gh-102140 : We've improved the heuristic of has_header() method in Lib/csv.py. We wanted to respect the methodology on how this function was created, even if the determining factor string length is meaningless . We added more verifications before deleting a column for its inconsistency :

similiratyWords is a dictionnary in which we stock the number of repetitions of words per column.

compareWords is a list in which we stock every word of an element by using regex. By comparing two lists which represent for example element_line1 and element_line2, we increment similarityWords if there are same values in element_line1 and element_line2.
Thanks to this, we respect, the methology of comparing each element of every row in a column.

We've made the average of string lengths and compared it to the header length to keep the consistency.

Checking the header if it's a single word.

We group up all of that in the vote at the end of the function.

Where: gh-102140

Contributors : @Drakariboo & @Vanille-22