HTML Character Entities

Anders-Bergqvist · March 10, 2026, 5:00pm

We have installed a fess version 15.5. In search results our Meta titles and description have the encoded Ö instead of the (Swedish) caracter Ö for example. We have checked that we have utf-8 encoding. Did we miss something?

We don’t have that problem with our Fess 14.10 in production.

shinsuke · March 12, 2026, 2:30am

Thank you for the information. It will be fixed in the next release.

Anders-Bergqvist · April 20, 2026, 11:48am

Is this fixed now? I did not see it in the release notes for 15.6.

shinsuke · April 23, 2026, 1:09am

github.com/codelibs/fess

refactor(crawler): replace URI with URL in XpathTransformer and ProtocolHelper

master ← refactor/replace-uri-with-url-in-transformer

opened 03:03PM - 11 Mar 26 UTC

marevol

+383 -114

## Summary - Replace `java.net.URI`/`URISyntaxException` with `java.net.URL`/`M…alformedURLException` in `FessXpathTransformer` - Simplify URL handling in `ProtocolHelper` by using `URL.getProtocol()` directly instead of converting to URI - Update corresponding test method names and assertions ## Changes Made - **`FessXpathTransformer.java`**: Migrated all URI-based URL resolution to use `java.net.URL`. The `getURI()` method is renamed to `getURL()`, `getBaseUri()` to `getBaseUrl()`. The `addChildUrlFromTagAttribute()` signature now accepts `URL` instead of `URI`, and the overridden method simplifies logic by leveraging `new URL(base, spec)` constructor which handles relative URL resolution natively. - **`ProtocolHelper.java`**: Removed unnecessary `URI` conversion when checking resource protocol — now reads `resource.getProtocol()` directly from the `java.net.URL` object. The URI conversion is still performed only when constructing a `File` from the resource (required by `File(URI)` constructor). - **`FessXpathTransformerTest.java`**: Renamed `test_getBaseUri()` to `test_getBaseUrl()` and updated all assertion calls to use `URL` API (`toExternalForm()` instead of `toString()`). ## Testing - Existing unit tests in `FessXpathTransformerTest` updated and cover the refactored methods - The path normalization for `/../` prefixes is retained in `addChildUrlFromTagAttribute()` ## Breaking Changes - `getBaseUri()` and `getURI()` protected methods are renamed to `getBaseUrl()` and `getURL()` — any subclasses overriding these methods will need to be updated - `addChildUrlFromTagAttribute()` parameter type changed from `URI` to `URL` ## Additional Notes - `java.net.URL` handles a wider range of URL formats that `URI.create()` would reject with `IllegalArgumentException`, making the crawler more resilient to non-standard URLs found in the wild - The simplified error handling in `addChildUrlFromTagAttribute()` removes the large fallback block that manually constructed absolute URLs, since `new URL(base, spec)` handles these cases natively