UNIX for Dummies Questions & Answers

Is there any way using the sort command to get something like a GROUP BY clause?

I'm trying to sort through race_event records and group them by those individuals that raced together in the same race on the same date at the same track, but i run into a problem when there were days when the track had qualifying races, plus regular races on the same day; so there could be two Race 1's, Race 2's, etc for the same date at the same track.
I tried a sort like:

Code:

sort -t':' -k3 -k4 -k9 -k29 -k8 -k28 ~/Desktop/1998/AllRaceEvents98.txt > ~/Desktop/sortedevents.txt

where
k3 is a date
k4 is the racetrack
k9 is the class of race (would show it was a qualifier or a regular race)
k29 is the race number
k8 is the purse of race (would show 0 for qualifier and dollar amount for regular race)
k28 is the names of the top three finishers

but this does not sort right: it does not group them together into their respective races correctly. Any suggestions?

Without seeing the data, I can only simulate. Perhaps the following example will help illustrate:

sample file

Code:

> cat file99
joe:1:alpha:4:5
jim:4:beta:16:9
bette:2:comma:99:0
dennis:3:per:12:12
harry:2:comma:12:13
olivia:1:beta:4:4

Trying to sort based on key field 3 then field 2. Field 3 should put the two beta people together and then the second field should put olivia before jim based on the 1 and 4 in their records. This did not happen.

Code:

> sort -t: -k3 -k2 file99
joe:1:alpha:4:5
jim:4:beta:16:9
olivia:1:beta:4:4
harry:2:comma:12:13
bette:2:comma:99:0
dennis:3:per:12:12

I then tried by putting explicit field definitions (start and end locations). So I am sorting on the first 3 characters of field 3 to get the beta people together. Then sort on field 2 - getting olivia before jim.

Code:

> sort -t: -k3.1,3.3 -k2 file99
joe:1:alpha:4:5
olivia:1:beta:4:4
jim:4:beta:16:9
harry:2:comma:12:13
bette:2:comma:99:0
dennis:3:per:12:12

i tried doing a sort using explicit definition as suggested on a couple of fields but again i could not get it to come out right.

I think i don't understand enough about how sort actually works: when i specify certain fields to sort on, what does it do sorting-wise with the fields i have not specified?

I'm beginning to think it's also sorting on a field i have not specified and that is why it is messing up the grouping.

Could you post sample input and example of the desired output?

Code:

Data records deleted

I'd like to get all the individuals that raced in the same race, on the same date at the same track together, but you'll notice after i do the sort in the instance above two of the individuals (bolded) ended up getting grouped with another race down on lines 616120 and 616121 when they should have followed the other individuals in their race at lines 616072 and 616073.

I hope that makes some sort of sense....

Sort is normally alphabetic - thus a field with a 12 will be before a 2 or 3. You will need the 'n' option on the field to make numeric.

To help determine where your script has gone astray, I would suggest you verify that your logic is fine for just one key. Then, add a second key; and so on until you discover the problem.

And from more investigation, I do not see any records improperly sorted.

Code:

> sort file200 -k3n -k4.1,4.3 -k9.1,9.10 -k29n -k8.1,8.2 -k28.1,28.10 | cut -d":" -f2-4,8-9,28-29

{data records here --- removed once issue solved --- not part of issue}

>

Thanks for this explanation, i think the 'n' option was the key to fixing my sorting problem.

Sorry i think my ". . ." between the records did not properly indicate that there were 49 other records between the two sets of records that actually needed to be together.

Anyhow, thanks for all your help, you steered me to a solution.