If we change the sort order from @ind_num_asc to @ind_num_desc we can generate the output in descending year order, ie:

alpaca   Intermediate Perl         2012   Schwatz, Randal
horse    Linux in a Nutshell       2009   Siever, Ellen
donkey   Cisco IOS in a Nutshell   2005   Boney, James

NOTES:

• GNU awk required for multi-dimensional array (aka array of arrays) support
• GNU awk required for the PROCINFO["sorted_in"] feature
• assumes the entire file can fit into memory (due to storing all lines in the array)

Answer 4

One way to do it is to copy the year to the start of each line with sed , sort the resulting output numerically, and then remove the year at the start of each line:

sed 's/^.*[[:space:]]\([12][09][0-9][0-9]\)[[:space:]].*$/\1 &/' animals.txt \
    | sort -n | sed 's/^.....//'

The output with the example animals.txt in the question is:

oryx    Writing Word Macros     1999    Roman, Steven
donkey  Cisco IOS in a Nutshell 2005    Boney, James
snail   SSH, The Secure Shell   2005    Barrett, Daniel
horse   Linux in a Nutshell     2009    Sievers, Ellen
python  Programming Python      2010    Lutz, Mark
alpaca  Intermediate Perl       2012    Schwartz, Randal
robin   MySQL High Availability 2014    Bell, Charles