Re: Deleting a subfield using MARC::Record
On May 1, 2006, at 4:41 PM, Leif Andersson wrote: +1 count can possibly be complemented or replaced with occurrence as suggested. It'd be nice to be able to denote last occurrence [-1]. And I suppose the indexing should be based on ordinary perl subscript indexing - i.e. governed by the value of special variable $[ $field-delete_subfield( code = $code, # of course occur = [0,2,3], # occur or pos or whatever... match = qr/pat/, # doesn't need to be repeatable ); I actually like 'pos' better than 'occur' -- but alas $field-delete_subfield(pos = 2); won't work because 'pos' is a perl keyword--which is why I like using it I suppose :-) How about: $field-delete_subfield(position = 2); A bit more wordy I guess, but I still like it better than occur. Nice tip on the use of $[ by the way! I also like Tim's suggestion to allow 'code' to take multiple values too: $field-delete_subfield(code = ['a','b','c']) So if you check out the CVS you should find this implemented. If you are interested in adding any tests or documentation let me know and I'll add you as a sf.net developer. The current documentation for the new method reads like this: -- delete_subfield() allows you to remove subfields from a field: # delete any subfield a in the field $field-delete_subfield(code = 'a'); # delete any subfield a or u in the field $field-delete_subfield(code = ['a', 'u']); If you want to only delete subfields at a particular position you can use the position parameter: # delete subfield u at the first position $field-delete_subfield(code = 'u', position = 0); # delete subfield u at first or second position $field-delete_subfield(code = 'u', position = [0,1]); You can specify a regex to for only deleting subfields that match: # delete any subfield u that matches zombo.com $field-delete_subfield(code = 'u', match = qr/zombo.com/); -- Sound ok? //Ed
Re: Deleting a subfield using MARC::Record
On May 3, 2006, at 6:28 AM, Edward Summers wrote: $field-delete_subfield(pos = 2); won't work because 'pos' is a perl keyword-- I should've tried it before I said this -- it works fine in that context, even though my perl syntax highlighter indicates otherwise. So I've changed the parameter name from 'position' to 'pos' keeping with Leif's original suggestion. //Ed
Re: Deleting a subfield using MARC::Record
Edward Summers wrote: The current documentation for the new method reads like this: -- delete_subfield() allows you to remove subfields from a field: # delete any subfield a in the field $field-delete_subfield(code = 'a'); # delete any subfield a or u in the field $field-delete_subfield(code = ['a', 'u']); If you want to only delete subfields at a particular position you can use the position parameter: # delete subfield u at the first position $field-delete_subfield(code = 'u', position = 0); # delete subfield u at first or second position $field-delete_subfield(code = 'u', position = [0,1]); If you implemented negative indexes, it would be nice to add an example: # delete subfield u at last position $field-delete_subfield(code = 'u', pos = [-1]); You can specify a regex to for only deleting subfields that match: # delete any subfield u that matches zombo.com $field-delete_subfield(code = 'u', match = qr/zombo.com/); The term position (pos) seems a little ambiguous to me on the face of it. Does (code = 'u', pos = 0) mean the first subfield u (which is what I take it to mean) or subfield u if it's the first subfield (which it might sound like outside the context of this discussion)? Mike -- Michael Kreyche Systems Librarian Associate Professor Kent State University Libraries and Media Services http://www.personal.kent.edu/~mkreyche 330-672-1918
Re: Deleting a subfield using MARC::Record
Brad Baxter wrote: On 5/3/06, Michael Kreyche [EMAIL PROTECTED] wrote: The term position (pos) seems a little ambiguous to me on the face of it. Does (code = 'u', pos = 0) mean the first subfield u (which is what I take it to mean) or subfield u if it's the first subfield (which it might sound like outside the context of this discussion)? I had the same thought. To me, 'occur' has a clearer meaning in that context, the zeroth occurrence of subfield 'u', while 'pos' has more of the ambiguity described above. So of the two, I'd prefer 'occur', but I can live with 'pos'. Synonyms perhaps? (Unless someone has a need to delete the third subfield regardless of code? I never have, so perhaps not.) -- Brad I think it should mean the zeroth occurrence of subfield 'u', since specifying which of a repeated group of subfields is a realistic task, as you say. For example, each record has two 'u's but all of the first ones are garbage. Mark -- Mark Jordan Head of Library Systems W.A.C. Bennett Library, Simon Fraser University Burnaby, British Columbia, V5A 1S6, Canada Phone (604) 291 5753 / Fax (604) 291 3023 [EMAIL PROTECTED] / http://www.sfu.ca/~mjordan/
