logo

utils-std

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

wc.c (6551B)

    

  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 <ctype.h> // isspace
  9. #include <errno.h>
  10. #include <fcntl.h>  // posix_fadvise
  11. #include <locale.h> // setlocale
  12. #include <stdbool.h>
  13. #include <stdint.h> // uint8_t
  14. #include <stdio.h>  // fprintf, fopen
  15. #include <stdlib.h> // abort
  16. #include <string.h> // strchr, strerror
  17. #include <sys/stat.h>
  18. #include <unistd.h> // getopt
  19. #include <wchar.h>
  20. #include <wctype.h> // iswspace
  21. #ifdef HAS_GETOPT_LONG
  22. #include <getopt.h>
  23. #endif
  25. #define WC_BUFSIZ 16320
  26. static char buf[WC_BUFSIZ] = "";
  28. static const char *argv0 = "wc";
  30. static enum {
  31. 	WC_OPT_C = 1 << 0,
  32. 	WC_OPT_L = 1 << 1,
  33. 	WC_OPT_W = 1 << 2,
  34. 	WC_OPT_ALL = WC_OPT_C | WC_OPT_L | WC_OPT_W,
  35. } wc_opts = 0;
  37. off_t total_bytes = 0, total_lines = 0, total_words = 0;
  39. static void
  40. print_counts(off_t lines, off_t words, off_t bytes, const char *filename)
  41. {
  42. 	const char *fmt = "%ld";
  43. 	if(FIELD_MATCH(wc_opts, WC_OPT_L))
  44. 	{
  45. 		printf(fmt, lines);
  46. 		fmt = " %ld";
  47. 	}
  48. 	if(FIELD_MATCH(wc_opts, WC_OPT_W))
  49. 	{
  50. 		printf(fmt, words);
  51. 		fmt = " %ld";
  52. 	}
  53. 	if(FIELD_MATCH(wc_opts, WC_OPT_C))
  54. 	{
  55. 		printf(fmt, bytes);
  56. 	}
  58. 	if(filename != NULL) printf(" %s", filename);
  59. 	printf("\n");
  60. }
  62. static int
  63. wc_file_bytes(int fd, char *filename)
  64. {
  65. 	off_t bytes = 0, lines = 0, words = 0, wordlen = 0;
  67. 	ssize_t nread = -1;
  68. 	if(FIELD_MATCH(wc_opts, WC_OPT_L) || FIELD_MATCH(wc_opts, WC_OPT_W))
  69. 	{
  70. 		while((nread = read(fd, buf, WC_BUFSIZ)) > 0)
  71. 		{
  72. 			bytes += nread;
  74. 			for(ssize_t i = 0; i < nread; i++)
  75. 			{
  76. 				int c = buf[i];
  78. 				if(c == '\n') lines++;
  80. 				if(isspace(c))
  81. 				{
  82. 					if(wordlen > 0)
  83. 					{
  84. 						words++;
  85. 						wordlen = 0;
  86. 					}
  87. 				}
  88. 				else
  89. 				{
  90. 					wordlen++;
  91. 				}
  92. 			}
  93. 		}
  94. 	}
  95. 	else
  96. 	{
  97. 		struct stat status;
  98. 		if(fstat(fd, &status) < 0)
  99. 		{
  100. 			fprintf(stderr,
  101. 			        "%s: error: Failed getting status for file '%s': %s\n",
  102. 			        argv0,
  103. 			        filename,
  104. 			        strerror(errno));
  105. 			return 1;
  106. 		}
  108. 		if(S_ISREG(status.st_mode) && status.st_size != 0)
  109. 		{
  110. 			bytes += status.st_size;
  111. 		}
  112. 		else
  113. 		{
  114. 			while((nread = read(fd, buf, WC_BUFSIZ)) > 0)
  115. 				bytes += nread;
  116. 		}
  117. 	}
  119. 	if(nread < 0 && errno != 0)
  120. 	{
  121. 		fprintf(stderr,
  122. 		        "%s: error: Failed reading from file '%s': %s\n",
  123. 		        argv0,
  124. 		        filename != NULL ? filename : "<stdin>",
  125. 		        strerror(errno));
  126. 		return -1;
  127. 	}
  129. 	if(wordlen > 0) words++;
  131. 	total_bytes += bytes, total_lines += lines, total_words += words;
  133. 	print_counts(lines, words, bytes, filename);
  135. 	return 0;
  136. }
  138. static int
  139. wc_file_chars(int fd, char *filename)
  140. {
  141. 	off_t chars = 0, lines = 0, words = 0, wordlen = 0;
  143. 	FILE *file = fdopen(fd, "r");
  144. 	if(file == NULL)
  145. 	{
  146. 		fprintf(stderr,
  147. 		        "%s: error: Failed getting file stream for file '%s': %s\n",
  148. 		        argv0,
  149. 		        filename,
  150. 		        strerror(errno));
  151. 		return -1;
  152. 	}
  154. 	if(setvbuf(file, buf, _IOFBF, WC_BUFSIZ) != 0)
  155. 	{
  156. 		fprintf(stderr,
  157. 		        "%s: warning: Failed setting a new buffer for <stdin>: %s\n",
  158. 		        argv0,
  159. 		        strerror(errno));
  160. 		errno = 0;
  161. 	}
  163. 	while(true)
  164. 	{
  165. 		wint_t c = getwc(file);
  166. 		if(c == WEOF)
  167. 		{
  168. 			if(errno != 0)
  169. 			{
  170. 				fprintf(stderr,
  171. 				        "%s: error: Failed reading from file '%s': %s\n",
