Intelligent Character Replacement question

  • Resolved H&S-Team

    (@hackespitze)


    4 years, 6 months ago

    First, thanks a lot for a very helpful plugin!

    In our project, we auto-create Image Titles and Alt-Text from image-file-names. We use custom diacritic word replacements to re-establish German Umlauts and special character.

    As our list is growing quickly, I would like to know whether it also works with partial matches. Say I would like to re-establish the letter ä, whenever the string aermel is found.

    Instead of explicitly adding every possible word and case (aermel, aermeln, aermels, spitzenaermel) … is there a way to add just the relevant subset of the word?

    In my sample, the snippet *aerm* would do – and it also worked for laerm and waermflasche. Hence – we could serve all cases with a far shorter list.

    • This topic was modified 4 years, 6 months ago by H&S-Team.
Viewing 6 replies - 1 through 6 (of 6 total)
  • Plugin Author pepe

    (@pputzer)

    4 years, 6 months ago

    Hi @hackespitze, the custom diacritics unfortunately only work with full words by design – otherwise, there would be far to many incorrect replacements.

    I think adding an upload filter to automatically rename images with “ae” etc. in the filenames would be a better solution for your usecase. (Or if you’ve already uploaded all the files, a database search & replace.). This would allow for subsequent manual fixes to incorrect replacements.

    Thread Starter H&S-Team

    (@hackespitze)

    4 years, 6 months ago

    Thanks for your reply! I now understand that limiting the tool to full words was a deliberate design decision. I’m curious – Have you actually tested failure rate at scale? Only considering the long explicit replacement list I already have, turning Umlauts in German language from ue, ae and oe to ü, ä and ö should work robustly. None of the list-entries we have so far would cause an error.

    An inverted list, which holds all exceptions to transcription rules (Carmen McRae, Dr. Oetker, Menuett, Hugo Strasser) should be greatly shorter.

    @Uploadfilter: We by all means need image names without umlauts and other German language characters, as these cause trouble in our Import Pipeline. While naming images offline, we therefore have removed all Umlauts, etc. As we still forgot a few, our Importer went haywire. For that reason, a script now skips wrongly named assets.

    What we want is recreate the special characters with image file-names as input while leaving image-names unchanged. Search and Replace could work – but one would not want to run this manually, each time, when uploading assets.

    Having these special characters right could give a competitive edge in Image SEO in E-Commerce for German language markets. For this to work, however, one needs to re-establish proper names, before assets get indexed by search engines.

    Plugin Author pepe

    (@pputzer)

    4 years, 6 months ago

    By “renaming” I meant the names of the attachment objects in WordPress, not (necessarily) the actual filenames. All this can be done with upload filter hooks. In any case, you can add custom fixes using the wp-Typography API, it is just not something I would recommend. However long your exception list is, it will always miss some words.

    Thread Starter H&S-Team

    (@hackespitze)

    4 years, 6 months ago

    Thanks again :o) I will discuss the options you propose with our developer.

    While our project hasn’t gone live yet I’m very happy how much better text looks on the staging site with your WP-Typography. Great that you make this available. 👋

    Plugin Author pepe

    (@pputzer)

    4 years, 4 months ago

    Hi @hackespitze, do you have more question or can we consider the issue resolved?

    Thread Starter H&S-Team

    (@hackespitze)

    4 years, 4 months ago

    Hi @pputzer,
    please excuse my belated reply! Thanks for asking – yes you can mark this issue as solved. 👋

Viewing 6 replies - 1 through 6 (of 6 total)

The topic ‘Intelligent Character Replacement question’ is closed to new replies.