Re: Deleting a subfield using MARC::Record
On May 3, 2006, at 8:55 AM, Mark Jordan wrote: I think it should mean the zeroth occurrence of subfield 'u', since specifying which of a repeated group of subfields is a realistic task, as you say. For example, each record has two 'u's but all of the first ones are garbage. Actually 'pos' as implemented will remove the subfield u if it is at position n in the field. So we could have occurrence too. I feel like I'm chasing windmills a bit. Do y'all really *need* all this functionality in delete_subfield() :-) I guess you do or else you wouldn't be so interested in asking for it. I didn't implement the -1 behavior because i wasn't quite sure how to do it quickly, and it seemed like too much somehow. //Ed
Re: Deleting a subfield using MARC::Record
Edward Summers wrote: On May 3, 2006, at 8:55 AM, Mark Jordan wrote: I think it should mean the zeroth occurrence of subfield 'u', since specifying which of a repeated group of subfields is a realistic task, as you say. For example, each record has two 'u's but all of the first ones are garbage. Actually 'pos' as implemented will remove the subfield u if it is at position n in the field. So we could have occurrence too. I feel like I'm chasing windmills a bit. Do y'all really *need* all this functionality in delete_subfield() :-) I guess you do or else you wouldn't be so interested in asking for it. Well, maybe this IS getting a little out of hand! I could live with the old-fashioned way myself. Being a newbie to the list I was surprised how fast you jumped in and provided the new functionality. Mike -- Michael Kreyche Systems Librarian Associate Professor Kent State University Libraries and Media Services http://www.personal.kent.edu/~mkreyche 330-672-1918
Re: Deleting a subfield using MARC::Record
Ed, the only problem I can see with position in the field is if a preceding subfield does not exist in every record. For example, in a given batch, most but not all records have an 856 subfield 3, followed by multiple subfield u's. If you ask to delete the first u using pos, then your target will be different determined by the presence of subfield 3. If you know that you want to eliminate u's (without regard to what else is in the field) then your target would be easier to hit. However, you raise a good point -- how much functionality do people need? Maybe some actual examples from the wild would be useful. I can supply some but probably not until tomorrow afternoon since I have a presentation to prepare for tomorrow. If other users have some examples of real records or use cases they might clarify the most common usage. I'll see what I can find tomorrow. Mark Edward Summers wrote: On May 3, 2006, at 8:55 AM, Mark Jordan wrote: I think it should mean the zeroth occurrence of subfield 'u', since specifying which of a repeated group of subfields is a realistic task, as you say. For example, each record has two 'u's but all of the first ones are garbage. Actually 'pos' as implemented will remove the subfield u if it is at position n in the field. So we could have occurrence too. I feel like I'm chasing windmills a bit. Do y'all really *need* all this functionality in delete_subfield() :-) I guess you do or else you wouldn't be so interested in asking for it. I didn't implement the -1 behavior because i wasn't quite sure how to do it quickly, and it seemed like too much somehow. //Ed -- Mark Jordan Head of Library Systems W.A.C. Bennett Library, Simon Fraser University Burnaby, British Columbia, V5A 1S6, Canada Phone (604) 291 5753 / Fax (604) 291 3023 [EMAIL PROTECTED] / http://www.sfu.ca/~mjordan/
Re: Deleting a subfield using MARC::Record
On May 3, 2006, at 11:25 AM, Mark Jordan wrote: For example, in a given batch, most but not all records have an 856 subfield 3, followed by multiple subfield u's. If you ask to delete the first u using pos, then your target will be different determined by the presence of subfield 3. If you know that you want to eliminate u's (without regard to what else is in the field) then your target would be easier to hit. Ok this I like, having a use case like this makes it much easier to decide about the API. How about we go back to occurrence and remove pos? However, you raise a good point -- how much functionality do people need? Maybe some actual examples from the wild would be useful. I can supply some but probably not until tomorrow afternoon since I have a presentation to prepare for tomorrow. If other users have some examples of real records or use cases they might clarify the most common usage. I'll see what I can find tomorrow. Yeah, you know if you have the interest/time it would be great if you could add a couple tests to the existing test file. The tests need not pass, but they should illustrate they should illustrate the use. Hop onto #code4lib and I can walk you through how to do this, and get your access set up if you are interested. //Ed