  172. 				        argv0,
  173. 				        filename != NULL ? filename : "<stdin>",
  174. 				        strerror(errno));
  175. 				return -1;
  176. 			}
  177. 			break;
  178. 		}
  179. 		chars++;
  181. 		if(c == '\n') lines++;
  183. 		if(iswspace(c))
  184. 		{
  185. 			if(wordlen > 0)
  186. 			{
  187. 				words++;
  188. 				wordlen = 0;
  189. 			}
  190. 		}
  191. 		else
  192. 		{
  193. 			wordlen++;
  194. 		}
  195. 	}
  197. 	if(wordlen > 0) words++;
  199. 	print_counts(lines, words, chars, filename);
  201. 	return 0;
  202. }
  204. static void
  205. usage(void)
  206. {
  207. 	fprintf(stderr, "Usage: wc [-c|-m] [-lw] [file...]\n");
  208. }
  210. int
  211. main(int argc, char *argv[])
  212. {
  213. 	setlocale(LC_ALL, "");
  214. 	errno = 0;
  215. 	int (*wc_file)(int, char *) = &wc_file_bytes;
  217. 	int c = -1;
  218. #ifdef HAS_GETOPT_LONG
  219. 	// Strictly for GNUisms compatibility so no long-only options
  220. 	// clang-format off
  221. 	static struct option opts[] = {
  222. 		{"bytes", no_argument, 0, 'c'},
  223. 		{"lines", no_argument, 0, 'l'},
  224. 		{"chars", no_argument, 0, 'm'},
  225. 		{"words", no_argument, 0, 'w'},
  226. 		{0, 0, 0, 0},
  227. 	};
  228. 	// clang-format on
  230. 	// Need + as first character to get POSIX-style option parsing
  231. 	while((c = getopt_long(argc, argv, "+:clmw", opts, NULL)) != -1)
  232. #else
  233. 	while((c = getopt(argc, argv, ":clmw")) != -1)
  234. #endif
  235. 	{
  236. 		switch(c)
  237. 		{
  238. 		case 'c':
  239. 			wc_opts |= WC_OPT_C;
  240. 			wc_file = wc_file_bytes;
  241. 			break;
  242. 		case 'l':
  243. 			wc_opts |= WC_OPT_L;
  244. 			break;
  245. 		case 'm':
  246. 			wc_opts |= WC_OPT_C;
  247. 			wc_file = wc_file_chars;
  248. 			break;
  249. 		case 'w':
  250. 			wc_opts |= WC_OPT_W;
  251. 			break;
  252. 		case ':':
  253. 			fprintf(stderr, "%s: error: Missing operand for option: '-%c'\n", argv0, optopt);
  254. 			usage();
  255. 			return 1;
  256. 		case '?':
  257. 			fprintf(stderr, "%s: error: Unrecognised option: '-%c'\n", argv0, optopt);
  258. 			usage();
  259. 			return 1;
  260. 		default:
  261. 			abort();
  262. 		}
  263. 	}
  264. 	if(wc_opts == 0) wc_opts = WC_OPT_ALL;
  266. 	argc -= optind;
  267. 	argv += optind;
  269. 	if((errno = posix_fadvise(STDIN_FILENO, 0, 0, POSIX_FADV_SEQUENTIAL)) != 0)
  270. 	{
  271. 		if(errno != ESPIPE)
  272. 		{
  273. 			fprintf(stderr,
  274. 			        "%s: warning: Failure from posix_fadvise sequential for <stdin>: %s\n",
  275. 			        argv0,
  276. 			        strerror(errno));
  277. 		}
  278. 		errno = 0;
  279. 	}
  281. 	if(argc < 1)
  282. 	{
  283. 		if(wc_file(STDIN_FILENO, NULL) < 0) return 1;
  284. 	}
  286. 	for(int i = 0; i < argc; i++)
  287. 	{
  288. 		char *path = argv[i];
  289. 		if(path[0] == '-' && path[1] == 0)
  290. 		{
  291. 			if(wc_file(STDIN_FILENO, NULL) < 0) return 1;
  293. 			continue;
  294. 		}
  296. 		// https://www.austingroupbugs.net/view.php?id=251
  297. 		if(strchr(path, '\n') != NULL)
  298. 			fprintf(
  299. 			    stderr,
  300. 			    "%s: warning: Filename '%s' contains a newline while wc(1) uses newlines as separators\n",
  301. 			    argv0,
  302. 			    path);
  304. 		int arg_fd = open(path, O_RDONLY | O_NOCTTY);
  305. 		if(arg_fd < 0)
  306. 		{
  307. 			fprintf(stderr, "%s: error: Failed opening file '%s': %s\n", argv0, path, strerror(errno));
  308. 			return 1;
  309. 		}
  311. 		if((errno = posix_fadvise(arg_fd, 0, 0, POSIX_FADV_SEQUENTIAL)) != 0)
  312. 		{
  313. 			fprintf(stderr,
  314. 			        "%s: warning: Failure from posix_fadvise sequential for file '%s': %s\n",
  315. 			        argv0,
  316. 			        path,
  317. 			        strerror(errno));
  318. 			errno = 0;
  319. 		}
  321. 		if(wc_file(arg_fd, path) < 0) return 1;
  323. 		if(close(arg_fd) < 0)
  324. 		{
  325. 			fprintf(stderr, "%s: error: Failed closing file '%s': %s\n", argv0, path, strerror(errno));
  326. 			return 1;
  327. 		}
  328. 	}
  330. 	if(argc > 1) print_counts(total_lines, total_words, total_bytes, "total");
  332. 	return 0;
  333. }