Add trim and trim_empty support to strsplit and update all callers

cmdline/selftest.c

@@ -1063,6 +1063,125 @@ static void test_wnmatch(void)
 	}
 }
 
+static void test_strsplit(void)
+{
+	char buf[64];
+	char* map[8];
+	unsigned n;
+
+	/* empty string */
+	strcpy(buf, "");
+	n = strsplit(map, 8, buf, ",", " ", 1);
+	if (n != 0) {
+		/* LCOV_EXCL_START */
+		log_fatal(EINTERNAL, "Failed strsplit empty string test with trim_empty=1\n");
+		exit(EXIT_FAILURE);
+		/* LCOV_EXCL_STOP */
+	}
+	strcpy(buf, "");
+	n = strsplit(map, 8, buf, ",", " ", 0);
+	if (n != 0) {
+		/* LCOV_EXCL_START */
+		log_fatal(EINTERNAL, "Failed strsplit empty string test with trim_empty=0\n");
+		exit(EXIT_FAILURE);
+		/* LCOV_EXCL_STOP */
+	}
+
+	/* trim_empty = 1 with delimiters only */
+	strcpy(buf, ",,,");
+	n = strsplit(map, 8, buf, ",", " ", 1);
+	if (n != 0) {
+		/* LCOV_EXCL_START */
+		log_fatal(EINTERNAL, "Failed strsplit consecutive delimiters with trim_empty=1\n");
+		exit(EXIT_FAILURE);
+		/* LCOV_EXCL_STOP */
+	}
+
+	/* trim_empty = 0 with delimiters only */
+	strcpy(buf, ",,,");
+	n = strsplit(map, 8, buf, ",", " ", 0);
+	if (n != 4 || strcmp(map[0], "") != 0 || strcmp(map[1], "") != 0 || strcmp(map[2], "") != 0 || strcmp(map[3], "") != 0) {
+		/* LCOV_EXCL_START */
+		log_fatal(EINTERNAL, "Failed strsplit consecutive delimiters with trim_empty=0\n");
+		exit(EXIT_FAILURE);
+		/* LCOV_EXCL_STOP */
+	}
+
+	/* trim_empty = 1 with mixed spaces and commas */
+	strcpy(buf, "  one, two,, , three  ");
+	n = strsplit(map, 8, buf, ",", " ", 1);
+	if (n != 3 || strcmp(map[0], "one") != 0 || strcmp(map[1], "two") != 0 || strcmp(map[2], "three") != 0) {
+		/* LCOV_EXCL_START */
+		log_fatal(EINTERNAL, "Failed strsplit mixed test with trim_empty=1\n");
+		exit(EXIT_FAILURE);
+		/* LCOV_EXCL_STOP */
+	}
+
+	/* trim_empty = 0 with mixed spaces and commas */
+	strcpy(buf, "  one, two,, , three  ");
+	n = strsplit(map, 8, buf, ",", " ", 0);
+	if (n != 5 || strcmp(map[0], "one") != 0 || strcmp(map[1], "two") != 0 || strcmp(map[2], "") != 0 || strcmp(map[3], "") != 0 || strcmp(map[4], "three") != 0) {
+		/* LCOV_EXCL_START */
+		log_fatal(EINTERNAL, "Failed strsplit mixed test with trim_empty=0\n");
+		exit(EXIT_FAILURE);
+		/* LCOV_EXCL_STOP */
+	}
+
+	/* trailing delimiter */
+	strcpy(buf, "a,");
+	n = strsplit(map, 8, buf, ",", 0, 1);
+	if (n != 1 || strcmp(map[0], "a") != 0) {
+		/* LCOV_EXCL_START */
+		log_fatal(EINTERNAL, "Failed strsplit trailing delimiter with trim_empty=1\n");
+		exit(EXIT_FAILURE);
+		/* LCOV_EXCL_STOP */
+	}
+	strcpy(buf, "a,");
+	n = strsplit(map, 8, buf, ",", 0, 0);
+	if (n != 2 || strcmp(map[0], "a") != 0 || strcmp(map[1], "") != 0) {
+		/* LCOV_EXCL_START */
+		log_fatal(EINTERNAL, "Failed strsplit trailing delimiter with trim_empty=0\n");
+		exit(EXIT_FAILURE);
+		/* LCOV_EXCL_STOP */
+	}
+
+	/* leading delimiter */
+	strcpy(buf, ",a");
+	n = strsplit(map, 8, buf, ",", 0, 1);
+	if (n != 1 || strcmp(map[0], "a") != 0) {
+		/* LCOV_EXCL_START */
+		log_fatal(EINTERNAL, "Failed strsplit leading delimiter with trim_empty=1\n");
+		exit(EXIT_FAILURE);
+		/* LCOV_EXCL_STOP */
+	}
+	strcpy(buf, ",a");
+	n = strsplit(map, 8, buf, ",", 0, 0);
+	if (n != 2 || strcmp(map[0], "") != 0 || strcmp(map[1], "a") != 0) {
+		/* LCOV_EXCL_START */
+		log_fatal(EINTERNAL, "Failed strsplit leading delimiter with trim_empty=0\n");
+		exit(EXIT_FAILURE);
+		/* LCOV_EXCL_STOP */
+	}
+
+	/* capacity limit (split_max) */
+	strcpy(buf, "a,b,c");
+	n = strsplit(map, 2, buf, ",", 0, 1);
+	if (n != 2 || strcmp(map[0], "a") != 0 || strcmp(map[1], "b") != 0) {
+		/* LCOV_EXCL_START */
+		log_fatal(EINTERNAL, "Failed strsplit split_max test with trim_empty=1\n");
+		exit(EXIT_FAILURE);
+		/* LCOV_EXCL_STOP */
+	}
+	strcpy(buf, "a,b,c");
+	n = strsplit(map, 2, buf, ",", 0, 0);
+	if (n != 2 || strcmp(map[0], "a") != 0 || strcmp(map[1], "b") != 0) {
+		/* LCOV_EXCL_START */
+		log_fatal(EINTERNAL, "Failed strsplit split_max test with trim_empty=0\n");
+		exit(EXIT_FAILURE);
+		/* LCOV_EXCL_STOP */
+	}
+}
+
 static void test_path(void)
 {
 	char dst[PATH_MAX];
@@ -2801,6 +2920,9 @@ void test(int argc, char* argv[])
 	/* pathname import and export */
 	test_path();
 
