//config: sort is used to sort lines of text in specified files.
//config:
//config:config FEATURE_SORT_BIG
-//config: bool "Full SuSv3 compliant sort (support -ktcbdfiozgM)"
+//config: bool "Full SuSv3 compliant sort (support -ktcbdfiogM)"
//config: default y
//config: depends on SORT
//config: help
-//config: Without this, sort only supports -r, -u, -s, and an integer version
+//config: Without this, sort only supports -rusz, and an integer version
//config: of -n. Selecting this adds sort keys, floating point support, and
//config: more. This adds a little over 3k to a nonstatic build on x86.
//config:
//usage: IF_FEATURE_SORT_BIG(
//usage: "\n -g General numerical sort"
//usage: "\n -M Sort month"
+//usage: "\n -V Sort version"
//usage: "\n -t CHAR Field separator"
//usage: "\n -k N[,M] Sort by Nth field"
//usage: )
//usage: "\n -r Reverse sort order"
//usage: "\n -s Stable (don't sort ties alphabetically)"
//usage: "\n -u Suppress duplicate lines"
-//usage: IF_FEATURE_SORT_BIG(
//usage: "\n -z Lines are terminated by NUL, not newline"
///////: "\n -m Ignored for GNU compatibility"
///////: "\n -S BUFSZ Ignored for GNU compatibility"
///////: "\n -T TMPDIR Ignored for GNU compatibility"
-//usage: )
//usage:
//usage:#define sort_example_usage
//usage: "$ echo -e \"e\\nf\\nb\\nd\\nc\\na\" | sort\n"
/* These are sort types */
enum {
- FLAG_n = 1, /* Numeric sort */
- FLAG_g = 2, /* Sort using strtod() */
- FLAG_M = 4, /* Sort date */
+ FLAG_n = 1 << 0, /* Numeric sort */
+ FLAG_g = 1 << 1, /* Sort using strtod() */
+ FLAG_M = 1 << 2, /* Sort date */
+ FLAG_V = 1 << 3, /* Sort version */
/* ucsz apply to root level only, not keys. b at root level implies bb */
- FLAG_u = 8, /* Unique */
- FLAG_c = 0x10, /* Check: no output, exit(!ordered) */
- FLAG_s = 0x20, /* Stable sort, no ascii fallback at end */
- FLAG_z = 0x40, /* Input and output is NUL terminated, not \n */
+ FLAG_u = 1 << 4, /* Unique */
+ FLAG_c = 1 << 5, /* Check: no output, exit(!ordered) */
+ FLAG_s = 1 << 6, /* Stable sort, no ascii fallback at end */
+ FLAG_z = 1 << 7, /* Input and output is NUL terminated, not \n */
/* These can be applied to search keys, the previous four can't */
- FLAG_b = 0x80, /* Ignore leading blanks */
- FLAG_r = 0x100, /* Reverse */
- FLAG_d = 0x200, /* Ignore !(isalnum()|isspace()) */
- FLAG_f = 0x400, /* Force uppercase */
- FLAG_i = 0x800, /* Ignore !isprint() */
- FLAG_m = 0x1000, /* ignored: merge already sorted files; do not sort */
- FLAG_S = 0x2000, /* ignored: -S, --buffer-size=SIZE */
- FLAG_T = 0x4000, /* ignored: -T, --temporary-directory=DIR */
- FLAG_o = 0x8000,
- FLAG_k = 0x10000,
- FLAG_t = 0x20000,
+ FLAG_b = 1 << 8, /* Ignore leading blanks */
+ FLAG_r = 1 << 9, /* Reverse */
+ FLAG_d = 1 << 10, /* Ignore !(isalnum()|isspace()) */
+ FLAG_f = 1 << 11, /* Force uppercase */
+ FLAG_i = 1 << 12, /* Ignore !isprint() */
+ FLAG_m = 1 << 13, /* ignored: merge already sorted files; do not sort */
+ FLAG_S = 1 << 14, /* ignored: -S, --buffer-size=SIZE */
+ FLAG_T = 1 << 15, /* ignored: -T, --temporary-directory=DIR */
+ FLAG_o = 1 << 16,
+ FLAG_k = 1 << 17,
+ FLAG_t = 1 << 18,
FLAG_bb = 0x80000000, /* Ignore trailing blanks */
FLAG_no_tie_break = 0x40000000,
};
static const char sort_opt_str[] ALIGN1 = "^"
- "ngMucszbrdfimS:T:o:k:*t:"
+ "ngMVucszbrdfimS:T:o:k:*t:"
"\0" "o--o:t--t"/*-t, -o: at most one of each*/;
/*
* OPT_STR must not be string literal, needs to have stable address:
y = *(char **)yarg;
#endif
/* Perform actual comparison */
- switch (flags & (FLAG_n | FLAG_M | FLAG_g)) {
+ switch (flags & (FLAG_n | FLAG_g | FLAG_M | FLAG_V)) {
default:
bb_error_msg_and_die("unknown sort type");
break;
+#if defined(HAVE_STRVERSCMP) && HAVE_STRVERSCMP == 1
+ case FLAG_V:
+ retval = strverscmp(x, y);
+ break;
+#endif
/* Ascii sort */
case 0:
#if ENABLE_LOCALE_SUPPORT
#if ENABLE_FEATURE_SORT_OPTIMIZE_MEMORY
bool can_drop_dups;
size_t prev_len = 0;
+ char *prev_line = (char*) "";
/* Postpone optimizing if the input is small, < 16k lines:
* even just free()ing duplicate lines takes time.
*/
if (count_to_optimize_dups == 0) {
size_t len;
char *new_line;
- char first = *line;
/* On kernel/linux/arch/ *.[ch] files,
* this reduces memory usage by 6%.
* yes | head -99999999 | sort
* goes down from 1900Mb to 380 Mb.
*/
- if (first == '\0' || first == '\n') {
- len = !(first == '\0');
- new_line = (char*)"\n" + 1 - len;
- goto replace;
- }
len = strlen(line);
if (len <= prev_len) {
- new_line = lines[linecount-1] + (prev_len - len);
+ new_line = prev_line + (prev_len - len);
if (strcmp(line, new_line) == 0) {
+ /* it's a tail of the prev line */
if (can_drop_dups && prev_len == len) {
+ /* it's identical to prev line */
free(line);
continue;
}
- replace:
free(line);
line = new_line;
+ /* continue using longer prev_line
+ * for future tail tests.
+ */
+ goto skip;
}
}
prev_len = len;
+ prev_line = line;
+ skip: ;
}
+#else
+//TODO: lighter version which only drops total dups if can_drop_dups == true
#endif
lines = xrealloc_vector(lines, 6, linecount);
lines[linecount++] = line;
}
return EXIT_SUCCESS;
}
-#else
-//TODO: lighter version which only drops total dups if can_drop_dups == true
#endif
/* For stable sort, store original line position beyond terminating NUL */
projects / oweals / busybox.git / blobdiff