12490Sjkh/*- 22490Sjkh * Copyright (c) 1989, 1993 32490Sjkh * The Regents of the University of California. All rights reserved. 42490Sjkh * 52490Sjkh * This code is derived from software contributed to Berkeley by 62490Sjkh * Ken Arnold. 72490Sjkh * 82490Sjkh * Redistribution and use in source and binary forms, with or without 92490Sjkh * modification, are permitted provided that the following conditions 102490Sjkh * are met: 112490Sjkh * 1. Redistributions of source code must retain the above copyright 122490Sjkh * notice, this list of conditions and the following disclaimer. 132490Sjkh * 2. Redistributions in binary form must reproduce the above copyright 142490Sjkh * notice, this list of conditions and the following disclaimer in the 152490Sjkh * documentation and/or other materials provided with the distribution. 16203926Suqs * 3. Neither the name of the University nor the names of its contributors 172490Sjkh * may be used to endorse or promote products derived from this software 182490Sjkh * without specific prior written permission. 192490Sjkh * 202490Sjkh * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 212490Sjkh * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 222490Sjkh * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 232490Sjkh * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 242490Sjkh * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 252490Sjkh * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 262490Sjkh * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 272490Sjkh * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 282490Sjkh * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 292490Sjkh * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 302490Sjkh * SUCH DAMAGE. 312490Sjkh */ 322490Sjkh 33114725Sobrien#if 0 342490Sjkh#ifndef lint 3515944Sachestatic const char copyright[] = 362490Sjkh"@(#) Copyright (c) 1989, 1993\n\ 372490Sjkh The Regents of the University of California. All rights reserved.\n"; 382490Sjkh#endif /* not lint */ 392490Sjkh 402490Sjkh#ifndef lint 4115944Sachestatic const char sccsid[] = "@(#)strfile.c 8.1 (Berkeley) 5/31/93"; 42114725Sobrien#endif /* not lint */ 4351287Speter#endif 44114725Sobrien#include <sys/cdefs.h> 45114725Sobrien__FBSDID("$FreeBSD: stable/11/usr.bin/fortune/strfile/strfile.c 317438 2017-04-26 14:43:21Z asomers $"); 462490Sjkh 47203926Suqs#include <sys/param.h> 48203926Suqs#include <sys/endian.h> 49203926Suqs#include <ctype.h> 50203926Suqs#include <locale.h> 51203926Suqs#include <stdbool.h> 52203926Suqs#include <stdio.h> 53203926Suqs#include <stdlib.h> 54203926Suqs#include <string.h> 55203926Suqs#include <time.h> 56203926Suqs#include <unistd.h> 572490Sjkh 58203926Suqs#include "strfile.h" 59203926Suqs 602490Sjkh/* 6172089Sasmodai * This program takes a file composed of strings separated by 622490Sjkh * lines starting with two consecutive delimiting character (default 632490Sjkh * character is '%') and creates another file which consists of a table 642490Sjkh * describing the file (structure from "strfile.h"), a table of seek 652490Sjkh * pointers to the start of the strings, and the strings, each terminated 662490Sjkh * by a null byte. Usage: 672490Sjkh * 682490Sjkh * % strfile [-iorsx] [ -cC ] sourcefile [ datafile ] 692490Sjkh * 7051864Sdcs * C - Allow comments marked by a double delimiter at line's beginning 712490Sjkh * c - Change delimiting character from '%' to 'C' 722490Sjkh * s - Silent. Give no summary of data processed at the end of 732490Sjkh * the run. 742490Sjkh * o - order the strings in alphabetic order 758856Srgrimes * i - if ordering, ignore case 762490Sjkh * r - randomize the order of the strings 772490Sjkh * x - set rotated bit 782490Sjkh * 792490Sjkh * Ken Arnold Sept. 7, 1978 -- 802490Sjkh * 812490Sjkh * Added ordering options. 822490Sjkh */ 832490Sjkh 84203926Suqs#define STORING_PTRS (Oflag || Rflag) 85203926Suqs#define CHUNKSIZE 512 862490Sjkh 87203926Suqs#define ALLOC(ptr, sz) do { \ 882490Sjkh if (ptr == NULL) \ 89203926Suqs ptr = malloc(CHUNKSIZE * sizeof(*ptr)); \ 902490Sjkh else if (((sz) + 1) % CHUNKSIZE == 0) \ 91203926Suqs ptr = realloc(ptr, ((sz) + CHUNKSIZE) * sizeof(*ptr)); \ 922490Sjkh if (ptr == NULL) { \ 932490Sjkh fprintf(stderr, "out of space\n"); \ 942490Sjkh exit(1); \ 952490Sjkh } \ 96203926Suqs } while (0) 972490Sjkh 982490Sjkhtypedef struct { 99123243Sdes int first; 100142022Sru off_t pos; 1012490Sjkh} STR; 1022490Sjkh 103123243Sdesstatic char *Infile = NULL, /* input file name */ 104123243Sdes Outfile[MAXPATHLEN] = "", /* output file name */ 105123243Sdes Delimch = '%'; /* delimiting character */ 1062490Sjkh 107203926Suqsstatic int Cflag = false; /* embedded comments */ 108203926Suqsstatic int Sflag = false; /* silent run flag */ 109203926Suqsstatic int Oflag = false; /* ordering flag */ 110203926Suqsstatic int Iflag = false; /* ignore case flag */ 111203926Suqsstatic int Rflag = false; /* randomize order flag */ 112203926Suqsstatic int Xflag = false; /* set rotated bit */ 113142022Srustatic uint32_t Num_pts = 0; /* number of pointers/strings */ 1142490Sjkh 115142022Srustatic off_t *Seekpts; 1162490Sjkh 117123243Sdesstatic FILE *Sort_1, *Sort_2; /* pointers for sorting */ 1182490Sjkh 119123243Sdesstatic STRFILE Tbl; /* statistics table */ 1202490Sjkh 121123243Sdesstatic STR *Firstch; /* first chars of each string */ 1222490Sjkh 123203926Suqsstatic void add_offset(FILE *, off_t); 124203926Suqsstatic int cmp_str(const void *, const void *); 125203926Suqsstatic int stable_collate_range_cmp(int, int); 126203926Suqsstatic void do_order(void); 127203926Suqsstatic void getargs(int, char **); 128203926Suqsstatic void randomize(void); 129203926Suqsstatic void usage(void); 1302490Sjkh 1312490Sjkh/* 1322490Sjkh * main: 1332490Sjkh * Drive the sucker. There are two main modes -- either we store 1342490Sjkh * the seek pointers, if the table is to be sorted or randomized, 1352490Sjkh * or we write the pointer directly to the file, if we are to stay 1362490Sjkh * in file order. If the former, we allocate and re-allocate in 1372490Sjkh * CHUNKSIZE blocks; if the latter, we just write each pointer, 1382490Sjkh * and then seek back to the beginning to write in the table. 