// SPDX-License-Identifier: MIT
// SPDX-FileCopyrightText: 2023 SASANO Takayoshi <uaa@uaa.org.uk>

#include <cstdio>
#include <clocale>
#include <cwchar>

#define MAX_TOKENS 65535 // should be odd value
#define CONV_BUFSIZE 65536

class convert {
public:
	int do_line(wchar_t *buf);
	wchar_t *token[MAX_TOKENS];

protected:
	int tokenize(wchar_t *tokens[], wchar_t *in_buf);
	int check_token(wchar_t *token);
	int convert_yomi(wchar_t *yomi);
  
};

int convert::do_line(wchar_t *buf)
{
	int i, num;

	if ((num = tokenize(token, buf)) & 1) {
		fprintf(stderr, "tokens not even (%d)\n", num);
		num &= ~1;
	}

	for (i = 0; i < num; i += 2) {
		if (check_token(token[i]) || convert_yomi(token[i + 1]))
			token[i] = token[i + 1] = NULL;
	}

	return num;
}

int convert::tokenize(wchar_t *tokens[], wchar_t *in)
{
	int i;
	wchar_t *wp;

	for (i = 0; i < MAX_TOKENS; i++) {
		tokens[i] = wcstok(i ? NULL : in, L"\t\n", &wp);
		if (tokens[i] == NULL)
			return i;
	}
	  
	fprintf(stderr, "too many tokens!\n");
	return i;
}

int convert::convert_yomi(wchar_t *yomi)
{
	size_t i, len = wcslen(yomi);
	int ret = -1;

	if (!len)
		goto fin;

	/* convert katakana -> hiragana */
	for (i = 0; i < wcslen(yomi); i++) {
		if (yomi[i] != L'ー' &&
		    !(yomi[i] >= L'ぁ' && yomi[i] <= L'ん') &&
		    !(yomi[i] >= L'ァ' && yomi[i] <= L'ヴ'))
			goto fin;

		/* convert to hiragana (except "ー" and "") */
		if (yomi[i] >= L'ァ' && yomi[i] < L'ヴ')
			yomi[i] -= 0x60;
	}

	ret = 0;
fin:
	return ret;
}

int convert::check_token(wchar_t *token)
{
	size_t i;
	int ret = -1;

	/* check first character */
	if (*token == L'ー' || *token == L'々')
		goto fin;

	/* some ideographic marks and hiragana, katakana, CJK is permitted */
	for (i = 0; i < wcslen(token); i++) {
		if (token[i] != L'ー' && token[i] != L'々' &&
		    !(token[i] >= L'ぁ' && token[i] <= L'ん') &&
		    !(token[i] >= L'ァ' && token[i] <= L'ヶ') &&
		    !(token[i] >= 0x4e00 && token[i] <= 0x9fff))
			goto fin;
	}

	ret = 0;
fin:
	return ret;
}

int main(int argc, char *argv[])
{
	int i, n;
	FILE *fpi = stdin, *fpo = stdout;
	convert c;
	wchar_t buf[CONV_BUFSIZE];

	setlocale(LC_ALL, "ja_JP.UTF-8");

	while (fgetws(buf, CONV_BUFSIZE, fpi) != NULL) {
		if ((n = c.do_line(buf)) < 2)
			continue;

		for (i = 0; i < n; i += 2) {
			if (c.token[i] == NULL) {
				fprintf(fpo, "<UNK> ");
			} else {
				fprintf(fpo, "%ls/%ls ",
					c.token[i + 1], c.token[i]);
			}
		}
		fprintf(fpo, "\n");
	}

	return 0;
}