+	/* string split */
+	test_strsplit();
+
 	/* smartctl output parsing */
 	test_parse_smartctl();
 

cmdline/state.c

@@ -837,7 +837,7 @@ void state_config(struct snapraid_state* state, const char* path, const char* co
 				/* LCOV_EXCL_STOP */
 			}
 
-			split_mac = strsplit(split_map, SPLIT_MAX + 1, buffer, ",");
+			split_mac = strsplit(split_map, SPLIT_MAX + 1, buffer, ",", " \t\r\n", 0);
 
 			if (split_mac > SPLIT_MAX) {
 				/* LCOV_EXCL_START */

cmdline/str.c

@@ -22,27 +22,42 @@ char* strpolish(char* s)
 	return s;
 }
 
-unsigned strsplit(char** split_map, unsigned split_max, char* str, const char* delimiters)
+unsigned strsplit(char** split_map, unsigned split_max, char* str, const char* delimiters, const char* trim, int trim_empty)
 {
 	unsigned mac = 0;
 
-	/* skip initial delimiters */
-	str += strspn(str, delimiters);
+	if (*str == 0)
+		return 0;
 
-	while (*str != 0 && mac < split_max) {
-		/* start of the token */
-		split_map[mac] = str;
-		++mac;
+	while (mac < split_max) {
+		char* begin = str;
 
-		/* find the first delimiter or the end of the string */
 		str += strcspn(str, delimiters);
 