Re: Deleting a subfield using MARC::Record
On 5/1/06, Edward Summers [EMAIL PROTECTED] wrote: On May 1, 2006, at 1:24 PM, Brad Baxter wrote: # delete first two subfield u $field-delete_subfield(code = 'u', count = 2); I don't think I like it this way. How would you delete just the second one? I'd rather see 'count' mean 'occurrence', so the above would mean delete the second subfield u. Yeah, I like your suggestion of using 'occurrence' over 'count' here. And ... # delete second and third subfield u $field-delete_subfield(code = 'u', count = 2, count = 3); This won't translate very well to passing in arguments as a hash, since the second count will stomp on the first. Have you actually had to any real need to delete this way in the past? Of course. I was thinking too shallowly. The suggestions for occur = [1,2] would do the trick. -- Brad
Re: Deleting a subfield using MARC::Record
++ sorry, couldn't resist
Re: Deleting a subfield using MARC::Record
+1 I like Leif's proposal. It also might be useful to allow code to accept multiple values. -Tim Timothy Prettyman LIT/LIbrary Systems University of Michigan On May 1, 2006, at 5:41 PM, Leif Andersson wrote: +1 count can possibly be complemented or replaced with occurrence as suggested. It'd be nice to be able to denote last occurrence [-1]. And I suppose the indexing should be based on ordinary perl subscript indexing - i.e. governed by the value of special variable $[ $field-delete_subfield( code = $code, # of course occur = [0,2,3], # occur or pos or whatever... match = qr/pat/, # doesn't need to be repeatable ); Leif == Leif Andersson, Systems Librarian Stockholm University Library SE-106 91 Stockholm SWEDEN Phone : +46 8 162769 Mobile: +46 70 6904281
Re: Deleting a subfield using MARC::Record
Edward Summers wrote:
On Apr 29, 2006, at 10:31 AM, Mark Jordan wrote:
Maybe other people should verify the usefulness of a delete subfield
function before anyone does anything about it, though. Would a half
dozen +1 votes from perl4libers validate its usefulness?
Yes it would...but to get the changes out on CPAN we'd all need to
convince Mike O'Regan who holds the keys to MARC-Record on CPAN. Right
now the version on CPAN is not the latest version that's available in
CVS on SourceForge. I think Mike has some performance concerns about the
Unicode handling code in v2.0, since MARC::Record is used in some of his
time critical applications. Hopefully we can benchmark v1.38 and v2.0
sometime and get the latest code pushed out to CPAN if things look OK.
Or if this is not an option there is always the possibility of creating
a MARC::Record2 on CPAN. That would be kind of ugly for backwards
compatibility though.
OK -- here's the call for a vote. All interested perl4lib members are
encouraged to participate by emailing the list.
Proposal: Incorporate the functions for deleting specific subfields into
MARC-Record v2.0, based on examples Ed supplies below.
Process: +1 = Yes for Proposal, anything else = No and must be
accompanied by reasons not to incorporate these features into v2.0.
Deadline: Midnight (Greenwich/UTC), Thursday May 4.
Mark
At any rate if you are open to using the latest/greatest code from
SourceForge there is nothing stopping us from getting a delete_subfield
method working.
$field-delete_subfield('a', 0);
In the interests of flexibility how about using hash key/value pairs?
Here's what I'm thinking:
# delete all subfield u
$field-delete_subfield(code = 'u');
# delete first two subfield u
$field-delete_subfield(code = 'u', count = 2);
# delete all subfield u that have 'zombo.com' in them
$field-delete_subfield(code = 'u', match = qr/zombo\.com/);
# delete only the first subfield u that has 'zombo.com' in it
$field-delete_subfield(code = 'u', match = qr/zombo\.com', count = 1);
# delete any subfield with a value that matches 'zombo.com'
$field-delete_subfield(match = qr/zombo\.com/);
So effectively there is AND boolean logic between any options that are
supplied. If this looks good I've got some code that does it and some
tests committed in SourceForge for you to take a look at [1]. I'm open
to suggestions on renaming parameters, etc... Here are the basics for
checking out the code if you haven't done it before.
