Open timohund opened 8 years ago
Discussion: 1 Updated by Ingo Renner about 3 years ago
2 Updated by Bernhard Kraft over 2 years ago Comment Edit
You could configure an additional solr field and make it a copy of the "content" field. Then let this solr field get tokenized by the mentioned tokenizer and configure the suggest feature to take this solr field into account. 3 Updated by Jigal van Hemert over 2 years ago Comment Edit
That's a workaround. There is a suitable tokenizer available, so why not use it? That way all users of EXT:solr have understandable suggestions without having to tinker with the server configuration.
Estimation:
Effort: 5 Value: 5
Mail addresses in content such as my.name@example.org are split by the StandardTokenizerFactory as "my.name" and "example.org" because according to http://unicode.org/reports/tr29/#Word_Boundaries certain punctuation between letter is not a word boundary. This causes "my.name" and "example.org" to show up in the autocomplete suggestions.
Alternative tokenizer: UAX29URLEmailTokenizerFactory this is similar to the StandardTokenizerFactory, but also recognizes URLs and mail addresses as entire tokens. This way the complete mail address would show up in the suggestions, which is more understandable to website users.
https://forge.typo3.org/issues/48990