Commit | Line | Data |
---|---|---|
a860ef9e KB |
1 | /*- |
2 | * Copyright (c) 1989 The Regents of the University of California. | |
3 | * All rights reserved. | |
4 | * | |
5 | * %sccs.include.proprietary.c% | |
6 | */ | |
7 | ||
8 | #ifndef lint | |
9 | char copyright[] = | |
10 | "@(#) Copyright (c) 1989 The Regents of the University of California.\n\ | |
11 | All rights reserved.\n"; | |
12 | #endif /* not lint */ | |
13 | ||
bef8bbbd | 14 | #ifndef lint |
a860ef9e | 15 | static char sccsid[] = "@(#)ptx.c 4.7 (Berkeley) %G%"; |
bef8bbbd | 16 | #endif /* not lint */ |
f125e2b0 BJ |
17 | |
18 | /* permuted title index | |
19 | ptx [-t] [-i ignore] [-o only] [-w num] [-f] [input] [output] | |
20 | Ptx reads the input file and permutes on words in it. | |
21 | It excludes all words in the ignore file. | |
22 | Alternately it includes words in the only file. | |
435e8dff | 23 | if neither is given it excludes the words in _PATH_EIGN. |
f125e2b0 BJ |
24 | |
25 | The width of the output line can be changed to num | |
26 | characters. If omitted 72 is default unless troff than 100. | |
27 | the -f flag tells the program to fold the output | |
28 | the -t flag says the output is for troff and the | |
29 | output is then wider. | |
30 | ||
f125e2b0 BJ |
31 | */ |
32 | ||
33 | #include <stdio.h> | |
34 | #include <ctype.h> | |
35 | #include <signal.h> | |
435e8dff KB |
36 | #include "pathnames.h" |
37 | ||
f125e2b0 | 38 | #define TILDE 0177 |
f125e2b0 BJ |
39 | #define N 30 |
40 | #define MAX N*BUFSIZ | |
41 | #define LMAX 200 | |
42 | #define MAXT 2048 | |
43 | #define MASK 03777 | |
44 | #define SET 1 | |
45 | ||
46 | #define isabreak(c) (btable[c]) | |
47 | ||
48 | extern char *calloc(), *mktemp(); | |
49 | extern char *getline(); | |
50 | int status; | |
51 | ||
52 | ||
53 | char *hasht[MAXT]; | |
54 | char line[LMAX]; | |
55 | char btable[128]; | |
56 | int ignore; | |
57 | int only; | |
58 | int llen = 72; | |
59 | int gap = 3; | |
60 | int gutter = 3; | |
61 | int mlen = LMAX; | |
62 | int wlen; | |
63 | int rflag; | |
64 | int halflen; | |
65 | char *strtbufp, *endbufp; | |
66 | char *empty = ""; | |
67 | ||
68 | char *infile; | |
69 | FILE *inptr = stdin; | |
70 | ||
71 | char *outfile; | |
72 | FILE *outptr = stdout; | |
73 | ||
435e8dff | 74 | char sortfile[] = _PATH_TMP; /* output of sort program */ |
f125e2b0 BJ |
75 | char nofold[] = {'-', 'd', 't', TILDE, 0}; |
76 | char fold[] = {'-', 'd', 'f', 't', TILDE, 0}; | |
77 | char *sortopt = nofold; | |
78 | FILE *sortptr; | |
79 | ||
80 | char *bfile; /*contains user supplied break chars */ | |
81 | FILE *bptr; | |
82 | ||
83 | main(argc,argv) | |
84 | int argc; | |
85 | char **argv; | |
86 | { | |
87 | register int c; | |
88 | register char *bufp; | |
89 | int pid; | |
90 | char *pend; | |
bc0c28c1 | 91 | extern void onintr(); |
f125e2b0 BJ |
92 | |
93 | char *xfile; | |