-		/* put the final terminator if missing */
-		if (*str != 0)
-			*str++ = 0;
+		char* end = str;
 
-		/* skip trailing delimiters */
-		str += strspn(str, delimiters);
+		int has_next = *str != 0;
+		if (has_next) {
+			*str = 0;
+			++str;
+		}
+
+		if (trim != 0) {
+			begin += strspn(begin, trim);
+
+			while (end > begin && strchr(trim, end[-1]) != 0) {
+				--end;
+				*end = 0;
+			}
+		}
+
+		if (!trim_empty || *begin != 0) {
+			split_map[mac] = begin;
+			++mac;
+		}
+
+		if (!has_next)
+			break;
 	}
 
 	return mac;

cmdline/str.h

@@ -26,8 +26,9 @@ char* strpolish(char* s);
  * null-terminates each token by overwriting its trailing delimiter,
  * and stores a pointer to each token's start in @split_map.
  *
- * Leading and consecutive delimiters are silently skipped, so the
- * output never contains empty tokens.
+ * If @trim_empty is non-zero, leading and consecutive delimiters are
+ * silently skipped, and tokens trimmed to empty are discarded.
+ * If @trim_empty is zero, empty tokens are preserved.
  *
  * The input string is modified in-place; @split_map entries point
  * directly into @str and are only valid for its lifetime.
@@ -36,15 +37,17 @@ char* strpolish(char* s);
  * \param split_max Capacity of @split_map; at most this many tokens are stored.
  * \param str Input string to tokenise (modified in-place).
  * \param delimiters Null-terminated set of delimiter characters.
+ * \param trim Null-terminated set of characters to trim from the start and end of each token.
+ * \param trim_empty Whether to trim empty tokens and consecutive delimiters.
  * \return Number of tokens stored in @split_map.
  *
  * Example:
- *   char  buf[] = "  one,two,,three  ";
+ *   char  buf[] = "  one, two,, , three  ";
  *   char* map[8];
- *   unsigned n = strsplit(map, 8, buf, " ,");
+ *   unsigned n = strsplit(map, 8, buf, ",", " ", 1);
  *   // n == 3, map[] == { "one", "two", "three" }
  */
-unsigned strsplit(char** split_map, unsigned split_max, char* str, const char* delimiters);
+unsigned strsplit(char** split_map, unsigned split_max, char* str, const char* delimiters, const char* trim, int trim_empty);
 
 /**
  * Split a command-line string into an argument vector,

cmdline/unixapp.c

@@ -873,8 +873,8 @@ static int devdereference_bcachefs(uint64_t device, const char* dir, tommy_list*
 		separator += 3;
 
 		/* split the line */
-		id_mac = strsplit(id_map, 8, s, " \t\n");
-		fs_mac = strsplit(fs_map, 8, separator, " \t\n");
+		id_mac = strsplit(id_map, 8, s, " \t\r\n", 0, 1);
+		fs_mac = strsplit(fs_map, 8, separator, " \t\r\n", 0, 1);
 
 		/* if too short, it's the wrong line */
 		if (id_mac < 5)
@@ -914,7 +914,7 @@ static int devdereference_bcachefs(uint64_t device, const char* dir, tommy_list*
 
 	/* split the device list */
 	char* dev_map[64];
-	unsigned dev_mac = strsplit(dev_map, 64, device_list, ":");
+	unsigned dev_mac = strsplit(dev_map, 64, device_list, ":", 0, 0);
 	for (unsigned i = 0; i < dev_mac; ++i) {
 		/* get major:minor, use stat on the path returned */
 		struct stat st;
@@ -998,7 +998,7 @@ static int extract_zfs(const char* dir, char* dataset, size_t dataset_size, char
 			break;
 
 		/* split the line */
-		mac = strsplit(map, 3, s, "\t\n");
+		mac = strsplit(map, 3, s, "\t\r\n", 0, 1);
 
 		if (mac < 3)
 			continue;