We've tried to reduce the ambiguity of parsing space-separated keywords by implementing the following rules in the new parser (once again, these changes aren't live yet, but will be once we finish testing):
If the keyword string doesn't contain a comma, semicolon, or pipe, then we treat spaces as delimiters unless the space is enclosed in quotes. So [zoo San Diego San Diego Zoo] becomes "zoo", "San", "Diego" after duplicates are removed, but [zoo "San Diego" "San Diego Zoo"] becomes "zoo", "San Diego", "San Diego Zoo".
There's a side effect when using Lightroom and the plug-in. Lightroom happily creates multi-word keywords with spaces and treats them as a unit. If I attach one such keyword to an image and upload via the plugin, it's going to be parsed into several single-word keywords. That's a mismatch between the Smugmug and Lightroom models for keywords. At the moment it's not a big deal for me -- I have exactly one such multiword keyword that gets used alone -- but a mismatch like this can be lurking bomb waiting for me or someone else to step on it.
So, it would seem there needs to be a patch somewhere to reconcile the behavior between these two models. For example, it would work if the plugin wrapped individual multi-word keywords in quotes before uploading. It would work if the plug-in had some other flag saying a single multiword keyword was coming. (If I attach two such keywords, the plugin inserts a comma, so there all is cool.) Perhaps a different adjustment could be made.
I just unveiled and I have a huge amount of combined keywords. All combined keywords were cleaned up
before I unveiled today. Plus I have a huge amount of number keywords extracted from filenames. I had and
still have it set not to do that in settings.
Sorry to hear that, Allen. The keyword changes we've been discussing here are still in testing and aren't yet live. As I said, these changes fix quite a few bugs, but they're large changes and we're testing them very thoroughly, which is why it's taking a while.
Sorry to hear that, Allen. The keyword changes we've been discussing here are still in testing and aren't yet live. As I said, these changes fix quite a few bugs, but they're large changes and we're testing them very thoroughly, which is why it's taking a while.
I don't know how things are organized at smugmug, but have they considered putting you in charge of... everything?
Smugmug and testing aren't words that seem to go together anymore.
Is there a delay due to indexing that will cause the fix to take some time to populate? Still seeing all the same errors in my keywords list, but I'm assuming they need to re-index.
Is there a delay due to indexing that will cause the fix to take some time to populate? Still seeing all the same errors in my keywords list, but I'm assuming they need to re-index.
We aren't automatically re-indexing existing keywords, but if you go back and edit broken keywords, then they'll be re-indexed.
Spoke too soon, just found 5 multi-keywords split into singles on the first keyword I checked. None
of the singles show in the keyword list though? /keyword Only show split under the photos.
Spoke too soon, just found 5 multi-keywords split into singles on the first keyword I checked. None
of the singles show in the keyword list though? /keyword Only shoe split under the photos.
Found five more split checking another keyword. These first two KW's I've found are new keywords added
since going into sandbox. Both of these KW's were copied from a text file so they were all entered
exactly the same. ALL had ending semi-colons. The split ones so far are single multi-word like this.
xxxx xxx xx xxxx;
In both smart galleries if more then the one KW I've not seen any splitting.
xxxx xxx xxxx; xxxx xxx;
Another smart gallery. Repeat post #74, single multi-words are splitting. 46 photos in gallery, 12 single mult-words split.
One multi-word did not split but it had "" around it. All others were multiple multi-words and ok.
All the splitting single mult-words probably ended with a semicolon with no double quotes.
single mult-words means only the one KW
xxx xxxx xxxx xx;
I think by default when adding keywords the trailing semicolon is removed so when going back into
edit it's not there. That's what I see on every edit, trailing ";" gone.
So your program probably did not see one and split the mult-words into singles.
In gallery edited split KW by removing all the ";" and saved. Went back into edit and
double quotes now around it all with no semicolon at the end.
All the splitting single mult-words probably ended with a semicolon with no double quotes.
single mult-words means only the one KW
xxx xxxx xxxx xx;
I think by default when adding keywords the trailing semicolon is removed so when going back into
edit it's not there. That's what I see on every edit, trailing ";" gone.
