Bulk Rename

[swestrup]

Am I right that there isn't a half-decent bulk rename command out there for Linux?  I did a search and found a bunch, but they all rely on regex which is not terribly useful when what you want to do is pull a number out of the middle of a file name, do math on it, and depending on the result replace it with a string containing a different embedded number.

(Okay, maybe this can be done with Perls ultra-extended regular expressions, but after rereading how they work a dozen times, and trying 30 different patterns, I gave up as the documentation is inadequate and I haven't been able to figure out how to express this.)

This, by-the-way is what you want to do if you have 1000 files numbered sequentially, from 10 years of a TV show, and you want to rename them as season and episode numbers, and the number of episodes per season was not a constant...

Okay, I know this is an edge case, but if I were to write a bulk rename, this would be one of the things I would, for certain, want to make easy.  Anything less complex and you can just throw together a one-line bash for-loop to handle it.

Comments

[sps.livejournal.com]

I'd use Emacs to do it quasi-interactively, or Perl—there's a fully functional rename, it wouldn't have to be a regexp at all. Or am I missing something?

Just how hairy is the mapping you need?

[swestrup]

> Or am I missing something?

Just that its on the border between what the command line stuff will handle easily, and what you need to write a script for.

Frankly, its annoying to have to write a script because I have to figure out how to use a few dozen commands and weave them together. Same with bash.

In the end I broke the problem into pieces by first moving each group into its own subdir and then running slightly-modified 1-line commands on each subdir. But frankly, when I write a for loop that passes something in turn through echo, cut, tr, and xargs 'mv', I start thinking there has to be a better way...

(see, I need to process the files in numerical sort order, despite there being a variable number of digits in the numbers...)

[sps.livejournal.com]

perl -e 'for (sort {$a<=>$b} <*>) /\(\d+\)\.sthg/ and rename $_, (100+$1).".sthgelse"' ishly?

[swestrup]

I hate perl -e because it causes me to have to worry about at least two levels of different escapes at the same time. As a result it takes me forever to get anything that works.

[joenotcharles.livejournal.com]

Why the fetish for doing everything on one line? Put your script in a text file, run the text file, fix mistakes, delete the text file when you're done, or save it somewhere if you think it'll be useful again.

If it's complicated enough that it takes you significant brain power to figure out how to shove it into a one-liner, or to work out the escaping involved, then the extra time taken to write to and delete a text file goes toward saving you from that, and it makes debugging the inevitable mistakes so much easier.

[skjalm.livejournal.com]

Agreed! Doing one liners (in whatever language) is fun, but sometimes it's not worth the extra hassle if the point of the exercise is to get something done rather than to do a nifty one liner ;)

That being said I couldn't help myself (see comment below) ;)

[swestrup]

Your mileage may vary, of course, but I find it MUCH faster to just shove everything onto one line, and have the final generated command echoed rather than run. No edit, save, try cycle. I just move the cursor to the middle of the line, change something and hit RETURN to try it out.

[swestrup]

Oh, BTW, I am now using the CRT you gave me before you left, and it has this annoying tendency to go Bzzzt! and fade the screen for a moment or two.

I vaguely recall you mentioning it when you gave it me, but not what it was you said. Should I be wearing lead underwear?

[sps.livejournal.com]

I'm afraid I'm not remembering the that. I never wore lead shorts, but then again, I haven't had any kids (that I know of) whereof to count the heads.

[sps.livejournal.com]

It has been a while.

[skjalm.livejournal.com]

This is not very pretty, but gets the job done (example files shown before and after the move)

steelbook:foo steellynx$ ls
abc0	abc14	abc2	abc25	abc30	abc36	abc41	abc47	abc52	abc58
abc1	abc15	abc20	abc26	abc31	abc37	abc42	abc48	abc53	abc59
abc10	abc16	abc21	abc27	abc32	abc38	abc43	abc49	abc54	abc6
abc11	abc17	abc22	abc28	abc33	abc39	abc44	abc5	abc55	abc7
abc12	abc18	abc23	abc29	abc34	abc4	abc45	abc50	abc56	abc8
abc13	abc19	abc24	abc3	abc35	abc40	abc46	abc51	abc57	abc9
steelbook:foo steellynx$ find . -type f|perl -p -e 's/([^\d]*)(\d+)(.*)$/mv $1$2$3 `printf series_s%02dx%02d \$[$2%12] \$[$2\/12]`$3/g' | /bin/bash
steelbook:foo steellynx$ ls
series_s00x00	series_s02x02	series_s04x04	series_s07x01	series_s09x03
series_s00x01	series_s02x03	series_s05x00	series_s07x02	series_s09x04
series_s00x02	series_s02x04	series_s05x01	series_s07x03	series_s10x00
series_s00x03	series_s03x00	series_s05x02	series_s07x04	series_s10x01
series_s00x04	series_s03x01	series_s05x03	series_s08x00	series_s10x02
series_s01x00	series_s03x02	series_s05x04	series_s08x01	series_s10x03
series_s01x01	series_s03x03	series_s06x00	series_s08x02	series_s10x04
series_s01x02	series_s03x04	series_s06x01	series_s08x03	series_s11x00
series_s01x03	series_s04x00	series_s06x02	series_s08x04	series_s11x01
series_s01x04	series_s04x01	series_s06x03	series_s09x00	series_s11x02
series_s02x00	series_s04x02	series_s06x04	series_s09x01	series_s11x03
series_s02x01	series_s04x03	series_s07x00	series_s09x02	series_s11x04

As mentioned it's not pretty, but does have the advantage, imho, that I could build the mv commands and see that they looked sane before throwing them into a new shell. Plus... it's late so I have apparently made the classic mistake of swapping mod and div in the command *head+desk*

I'll shut up and go to bed now ;)