usr/src/cmd/refer/sortbib.c

   1 /*
   2  * Copyright 2005 Sun Microsystems, Inc.  All rights reserved.
   3  * Use is subject to license terms.
   4  */
   5
   6 /*      Copyright (c) 1984, 1986, 1987, 1988, 1989 AT&T */
   7 /*        All Rights Reserved   */
   8
   9 /*
  10  * Copyright (c) 1980 Regents of the University of California.
  11  * All rights reserved. The Berkeley software License Agreement
  12  * specifies the terms and conditions for redistribution.
  13  */
  14
  15 #pragma ident   "%Z%%M% %I%     %E% SMI"
  16
  17 #include <locale.h>
  18 #include <stdio.h>
  19 #include <signal.h>
  20 #include <stdlib.h>
  21
  22 #define BUF BUFSIZ
  23 #define MXFILES 16
  24
  25 char tempfile[32];              /* temporary file for sorting keys */
  26 int tmpfd = -1;
  27 char *keystr = "AD";            /* default sorting on author and date */
  28 int multauth = 0;               /* by default sort on senior author only */
  29 int oneauth;                    /* has there been author in the record? */
  30
  31 static int article(char *);
  32 static void deliver(FILE *[], FILE *);
  33 static int endcomma(char *);
  34 static void error(char *);
  35 static void eval(char []);
  36 static void parse(char [], char fld[][BUF]);
  37 static void sortbib(FILE *, FILE *, int);
  38 static void onintr(void);
  39
  40 /* sortbib: sort bibliographic database in place */
  41 int
  42 main(int argc, char *argv[])
  43 {
  44         FILE *fp[MXFILES], *tfp;
  45         int i;
  46
  47         (void) setlocale(LC_ALL, "");
  48
  49 #if !defined(TEXT_DOMAIN)
  50 #define TEXT_DOMAIN "SYS_TEST"
  51 #endif
  52         (void) textdomain(TEXT_DOMAIN);
  53
  54         if (argc == 1) {                /* can't use stdin for seeking anyway */
  55                 puts(gettext("Usage:  sortbib [-sKEYS] database [...]\n\
  56 \t-s: sort by fields in KEYS (default is AD)"));
  57                 exit(1);
  58         }
  59         if (argc > 2 && argv[1][0] == '-' && argv[1][1] == 's') {
  60                 /* if a key is specified use it, otherwise use default key */
  61                 if (argv[1][2] != '\0')
  62                         keystr = argv[1] + 2;
  63                 eval(keystr);           /* evaluate A+ for multiple authors */
  64                 argv++; argc--;
  65         }
  66         if (argc > MXFILES+1) { /* too many open file streams */
  67                 fprintf(stderr,
  68                 gettext("sortbib: More than %d databases specified\n"),
  69                     MXFILES);
  70                 exit(1);
  71         }
  72         for (i = 1; i < argc; i++)              /* open files in arg list */
  73                 if ((fp[i-1] = fopen(argv[i], "r")) == NULL)
  74                         error(argv[i]);
  75         strcpy(tempfile, "/tmp/SbibXXXXXX");    /* tempfile for sorting keys */
  76         if ((tmpfd = mkstemp(tempfile)) == -1)
  77                 error(tempfile);
  78
  79         (void) close(tmpfd);
  80         if (signal(SIGINT, SIG_IGN) != SIG_IGN) /* remove if interrupted */
  81                 signal(SIGINT, (void(*)())onintr);
  82         if ((tfp = fopen(tempfile, "w")) == NULL) {
  83                 (void) unlink(tempfile);
  84                 error(tempfile);
  85         }
  86         for (i = 0; i < argc-1; i++)            /* read keys from bib files */
  87                 sortbib(fp[i], tfp, i);
  88         fclose(tfp);
  89         deliver(fp, tfp);       /* do disk seeks and read from biblio files */
  90         (void) unlink(tempfile);
  91         return (0);
  92 }
  93
  94 int rsmode = 0;         /* record separator: 1 = null line, 2 = bracket */
  95
  96 /* read records, prepare list for sorting */
  97 static void
  98 sortbib(FILE *fp, FILE *tfp, int i)
  99 {
 100         long offset, lastoffset = 0, ftell();   /* byte offsets in file */
 101         int length, newrec, recno = 0;          /* reclen, new rec'd?, number */
 102         char line[BUF], fld[4][BUF];            /* one line, the sort fields */
 103
 104         /* measure byte offset, then get new line */
 105         while (offset = ftell(fp), fgets(line, BUF, fp)) {
 106                 if (recno == 0)         /* accept record w/o initial newline */
 107                         newrec = 1;
 108                 if (line[0] == '\n') {  /* accept null line record separator */
 109                         if (!rsmode)
 110                                 rsmode = 1;     /* null line mode */
 111                         if (rsmode == 1)
 112                                 newrec = 1;
 113                 }
 114                 if (line[0] == '.' && line[1] == '[') { /* also accept .[ .] */
