logo

utils-std

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

wc.c (6744B)

    

  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 "../lib/bitmasks.h"
  8. #include <assert.h>
  9. #include <ctype.h> // isspace
  10. #include <errno.h>
  11. #include <fcntl.h>  // posix_fadvise
  12. #include <locale.h> // setlocale
  13. #include <stdbool.h>
  14. #include <stdint.h> // uint8_t
  15. #include <stdio.h>  // fprintf, fopen
  16. #include <stdlib.h> // abort
  17. #include <string.h> // strchr, strerror
  18. #include <sys/stat.h>
  19. #include <unistd.h> // getopt
  20. #include <wchar.h>
  21. #include <wctype.h> // iswspace
  22. #ifdef HAS_GETOPT_LONG
  23. #include <getopt.h>
  24. #endif
  26. #define WC_BUFSIZ 16320
  27. static char buf[WC_BUFSIZ] = "";
  29. static const char *argv0 = "wc";
  31. static enum {
  32. 	WC_OPT_C = 1 << 0,
  33. 	WC_OPT_L = 1 << 1,
  34. 	WC_OPT_W = 1 << 2,
  35. 	WC_OPT_ALL = WC_OPT_C | WC_OPT_L | WC_OPT_W,
  36. } wc_opts = 0;
  38. off_t total_bytes = 0, total_lines = 0, total_words = 0;
  40. static void
  41. print_counts(off_t lines, off_t words, off_t bytes, const char *filename)
  42. {
  43. 	const char *fmt = "%ld";
  44. 	if(FIELD_MATCH(wc_opts, WC_OPT_L))
  45. 	{
  46. 		printf(fmt, lines);
  47. 		fmt = " %ld";
  48. 	}
  49. 	if(FIELD_MATCH(wc_opts, WC_OPT_W))
  50. 	{
  51. 		printf(fmt, words);
  52. 		fmt = " %ld";
  53. 	}
  54. 	if(FIELD_MATCH(wc_opts, WC_OPT_C))
  55. 	{
  56. 		printf(fmt, bytes);
  57. 	}
  59. 	if(filename != NULL) printf(" %s", filename);
  60. 	printf("\n");
  61. }
  63. static int
  64. wc_file_bytes(int fd, char *filename)
  65. {
  66. 	off_t bytes = 0, lines = 0, words = 0, wordlen = 0;
  68. 	ssize_t nread = -1;
  69. 	if(FIELD_MATCH(wc_opts, WC_OPT_L) || FIELD_MATCH(wc_opts, WC_OPT_W))
  70. 	{
  71. 		while((nread = read(fd, buf, WC_BUFSIZ)) > 0)
  72. 		{
  73. 			bytes += nread;
  75. 			for(ssize_t i = 0; i < nread; i++)
  76. 			{
  77. 				int c = buf[i];
  79. 				if(c == '\n') lines++;
  81. 				if(isspace(c))
  82. 				{
  83. 					if(wordlen > 0)
  84. 					{
  85. 						words++;
  86. 						wordlen = 0;
  87. 					}
  88. 				}
  89. 				else
  90. 				{
  91. 					wordlen++;
  92. 				}
  93. 			}
  94. 		}
  95. 	}
  96. 	else
  97. 	{
  98. 		struct stat status;
  99. 		if(fstat(fd, &status) < 0)
  100. 		{
  101. 			fprintf(stderr,
  102. 			        "%s: error: Failed getting status for file '%s': %s\n",
  103. 			        argv0,
  104. 			        filename,
  105. 			        strerror(errno));
  106. 			return 1;
  107. 		}
  109. 		if(S_ISREG(status.st_mode) && status.st_size != 0)
  110. 		{
  111. 			bytes += status.st_size;
  112. 		}
  113. 		else
  114. 		{
  115. 			while((nread = read(fd, buf, WC_BUFSIZ)) > 0)
  116. 				bytes += nread;
  117. 		}
  118. 	}
  120. 	if(nread < 0 && errno != 0)
  121. 	{
  122. 		fprintf(stderr,
  123. 		        "%s: error: Failed reading from file '%s': %s\n",
  124. 		        argv0,
  125. 		        filename != NULL ? filename : "<stdin>",
  126. 		        strerror(errno));
  127. 		return -1;
  128. 	}
  130. 	if(wordlen > 0) words++;
  132. 	total_bytes += bytes, total_lines += lines, total_words += words;
  134. 	print_counts(lines, words, bytes, filename);
  136. 	return 0;
  137. }
  139. static int
  140. wc_file_chars(int fd, char *filename)
  141. {
  142. 	off_t chars = 0, lines = 0, words = 0, wordlen = 0;
  144. 	FILE *file = fdopen(fd, "r");
  145. 	if(file == NULL)
  146. 	{
  147. 		fprintf(stderr,
  148. 		        "%s: error: Failed getting file stream for file '%s': %s\n",
  149. 		        argv0,
  150. 		        filename,
  151. 		        strerror(errno));
  152. 		return -1;
  153. 	}
  155. 	if(setvbuf(file, buf, _IOFBF, WC_BUFSIZ) != 0)
  156. 	{
  157. 		fprintf(stderr,
