11590Srgrimes/*- 21590Srgrimes * Copyright (c) 1991, 1993 31590Srgrimes * The Regents of the University of California. All rights reserved. 41590Srgrimes * 51590Srgrimes * This code is derived from software contributed to Berkeley by 61590Srgrimes * David Hitz of Auspex Systems, Inc. 71590Srgrimes * 81590Srgrimes * Redistribution and use in source and binary forms, with or without 91590Srgrimes * modification, are permitted provided that the following conditions 101590Srgrimes * are met: 111590Srgrimes * 1. Redistributions of source code must retain the above copyright 121590Srgrimes * notice, this list of conditions and the following disclaimer. 131590Srgrimes * 2. Redistributions in binary form must reproduce the above copyright 141590Srgrimes * notice, this list of conditions and the following disclaimer in the 151590Srgrimes * documentation and/or other materials provided with the distribution. 161590Srgrimes * 4. Neither the name of the University nor the names of its contributors 171590Srgrimes * may be used to endorse or promote products derived from this software 181590Srgrimes * without specific prior written permission. 191590Srgrimes * 201590Srgrimes * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 211590Srgrimes * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 221590Srgrimes * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 231590Srgrimes * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 241590Srgrimes * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 251590Srgrimes * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 261590Srgrimes * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 271590Srgrimes * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 281590Srgrimes * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 291590Srgrimes * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 301590Srgrimes * SUCH DAMAGE. 311590Srgrimes */ 321590Srgrimes 331590Srgrimes#ifndef lint 3427623Scharnierstatic const char copyright[] = 351590Srgrimes"@(#) Copyright (c) 1991, 1993\n\ 361590Srgrimes The Regents of the University of California. All rights reserved.\n"; 371590Srgrimes#endif /* not lint */ 381590Srgrimes 391590Srgrimes#ifndef lint 4027623Scharnier#if 0 4123693Speterstatic char sccsid[] = "@(#)look.c 8.2 (Berkeley) 5/4/95"; 4227623Scharnier#endif 431590Srgrimes#endif /* not lint */ 4499112Sobrien#include <sys/cdefs.h> 4599112Sobrien__FBSDID("$FreeBSD$"); 461590Srgrimes 471590Srgrimes/* 481590Srgrimes * look -- find lines in a sorted list. 498874Srgrimes * 501590Srgrimes * The man page said that TABs and SPACEs participate in -d comparisons. 511590Srgrimes * In fact, they were ignored. This implements historic practice, not 521590Srgrimes * the manual page. 531590Srgrimes */ 541590Srgrimes 551590Srgrimes#include <sys/types.h> 561590Srgrimes#include <sys/mman.h> 571590Srgrimes#include <sys/stat.h> 581590Srgrimes 5927623Scharnier#include <err.h> 601590Srgrimes#include <errno.h> 611590Srgrimes#include <fcntl.h> 6223693Speter#include <limits.h> 6327623Scharnier#include <locale.h> 64226468Sed#include <stdint.h> 651590Srgrimes#include <stdio.h> 661590Srgrimes#include <stdlib.h> 671590Srgrimes#include <string.h> 6823693Speter#include <unistd.h> 69132394Stjr#include <wchar.h> 70132394Stjr#include <wctype.h> 7123693Speter 721590Srgrimes#include "pathnames.h" 731590Srgrimes 7495646Smarkmstatic char _path_words[] = _PATH_WORDS; 7595646Smarkm 761590Srgrimes#define EQUAL 