getgooglequery does not work correctly with Google Forms links

[jakubklimek]

Description

When I access a dokuwiki page on some dokuwikis and the HTTP request has referer HTTP header set to, e.g. referer: https://www.google.com/url?q=https://opendata.gov.cz/edu:konference:2019&sa=D&ust=1567692046621000&usg=AFQjCNG_0v_FF28O_RTQNJ0RZqqp49fGig - this is what Google Forms does when a link to https://opendata.gov.cz/edu:konference:2019 is used - the page has words from the https://opendata.gov.cz/edu:konference:2019 URL marked as search results, even though I did no search.

Apparently (giterlizzi/dokuwiki-template-bootstrap3#438 (comment)) this is a Dokuwiki getgooglequery functionality for highlighting search results from search engines, which, however, does not work correctly when placing a link into a Google Form.

Steps to reproduce

1. See this example, or this one: curl https://wiki.bash-hackers.org/ -H 'referer: https://www.google.com/url?q=https://opendata.wiki.gov.cz/edu:konference:2019&sa=D&ust=1567692046621000&usg=AFQjCNG_0v_FF28O_RTQNJ0RZqqp49fGig' | grep mark

Expected behavior: [What you expected to happen]
No search results marked with <span class="mark">

Actual behavior: [What actually happened]

Words from the referer header marked as <span class="mark">

Versions

[DokuWiki] Release 2018-04-22b "Greebo"

Screenshots or Logs

[Paste your logs or attach the screenshot]

[phy25]

This one is interesting. I don't think we can get user query from referrer anymore from Google, if users are being redirected from the Google /url endpoint.

One example: https://www.google.com/url?sa=t&source=web&rct=j&url=https://www.dokuwiki.org/&ved=2a-long-string-that-may-include-my-search-info

Maybe we should ignore it if it's from google.com/url? I am not that familiar with how Google's recent referrer works, so we need more input.

[phy25]

We can:

1. Remove Google (since they never include query in url)
2. or drop the query if query is a URL (we can just check if // is included)

I decided to implement 2.