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

#include <cstdio>
#include <cstring>
#include <clocale>
#include <cwchar>
#include <cinttypes>
#include <string>
#include <map>
#include <unistd.h>

template <typename T> struct data_entry {
	std::wstring key;
	T value;
};

template <typename T> class data_store {
public:
	std::map<std::wstring, data_entry<T>> db;

	void set(wchar_t *key, T value) {
		data_entry<T> d;

		d.key = key;
		d.value = value;

		if (!db.try_emplace(d.key, d).second) {
			db[key].value += value;
			fprintf(stderr,
				"update \"%ls\": %" PRId64 " -> %" PRId64 "\n",
				key, db[key].value - value, db[key].value);
		}
	};

	void init(void) {
		db.clear();
	};
};

#define CONV_BUFSIZE 65536

static int64_t count_char(wchar_t *buf, wchar_t c)
{
	int64_t count = 0;

	for (; *buf; buf++) {
		if (*buf == c)
			count++;
	}

	return count;
}

static int64_t do_line(wchar_t *buf)
{
	int64_t ret = -1;
	wchar_t *p;

	if ((p = wcsrchr(buf, L'\t')) == NULL)
		goto fin;
	*p = L'\0';

	if (wcsstr(buf, L"<UNK>") != NULL || count_char(buf, L' ') != 2)
		goto fin;

	ret = wcstoll(p + 1, NULL, 0);
fin:
	return ret;
}

static wchar_t *read_line(wchar_t *buf, int n, FILE *fp)
{
	wchar_t *p;
	int c;
	off_t pos, len;

	while (1) {
		if ((p = fgetws(buf, n, fp)) != NULL)
			return p;

		/* garbage detected, seek next line */
		pos = ftell(fp);
		for (len = 0; ; len++) {
			c = fgetc(fp);
			if (c == EOF || c == '\n')
				break;
		}

		fprintf(stderr, "garbage at % " PRId64 ", %" PRId64 " bytes\n",
			pos, len);

		if (c == EOF)
			break;
	}

	return NULL;
}

int main(int argc, char *argv[])
{
	char *infile = NULL, *outfile = NULL;
	FILE *fpi, *fpo;
	int64_t limit = 0;
	int ch;
	data_store<int64_t> d;
	wchar_t buf[CONV_BUFSIZE];

	while ((ch = getopt(argc, argv, "l:i:o:")) != -1) {
		switch (ch) {
		case 'l':
			limit = atoll(optarg);
			break;
		case 'i':
			infile = optarg;
			break;
		case 'o':
			outfile = optarg;
			break;
		}
	}

	if (infile == NULL || outfile == NULL) {
		fprintf(stderr, "usage: %s -i [infile] -o [outfile] "
			"-l [limit]\n", argv[0]);
		goto fin0;
	}

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

	fpi = !strcmp(infile, "-") ? stdin : fopen(infile, "r");
	if (fpi == NULL) {
		fprintf(stderr, "file open error (infile)\n");
		goto fin0;
	}

	fpo = !strcmp(outfile, "-") ? stdout : fopen(outfile, "w");
	if (fpo == NULL) {
		fprintf(stderr, "file open error (outfile)\n");
		goto fin1;
	}

	/* pass 1 */
	d.init();
	while (read_line(buf, CONV_BUFSIZE, fpi) != NULL) {
		int64_t c;
		if ((c = do_line(buf)) <= 0)
			continue;
		d.set(buf, c);
	}

	/* pass 2 */
	for (auto i = d.db.begin(); i != d.db.end(); i++) {
		if (d.db[i->first].value < limit)
			continue;

		fprintf(fpo, "%ls\t%" PRId64 "\n",
			d.db[i->first].key.c_str(), d.db[i->first].value);
	}

	fclose(fpo);
fin1:
	fclose(fpi);
fin0:
	return 0;
}