0 771590Srgrimes#define GREATER 1 781590Srgrimes#define LESS (-1) 791590Srgrimes 80227171Sedstatic int dflag, fflag; 811590Srgrimes 82227171Sedstatic char *binary_search(wchar_t *, unsigned char *, unsigned char *); 83227171Sedstatic int compare(wchar_t *, unsigned char *, unsigned char *); 84227171Sedstatic char *linear_search(wchar_t *, unsigned char *, unsigned char *); 85227171Sedstatic int look(wchar_t *, unsigned char *, unsigned char *); 86227171Sedstatic wchar_t *prepkey(const char *, wchar_t); 87227171Sedstatic void print_from(wchar_t *, unsigned char *, unsigned char *); 881590Srgrimes 8992920Simpstatic void usage(void); 901590Srgrimes 9127623Scharnierint 92102944Sdwmalonemain(int argc, char *argv[]) 931590Srgrimes{ 941590Srgrimes struct stat sb; 95132394Stjr int ch, fd, match; 96132394Stjr wchar_t termchar; 97132394Stjr unsigned char *back, *front; 9887288Sdwmalone unsigned const char *file; 99132394Stjr wchar_t *key; 1001590Srgrimes 10111900Sache (void) setlocale(LC_CTYPE, ""); 10211900Sache 10395646Smarkm file = _path_words; 104132394Stjr termchar = L'\0'; 10524360Simp while ((ch = getopt(argc, argv, "dft:")) != -1) 1061590Srgrimes switch(ch) { 1071590Srgrimes case 'd': 1081590Srgrimes dflag = 1; 1091590Srgrimes break; 1101590Srgrimes case 'f': 1111590Srgrimes fflag = 1; 1121590Srgrimes break; 1131590Srgrimes case 't': 114132394Stjr if (mbrtowc(&termchar, optarg, MB_LEN_MAX, NULL) != 115132394Stjr strlen(optarg)) 116132394Stjr errx(2, "invalid termination character"); 1171590Srgrimes break; 1181590Srgrimes case '?': 1191590Srgrimes default: 1201590Srgrimes usage(); 1211590Srgrimes } 1221590Srgrimes argc -= optind; 1231590Srgrimes argv += optind; 1241590Srgrimes 12538701Swosch if (argc == 0) 12638701Swosch usage(); 12738701Swosch if (argc == 1) /* But set -df by default. */ 1281590Srgrimes dflag = fflag = 1; 129132394Stjr key = prepkey(*argv++, termchar); 13038701Swosch if (argc >= 2) 13138701Swosch file = *argv++; 1321590Srgrimes 13338701Swosch match = 1; 1341590Srgrimes 13538701Swosch do { 13638701Swosch if ((fd = open(file, O_RDONLY, 0)) < 0 || fstat(fd, &sb)) 13738701Swosch err(2, "%s", file); 138226444Sed if ((uintmax_t)sb.st_size > (uintmax_t)SIZE_T_MAX) 13938701Swosch errx(2, "%s: %s", file, strerror(EFBIG)); 140196558Scperciva if (sb.st_size == 0) { 141196558Scperciva close(fd); 142196558Scperciva continue; 143196558Scperciva } 14438701Swosch if ((front = mmap(NULL, (size_t)sb.st_size, PROT_READ, MAP_SHARED, fd, (off_t)0)) == MAP_FAILED) 14538701Swosch err(2, "%s", file); 14638701Swosch back = front + sb.st_size; 147132394Stjr match *= (look(key, front, back)); 14838701Swosch close(fd); 14938701Swosch } while (argc-- > 2 && (file = *argv++)); 15038701Swosch 15138701Swosch exit(match); 1521590Srgrimes} 1531590Srgrimes 154227171Sedstatic wchar_t * 155132394Stjrprepkey(const char *string, wchar_t termchar) 1561590Srgrimes{ 157132394Stjr const char *readp; 158132394Stjr wchar_t *key, *writep; 159132394Stjr wchar_t ch; 160132394Stjr size_t clen; 1611590Srgrimes 162132394Stjr /* 163132394Stjr * Reformat search string and convert to wide character representation 164132394Stjr * to avoid doing it multiple times later. 