**List of steps to reproduce** (step by step, including full links if applicable):
Here is a piece of python code that uses Pywikibot to query Wikidata and find which Persian Wikipedia page corresponds with the English Wikipedia page [[https://en.wikipedia.org/wiki/Category:Companies_based_in_Edmonton|Category:Companies based in Edmonton]]
```lang=python
import pywikibot
page = pywikibot.Page(pywikibot.Site('en'), 'Category:Companies based in Edmonton')
print(page.langlinks())
```
**What happens?**:
Here is the response you get:
```
[pywikibot.page.Link('شركات مقرها في إدمونتون', APISite("ar", "wikipedia")), pywikibot.page.Link('Unternehmen (Edmonton)', APISite("de", "wikipedia")), pywikibot.page.Link('شرکت\u200cهای مستقر در ادمنتن', APISite("fa", "wikipedia"))]
```
This indicates that the `fa` equivalent page would be titled "شرکتهای مستقر در ادمنتن" (don't be confused by \u200c which is the [[https://en.wikipedia.org/wiki/Zero-width_non-joiner|zero-width-non-joiner]] character)
**What should have happened instead?**:
Per the [[https://www.wikidata.org/wiki/Q6986508|Wikidata page]] and as shown on the right side of the screenshot below, the expected answer would have been "شرکتهای ادمنتن" (more specifically, "رده:شرکتهای ادمنتن").
{F34856537, size=full}
It seems like `langlinks()` is not returning the links, but rather the labels of the descriptions in different languages. Case in point, the namespace prefix is also missing in the responses.
**Software version (if not a Wikimedia wiki), browser information, screenshots, other information, etc**:
Tested on 05ce190809b