logo

utils-std

Collection of commonly available Unix tools git clone https://anongit.hacktivis.me/git/utils-std.git/

wc.c (6828B)

    

  1. // utils-std: Collection of commonly available Unix tools
  2. // SPDX-FileCopyrightText: 2017 Haelwenn (lanodan) Monnier <contact+utils@hacktivis.me>
  3. // SPDX-License-Identifier: MPL-2.0
  5. #define _POSIX_C_SOURCE 200809L
  6. #include "../config.h" // HAS_*
  7. #include "../lib/bitmasks.h"
  8. #include "../lib/getopt_nolong.h"
  10. #include <ctype.h> // isspace
  11. #include <errno.h>
  12. #include <fcntl.h>  // posix_fadvise
  13. #include <locale.h> // setlocale
  14. #include <stdbool.h>
  15. #include <stdint.h> // uint8_t
  16. #include <stdio.h>  // fprintf, fopen
  17. #include <stdlib.h> // abort
  18. #include <string.h> // strchr, strerror
  19. #include <sys/stat.h>
  20. #include <unistd.h> // getopt
  21. #include <wchar.h>
  22. #include <wctype.h> // iswspace
  23. #ifdef HAS_GETOPT_LONG
  24. #include <getopt.h>
  25. #endif
  27. #define WC_BUFSIZ 16320
  28. static char buf[WC_BUFSIZ] = "";
  30. const char *argv0 = "wc";
  32. static enum {
  33. 	WC_OPT_C = 1 << 0,
  34. 	WC_OPT_L = 1 << 1,
  35. 	WC_OPT_W = 1 << 2,
  36. 	WC_OPT_ALL = WC_OPT_C | WC_OPT_L | WC_OPT_W,
  37. } wc_opts = 0;
  39. off_t total_bytes = 0, total_lines = 0, total_words = 0;
  41. static void
  42. print_counts(off_t lines, off_t words, off_t bytes, const char *filename)
  43. {
  44. 	const char *fmt = "%ld";
  45. 	if(FIELD_MATCH(wc_opts, WC_OPT_L))
  46. 	{
  47. 		printf(fmt, lines);
  48. 		fmt = " %ld";
  49. 	}
  50. 	if(FIELD_MATCH(wc_opts, WC_OPT_W))
  51. 	{
  52. 		printf(fmt, words);
  53. 		fmt = " %ld";
  54. 	}
  55. 	if(FIELD_MATCH(wc_opts, WC_OPT_C))
  56. 	{
  57. 		printf(fmt, bytes);
  58. 	}
  60. 	if(filename != NULL) printf(" %s", filename);
  61. 	printf("\n");
  62. }
  64. static int
  65. wc_file_bytes(int fd, char *filename)
  66. {
  67. 	off_t bytes = 0, lines = 0, words = 0, wordlen = 0;
  69. 	ssize_t nread = -1;
  70. 	if(FIELD_MATCH(wc_opts, WC_OPT_L) || FIELD_MATCH(wc_opts, WC_OPT_W))
  71. 	{
  72. 		while((nread = read(fd, buf, WC_BUFSIZ)) > 0)
  73. 		{
  74. 			bytes += nread;
  76. 			for(ssize_t i = 0; i < nread; i++)
  77. 			{
  78. 				int c = buf[i];
  80. 				if(c == '\n') lines++;
  82. 				if(isspace(c))
  83. 				{
  84. 					if(wordlen > 0)
  85. 					{
  86. 						words++;
  87. 						wordlen = 0;
  88. 					}
  89. 				}
  90. 				else
  91. 				{
  92. 					wordlen++;
  93. 				}
  94. 			}
  95. 		}
  96. 	}
  97. 	else
  98. 	{
  99. 		struct stat status;
  100. 		if(fstat(fd, &status) < 0)
  101. 		{
  102. 			fprintf(stderr,
  103. 			        "%s: error: Failed getting status for file '%s': %s\n",
  104. 			        argv0,
  105. 			        filename,
  106. 			        strerror(errno));
  107. 			return 1;
  108. 		}
  110. 		if(S_ISREG(status.st_mode) && status.st_size != 0)
  111. 		{
  112. 			bytes += status.st_size;