165132394Stjr */ 166132394Stjr if ((key = malloc(sizeof(wchar_t) * (strlen(string) + 1))) == NULL) 167132394Stjr err(2, NULL); 168132394Stjr readp = string; 169132394Stjr writep = key; 170132394Stjr while ((clen = mbrtowc(&ch, readp, MB_LEN_MAX, NULL)) != 0) { 171132394Stjr if (clen == (size_t)-1 || clen == (size_t)-2) 172132394Stjr errc(2, EILSEQ, NULL); 1731590Srgrimes if (fflag) 174132394Stjr ch = towlower(ch); 175132394Stjr if (!dflag || iswalnum(ch)) 176132394Stjr *writep++ = ch; 177132394Stjr readp += clen; 1781590Srgrimes } 179132394Stjr *writep = L'\0'; 180132394Stjr if (termchar != L'\0' && (writep = wcschr(key, termchar)) != NULL) 181132394Stjr *++writep = L'\0'; 182132394Stjr return (key); 183132394Stjr} 1841590Srgrimes 185227171Sedstatic int 186132394Stjrlook(wchar_t *string, unsigned char *front, unsigned char *back) 187132394Stjr{ 188132394Stjr 1891590Srgrimes front = binary_search(string, front, back); 1901590Srgrimes front = linear_search(string, front, back); 1911590Srgrimes 1921590Srgrimes if (front) 1931590Srgrimes print_from(string, front, back); 1941590Srgrimes return (front ? 0 : 1); 1951590Srgrimes} 1961590Srgrimes 1971590Srgrimes 1981590Srgrimes/* 1991590Srgrimes * Binary search for "string" in memory between "front" and "back". 2008874Srgrimes * 2011590Srgrimes * This routine is expected to return a pointer to the start of a line at 2021590Srgrimes * *or before* the first word matching "string". Relaxing the constraint 2031590Srgrimes * this way simplifies the algorithm. 2048874Srgrimes * 2051590Srgrimes * Invariants: 2068874Srgrimes * front points to the beginning of a line at or before the first 2071590Srgrimes * matching string. 2088874Srgrimes * 2098874Srgrimes * back points to the beginning of a line at or after the first 2101590Srgrimes * matching line. 2118874Srgrimes * 2121590Srgrimes * Base of the Invariants. 2138874Srgrimes * front = NULL; 2141590Srgrimes * back = EOF; 2158874Srgrimes * 2161590Srgrimes * Advancing the Invariants: 2178874Srgrimes * 2181590Srgrimes * p = first newline after halfway point from front to back. 2198874Srgrimes * 2208874Srgrimes * If the string at "p" is not greater than the string to match, 2211590Srgrimes * p is the new front. Otherwise it is the new back. 2228874Srgrimes * 2231590Srgrimes * Termination: 2248874Srgrimes * 2258874Srgrimes * The definition of the routine allows it return at any point, 2261590Srgrimes * since front is always at or before the line to print. 2278874Srgrimes * 2288874Srgrimes * In fact, it returns when the chosen "p" equals "back". This 2298874Srgrimes * implies that there exists a string is least half as long as 2308874Srgrimes * (back - front), which in turn implies that a linear search will 2311590Srgrimes * be no more expensive than the cost of simply printing a string or two. 2328874Srgrimes * 2338874Srgrimes * Trying to continue with binary search at this point would be 2341590Srgrimes * more trouble than it's worth. 2351590Srgrimes */ 2361590Srgrimes#define SKIP_PAST_NEWLINE(p, back) \ 2371590Srgrimes while (p < back && *p++ != '\n'); 2381590Srgrimes 239227171Sedstatic char * 240132394Stjrbinary_search(wchar_t *string, unsigned char *front, unsigned char *back) 2411590Srgrimes{ 242102944Sdwmalone unsigned char *p; 2431590Srgrimes 2441590Srgrimes p = front + (back - front) / 2; 2451590Srgrimes SKIP_PAST_NEWLINE(p, back); 2461590Srgrimes 2471590Srgrimes /* 2481590Srgrimes * If the file changes underneath us, make sure we don't 2491590Srgrimes * infinitely loop. 2501590Srgrimes */ 2511590Srgrimes while (p < back && back > front) { 2521590Srgrimes if (compare(string, p, back) == GREATER) 2531590Srgrimes front = p; 2541590Srgrimes else 2551590Srgrimes back = p; 2561590Srgrimes