cvs -d:pserver:[EMAIL PROTECTED]:/cvsroot/marcpm login
cvs -z3 -d:pserver:[EMAIL PROTECTED]:/cvsroot/marcpm co
-P marc-record
//Ed
[1] http://sourceforge.net/cvs/?group_id=1254
RE: Deleting a subfield using MARC::Record
OK -- here's the call for a vote. All interested perl4lib members are encouraged to participate by emailing the list. +1 Bryan Baldus [EMAIL PROTECTED] [EMAIL PROTECTED] http://home.inwave.com/eija
Re: Deleting a subfield using MARC::Record
+1 :-) //Ed
Re: Deleting a subfield using MARC::Record
+1 //Gary -- === Gary Bertchume Library Systems Analyst Columbia University Libraries 212 854-8582 [EMAIL PROTECTED] ===
Re: Deleting a subfield using MARC::Record
On 5/1/06, Bryan Baldus [EMAIL PROTECTED] wrote: On Monday, May 01, 2006 1:24 PM, Brad Baxter wrote: On 4/30/06, Edward Summers [EMAIL PROTECTED] wrote: [snip] # delete first two subfield u $field-delete_subfield(code = 'u', count = 2); I don't think I like it this way. How would you delete just the second one? I'd rather see 'count' mean 'occurrence', so the above would mean delete the second subfield u. And ... # delete second and third subfield u $field-delete_subfield(code = 'u', count = 2, count = 3); When I looked at it, I also wondered about 'count' being 'position' or something like that, to be able to note which occurrence. It might be useful to have both--if you know it is always the 1st 2 occurrences of the subfields, use 'count', but if you know it is the 1st and 3rd, then use 'position' or 'occurrence'. examples: #remove 1st 2 subfield 'u' $field-delete_subfield(code = 'u', count = 2); #remove 1st and 3rd subfield 'u' $field-delete_subfield(code = 'u', occurence = (0, 2)); #or (1, 3) #remove last subfield u $field-delete_subfield(code = 'u', occurence = (-1)); I guess those work for me, as long as it's 'occurrence'. :-) Or perhaps just 'occur'. (If we allow -1, then I agree it should start at 0.) -- Brad
Re: Deleting a subfield using MARC::Record
On May 1, 2006, at 1:24 PM, Brad Baxter wrote: # delete first two subfield u $field-delete_subfield(code = 'u', count = 2); I don't think I like it this way. How would you delete just the second one? I'd rather see 'count' mean 'occurrence', so the above would mean delete the second subfield u. Yeah, I like your suggestion of using 'occurrence' over 'count' here. And ... # delete second and third subfield u $field-delete_subfield(code = 'u', count = 2, count = 3); This won't translate very well to passing in arguments as a hash, since the second count will stomp on the first. Have you actually had to any real need to delete this way in the past? //Ed
Re: Deleting a subfield using MARC::Record
On Apr 29, 2006, at 10:31 AM, Mark Jordan wrote:
Maybe other people should verify the usefulness of a delete
subfield function before anyone does anything about it, though.
Would a half dozen +1 votes from perl4libers validate its usefulness?
Yes it would...but to get the changes out on CPAN we'd all need to
convince Mike O'Regan who holds the keys to MARC-Record on CPAN.
Right now the version on CPAN is not the latest version that's
available in CVS on SourceForge. I think Mike has some performance
concerns about the Unicode handling code in v2.0, since MARC::Record
is used in some of his time critical applications. Hopefully we can
benchmark v1.38 and v2.0 sometime and get the latest code pushed out
to CPAN if things look OK. Or if this is not an option there is
always the possibility of creating a MARC::Record2 on CPAN. That
would be kind of ugly for backwards compatibility though.
At any rate if you are open to using the latest/greatest code from
SourceForge there is nothing stopping us from getting a
delete_subfield method working.
$field-delete_subfield('a', 0);
In the interests of flexibility how about using hash key/value pairs?
Here's what I'm thinking:
# delete all subfield u
$field-delete_subfield(code = 'u');
# delete first two subfield u
$field-delete_subfield(code = 'u', count = 2);
# delete all subfield u that have 'zombo.com' in them
$field-delete_subfield(code = 'u', match = qr/zombo\.com/);
# delete only the first subfield u that has 'zombo.com' in it
$field-delete_subfield(code = 'u', match = qr/zombo\.com', count
= 1);
# delete any subfield with a value that matches 'zombo.com'
$field-delete_subfield(match = qr/zombo\.com/);
So effectively there is AND boolean logic between any options that
are supplied. If this looks good I've got some code that does it and
some tests committed in SourceForge for you to take a look at [1].
I'm open to suggestions on renaming parameters, etc... Here are the
basics for checking out the code if you haven't done it before.
cvs -d:pserver:[EMAIL PROTECTED]:/cvsroot/marcpm login
cvs -z3 -d:pserver:[EMAIL PROTECTED]:/cvsroot/marcpm co
-P marc-record
//Ed
[1] http://sourceforge.net/cvs/?group_id=1254
Re: Deleting a subfield using MARC::Record
Edward Summers wrote:
Deleting subfields is a bit tricky since subfields may
repeat, and sometimes people just want to delete one of them. An
unfortunate state of affairs perhaps.
