Fwd: Re: Batch import of images

Tue Nov 1 10:40:12 GMT 2022

Lovely!
Thank you very much! It will be an interesting operation.
A seemingly complete road map for me.

I will definitely need another beer when done.
Best not buy before the finish line is crossed though :0)

Besides I do like the data format and I was surprised the XML is processed
so fast.
What's the max number of images that is possible for KPA to handle?
Or phrased it another way: What's the highest number of images you have
encountered up to now?
I have some 25000 more images...
It could be useful to have an idea (if I have to create another
instance/initfile).

Per

On Mon, Oct 31, 2022 at 11:39 PM Johannes Zarl-Zierl <johannes at zarl-zierl.at>
wrote:

> Hi,
>
> Thanks for the detailed response...
>
> Am Montag, 31. Oktober 2022, 15:40:39 CET schrieb Per Funke:
> > Structure:
> > =======
> >
> >    1.
> >    There is one file for exif data, “data”
> >    2.
> >    There are three files for tags called “what”,”when” and “where”
> >    3.
> >    To connect “data” with the tag files there is an index called “datab”
> >    4.
> >    The record id of the datab records are hashed, creating a file
> structure
> >    called “more_txt”. An empty file is created for each record when
> > imported. Annotating will later fill these already existing records with
> > text. The hashing mechanism means there is no index to the texts.
> >    5.
> >    For searching purposes there is a index connecting all existing tags
> >    (tags that are used, not empty ones) to the “datab” file’s record(s)
> > where that tag is used..
> >    6.
> >    All images directories are named as the current date when created.
> >    This means the dir contains images from the name (date) of the dir
> >    before it, up to the date that constitutes it’s own name.
> >    When you know the name of the dir where an image resides you also know
> >    the approximate date of it's creation.
> > [...]
> > As you see, importing from this troglodyte of a database can only be done
> > by an expert! Me!
>
> Ok, that sounds like you have all the data you need in a usable format.
> You
> also have an expert at hand, which is always good ;-)
>
> > If I were employed, this would have created my total job security. I'm
> like
> > Sid in User Friendly, though not quite so skilled.
>
> I understood that reference ;-)
>
> > Also I do not use punch cards since my last university course where I
> > programmed in Cobol.
>
> ... and while I never learned Cobol I do find punch cards handy for taking
> notes...
>
>
> > It was kind of fun doing this description. That's something I've never
> > done, ever. I had it all in my head from the beginning. Now it's in
> writing.
> > I think I'll go and buy a fat trappist beer, I'm worth it!
>
> Joking aside: writing down stuff is actually a great start.
>
> So, in summary: you already know your data, and probably know how to
> 1) create a list of all tags
> 2) create a list of all images and their associated tags
>
> Also, it looks like your category names are plain ASCII, so you won't have
> to
> escape any special characters (before you ask: having non-ASCII characters
> in
> your tag names is not a problem). If your data is not already encoded in
> UTF-8, I would advise you to make sure that you tag names are valid UTF-8
> before the import.
>
> If you need a template for creating the index.xml file for KPhotoAlbum,
> you can
> use this one:
>
> <?xml version="1.0" encoding="UTF-8"?>
> <KPhotoAlbum version="7" compressed="1">
>     <Categories>
>         <Category name="Who" icon="system-users" show="1" viewtype="0"
> positionable="1">
>            <value value="Person A"/>
>            <value value="Person B"/>
>         </Category>
>         <Category name="Where" icon="applications-internet" show="1"
> viewtype="0">
>            <value value="Place A"/>
>         </Category>
>         <Category name="What" icon="favorites" show="1" viewtype="0">
>             <value value="untagged" id="1"/>
>             <value value="The What"/>
>         </Category>
>     </Categories>
>     <images>
>        <image file="relative/path/to/file.jpg"
> startDate="2000-01-01T00:00:00"
> endDate="2000-12-31T23:59:59" >
>             <options>
>                 <option name="Who">
>                     <value value="Person A"/>
>                     <value value="Person B"/>
>                  </option>
>                  <option name="Where">
>                     <value value="Place A"/>
>                  </option>
>                  <option name="What">
>                     <value value="The What"/>
>                  </option>
>             </options>
>         </image>
>     </images>
> </KPhotoAlbum>
>
> That means you need to convert your list of Who/Where/What tags into value
> XML-element under the appropriate Category element.
> Then you'll need to create an image element for each image, possessing
> correct
> startDate and endDate attributes. Below the image element, create
> Who/Where/
> What entries for each tag that applies to the image.
>
> This will result in a very verbose file that has enough information for
> kphotoalbum to repair it to get a valid index.xml file. Using this verbose
> (or
> in kphotoalbum parlance "uncompressed") file format allows you to not care
> about tag ids. Having the "compressed=1" attribute tells kphotoalbum to
> save
> the database in a less verbose and more performance-friendly format.
>
> Once you have created an index.xml file in this manner, you can open the
> database/image directory in kphotoalbum:
> KPhotoAlbum will complain about missing image size information, and will
> tell
> you that you should pick a tag to mark untagged images. You should start
> with
> selecting the selecting the "untagged" tag and save the database.
> After that, you need to run "Recreate MD5 sums..." and "Recreate EXIF
> database" from the maintenance menu. Make sure that the "Update image
> time"
> checkbox is not active.
>
> When that is done, save your database again, recreate the thumbnails
> (maintenance menu) and go for another trappist beer. If all went well, you
> now
> have all your images in kphotoalbum.
>
> Hope that helps,
>   Johannes
>
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kphotoalbum/attachments/20221101/39caefd5/attachment.htm>