94 | FILE *xptr; | |
95 | ||
96 | if(signal(SIGHUP,onintr)==SIG_IGN) | |
97 | signal(SIGHUP,SIG_IGN); | |
98 | if(signal(SIGINT,onintr)==SIG_IGN) | |
99 | signal(SIGINT,SIG_IGN); | |
100 | signal(SIGPIPE,onintr); | |
101 | signal(SIGTERM,onintr); | |
102 | ||
103 | /* argument decoding */ | |
104 | ||
435e8dff | 105 | xfile = _PATH_EIGN; |
f125e2b0 BJ |
106 | argv++; |
107 | while(argc>1 && **argv == '-') { | |
108 | switch (*++*argv){ | |
109 | ||
110 | case 'r': | |
111 | rflag++; | |
112 | break; | |
113 | case 'f': | |
114 | sortopt = fold; | |
115 | break; | |
116 | ||
117 | case 'w': | |
118 | if(argc >= 2) { | |
119 | argc--; | |
120 | wlen++; | |
121 | llen = atoi(*++argv); | |
122 | if(llen == 0) | |
123 | diag("Wrong width:",*argv); | |
124 | if(llen > LMAX) { | |
125 | llen = LMAX; | |
126 | msg("Lines truncated to 200 chars.",empty); | |
127 | } | |
128 | break; | |
129 | } | |
130 | ||
131 | case 't': | |
132 | if(wlen == 0) | |
133 | llen = 100; | |
134 | break; | |
135 | case 'g': | |
136 | if(argc >=2) { | |
137 | argc--; | |
138 | gap = gutter = atoi(*++argv); | |
139 | } | |
140 | break; | |
141 | ||
142 | case 'i': | |
143 | if(only) | |
144 | diag("Only file already given.",empty); | |
145 | if (argc>=2){ | |
146 | argc--; | |
147 | ignore++; | |
148 | xfile = *++argv; | |
149 | } | |
150 | break; | |
151 | ||
152 | case 'o': | |
153 | if(ignore) | |
154 | diag("Ignore file already given",empty); | |
155 | if (argc>=2){ | |
156 | only++; | |
157 | argc--; | |
158 | xfile = *++argv; | |
159 | } | |
160 | break; | |
161 | ||
162 | case 'b': | |
163 | if(argc>=2) { | |
164 | argc--; | |
165 | bfile = *++argv; | |
166 | } | |
167 | break; | |
168 | ||
169 | default: | |
170 | msg("Illegal argument:",*argv); | |
171 | } | |
172 | argc--; | |
173 | argv++; | |
174 | } | |
175 | ||
176 | if(argc>3) | |
177 | diag("Too many filenames",empty); | |
178 | else if(argc==3){ | |
179 | infile = *argv++; | |
180 | outfile = *argv; | |
181 | if((outptr = fopen(outfile,"w")) == NULL) | |
182 | diag("Cannot open output file:",outfile); | |
183 | } else if(argc==2) { | |
184 | infile = *argv; | |
185 | outfile = 0; | |
186 | } | |
187 | ||
188 | ||
189 | /* Default breaks of blank, tab and newline */ | |
190 | btable[' '] = SET; | |
191 | btable['\t'] = SET; | |
192 | btable['\n'] = SET; | |
193 | if(bfile) { | |
194 | if((bptr = fopen(bfile,"r")) == NULL) | |
195 | diag("Cannot open break char file",bfile); | |
196 | ||
197 | while((c = getc(bptr)) != EOF) | |
198 | btable[c] = SET; | |
199 | } | |
200 | ||
201 | /* Allocate space for a buffer. If only or ignore file present | |
202 | read it into buffer. Else read in default ignore file | |
203 | and put resulting words in buffer. | |
204 | */ | |
205 | ||
206 | ||
207 | if((strtbufp = calloc(N,BUFSIZ)) == NULL) | |
208 | diag("Out of memory space",empty); | |