1392490Sjkh */ 140201175Sedint 141201175Sedmain(int ac, char *av[]) 1422490Sjkh{ 143203926Suqs char *sp, *nsp, dc; 144203926Suqs FILE *inf, *outf; 145203926Suqs off_t last_off, pos, *p; 146203926Suqs size_t length; 147203926Suqs int first; 148203926Suqs uint32_t cnt; 149203926Suqs STR *fp; 150203926Suqs static char string[257]; 1512490Sjkh 152203926Suqs setlocale(LC_ALL, ""); 15315944Sache 1542490Sjkh getargs(ac, av); /* evalute arguments */ 1552490Sjkh dc = Delimch; 1562490Sjkh if ((inf = fopen(Infile, "r")) == NULL) { 1572490Sjkh perror(Infile); 1582490Sjkh exit(1); 1592490Sjkh } 1602490Sjkh 1612490Sjkh if ((outf = fopen(Outfile, "w")) == NULL) { 1622490Sjkh perror(Outfile); 1632490Sjkh exit(1); 1642490Sjkh } 1652490Sjkh if (!STORING_PTRS) 166203926Suqs fseek(outf, (long)sizeof(Tbl), SEEK_SET); 1672490Sjkh 1682490Sjkh /* 1692490Sjkh * Write the strings onto the file 1702490Sjkh */ 1712490Sjkh 1722490Sjkh Tbl.str_longlen = 0; 173142022Sru Tbl.str_shortlen = 0xffffffff; 1742490Sjkh Tbl.str_delim = dc; 1752490Sjkh Tbl.str_version = VERSION; 1762490Sjkh first = Oflag; 177142022Sru add_offset(outf, ftello(inf)); 1782490Sjkh last_off = 0; 1792490Sjkh do { 1802490Sjkh sp = fgets(string, 256, inf); 18115944Sache if (sp == NULL || (sp[0] == dc && sp[1] == '\n')) { 182142022Sru pos = ftello(inf); 183142022Sru length = (size_t)(pos - last_off) - 184142022Sru (sp != NULL ? strlen(sp) : 0); 1852490Sjkh last_off = pos; 186142022Sru if (length == 0) 1872490Sjkh continue; 1882490Sjkh add_offset(outf, pos); 189142022Sru if ((size_t)Tbl.str_longlen < length) 1902490Sjkh Tbl.str_longlen = length; 191142022Sru if ((size_t)Tbl.str_shortlen > length) 1922490Sjkh Tbl.str_shortlen = length; 1932490Sjkh first = Oflag; 1942490Sjkh } 1952490Sjkh else if (first) { 19615944Sache for (nsp = sp; !isalnum((unsigned char)*nsp); nsp++) 1972490Sjkh continue; 1982490Sjkh ALLOC(Firstch, Num_pts); 1992490Sjkh fp = &Firstch[Num_pts - 1]; 20015944Sache if (Iflag && isupper((unsigned char)*nsp)) 20115944Sache fp->first = tolower((unsigned char)*nsp); 2022490Sjkh else 2032490Sjkh fp->first = *nsp; 2042490Sjkh fp->pos = Seekpts[Num_pts - 1]; 205203926Suqs first = false; 2062490Sjkh } 2072490Sjkh } while (sp != NULL); 2082490Sjkh 2092490Sjkh /* 2102490Sjkh * write the tables in 2112490Sjkh */ 2122490Sjkh 213203926Suqs fclose(inf); 21433633Ssteve Tbl.str_numstr = Num_pts - 1; 2152490Sjkh 21651864Sdcs if (Cflag) 21751864Sdcs Tbl.str_flags |= STR_COMMENTS; 21851864Sdcs 2192490Sjkh if (Oflag) 2202490Sjkh do_order(); 2212490Sjkh else if (Rflag) 2222490Sjkh randomize(); 2232490Sjkh 2242490Sjkh if (Xflag) 2252490Sjkh Tbl.str_flags |= STR_ROTATED; 2262490Sjkh 2272490Sjkh if (!Sflag) { 2282490Sjkh printf("\"%s\" created\n", Outfile); 2292490Sjkh if (Num_pts == 2) 2302490Sjkh puts("There was 1 string"); 2312490Sjkh else 232142022Sru printf("There were %u strings\n", Num_pts - 1); 233142022Sru printf("Longest string: %u byte%s\n", Tbl.str_longlen, 2342490Sjkh Tbl.str_longlen == 1 ? "" : "s"); 235142022Sru printf("Shortest string: %u byte%s\n", Tbl.str_shortlen, 2362490Sjkh Tbl.str_shortlen == 1 ? "" : "s"); 2372490Sjkh } 2382490Sjkh 23915944Sache rewind(outf); 240142022Sru Tbl.str_version = htobe32(Tbl.str_version); 241142022Sru Tbl.str_numstr = htobe32(Tbl.str_numstr); 242142022Sru Tbl.str_longlen = htobe32(Tbl.str_longlen); 243142022Sru Tbl.str_shortlen = htobe32(Tbl.str_shortlen); 244142022Sru Tbl.str_flags = htobe32(Tbl.str_flags); 245203926Suqs fwrite((char *)&Tbl, sizeof(Tbl), 1, outf); 2462490Sjkh if (STORING_PTRS) { 2472490Sjkh for (p = Seekpts, cnt = Num_pts; cnt--; ++p) 248142022Sru *p = htobe64(*p); 249203926Suqs fwrite(Seekpts, sizeof(*Seekpts), (size_t)Num_pts, outf); 2502490Sjkh } 251203926Suqs fclose(outf); 2522490Sjkh exit(0); 2532490Sjkh} 2542490Sjkh 2552490Sjkh/* 2562490Sjkh * This routine evaluates arguments from the command line 2572490Sjkh */ 258201175Sedvoid 259201175Sedgetargs(int argc, char **argv) 2602490Sjkh{ 261203926Suqs int ch; 2622490Sjkh 263176407Sru while ((ch = getopt(argc, argv, "Cc:iorsx")) != -1) 2642490Sjkh switch(ch) { 26551864Sdcs case 'C': /* embedded comments */ 26651864Sdcs Cflag++; 26751864Sdcs break; 2682490Sjkh case 'c': /* new delimiting char */ 2692490Sjkh Delimch = *optarg; 2702490Sjkh if (!isascii(Delimch)) { 2712490Sjkh printf("bad delimiting character: '\\%o\n'", 27215944Sache (unsigned char)Delimch); 2732490Sjkh } 2742490Sjkh break; 2752490Sjkh case 'i': /* ignore case in ordering */ 2762490Sjkh Iflag++; 2772490Sjkh break; 2782490Sjkh case 'o': /* order strings */ 2792490Sjkh Oflag++; 2802490Sjkh break; 2812490Sjkh case 'r': /* randomize pointers */ 2822490Sjkh Rflag++; 2832490Sjkh break; 2842490Sjkh case 's': /* silent */ 2852490Sjkh Sflag++; 2862490Sjkh break; 2872490Sjkh case 'x': /* set the rotated bit */ 2882490Sjkh Xflag++; 2892490Sjkh break; 2902490Sjkh case '?': 2912490Sjkh default: 2922490Sjkh usage(); 2932490Sjkh } 2942490Sjkh argv += optind; 2952490Sjkh 2962490Sjkh if (*argv) { 2972490Sjkh Infile = *argv; 2982490Sjkh if (*++argv) 299203926Suqs strcpy(Outfile, *argv); 3002490Sjkh } 3012490Sjkh if (!Infile) { 3022490Sjkh puts("No input file name"); 3032490Sjkh usage(); 3042490Sjkh } 3052490Sjkh if (*Outfile == '\0') { 306317438Sasomers strlcpy(Outfile, Infile, sizeof(Outfile)); 307317438Sasomers strlcat(Outfile, ".dat", sizeof(Outfile)); 3082490Sjkh } 3092490Sjkh} 3102490Sjkh 311201175Sedvoid 312201175Sedusage(void) 3132490Sjkh{ 314203926Suqs fprintf(stderr, 315141581Sru "strfile [-Ciorsx] [-c char] source_file [output_file]\n"); 3162490Sjkh exit(1); 3172490Sjkh} 3182490Sjkh 3192490Sjkh/* 3202490Sjkh * add_offset: 3212490Sjkh * Add an offset to the list, or write it out, as appropriate. 3222490Sjkh */ 323201175Sedvoid 324201175Sedadd_offset(FILE *fp, off_t off) 3252490Sjkh{ 326142022Sru off_t beoff; 3272490Sjkh 3282490Sjkh if (!STORING_PTRS) { 329142022Sru beoff = htobe64(off); 330203926Suqs fwrite(&beoff, 1, sizeof(beoff), fp); 3312490Sjkh } else { 3322490Sjkh ALLOC(Seekpts, Num_pts + 1); 3332490Sjkh Seekpts[Num_pts] = off; 3342490Sjkh } 3352490Sjkh Num_pts++; 3362490Sjkh} 3372490Sjkh 3382490Sjkh/* 3392490Sjkh * do_order: 3402490Sjkh * Order the strings alphabetically (possibly ignoring case). 