Issues with xpath processing along the "FullText" path template recognition.

[krstp]

Here is a sample HTML code:

<div class="FulltextWrapper">
    ...
</div>    

Where Trafilatura fails in processing it correcty.

The root issue appears to be in FulltextWrapper in particular in Fulltext pattern recognition in too general XPATH rule processing, such as:

trafilatura/trafilatura/xpaths.py

Line 45 in 42ada5a

contains(translate(@class, "FULTEX","fultex"), "fulltext")

Not sure what is the percentage of polluted websites, but seems to me this can be more done in a more fail-safe way.

The particular call that results in the error is: trafilatura.bare_extraction(html_content)

Would be great to see those cases handled. Thanks for all the great work!

[krstp]

Additional notes:

It seems the issue is directly related to the neighboring code structure, such as:

            <div class="container">
                <div id="main" class="layout">
                    <div class="layout__main--wide" id="main-content">
                        <div class="block" id="Test-ImgSrc">

<!--- WRONG ----->
                            <div class="FulltextWrapper">
                               ...
                            </div>
<!--- WRONG ----->

Also note the follow up block with similar class name, but different content just works fine:

                            <div class="FulltextWrapper">
                                <div xmlns="" xmlns:fn="http://www.w3.org/2005/xpath-functions" xmlns:meta="http://www.springer.com/app/meta" class="MainTitleSection">
                                    <h1 xmlns="http://www.w3.org/1999/xhtml" class="ArticleTitle" lang="en">The linguistic validation of the gut feelings questionnaire in three European languages</h1>
                                </div>
...

So I do not have exact explanation what combination of XML/HTML blocks intersect in XPATH processing, but ultimately it goes to fulltext naming.