From dc462165b8ab0033b78350126be90a1dd7eb9857 Mon Sep 17 00:00:00 2001
From: =?utf8?q?Luk=C3=A1=C5=A1=20Ji=C5=99i=C5=A1t=C4=9B?=
 <gymnazium.jiriste@gmail.com>
Date: Sat, 21 Oct 2023 15:19:00 +0200
Subject: [PATCH] Create the base for Byte Pair encoding project

---
 main.c | 121 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 121 insertions(+)
 create mode 100644 main.c

diff --git a/main.c b/main.c
new file mode 100644
index 0000000..caa4b81
--- /dev/null
+++ b/main.c
@@ -0,0 +1,121 @@
+
+#include <unistd.h>
+#include <fcntl.h>
+
+struct s_options
+{
+	int		compress;
+	size_t	buffer_size;
+}
+typedef struct s_options t_options;
+
+struct s_input
+{
+	int			valid;
+	const char	*ipath;
+	const char	*opath;
+	t_options	opt;
+}
+
+typedef struct s_input t_input;
+
+t_input	*parse_argv(int argc, const char **argv);
+
+void	rewrite_pairs(const t_input *inp, int count_mat[255][255], char *buff)
+{
+	int				ifd;
+	int				ofd;
+	unsigned char	sub;
+	t_uchar_pair	pair;
+	int				len;
+
+	pair = get_pair(count_mat);
+	sub = get_free_substitute(count_mat);
+	if (sub == '\0')
+		return ;
+	ifd = open(inp->ipath, O_RDONLY);
+	ofd = open(inp->opath, O_WRONLY | O_CREAT | O_TRUNC,
+		S_IRUSR | S_IRGRP | S_IROTH | S_IWUSR);
+	if (read(ifd, buff, 1) <= 0)
+		return ;
+	while (1)
+	{
+		j = read(ifd, buff + 1, inp->opt.buffer_size - 1);
+		if (j <= 0)
+			break ;
+		len = substitute(buff, pair, sub, j);
+		write(ofd, buff, len - 1);
+		buff[0] = buff[len];
+	}
+	write(ofd, buff, 1);
+	close(ifd);
+	close(ofd);
+	return ;
+}
+
+void	construct_count_mat(const int fd, int *count_mat, char *buff)
+{
+	int		i;
+	int		j;
+	char	last;
+	
+	last = '\0';
+	while (1)
+	{
+		j = read(fd, buff, inp->opt.buffer_size);
+		if (j <= 0)
+			break;
+		++(*get_elem(count_mat, last, buff[0]));
+		i = 0;
+		while (i < j)
+		{
+			++(*get_elem(count_mat, buff[i], buff[i + 1]));
+			++i;
+		}
+		last = buff[i];
+	}
+	return ;
+}
+
+void	compress(t_input *inp, char *buff)
+{
+	int		*count_mat;
+	int		i;
+	int		ifd;
+
+	count_mat = ft_calloc(255 * 255, sizeof(int));
+	if (!count_mat)
+		return ;
+	while (can_cont)
+	{
+		zero_out(count_mat);
+		ifd = open(inp->ipath, O_RDONLY);
+		construct_count_mat(ifd, count_mat, buff);
+		close(ifd);
+		rewrite_pairs(inp, count_mat, buff);
+	}
+	free(count_mat);
+}
+
+int	main(int argc, char **argv)
+{
+	t_input	*inp;
+	char	*buff;
+
+	inp = parse_argv(argc, argv);
+	if (!inp->valid)
+	{
+		ft_putstr_fd("Error\n", 2);
+		return (1);
+	}
+	buff = malloc(inp->opt.buffer_size * sizeof(char));
+	if (!buff)
+		return ;
+	if (inp->opt.compress)
+		compress(inp, buff);
+	else
+		decompress(inp);
+	free(buff);
+	free(inp);
+	return (0);
+}
-- 
2.30.2