clean.clj

(defn space-in-emph [[match g1 g2]]
  (let [l (if (= " " g2) "" g2)]
    (str "{" g1 l "}")))

(defn footnote [[match g1]]
  (if (#{"..." "…"} g1)
    " \\el "
    (str "\\footnote{" g1 "}")))

(defn final-clean [txt]
  (-> txt
      (s/replace #"[ \t\xA0]+" " ") ; non breaking space
      (s/replace #"\n +" "\n")
      (s/replace #"[\r\n]{2,}" "\n\n")
      (s/replace #"\[ " "[")
      (s/replace #"\( " "(")
      (s/replace #" \]" "]")
      (s/replace #" \)" ")")
      (s/replace #"\} ," "},")
      (s/replace #" »" "\"")
      (s/replace #"« " "\"")
      (s/replace #" *\[([^\]]+)\]" footnote) ; [] -> footnote
      (s/replace #"\" ?\\emph\{([^\}]+)\} ?\"" "\"$1\"") ; double quote
      (s/replace #"\{ *([^\}]+)([^\}])\}" space-in-emph)
      (s/replace #" ([,\.])" "$1") ; "hi , alex -> hi, alex"
      (s/replace #" - " " \u2014 ")
      (s/replace #"%" "\\%")))