Javanese Cryptanalysis

Javanese is the language of the people living in the central and eastern parts of the island of Java, Indonesia.

In 1926 a standard orthography was created for Javanese using the English alphabet. It uses all 26 letters from A to Z. The five letters A, E, I, O, and U are vowels; the other 21 letters are consonants. In every Javanese word vowels and consonants always alternate — no two vowels and no two consonants are ever next to each other. This property is very useful when deciphering encrypted Javanese text.

A text $s$ is made of words, and each word contains only uppercase letters. Call $s$ legitimate if, in every word of $s$, vowels and consonants alternate (no two vowels are adjacent and no two consonants are adjacent).

A simple substitution cipher is applied to $s$: a bijection $f$ on the 26 uppercase letters is chosen, and the encoded text $t$ is obtained from $s$ by replacing every letter $c$ with $f(c)$. Because $f$ is a bijection, exactly 5 letters of the alphabet are images of vowels and the other 21 are images of consonants.

You are given the encoded text $t$. Decide whether some legitimate text $s$ could have been encoded into $t$. If several legitimate texts are possible, you must report the single canonical one defined in the Output section.

The input is the encoded text $t$: a list of words separated by single spaces and/or line breaks. Each word consists only of uppercase letters (A–Z).

The input contains at most $100\,000$ characters.

If no legitimate text $s$ can be encoded into $t$, print the single word impossible.

Otherwise print the decoded legitimate text $s$, keeping exactly the same layout as $t$: replace every letter, but leave every space and line break unchanged and in its original position. Several legitimate texts may encode into $t$; print the lexicographically smallest one. To compare two candidate texts, read them character by character from the beginning; at the first position where they differ, the text with the smaller letter (with $A < B < \dots < Z$) is the smaller text. Every letter of $s$ is uppercase.