 115                         if (!rsmode)
 116                                 rsmode = 2;     /* bracket pair mode */
 117                         if (rsmode == 2)
 118                                 newrec = 1;
 119                 }
 120                 if (newrec) {           /* by whatever means above */
 121                         newrec = 0;
 122                         length = offset - lastoffset;   /* measure rec len */
 123                         if (length > BUF*8) {
 124                                 fprintf(stderr,
 125                                 gettext("sortbib: record %d longer than %d "
 126                                     "(%d)\n"), recno, BUF*8, length);
 127                                 (void) unlink(tempfile);
 128                                 exit(1);
 129                         }
 130                         if (recno++) {                  /* info for sorting */
 131                                 fprintf(tfp, "%d %d %d : %s %s %s %s\n",
 132                                     i, lastoffset, length,
 133                                     fld[0], fld[1], fld[2], fld[3]);
 134                                 if (ferror(tfp)) {
 135                                         (void) unlink(tempfile);
 136                                         error(tempfile);
 137                                 }
 138                         }
 139                         *fld[0] = *fld[1] = *fld[2] = *fld[3] = NULL;
 140                         oneauth = 0;            /* reset number of authors */
 141                         lastoffset = offset;    /* save for next time */
 142                 }
 143                 if (line[0] == '%')     /* parse out fields to be sorted */
 144                         parse(line, fld);
 145         }
 146         offset = ftell(fp);             /* measure byte offset at EOF */
 147         length = offset - lastoffset;   /* measure final record length */
 148         if (length > BUF*8) {
 149                 fprintf(stderr,
 150                     gettext("sortbib: record %d longer than %d (%d)\n"),
 151                     recno, BUF*8, length);
 152                 (void) unlink(tempfile);
 153                 exit(1);
 154         }
 155         if (line[0] != '\n') {          /* ignore null line just before EOF */
 156                 fprintf(tfp, "%d %d %d : %s %s %s %s\n",
 157                     i, lastoffset, length, fld[0], fld[1], fld[2], fld[3]);
 158                 if (ferror(tfp)) {
 159                         (void) unlink(tempfile);
 160                         error(tempfile);        /* disk error in /tmp */
 161                 }
 162         }
 163 }
 164
 165 /* deliver sorted entries out of database(s) */
 166 static void
 167 deliver(FILE *fp[], FILE *tfp)
 168 {
 169         char str[BUF], buff[BUF*8];     /* for tempfile & databases */
 170         char cmd[80];                   /* for using system sort command */
 171         long int offset;
 172         int i, length;
 173
 174         /* when sorting, ignore case distinctions; tab char is ':' */
 175         sprintf(cmd, "sort +4f +0n +1n %s -o %s", tempfile, tempfile);
 176         if (system(cmd) == 127) {
 177                 (void) unlink(tempfile);
 178                 error("sortbib");
 179         }
 180         tfp = fopen(tempfile, "r");
 181         while (fgets(str, sizeof (str), tfp)) {
 182                 /* get file pointer, record offset, and length */
 183                 if (sscanf(str, "%d %d %d :", &i, &offset, &length) != 3)
 184                         error(gettext("sortbib: sorting error"));
 185                 /* seek to proper disk location in proper file */
 186                 if (fseek(fp[i], offset, 0) == -1) {
 187                         (void) unlink(tempfile);
 188                         error("sortbib");
 189                 }
 190                 /* read exactly one record from bibliography */
 191                 if (fread(buff, sizeof (*buff), length, fp[i]) == 0) {
 192                         (void) unlink(tempfile);
 193                         error("sortbib");
 194                 }
 195                 /* add newline between unseparated records */
 196                 if (buff[0] != '\n' && rsmode == 1)
 197                         putchar('\n');
 198                 /* write record buffer to standard output */
 199                 if (fwrite(buff, sizeof (*buff), length, stdout) == 0) {
 200                         (void) unlink(tempfile);
 201                         error("sortbib");
 202                 }
 203         }
 204 }
 205
 206 /* get fields out of line, prepare for sorting */