p = front + (back - front) / 2; 2571590Srgrimes SKIP_PAST_NEWLINE(p, back); 2581590Srgrimes } 2591590Srgrimes return (front); 2601590Srgrimes} 2611590Srgrimes 2621590Srgrimes/* 2631590Srgrimes * Find the first line that starts with string, linearly searching from front 2641590Srgrimes * to back. 2658874Srgrimes * 2661590Srgrimes * Return NULL for no such line. 2678874Srgrimes * 2681590Srgrimes * This routine assumes: 2698874Srgrimes * 2708874Srgrimes * o front points at the first character in a line. 2711590Srgrimes * o front is before or at the first line to be printed. 2721590Srgrimes */ 273227171Sedstatic char * 274132394Stjrlinear_search(wchar_t *string, unsigned char *front, unsigned char *back) 2751590Srgrimes{ 2761590Srgrimes while (front < back) { 2771590Srgrimes switch (compare(string, front, back)) { 2781590Srgrimes case EQUAL: /* Found it. */ 2791590Srgrimes return (front); 2801590Srgrimes case LESS: /* No such string. */ 2811590Srgrimes return (NULL); 2821590Srgrimes case GREATER: /* Keep going. */ 2831590Srgrimes break; 2841590Srgrimes } 2851590Srgrimes SKIP_PAST_NEWLINE(front, back); 2861590Srgrimes } 2871590Srgrimes return (NULL); 2881590Srgrimes} 2891590Srgrimes 2901590Srgrimes/* 2911590Srgrimes * Print as many lines as match string, starting at front. 2921590Srgrimes */ 293227171Sedstatic void 294132394Stjrprint_from(wchar_t *string, unsigned char *front, unsigned char *back) 2951590Srgrimes{ 2961590Srgrimes for (; front < back && compare(string, front, back) == EQUAL; ++front) { 2971590Srgrimes for (; front < back && *front != '\n'; ++front) 2981590Srgrimes if (putchar(*front) == EOF) 29927623Scharnier err(2, "stdout"); 3001590Srgrimes if (putchar('\n') == EOF) 30127623Scharnier err(2, "stdout"); 3021590Srgrimes } 3031590Srgrimes} 3041590Srgrimes 3051590Srgrimes/* 3061590Srgrimes * Return LESS, GREATER, or EQUAL depending on how the string1 compares with 3071590Srgrimes * string2 (s1 ??? s2). 3088874Srgrimes * 3098874Srgrimes * o Matches up to len(s1) are EQUAL. 3101590Srgrimes * o Matches up to len(s2) are GREATER. 3118874Srgrimes * 3121590Srgrimes * Compare understands about the -f and -d flags, and treats comparisons 3131590Srgrimes * appropriately. 3148874Srgrimes * 3151590Srgrimes * The string "s1" is null terminated. The string s2 is '\n' terminated (or 3161590Srgrimes * "back" terminated). 3171590Srgrimes */ 318227171Sedstatic int 319132394Stjrcompare(wchar_t *s1, unsigned char *s2, unsigned char *back) 3201590Srgrimes{ 321132394Stjr wchar_t ch1, ch2; 322132394Stjr size_t len2; 3231590Srgrimes 324132394Stjr for (; *s1 && s2 < back && *s2 != '\n'; ++s1, s2 += len2) { 325132394Stjr ch1 = *s1; 326132394Stjr len2 = mbrtowc(&ch2, s2, back - s2, NULL); 327132394Stjr if (len2 == (size_t)-1 || len2 == (size_t)-2) { 328132394Stjr ch2 = *s2; 329132394Stjr len2 = 1; 330132394Stjr } 3311590Srgrimes if (fflag) 332132394Stjr ch2 = towlower(ch2); 333132394Stjr if (dflag && !iswalnum(ch2)) { 334132394Stjr /* Ignore character in comparison. */ 335132394Stjr --s1; 3361590Srgrimes continue; 3371590Srgrimes } 338132394Stjr if (ch1 != ch2) 339132394Stjr return (ch1 < ch2 ? LESS : GREATER); 3401590Srgrimes } 3411590Srgrimes return (*s1 ? GREATER : EQUAL); 3421590Srgrimes} 3431590Srgrimes 3441590Srgrimesstatic void 345102944Sdwmaloneusage(void) 3461590Srgrimes{ 34738701Swosch (void)fprintf(stderr, "usage: look [-df] [-t char] string [file ...]\n"); 3481590Srgrimes exit(2); 3491590Srgrimes} 350