209 | bufp = strtbufp; | |
210 | endbufp = strtbufp+MAX; | |
211 | ||
212 | if((xptr = fopen(xfile,"r")) == NULL) | |
213 | diag("Cannot open file",xfile); | |
214 | ||
215 | while(bufp < endbufp && (c = getc(xptr)) != EOF) { | |
216 | if(isabreak(c)) { | |
217 | if(storeh(hash(strtbufp,bufp),strtbufp)) | |
218 | diag("Too many words",xfile); | |
219 | *bufp++ = '\0'; | |
220 | strtbufp = bufp; | |
221 | } | |
222 | else { | |
223 | *bufp++ = (isupper(c)?tolower(c):c); | |
224 | } | |
225 | } | |
226 | if (bufp >= endbufp) | |
227 | diag("Too many words in file",xfile); | |
228 | endbufp = --bufp; | |
229 | ||
230 | /* open output file for sorting */ | |
231 | ||
83583e30 | 232 | mktemp(sortfile); |
f125e2b0 BJ |
233 | if((sortptr = fopen(sortfile, "w")) == NULL) |
234 | diag("Cannot open output for sorting:",sortfile); | |
235 | ||
236 | /* get a line of data and compare each word for | |
237 | inclusion or exclusion in the sort phase | |
238 | */ | |
239 | ||
240 | if (infile!=0 && (inptr = fopen(infile,"r")) == NULL) | |
241 | diag("Cannot open data: ",infile); | |
242 | while(pend=getline()) | |
243 | cmpline(pend); | |
244 | fclose(sortptr); | |
245 | ||
246 | switch (pid = fork()){ | |
247 | ||
248 | case -1: /* cannot fork */ | |
249 | diag("Cannot fork",empty); | |
250 | ||
251 | case 0: /* child */ | |
435e8dff | 252 | execl(_PATH_SORT, "sort", sortopt, "+0", "-1", "+1", |
f125e2b0 BJ |
253 | sortfile, "-o", sortfile, 0); |
254 | ||
255 | default: /* parent */ | |
256 | while(wait(&status) != pid); | |
257 | } | |
258 | ||
259 | ||
260 | getsort(); | |
83583e30 | 261 | unlink(sortfile); |
62939760 | 262 | exit(0); |
f125e2b0 BJ |
263 | } |
264 | ||
265 | msg(s,arg) | |
266 | char *s; | |
267 | char *arg; | |
268 | { | |
269 | fprintf(stderr,"%s %s\n",s,arg); | |
270 | return; | |
271 | } | |
272 | diag(s,arg) | |
273 | char *s, *arg; | |
274 | { | |
275 | ||
276 | msg(s,arg); | |
277 | exit(1); | |
278 | } | |
279 | ||
280 | ||
281 | char *getline() | |
282 | { | |
283 | ||
284 | register c; | |
285 | register char *linep; | |
286 | char *endlinep; | |
287 | ||
288 | ||
289 | endlinep= line + mlen; | |
290 | linep = line; | |
291 | /* Throw away leading white space */ | |
292 | ||
293 | while(isspace(c=getc(inptr))) | |
294 | ; | |
295 | if(c==EOF) | |
296 | return(0); | |
297 | ungetc(c,inptr); | |
298 | while(( c=getc(inptr)) != EOF) { | |
299 | switch (c) { | |
300 | ||
301 | case '\t': | |
302 | if(linep<endlinep) | |
303 | *linep++ = ' '; | |
304 | break; | |
305 | case '\n': | |
306 | while(isspace(*--linep)); | |
307 | *++linep = '\n'; | |
308 | return(linep); | |
309 | default: | |
310 | if(linep < endlinep) | |
311 | *linep++ = c; | |
312 | } | |
313 | } | |
314 | return(0); | |
315 | } | |
316 | ||
317 | cmpline(pend) | |
318 | char *pend; | |
319 | { | |
320 | ||
321 | char *pstrt, *pchar, *cp; | |