So your program probably did not see one and split the mult-words into singles.
In gallery edited split KW by removing all the ";" and saved. Went back into edit and
double quotes now around it all with no semicolon at the end.
That's correct. When displaying keywords in the edit field, we won't display a trailing semicolon, but will use quotes to enclose a single multi-word keyword. As long as a single multi-word keyword is enclosed in quotes, it will be preserved. If you have multiple multi-word keywords, you can either enclose each of them in quotes and separate them with spaces, or separate them with semicolons, commas, or pipes.
I was hoping the keyword fixes would include the problem of keywords on videos not working...any news on when that might be fixed?
Sorry, these fixes didn't include any changes to parsing video metadata. That's something I'd like to work on, but I can't offer a timeline (lots of stuff on my plate, and metadata is tricky -- especially videos).
That's correct. When displaying keywords in the edit field, we won't display a trailing semicolon, but will use quotes to enclose a single multi-word keyword. As long as a single multi-word keyword is enclosed in quotes, it will be preserved. If you have multiple multi-word keywords, you can either enclose each of them in quotes and separate them with spaces, or separate them with semicolons, commas, or pipes...
Every keyword I've added in the last year or so is separated by semicolons. This includes multi-word. If
only one KW and it's a multi-word I ended with a semi-colon. This is how we were directed.
Now a lot all the single multi-word KW's are split. I suggest when performing any kind of operation on
photo keywords add a trailing comma or semicolon to it first.
The bulk edit needs to be fixed, when adding a multi-word keyword from "add" box at top it adds
, one two (if another KW exists, ok when saved)
or if none
one two (blank, splits when saved)
This means when saved it auto splits the single multi-word into two keywords from the blank. The bulk edit should add
multi-words with quotes.
I'm 1/3 way thru my ~190 smart galleries fixing split KW's. Weird part is almost all of the split KW's do
not show on the /keyword page.
Every keyword I've added in the last year or so is separated by semicolons. This includes multi-word. If
only one KW and it's a multi-word I ended with a semi-colon. This is how we were directed.
Now a lot all the single multi-word KW's are split. I suggest when performing any kind of operation on
photo keywords add a trailing comma or semicolon to it first.
You can continue doing this, and it will work fine when saving keywords. However, trailing delimiters will be removed before your keywords are displayed back to you. This is part of the normalization step the new parser does to ensure that keywords are always saved in a consistent format.
If you enter a single multi-word keyword followed by a trailing delimiter, like this:
birthday party;
...then after you save the keywords and open the editor again, you'll see:
"birthday party"
If you enter a single-word keyword followed by a trailing delimiter:
birthday;
...then the trailing delimiter will simply be removed:
birthday
The bulk edit needs to be fixed, when adding a multi-word keyword from "add" box at top it adds
, one two (if another KW exists, ok when saved)
or if none
one two (blank, splits when saved)
Yeah, the legacy bulk editor has some bugs. We're working on a modern replacement for it, but that's a separate project from the new keyword parser.
Yeah, the legacy bulk editor has some bugs. We're working on a modern replacement for it, but that's a separate project from the new keyword parser.
I've used this trying to remove old <font> or any tags, etc, in captions. But doesn't work. I can't see
why it doesn't recognize all characters. This is needed to clean up captions.
The old bulk doesn't work for /keyword pages forcing me to use a smart gallery to edit them. This
needs to be fixed.
I was hoping the keyword fixes would include the problem of keywords on videos not working...any news on when that might be fixed?
So, I looked into this and uncovered some details for you.
Turns out we do support XMP ("Subject" field only), IPTC, and QuickTime keyword metadata on videos. But I was surprised to discover that Lightroom doesn't actually write keyword metadata when it exports a video, and it's possible other image/video editing tools don't as well.
One solution to this, if you use Lightroom, is to use the official SmugMug Lightroom plugin to upload your videos to SmugMug. It adds metadata in a separate step from uploading, to ensure that keywords and other metadata are preserved.