Yeah, I can see what you're saying, but doesn't that also apply to
repeatable fields? If a particular subfield that is one of a repeated
set needed to be deleted, it could be identified by a regex or by its
order in an array (following object syntax probably not correct):
@subfields = $subject-subfields();
foreach $notwanted (@subfields) {
if ($notwanted =~ /badsubject/) {
$notwanted-delete_subfield();
}
}
Mark
--
Mark Jordan
Head of Library Systems
W.A.C. Bennett Library, Simon Fraser University
Burnaby, British Columbia, V5A 1S6, Canada
Phone (604) 291 5753 / Fax (604) 291 3023
[EMAIL PROTECTED] / http://www.sfu.ca/~mjordan/
Re: Deleting a subfield using MARC::Record
On Apr 29, 2006, at 1:08 AM, Mark Jordan wrote:
Edward Summers wrote:
Deleting subfields is a bit tricky since subfields may repeat, and
sometimes people just want to delete one of them. An unfortunate
state of affairs perhaps.
Yeah, I can see what you're saying, but doesn't that also apply to
repeatable fields?
Well yeah it does, and you're right there is a
MARC::Record::delete_field isn't there. Would having something
similar to
that in MARC::Field be useful to you?
If a particular subfield that is one of a repeated set needed to be
deleted, it could be identified by a regex or by its order in an
array (following object syntax probably not correct):
@subfields = $subject-subfields();
foreach $notwanted (@subfields) {
if ($notwanted =~ /badsubject/) {
$notwanted-delete_subfield();
}
}
That could work if subfields were objects, but they're just strings.
It could simply delete all of them unless a second parameter is
passed in, which would basically act like a filter:
$field-delete_subfield('a', qr/badsubject/);
An alternative of course is to play fast and loose with the Object
model and twiddle with $field-_subfields ... this is an array that
looks like:
['a', 'foo', 'b', 'bar']
Of course this opens you up to future failure if the internals of
MARC::Field change at some point in the future. Which may not be all
that likely :-)
//Ed
Re: Deleting a subfield using MARC::Record
Edward Summers wrote:
That could work if subfields were objects, but they're just strings. It
could simply delete all of them unless a second parameter is passed in,
which would basically act like a filter:
$field-delete_subfield('a', qr/badsubject/);
That sounds pretty good, though I'm working with a database where I
found a 6xx field with two identical occurrences of subfield 2. So what
you propose would, I suppose, delete both of them and I'd have to put
one back in. Add another parameter to specify which occurrence?
Mike
Re: Deleting a subfield using MARC::Record
Edward Summers wrote:
On Apr 29, 2006, at 1:08 AM, Mark Jordan wrote:
Edward Summers wrote:
Deleting subfields is a bit tricky since subfields may repeat, and
sometimes people just want to delete one of them. An unfortunate
state of affairs perhaps.
Yeah, I can see what you're saying, but doesn't that also apply to
repeatable fields?
Well yeah it does, and you're right there is a
MARC::Record::delete_field isn't there. Would having something similar to
that in MARC::Field be useful to you?
Strangely enough, I've encountered two separate situations in the last
two weeks (both on Friday afternoons... I think I've fallen into some
kind of recursive metadata loop) where I wanted to delete just a
specific subfield. Maybe other people should verify the usefulness of a
delete subfield function before anyone does anything about it, though.
Would a half dozen +1 votes from perl4libers validate its usefulness?
If a particular subfield that is one of a repeated set needed to be
deleted, it could be identified by a regex or by its order in an array
(following object syntax probably not correct):
@subfields = $subject-subfields();
foreach $notwanted (@subfields) {
if ($notwanted =~ /badsubject/) {
$notwanted-delete_subfield();
}
}
That could work if subfields were objects, but they're just strings. It
could simply delete all of them unless a second parameter is passed in,
which would basically act like a filter:
$field-delete_subfield('a', qr/badsubject/);
Yeah, I knew I had the syntax wrong. The revised one you supply looks
intuitive. Could the second parameter also accommodate the order (in an
array) of the target subfield if the subfield repeated, for when you
know your records are consistent enough to use it (such as when you get
vendor records that have two 856 u subfields but you only want the
second one)? For example:
$field-delete_subfield('a', 0);
Mark
Re: Deleting a subfield using MARC::Record
On Apr 28, 2006, at 8:20 PM, Michael Kreyche wrote:
my $new856f = MARC::Field-new('856',$i1,$i2,@new856s);
$field-replace_with($new856f);
If there's an easier way, I'd like to know!
Creating a new field and replacing the old one with the new one is
the way to go. Deleting subfields is a bit tricky since subfields may
repeat, and sometimes people just want to delete one of them. An
unfortunate state of affairs perhaps. If someone can suggest an
easier API for doing this I'm open to suggestions :-)
//Ed