  113. 		}
  114. 		else
  115. 		{
  116. 			while((nread = read(fd, buf, WC_BUFSIZ)) > 0)
  117. 				bytes += nread;
  118. 		}
  119. 	}
  121. 	if(nread < 0 && errno != 0)
  122. 	{
  123. 		fprintf(stderr,
  124. 		        "%s: error: Failed reading from file '%s': %s\n",
  125. 		        argv0,
  126. 		        filename != NULL ? filename : "<stdin>",
  127. 		        strerror(errno));
  128. 		return -1;
  129. 	}
  131. 	if(wordlen > 0) words++;
  133. 	total_bytes += bytes, total_lines += lines, total_words += words;
  135. 	print_counts(lines, words, bytes, filename);
  137. 	return 0;
  138. }
  140. static int
  141. wc_file_chars(int fd, char *filename)
  142. {
  143. 	off_t chars = 0, lines = 0, words = 0, wordlen = 0;
  145. 	FILE *file = fdopen(fd, "r");
  146. 	if(file == NULL)
  147. 	{
  148. 		fprintf(stderr,
  149. 		        "%s: error: Failed getting file stream for file '%s': %s\n",
  150. 		        argv0,
  151. 		        filename,
  152. 		        strerror(errno));
  153. 		return -1;
  154. 	}
  156. 	if(setvbuf(file, buf, _IOFBF, WC_BUFSIZ) != 0)
  157. 	{
  158. 		fprintf(stderr,
  159. 		        "%s: warning: Failed setting a new buffer for <stdin>: %s\n",
  160. 		        argv0,
  161. 		        strerror(errno));
  162. 		errno = 0;
  163. 	}
  165. 	while(true)
  166. 	{
  167. 		wint_t c = getwc(file);
  168. 		if(c == WEOF)
  169. 		{
  170. 			if(errno != 0)
  171. 			{
  172. 				fprintf(stderr,
  173. 				        "%s: error: Failed reading from file '%s': %s\n",
  174. 				        argv0,
  175. 				        filename != NULL ? filename : "<stdin>",
  176. 				        strerror(errno));
  177. 				return -1;
  178. 			}
  179. 			break;
  180. 		}
  181. 		chars++;
  183. 		if(c == '\n') lines++;
  185. 		if(iswspace(c))
  186. 		{
  187. 			if(wordlen > 0)
  188. 			{
  189. 				words++;
  190. 				wordlen = 0;
  191. 			}
  192. 		}
  193. 		else
  194. 		{
  195. 			wordlen++;
  196. 		}
  197. 	}
  199. 	if(wordlen > 0) words++;
  201. 	print_counts(lines, words, chars, filename);
  203. 	return 0;
  204. }
  206. static void
  207. usage(void)
  208. {
  209. 	fprintf(stderr, "Usage: wc [-c|-m] [-lw] [file...]\n");
  210. }
  212. int
  213. main(int argc, char *argv[])
  214. {
  215. 	char *lc_all = setlocale(LC_ALL, "");
  216. 	if(lc_all == NULL)
  217. 	{
  218. 		fprintf(stderr,
  219. 		        "%s: warning: Failed loading locales. setlocale(LC_ALL, \"\"): %s\n",
  220. 		        argv0,
  221. 		        strerror(errno));
  222. 	}
  223. 	errno = 0;
  224. 	int (*wc_file)(int, char *) = &wc_file_bytes;
  226. #ifdef HAS_GETOPT_LONG
  227. 	// Strictly for GNUisms compatibility so no long-only options
  228. 	// clang-format off
  229. 	static struct option opts[] = {
  230. 		{"bytes", no_argument, 0, 'c'},
  231. 		{"lines", no_argument, 0, 'l'},
  232. 		{"chars", no_argument, 0, 'm'},
  233. 		{"words", no_argument, 0, 'w'},
  234. 		{0, 0, 0, 0},
  235. 	};
