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