asked on

perl parsing file - conditionally splitting to new file(s)

I have a text file which is as follows

000001      $$1orange$$2apple2$$3pear
000002      $$1orange$$2apple2$$3pear
000003      $$1orange$$2apple2$$3pear
000002      $$1pear$$2apple2$$3orange
000003      $$1elephant$$2tiger$$3giraffe
000003      $$1elephant$$2tiger$$lion
000003      $$1elephant$$2monkey$$lion

The first column (i.e. 000001) is the record ID. The subfields are then denoted with the $$. So, in the first line, subfield1 is orange, subfield2 is apple, subfield3 is pear

I'm wanting (and the file could be a few hundred lines or just a couple) to split the file based on the record ID, and subfield 2.

So, the first time it encounters a unique ID/subfield 2 pair (e.g. 000001 / apple) it should write out to one. The second (and subsequent) times it finds this combo, it should write out to a different file.

Also, if it finds a repeat of the ID, but subfield 2 is different - each occurrence of this should write to a separate file.

For example, the file above would be parsed into the following files:

File1 (the first occurrence of ID/Subfield2 being seen):

000001      $$1orange$$2apple2$$3pea
000002      $$1orange$$2apple2$$3pear

File2 (second occurrence of an ID being seen, but subfield2 is different):
000003      $$1elephant$$2tiger$$3giraffe

File3 (third occurrence of an ID being seen, but subfield2 is different):
000003      $$1elephant$$2monkey$$3giraffe

LastFile (2nd occurrence of ID/Subfield2 being seen):
000003      $$1orange$$2apple2$$3pear
000002      $$1pear$$2apple2$$3orange
000003      $$1elephant$$2tiger$$lion

How can this be achieved in perl?

any help much appreciated (as always!)

ozo

perl -F'/\s*\$\$/' -ape '$f=++$f{"@F[0,2]"};open STDOUT,">>$f" or warn "$f $!"' file

yelbow

ASKER

Hi Ozo,

Thanks for this - however, it doesn't quite work (and looking back my example may have been a little dodgy)

It gives me the following five files:

'000001$$1orange'

000001 $$1orange$$2apple2$$3pear

cat '000002$$1orange'

000002 $$1orange$$2apple2$$3pear

cat '000003$$1orange'

000003 $$1orange$$2apple2$$3pear

cat '000002$$1pear'

000002 $$1pear$$2apple2$$3orange

cat '000003$$1elephant'

000003 $$1elephant$$2tiger$$3giraffe
000003 $$1elephant$$2tiger$$lion
000003 $$1elephant$$2monkey$$lion

The output of the first 3 files should all be combined in 1, as it's the first time each of them has been seen.

The fourth file is fine, as it's the second time the ID 000002 has appeared, but subfield 2 is different.

The 5th file has "000003 / tiger" which hasn't been seen before, so should be in the first file etc

ozo

I think http:#a39719995 is now corrected, but I'm still a little confused about the example.

yelbow

ASKER

Hi,

That now gives:
cat 2

000002 $$1pear$$2apple2$$3orange
000003 $$1elephant$$2tiger$$lion

cat 1

000001 $$1orange$$2apple2$$3pear
000002 $$1orange$$2apple2$$3pear
000003 $$1orange$$2apple2$$3pear
000003 $$1elephant$$2tiger$$3giraffe
000003 $$1elephant$$2monkey$$lion

which isn't quite right. You're absolutely right about the example though - so, I'll try that again (sorry!)

Original File:

000001 $$1orange$$2apple2$$3pear
000002 $$1orange$$2apple2$$3pear
000003 $$1orange$$2apple2$$3pear
000002 $$1pear$$2apple2$$3orange
000003 $$1elephant$$2tiger$$3giraffe
000003 $$1elephant$$2tiger$$3lion
000003 $$1elephant$$2monkey$$3lion
000003 $$1lion$$2monkey$$3giraffe

Resulting files:

File 1 (the first occurrence of each pair of ID/subfield2):

000001 $$1orange$$2apple2$$3pear
000002 $$1orange$$2apple2$$3pear
000003 $$1orange$$2apple2$$3pear

File 2 (second occurrence of an ID being seen, but subfield2 is different)

000003 $$1elephant$$2tiger$$3giraffe

File 3 (third occurrence of an ID being seen, but subfield2 is different)

000003 $$1elephant$$2monkey$$lion

File 4 (the second (and subsequent) occurences of an ID/subfield2 pair)

000002 $$1pear$$2apple2$$3orange
000003 $$1elephant$$2tiger$$lion
000003 $$1lion$$2monkey$$3giraffe

Does that help? Sorry if I'm being unclear.

The ultimate aim is to only have the ID appear once in each file, unless it's in the final file which is a catchall of the second and subsequent occurrences of the ID/Subfield2 pairings.

ozo

Now it sounds like you are describing something like
perl -F'/\s*\$\$/' -ape '$i=++$i{$F[0]};$f=++$f{"@F[0,2]"};open STDOUT,">>$i.$f" or warn "$i.$f $!"'
But I'm still confused
00003 $$1elephant$$2tiger$$3giraffe is the first occurrence of the pair 000003/2tiger, so why isn't it in the first file?

yelbow

ASKER

Hi,

It's because it's the second occurrence of the ID 00003 - and in all files bar the "catchall" file (where its seen both the id and subfield2 before), the ID can only exist once.

I'm, roughly, trying to achieve this:

For each line
	if I've seen this ID before then
		if I've seen this ID/Subfield2 combo before then
			write out to a single catchall file
		else
			write out to new filex
		endif
	else
		Write out to file1
	endif
end for

Open in new window

ozo

Here's another interpretation:
perl -F'/\s*\$\$/' -ape '$f=$f{"@F[0,2]"}++||$F[2];open STDOUT,">>$f" or warn "$f $!"'

yelbow

ASKER

Ooh, that's nearly there (thanks so much for your patience), only it writes out *all* of the lines to the "catchall" file?

ASKER CERTIFIED SOLUTION

ozo

membership

This solution is only available to members.

To access this solution, you must be a member of Experts Exchange.

Start Free Trial

ozo

Are you sure catchall file and new file aren't switched in http:#a39720116?
It matches your most recent example better when they are switched.

yelbow

ASKER

That seems perfect, thanks so much for your help and patience

ozo

perl -F'/\s*\$\$/' -ape '$s=$f{"@F[0,2]"}++;$f=$f{$F[0]}++?$s?"catchcall":++$n:"file1";open STDOUT,">>$f" or warn "$f $!"'