  236. 	// clang-format on
  238. 	// Need + as first character to get POSIX-style option parsing
  239. 	for(int c = -1; (c = getopt_long(argc, argv, "+:clmw", opts, NULL)) != -1;)
  240. #else
  241. 	for(int c = -1; (c = getopt_nolong(argc, argv, ":clmw")) != -1;)
  242. #endif
  243. 	{
  244. 		switch(c)
  245. 		{
  246. 		case 'c':
  247. 			wc_opts |= WC_OPT_C;
  248. 			wc_file = wc_file_bytes;
  249. 			break;
  250. 		case 'l':
  251. 			wc_opts |= WC_OPT_L;
  252. 			break;
  253. 		case 'm':
  254. 			wc_opts |= WC_OPT_C;
  255. 			wc_file = wc_file_chars;
  256. 			break;
  257. 		case 'w':
  258. 			wc_opts |= WC_OPT_W;
  259. 			break;
  260. 		case ':':
  261. 			fprintf(stderr, "%s: error: Missing operand for option: '-%c'\n", argv0, optopt);
  262. 			usage();
  263. 			return 1;
  264. 		case '?':
  265. 			if(!got_long_opt) fprintf(stderr, "%s: error: Unrecognised option: '-%c'\n", argv0, optopt);
  266. 			usage();
  267. 			return 1;
  268. 		default:
  269. 			abort();
  270. 		}
  271. 	}
  272. 	if(wc_opts == 0) wc_opts = WC_OPT_ALL;
  274. 	argc -= optind;
  275. 	argv += optind;
  277. 	if((errno = posix_fadvise(STDIN_FILENO, 0, 0, POSIX_FADV_SEQUENTIAL)) != 0)
  278. 	{
  279. 		if(errno != ESPIPE)
  280. 		{
  281. 			fprintf(stderr,
  282. 			        "%s: warning: Failure from posix_fadvise sequential for <stdin>: %s\n",
  283. 			        argv0,
  284. 			        strerror(errno));
  285. 		}
  286. 		errno = 0;
  287. 	}
  289. 	if(argc < 1)
  290. 	{
  291. 		if(wc_file(STDIN_FILENO, NULL) < 0) return 1;
  292. 	}
  294. 	for(int i = 0; i < argc; i++)
  295. 	{
  296. 		char *path = argv[i];
  297. 		if(path[0] == '-' && path[1] == 0)
  298. 		{
  299. 			if(wc_file(STDIN_FILENO, NULL) < 0) return 1;
  301. 			continue;
  302. 		}
  304. 		// https://www.austingroupbugs.net/view.php?id=251
  305. 		if(strchr(path, '\n') != NULL)
  306. 			fprintf(
  307. 			    stderr,
  308. 			    "%s: warning: Filename '%s' contains a newline while wc(1) uses newlines as separators\n",
  309. 			    argv0,
  310. 			    path);
  312. 		int arg_fd = open(path, O_RDONLY | O_NOCTTY);
  313. 		if(arg_fd < 0)
  314. 		{
  315. 			fprintf(stderr, "%s: error: Failed opening file '%s': %s\n", argv0, path, strerror(errno));
  316. 			return 1;
  317. 		}
  319. 		if((errno = posix_fadvise(arg_fd, 0, 0, POSIX_FADV_SEQUENTIAL)) != 0)
  320. 		{
  321. 			fprintf(stderr,
  322. 			        "%s: warning: Failure from posix_fadvise sequential for file '%s': %s\n",
  323. 			        argv0,
  324. 			        path,
  325. 			        strerror(errno));
  326. 			errno = 0;
  327. 		}
  329. 		if(wc_file(arg_fd, path) < 0) return 1;
  331. 		if(close(arg_fd) < 0)
  332. 		{
  333. 			fprintf(stderr, "%s: error: Failed closing file '%s': %s\n", argv0, path, strerror(errno));
  334. 			return 1;
  335. 		}
  336. 	}
  338. 	if(argc > 1) print_counts(total_lines, total_words, total_bytes, "total");
  340. 	return 0;
  341. }