 207 static void
 208 parse(char line[], char fld[][BUF])
 209 {
 210         char wd[8][BUF/4], *strcat();
 211         int n, i, j;
 212
 213         for (i = 0; i < 8; i++)         /* zap out old strings */
 214                 *wd[i] = NULL;
 215         n = sscanf(line, "%s %s %s %s %s %s %s %s",
 216             wd[0], wd[1], wd[2], wd[3], wd[4], wd[5], wd[6], wd[7]);
 217         for (i = 0; i < 4; i++) {
 218                 if (wd[0][1] == keystr[i]) {
 219                         if (wd[0][1] == 'A') {
 220                                 if (oneauth && !multauth)       /* no repeat */
 221                                         break;
 222                                 else if (oneauth)               /* mult auths */
 223                                         strcat(fld[i], "~~");
 224                                 if (!endcomma(wd[n-2]))         /* surname */
 225                                         strcat(fld[i], wd[n-1]);
 226                                 else {                          /* jr. or ed. */
 227                                         strcat(fld[i], wd[n-2]);
 228                                         n--;
 229                                 }
 230                                 strcat(fld[i], " ");
 231                                 for (j = 1; j < n-1; j++)
 232                                         strcat(fld[i], wd[j]);
 233                                 oneauth = 1;
 234                         } else if (wd[0][1] == 'D') {
 235                                 strcat(fld[i], wd[n-1]);        /* year */
 236                                 if (n > 2)
 237                                         strcat(fld[i], wd[1]);  /* month */
 238                         } else if (wd[0][1] == 'T' || wd[0][1] == 'J') {
 239                                 j = 1;
 240                                 if (article(wd[1]))     /* skip article */
 241                                         j++;
 242                                 for (; j < n; j++)
 243                                         strcat(fld[i], wd[j]);
 244                         } else  /* any other field */
 245                                 for (j = 1; j < n; j++)
 246                                         strcat(fld[i], wd[j]);
 247                 }
 248                 /* %Q quorporate or queer author - unreversed %A */
 249                 else if (wd[0][1] == 'Q' && keystr[i] == 'A')
 250                         for (j = 1; j < n; j++)
 251                                 strcat(fld[i], wd[j]);
 252         }
 253 }
 254
 255 /* see if string contains an article */
 256 static int
 257 article(char *str)
 258 {
 259         if (strcmp("The", str) == 0)    /* English */
 260                 return (1);
 261         if (strcmp("A", str) == 0)
 262                 return (1);
 263         if (strcmp("An", str) == 0)
 264                 return (1);
 265         if (strcmp("Le", str) == 0)     /* French */
 266                 return (1);
 267         if (strcmp("La", str) == 0)
 268                 return (1);
 269         if (strcmp("Der", str) == 0)    /* German */
 270                 return (1);
 271         if (strcmp("Die", str) == 0)
 272                 return (1);
 273         if (strcmp("Das", str) == 0)
 274                 return (1);
 275         if (strcmp("El", str) == 0)     /* Spanish */
 276                 return (1);
 277         if (strcmp("Den", str) == 0)    /* Scandinavian */
 278                 return (1);
 279         return (0);
 280 }
 281
 282 /* evaluate key string for A+ marking */
 283 static void
 284 eval(char keystr[])
 285 {
 286         int i, j;
 287
 288         for (i = 0, j = 0; keystr[i]; i++, j++) {
 289                 if (keystr[i] == '+') {
 290                         multauth = 1;
 291                         i++;
 292                 }
 293                 if (keystr[i] == NULL)
 294                         break;
 295                 keystr[j] = keystr[i];
 296         }
 297         keystr[j] = NULL;
 298 }
 299
 300 /* exit in case of various system errors */
 301 static void
 302 error(char *s)
 303 {
 304         perror(s);
 305         exit(1);
 306 }
 307
 308 /* remove tempfile in case of interrupt */
 309 static void
 310 onintr(void)
 311 {
 312         fprintf(stderr, gettext("\nInterrupt\n"));
 313         unlink(tempfile);
 314         exit(1);
 315 }
 316
 317 static int
 318 endcomma(char *str)
 319 {
 320         int n;
 321
 322         n = strlen(str) - 1;
 323         if (str[n] == ',') {
 324                 str[n] = NULL;
 325                 return (1);
 326         }
 327         return (0);
 328 }