Announcement

Collapse
No announcement yet.

UTF8 migration and URL redirection

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • UTF8 migration and URL redirection

    Hi,

    I finally managed to migrate my test site in full UTF8, but i think i can't go live because the URL are now with weird characters.
    I know i can resolve this by using route by node as default.

    But still : google knows the old URL, and only the "new" URL (with utf8 encoding) will be redirected to the route by node url. So this would mean all my SEO would be lost...
    The only option i can see is to switch to route by node and wait for a few months to migrate to UTF8.
    Or is there a way to redirect old "non UTF8" URLs either to UTF8 URLs or route by node URLs ?

    I am not sure i am very clear so feel free to ask for precisions.

  • #2
    URLs shouldn't have changed. They would be stored as encoded characters if the character was UTF-8. These should still work after changing the database to UTF-8.
    Translations provided by Google.

    Wayne Luke
    The Rabid Badger - a vBulletin Cloud customization and demonstration site.
    vBulletin 5 Documentation - Updated every Friday. Report issues here.
    vBulletin 5 API - Full / Mobile
    I am not currently available for vB Messenger Chats.

    Comment


    • #3
      I did migrate from Latin1 to UTF8 (or almost, the DB was partially utf8 and partially latin1, the site was latin1) :

      The migration is only done on the test site, when i look at the same topic on my production site and my test site, the url did change :
      Test site (utf8) : https://beta.plongeur.com/forums/for...oit-de-reponse
      prod site (Latin1) : https://www.plongeur.com/forums/foru...oit-de-reponse

      So i guess my SEO would be totally lost because there is no redirection between the url (if i just change the www by beta in the production URL, i end up on a 404 error; this used to work when both site were in the same charset).

      Did i miss something during my migration (route rebuild for instance) ?
      Is there a solution other than migrating my prod site to "route by node", wait a few months so google update its indexes, and then migrate to UTF8 ?

      Comment


      • #4
        They shouldn't change. You'll have to log this as a bug in the tracker.

        http://tracker.vbulletin.com/secure/...91&issuetype=1
        Translations provided by Google.

        Wayne Luke
        The Rabid Badger - a vBulletin Cloud customization and demonstration site.
        vBulletin 5 Documentation - Updated every Friday. Report issues here.
        vBulletin 5 API - Full / Mobile
        I am not currently available for vB Messenger Chats.

        Comment


        • #5
          OK it is created http://tracker.vbulletin.com/browse/VBV-17836

          In the meantime i will try rebuilding the routes from the tools.php file.
          And mybe try to migrate my DB to latin1 so it can at least be coherent, but it really does not seem a step in the good direction.

          Can you tell how the URLs are built ? maybe i have double converted a field that was already utf8 encoded...

          Comment


          • #6
            That portion of the URL is stored in the urlident field of the node table.

            URLs are created by combining the route prefix from the routenew table with the urlident. Each type of content has a record in the route table. For channels, there are two routes. One for the channel and one for the topics within the channel.
            Translations provided by Google.

            Wayne Luke
            The Rabid Badger - a vBulletin Cloud customization and demonstration site.
            vBulletin 5 Documentation - Updated every Friday. Report issues here.
            vBulletin 5 API - Full / Mobile
            I am not currently available for vB Messenger Chats.

            Comment


            • #7
              OK, think i have a working solution for short term :
              - Keep the full UTF-8 DB
              - Change the $config['Mysqli']['charset'] to 'latin1'
              - Change the laguage setting to latin1

              This way everything is working and i keep my existing urls.
              Would this be a supported configuration ?

              You really need to remove all specials characters from the URLs and replace them with their basic equivalent (e for ).
              I know you advise to use route by nodes instead but it is not the same, with long urls we can build trafic report by channel and we cannot do it with route by node.

              Edit : it is not so great, everything is working for the search engine who does not find any word having a special character (just 0 post found).
              Last edited by plongeur.com; Sat 28th Oct '17, 1:36am.

              Comment


              • #8
                OK, i'm still working on it, looks like some colums are double encoded from the beginning, so it worked OK in latin1 (the borwer treated it as UTF8) but now it is a mess.

                The prefix and regex fields of the routenew table
                The urlident of the node table.

                What is weird is that other columns of the node table, with basically the same content, are ok.
                This has to be a consequence of my twisted upgrade path ...

                Now i ran :
                UPDATE routenew SET prefix = CONVERT(CAST(CONVERT(prefix USING latin1) AS BINARY) USING utf8)
                UPDATE routenew SET regex = CONVERT(CAST(CONVERT(regex USING latin1) AS BINARY) USING utf8)
                UPDATE node SET urlident = CONVERT(CAST(CONVERT(urlident USING latin1) AS BINARY) USING utf8)

                And everything seems to be fine, on the URL front at least. (i secretely hped it would solve my notification issue).
                Last edited by plongeur.com; Mon 30th Oct '17, 3:00am.

                Comment


                • #9
                  I will go live with these change hoping not to do a mess.

                  On my test for it seems to be working so it should be ok.

                  Comment


                  • #10
                    Originally posted by Wayne Luke View Post
                    They shouldn't change. You'll have to log this as a bug in the tracker.

                    http://tracker.vbulletin.com/secure/...91&issuetype=1
                    Hello Wayne,

                    It is the exact problem as stated here

                    Wayne Luke

                    Comment

                    Related Topics

                    Collapse

                    Working...
                    X