Strip arbitrary characters from a string?

Question

Q: How can I strip non-word characters from a string?

For an arbitrary string, how do I strip all characters with non-word syntax? Alternately, using a regexp, how do I strip out the non-alphabetic characters?

Hypothetical example:

(some-fnx "#here's a string!") ; => "heres a string"

Specific application:

I want to get a list of the LaTeX font commands without the markup syntax. The value of LaTeX-font-list is:

((1 "" "" "\\mathcal{" "}")
 (2 "\\textbf{" "}" "\\mathbf{" "}")
 (3 "\\textsc{" "}")
 (5 "\\emph{" "}")
 (6 "\\textsf{" "}" "\\mathsf{" "}")
 (9 "\\textit{" "}" "\\mathit{" "}")
 (13 "\\textmd{" "}")
 (14 "\\textnormal{" "}" "\\mathnormal{" "}")
 (18 "\\textrm{" "}" "\\mathrm{" "}")
 (19 "\\textsl{" "}" "\\mathbb{" "}")
 (20 "\\texttt{" "}" "\\mathtt{" "}")
 (21 "\\textup{" "}")
 (4 "" "" t))

I want to iterate through the nested lists in order to get returned values like "textbf", "textrm", etc., without the rest of the markup.

Answer 1

I'll go for the regex route:

(mapcar (lambda (item) (replace-regexp-in-string "\\W" "" (cadr item)))
        LaTeX-font-list)