If you're sure that your keywords are actually being stored in the video metadata but SmugMug still isn't extracting them, send me a link to the affected videos and I'll take a look.
Thanks Allen! This helped me find the bug that's preventing videos from showing up in keyword indexes. It's unrelated to the metadata actually embedded in videos, but I'll get this fixed. You don't need to do anything on your end; once the fix goes out, your videos will just start showing up on keyword pages where they should be.
Thanks Allen! This helped me find the bug that's preventing videos from showing up in keyword indexes. It's unrelated to the metadata actually embedded in videos, but I'll get this fixed. You don't need to do anything on your end; once the fix goes out, your videos will just start showing up on keyword pages where they should be.
Looks like you fixed it!
Yaypie, how are you with UI design? We have some more projects for you!
Yesterday I found 10 different keywords that were split. I have not touched the photo containing
galleries or these keywords for a long time. I have cleaned up all split keywords weeks ago, so
this is something new Smug is screwing up.
Edit: Has Smug re-indexed all my keywords since my last post?
Comments
There's a side effect when using Lightroom and the plug-in. Lightroom happily creates multi-word keywords with spaces and treats them as a unit. If I attach one such keyword to an image and upload via the plugin, it's going to be parsed into several single-word keywords. That's a mismatch between the Smugmug and Lightroom models for keywords. At the moment it's not a big deal for me -- I have exactly one such multiword keyword that gets used alone -- but a mismatch like this can be lurking bomb waiting for me or someone else to step on it.
So, it would seem there needs to be a patch somewhere to reconcile the behavior between these two models. For example, it would work if the plugin wrapped individual multi-word keywords in quotes before uploading. It would work if the plug-in had some other flag saying a single multiword keyword was coming. (If I attach two such keywords, the plugin inserts a comma, so there all is cool.) Perhaps a different adjustment could be made.
Mr. Parry (devbobo), can you save the day?
Jim Ringland
jtringl.smugmug.com
Portfolio • Workshops • Facebook • Twitter
before I unveiled today. Plus I have a huge amount of number keywords extracted from filenames. I had and
still have it set not to do that in settings.
What gives Smug?
My Website index | My Blog
I don't know how things are organized at smugmug, but have they considered putting you in charge of... everything?
Smugmug and testing aren't words that seem to go together anymore.
Dave
Is there a delay due to indexing that will cause the fix to take some time to populate? Still seeing all the same errors in my keywords list, but I'm assuming they need to re-index.
Dave
We aren't automatically re-indexing existing keywords, but if you go back and edit broken keywords, then they'll be re-indexed.
Thanks
My Website index | My Blog
of the singles show in the keyword list though? /keyword Only show split under the photos.
My Website index | My Blog
Link?
My Website index | My Blog
since going into sandbox. Both of these KW's were copied from a text file so they were all entered
exactly the same. ALL had ending semi-colons. The split ones so far are single multi-word like this.
xxxx xxx xx xxxx;
In both smart galleries if more then the one KW I've not seen any splitting.
xxxx xxx xxxx; xxxx xxx;
My Website index | My Blog
Reset modem and router, good now,
My Website index | My Blog
One multi-word did not split but it had "" around it. All others were multiple multi-words and ok.
All the splitting single mult-words probably ended with a semicolon with no double quotes.
single mult-words means only the one KW
xxx xxxx xxxx xx;
I think by default when adding keywords the trailing semicolon is removed so when going back into
edit it's not there. That's what I see on every edit, trailing ";" gone.
So your program probably did not see one and split the mult-words into singles.
In gallery edited split KW by removing all the ";" and saved. Went back into edit and
double quotes now around it all with no semicolon at the end.
My Website index | My Blog
I was hoping the keyword fixes would include the problem of keywords on videos not working...any news on when that might be fixed?
That's correct. When displaying keywords in the edit field, we won't display a trailing semicolon, but will use quotes to enclose a single multi-word keyword. As long as a single multi-word keyword is enclosed in quotes, it will be preserved. If you have multiple multi-word keywords, you can either enclose each of them in quotes and separate them with spaces, or separate them with semicolons, commas, or pipes.