  158. 		        "%s: warning: Failed setting a new buffer for <stdin>: %s\n",
  159. 		        argv0,
  160. 		        strerror(errno));
  161. 		errno = 0;
  162. 	}
  164. 	while(true)
  165. 	{
  166. 		assert(errno == 0);
  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. 	errno = 0;
  216. 	setlocale(LC_ALL, "");
  217. 	if(errno != 0)
  218. 	{
  219. 		fprintf(stderr, "%s: warning: Failed to initialize locales: %s\n", argv0, strerror(errno));
  220. 		errno = 0;
  221. 	}
  222. 	int (*wc_file)(int, char *) = &wc_file_bytes;
  224. 	int c = -1;
  225. #ifdef HAS_GETOPT_LONG
  226. 	// Strictly for GNUisms compatibility so no long-only options
  227. 	// clang-format off
  228. 	static struct option opts[] = {
  229. 		{"bytes", no_argument, 0, 'c'},
  230. 		{"lines", no_argument, 0, 'l'},
  231. 		{"chars", no_argument, 0, 'm'},
  232. 		{"words", no_argument, 0, 'w'},
  233. 		{0, 0, 0, 0},
  234. 	};
  235. 	// clang-format on
  237. 	// Need + as first character to get POSIX-style option parsing
  238. 	while((c = getopt_long(argc, argv, "+:clmw", opts, NULL)) != -1)
  239. #else
  240. 	while((c = getopt(argc, argv, ":clmw")) != -1)
  241. #endif
  242. 	{
  243. 		switch(c)
  244. 		{
  245. 		case 'c':
  246. 			wc_opts |= WC_OPT_C;
  247. 			wc_file = wc_file_bytes;
  248. 			break;
  249. 		case 'l':
  250. 			wc_opts |= WC_OPT_L;
  251. 			break;
  252. 		case 'm':
  253. 			wc_opts |= WC_OPT_C;
  254. 			wc_file = wc_file_chars;
  255. 			break;
  256. 		case 'w':
  257. 			wc_opts |= WC_OPT_W;
  258. 			break;
  259. 		case ':':
  260. 			fprintf(stderr, "%s: error: Missing operand for option: '-%c'\n", argv0, optopt);
  261. 			usage();
  262. 			return 1;
  263. 		case '?':
  264. 			fprintf(stderr, "%s: error: Unrecognised option: '-%c'\n", argv0, optopt);
  265. 			usage();
  266. 			return 1;
  267. 		default:
  268. 			abort();
  269. 		}
  270. 	}
  271. 	if(wc_opts == 0) wc_opts = WC_OPT_ALL;
  273. 	assert(errno == 0);
  275. 	argc -= optind;
  276. 	argv += optind;
  278. 	if((errno = posix_fadvise(STDIN_FILENO, 0, 0, POSIX_FADV_SEQUENTIAL)) != 0)
  279. 	{
  280. 		if(errno != ESPIPE)
  281. 		{
  282. 			fprintf(stderr,
  283. 			        "%s: warning: Failure from posix_fadvise sequential for <stdin>: %s\n",
  284. 			        argv0,
  285. 			        strerror(errno));
  286. 		}
  287. 		errno = 0;
  288. 	}
  290. 	if(argc < 1)
  291. 	{
  292. 		if(wc_file(STDIN_FILENO, NULL) < 0) return 1;
  293. 	}
  295. 	for(int i = 0; i < argc; i++)
  296. 	{
  297. 		char *path = argv[i];
  298. 		if(path[0] == '-' && path[1] == 0)
  299. 		{
  300. 			if(wc_file(STDIN_FILENO, NULL) < 0) return 1;
  302. 			continue;
  303. 		}
  305. 		// https://www.austingroupbugs.net/view.php?id=251
  306. 		if(strchr(path, '\n') != NULL)
  307. 			fprintf(
  308. 			    stderr,
  309. 			    "%s: warning: Filename '%s' contains a newline while wc(1) uses newlines as separators\n",
  310. 			    argv0,
  311. 			    path);
  313. 		int arg_fd = open(path, O_RDONLY | O_NOCTTY);
  314. 		if(arg_fd < 0)
  315. 		{
  316. 			fprintf(stderr, "%s: error: Failed opening file '%s': %s\n", argv0, path, strerror(errno));
  317. 			return 1;
  318. 		}
  320. 		if((errno = posix_fadvise(arg_fd, 0, 0, POSIX_FADV_SEQUENTIAL)) != 0)
  321. 		{
  322. 			fprintf(stderr,
  323. 			        "%s: warning: Failure from posix_fadvise sequential for file '%s': %s\n",
  324. 			        argv0,
  325. 			        path,
  326. 			        strerror(errno));
  327. 			errno = 0;
  328. 		}
  330. 		if(wc_file(arg_fd, path) < 0) return 1;
  332. 		if(close(arg_fd) < 0)
  333. 		{
  334. 			fprintf(stderr, "%s: error: Failed closing file '%s': %s\n", argv0, path, strerror(errno));
  335. 			return 1;
  336. 		}
  337. 	}
  339. 	if(argc > 1) print_counts(total_lines, total_words, total_bytes, "total");
  341. 	return 0;
  342. }