3412490Sjkh */ 342201175Sedvoid 343201175Seddo_order(void) 3442490Sjkh{ 345142022Sru uint32_t i; 346203926Suqs off_t *lp; 347203926Suqs STR *fp; 3482490Sjkh 3492490Sjkh Sort_1 = fopen(Infile, "r"); 3502490Sjkh Sort_2 = fopen(Infile, "r"); 351203926Suqs qsort(Firstch, (size_t)Tbl.str_numstr, sizeof(*Firstch), cmp_str); 3522490Sjkh i = Tbl.str_numstr; 3532490Sjkh lp = Seekpts; 3542490Sjkh fp = Firstch; 3552490Sjkh while (i--) 3562490Sjkh *lp++ = fp++->pos; 357203926Suqs fclose(Sort_1); 358203926Suqs fclose(Sort_2); 3592490Sjkh Tbl.str_flags |= STR_ORDERED; 3602490Sjkh} 3612490Sjkh 362201175Sedstatic int 363201175Sedstable_collate_range_cmp(int c1, int c2) 36419288Sache{ 36519288Sache static char s1[2], s2[2]; 36619288Sache int ret; 36719288Sache 36819288Sache s1[0] = c1; 36919288Sache s2[0] = c2; 37019288Sache if ((ret = strcoll(s1, s2)) != 0) 37119288Sache return (ret); 37219288Sache return (c1 - c2); 37319288Sache} 37419288Sache 3752490Sjkh/* 3762490Sjkh * cmp_str: 3772490Sjkh * Compare two strings in the file 3782490Sjkh */ 379201175Sedint 380201175Sedcmp_str(const void *s1, const void *s2) 3812490Sjkh{ 382203926Suqs const STR *p1, *p2; 383203926Suqs int c1, c2, n1, n2, r; 3842490Sjkh 385203926Suqs#define SET_N(nf,ch) (nf = (ch == '\n')) 386203926Suqs#define IS_END(ch,nf) (ch == EOF || (ch == (unsigned char)Delimch && nf)) 3872490Sjkh 388203926Suqs p1 = (const STR *)s1; 389203926Suqs p2 = (const STR *)s2; 390203926Suqs 391203926Suqs c1 = (unsigned char)p1->first; 392203926Suqs c2 = (unsigned char)p2->first; 393118397Sache if ((r = stable_collate_range_cmp(c1, c2)) != 0) 394118397Sache return (r); 3952490Sjkh 396203926Suqs fseeko(Sort_1, p1->pos, SEEK_SET); 397203926Suqs fseeko(Sort_2, p2->pos, SEEK_SET); 3982490Sjkh 399203926Suqs n1 = false; 400203926Suqs n2 = false; 40115944Sache while (!isalnum(c1 = getc(Sort_1)) && c1 != '\0' && c1 != EOF) 4022490Sjkh SET_N(n1, c1); 40315944Sache while (!isalnum(c2 = getc(Sort_2)) && c2 != '\0' && c2 != EOF) 4042490Sjkh SET_N(n2, c2); 4052490Sjkh 4062490Sjkh while (!IS_END(c1, n1) && !IS_END(c2, n2)) { 4072490Sjkh if (Iflag) { 4082490Sjkh if (isupper(c1)) 4092490Sjkh c1 = tolower(c1); 4102490Sjkh if (isupper(c2)) 4112490Sjkh c2 = tolower(c2); 4122490Sjkh } 413118397Sache if ((r = stable_collate_range_cmp(c1, c2)) != 0) 414118397Sache return (r); 4152490Sjkh SET_N(n1, c1); 4162490Sjkh SET_N(n2, c2); 4172490Sjkh c1 = getc(Sort_1); 4182490Sjkh c2 = getc(Sort_2); 4192490Sjkh } 4202490Sjkh if (IS_END(c1, n1)) 4212490Sjkh c1 = 0; 4222490Sjkh if (IS_END(c2, n2)) 4232490Sjkh c2 = 0; 424203926Suqs 425118397Sache return (stable_collate_range_cmp(c1, c2)); 4262490Sjkh} 4272490Sjkh 4282490Sjkh/* 4292490Sjkh * randomize: 4302490Sjkh * Randomize the order of the string table. We must be careful 4312490Sjkh * not to randomize across delimiter boundaries. All 4322490Sjkh * randomization is done within each block. 4332490Sjkh */ 434201175Sedvoid 435201175Sedrandomize(void) 4362490Sjkh{ 437142022Sru uint32_t cnt, i; 438203926Suqs off_t tmp; 439203926Suqs off_t *sp; 4402490Sjkh 4412490Sjkh Tbl.str_flags |= STR_RANDOM; 4422490Sjkh cnt = Tbl.str_numstr; 4432490Sjkh 4442490Sjkh /* 4452490Sjkh * move things around randomly 4462490Sjkh */ 4472490Sjkh 4482490Sjkh for (sp = Seekpts; cnt > 0; cnt--, sp++) { 449181386Sache i = arc4random_uniform(cnt); 4502490Sjkh tmp = sp[0]; 4512490Sjkh sp[0] = sp[i]; 4522490Sjkh sp[i] = tmp; 4532490Sjkh } 4542490Sjkh} 455