Handling stacked Grascii affixes

In the previous post we used regular expressions to rewrite word affixes in Grascii notation. Our solution mostly works, but there are a couple further considerations.

One case we didn't explore is when a word has both a disjoined affix and a joined affix on the same end. For example, the word relationship has the disjoined suffix -ship as well as the joined suffix -tion. Our regexes only match at the beginning and end of words, so relationship becomes relation^SH instead of relaSH^SH. We can account for disjoined Grascii affixes when we generate regexes:

(defn gregg [s]
  (-> (name s)
      (str/replace #"\*[aeiouy]|[aeiouy]\*" "[aeiouy]+")
      (str/replace #"^(.+)-$" "^([A-Z()|',._~-]+\\\\^)?$1")
      (str/replace #"^-(.+)$" "$1(\\\\^[A-Z()|',._~-]+)?\\$")
      re-pattern))

This is another reason to denote Grascii forms with upper-case, distinct from the original word's lower-case characters.

With this change we can match prefixes that are preceded by Grascii and suffixes that are followed by Grascii:

(gregg '-tion)

#"tion(\^[A-Z()|',._~-]+)?$"

(re-find (gregg '-tion) "relation^SH")

["tion^SH" "^SH"]

We want to include any captured affix and include it in the replacement, which we can do with $1 in the replacement string:

(str/replace "relation^SH" (gregg '-tion) "SH$1")

"relaSH^SH"

Here's a revised grascii function:

(defn grascii [word]
  (reduce (fn [word [affix replacement]]
            (str/replace word (gregg affix)
                         (str replacement "$1")))
          (str/lower-case word)
          (partition 2 replacements)))

(grascii "relationship")

"relaSH^SH"

While this works for suffixes, prefixes end up in the wrong order:

(grascii "counterdecree")

"DK^cree"

We also have to check for whether an affix is a prefix or a suffix and put the captured group on the correct side in the replacement string:

(defn grascii [word]
  (reduce (fn [word [affix replacement]]
            (str/replace word (gregg affix)
                         (if (str/ends-with? (name affix) "-")
                           (str "$1" replacement)
                           (str replacement "$1"))))
          (str/lower-case word)
          (partition 2 replacements)))

(grascii "counterdecree")

"K^Dcree"