Sorry, these fixes didn't include any changes to parsing video metadata. That's something I'd like to work on, but I can't offer a timeline (lots of stuff on my plate, and metadata is tricky -- especially videos).
only one KW and it's a multi-word I ended with a semi-colon. This is how we were directed.
Now a lot all the single multi-word KW's are split. I suggest when performing any kind of operation on
photo keywords add a trailing comma or semicolon to it first.
The bulk edit needs to be fixed, when adding a multi-word keyword from "add" box at top it adds
, one two (if another KW exists, ok when saved)
or if none
one two (blank, splits when saved)
This means when saved it auto splits the single multi-word into two keywords from the blank. The bulk edit should add
multi-words with quotes.
I'm 1/3 way thru my ~190 smart galleries fixing split KW's. Weird part is almost all of the split KW's do
not show on the /keyword page.
My Website index | My Blog
You can continue doing this, and it will work fine when saving keywords. However, trailing delimiters will be removed before your keywords are displayed back to you. This is part of the normalization step the new parser does to ensure that keywords are always saved in a consistent format.
If you enter a single multi-word keyword followed by a trailing delimiter, like this:
...then after you save the keywords and open the editor again, you'll see:
If you enter a single-word keyword followed by a trailing delimiter:
...then the trailing delimiter will simply be removed:
Yeah, the legacy bulk editor has some bugs. We're working on a modern replacement for it, but that's a separate project from the new keyword parser.
why it doesn't recognize all characters. This is needed to clean up captions.
The old bulk doesn't work for /keyword pages forcing me to use a smart gallery to edit them. This
needs to be fixed.
My Website index | My Blog
So, I looked into this and uncovered some details for you.
Turns out we do support XMP ("Subject" field only), IPTC, and QuickTime keyword metadata on videos. But I was surprised to discover that Lightroom doesn't actually write keyword metadata when it exports a video, and it's possible other image/video editing tools don't as well.
One solution to this, if you use Lightroom, is to use the official SmugMug Lightroom plugin to upload your videos to SmugMug. It adds metadata in a separate step from uploading, to ensure that keywords and other metadata are preserved.
If you're sure that your keywords are actually being stored in the video metadata but SmugMug still isn't extracting them, send me a link to the affected videos and I'll take a look.
Nothing is picked up especially the date taken.
I add keywords to these videos on Smug and none will show on a keyword page.
My Website index | My Blog
Can you give me a link to one of these videos so I can take a look?
http://www.photosbyat.com/MyKeywords/Bird-Videos/n-gF9bt/i-Hs3j9Hz
/keyword page
http://www.photosbyat.com/keyword/bird_videos
My Website index | My Blog
Thanks Allen! This helped me find the bug that's preventing videos from showing up in keyword indexes. It's unrelated to the metadata actually embedded in videos, but I'll get this fixed. You don't need to do anything on your end; once the fix goes out, your videos will just start showing up on keyword pages where they should be.
My Website index | My Blog
http://www.photosbyat.com/MyKeywords/Bird-Videos/n-gF9bt/i-Dtwbj5F
But I think it's impossible to modify before it was taken. Notice img/mov file numbers.
Name IMG_1496_7D2.jpg
Size 2490 x 1660
Date Taken 2013-01-09 13:43:19
Date Modified 2013-01-28 16:50:26
Name MVI_1497.MOV
Size 1920 x 1080
Date Modified 2013-01-09 05:43:23
Name IMG_1499_7D2.jpg
Size 2490 x 1660
Date Taken 2013-01-09 13:44:38
Date Modified 2013-01-28 16:51:22
My Website index | My Blog
Looks like you fixed it!
Yaypie, how are you with UI design? We have some more projects for you!
Dave
My Website index | My Blog
galleries or these keywords for a long time. I have cleaned up all split keywords weeks ago, so
this is something new Smug is screwing up.
Edit: Has Smug re-indexed all my keywords since my last post?
My Website index | My Blog