sled

чување 31dcd2bcf4114cf0e16827181252faa1126bc58b
родитељ 83e5082fdcd3661c838f93e559032e055803a1cb
Аутор: Страхиња Радић <sr@strahinja.org>
Датум:   Thu, 11 Jun 2026 09:22:05 +0000

Add -O and -o; handle invalid UTF-8 similar to table(1)

Diffstat:
M
TODO
 | 
3
++-
M
sled.1.in
 | 
10
++++++++++
M
sled.c
 | 
18
+++++++++++++++++-
M
util.c
 | 
27
++++++++++++++-------------
M
util.h
 | 
6
++++--

измењених датотека: 5, додавања: 47(+), брисања: 17(-)
diff --git a/TODO b/TODO
@@ -3,8 +3,9 @@ TODO
 
 [~] Code optimization (struct coord...)
 
-[ ] Add changes from table(1) regarding the drawing of invalid UTF-8. 
+[x] Add changes from table(1) regarding the drawing of invalid UTF-8. 
     That would perhaps also make open_binary obsolete.
+	- Add -O/-o parameters
 
 [ ] Implement proper drawing of wide characters covering dialogs, now 
     that termbox has a function for getting characters from the display 
diff --git a/sled.1.in b/sled.1.in
@@ -16,6 +16,8 @@
 .Nm
 .Op Fl B | Fl \-no\-backup
 .Op Fl b | Fl \-backup
+.Op Fl O | Fl \-no\-open\-binary
+.Op Fl o | Fl \-open\-binary
 .Ar filename
 .Sh DESCRIPTION
 .Nm
@@ -29,6 +31,14 @@ This option overrides the setting in
 Enables backups.
 This option overrides the setting in
 .Pa config.h .
+.It Fl O , \-no\-open\-binary
+Disables opening of binary files (containing the NUL character).
+This option overrides the setting in
+.Pa config.h .
+.It Fl o , \-open\-binary
+Enables opening of binary files (containing the NUL character).
+This option overrides the setting in
+.Pa config.h .
 .It Fl h , \-help
 Show option summary.
 .It Fl V , \-full\-version
diff --git a/sled.c b/sled.c
@@ -2901,7 +2901,9 @@ main_options_loop:
 	{
 		printf("Usage:\t%s -h | --help | -V | --full-version | "
 		       "-v | --version\n"
-		       "\t%s [-B | --no-backup] [-b | --backup] filename\n",
+		       "\t%s [-B | --no-backup] [-b | --backup] "
+		       "[-O | --no-open-binary] [-o | --open-binary] "
+		       "filename\n",
 			PROGRAM_NAME, PROGRAM_NAME);
 		return 0;
 	}
@@ -2941,6 +2943,20 @@ main_options_loop:
 		optind++;
 		goto main_options_loop;
 	}
+	else if (!strcmp(argv[optind], "-O")
+		|| !strcmp(argv[optind], "--no-open-binary"))
+	{
+		open_binary = 0;
+		optind++;
+		goto main_options_loop;
+	}
+	else if (!strcmp(argv[optind], "-o")
+		|| !strcmp(argv[optind], "--open-binary"))
+	{
+		open_binary = 1;
+		optind++;
+		goto main_options_loop;
+	}
 
 	/*
 	 * Chicken and egg problem: to get fpathconf(dirfd, _PC_PATH_MAX)
diff --git a/util.c b/util.c
@@ -18,32 +18,33 @@
 #define B_FULL " \v\t()[]{},.;:/\\"
 
 size_t
-u8_string_to_unicode(uint32_t* us, const char* s, const size_t max)
+u8_string_to_unicode(uint32_t* us, const char* s, const ssize_t max)
 {
 	uint32_t uch;
 	uint32_t* pus = us;
+	const char* ps = s;
 	int u8_len    = 0;
 	size_t added  = 0;
 
 	assert(s != NULL);
-	while (*s)
+	while (ps - s < max && *ps)
 	{
-		u8_len = tb_utf8_char_to_unicode(&uch, s);
-		if (added + 1 > max)
-			break;
-		if (u8_len > 0)
+		u8_len = tb_utf8_char_to_unicode(&uch, ps);
+		if (!u8_len)
 		{
-			*pus++ = uch;
-			s += u8_len;
-			added++;
+			uch = U32_REPL_CHAR;
+			u8_len = 1;
 		}
+		added++;
+		ps += u8_len;
+		*pus++ = uch;
 	}
-	*pus++ = 0;
+	*pus = 0;
 	return added;
 }
 
 size_t
-unicode_string_to_u8(char* s, const uint32_t* us, const size_t max)
+unicode_string_to_u8(char* s, const uint32_t* us, const ssize_t max)
 {
 	char ch[8];
 	char* ps	    = NULL;
@@ -53,10 +54,10 @@ unicode_string_to_u8(char* s, const uint32_t* us, const size_t max)
 
 	assert((s != NULL) && (us != NULL));
 	ps = s;
-	while (*pus)
+	while (pus - us < max && *pus)
 	{
 		len = tb_utf8_unicode_to_char(ch, *pus);
-		if (added + 1 >= max)
+		if (len < 0)
 			return added;
 		memccpy(ps, ch, 0, len);
 		ps += len;
diff --git a/util.h b/util.h
@@ -17,6 +17,8 @@
 #define UNUSED(x)	((void)(x))
 /* clang-format on */
 
+#define U32_REPL_CHAR ((uint32_t)0xFFFD)
+
 #define MEMCCPY(to, from, tolen, temp)                     \
 	do                                                 \
 	{                                                  \
@@ -116,8 +118,8 @@ enum {
 
 enum { ACT_IDLE, ACT_WRITING, ACT_READING, ACT_WRAPPING };
 
-size_t u8_string_to_unicode(uint32_t* us, const char* s, const size_t max);
-size_t unicode_string_to_u8(char* s, const uint32_t* us, const size_t max);
+size_t u8_string_to_unicode(uint32_t* us, const char* s, const ssize_t max);
+size_t unicode_string_to_u8(char* s, const uint32_t* us, const ssize_t max);
 int u32_has_nonblanks(const uint32_t* s);
 uint32_t* u32_memccpy(uint32_t* dst, const uint32_t* src, uint32_t c,
 	size_t len);