322 | char **hp; | |
323 | int flag; | |
324 | ||
325 | pchar = line; | |
326 | if(rflag) | |
327 | while(pchar<pend&&!isspace(*pchar)) | |
328 | pchar++; | |
329 | while(pchar<pend){ | |
330 | /* eliminate white space */ | |
331 | if(isabreak(*pchar++)) | |
332 | continue; | |
333 | pstrt = --pchar; | |
334 | ||
335 | flag = 1; | |
336 | while(flag){ | |
337 | if(isabreak(*pchar)) { | |
338 | hp = &hasht[hash(pstrt,pchar)]; | |
339 | pchar--; | |
340 | while(cp = *hp++){ | |
341 | if(hp == &hasht[MAXT]) | |
342 | hp = hasht; | |
343 | /* possible match */ | |
344 | if(cmpword(pstrt,pchar,cp)){ | |
345 | /* exact match */ | |
346 | if(!ignore && only) | |
347 | putline(pstrt,pend); | |
348 | flag = 0; | |
349 | break; | |
350 | } | |
351 | } | |
352 | /* no match */ | |
353 | if(flag){ | |
354 | if(ignore || !only) | |
355 | putline(pstrt,pend); | |
356 | flag = 0; | |
357 | } | |
358 | } | |
359 | pchar++; | |
360 | } | |
361 | } | |
362 | } | |
363 | ||
364 | cmpword(cpp,pend,hpp) | |
365 | char *cpp, *pend, *hpp; | |
366 | { | |
367 | char c; | |
368 | ||
369 | while(*hpp != '\0'){ | |
370 | c = *cpp++; | |
371 | if((isupper(c)?tolower(c):c) != *hpp++) | |
372 | return(0); | |
373 | } | |
374 | if(--cpp == pend) return(1); | |
375 | return(0); | |
376 | } | |
377 | ||
378 | putline(strt, end) | |
379 | char *strt, *end; | |
380 | { | |
381 | char *cp; | |
382 | ||
383 | for(cp=strt; cp<end; cp++) | |
384 | putc(*cp, sortptr); | |
385 | /* Add extra blank before TILDE to sort correctly | |
386 | with -fd option */ | |
387 | putc(' ',sortptr); | |
388 | putc(TILDE,sortptr); | |
389 | for (cp=line; cp<strt; cp++) | |
390 | putc(*cp,sortptr); | |
391 | putc('\n',sortptr); | |
392 | } | |
393 | ||
394 | getsort() | |
395 | { | |
396 | register c; | |
397 | register char *tilde, *linep, *ref; | |
398 | char *p1a,*p1b,*p2a,*p2b,*p3a,*p3b,*p4a,*p4b; | |
399 | int w; | |
400 | char *rtrim(), *ltrim(); | |
401 | ||
402 | if((sortptr = fopen(sortfile,"r")) == NULL) | |
403 | diag("Cannot open sorted data:",sortfile); | |
404 | ||
405 | halflen = (llen-gutter)/2; | |
406 | linep = line; | |
407 | while((c = getc(sortptr)) != EOF) { | |
408 | switch(c) { | |
409 | ||
410 | case TILDE: | |
411 | tilde = linep; | |
412 | break; | |
413 | ||
414 | case '\n': | |
415 | while(isspace(linep[-1])) | |
416 | linep--; | |
417 | ref = tilde; | |
418 | if(rflag) { | |
419 | while(ref<linep&&!isspace(*ref)) | |
420 | ref++; | |
421 | *ref++ = 0; | |
422 | } | |
423 | /* the -1 is an overly conservative test to leave | |
424 | space for the / that signifies truncation*/ | |
425 | p3b = rtrim(p3a=line,tilde,halflen-1); | |
426 | if(p3b-p3a>halflen-1) | |
427 | p3b = p3a+halflen-1; | |
428 | p2a = ltrim(ref,p2b=linep,halflen-1); | |
429 | if(p2b-p2a>halflen-1) | |
430 | p2a = p2b-halflen-1; | |
431 | p1b = rtrim(p1a=p3b+(isspace(p3b[0])!=0),tilde, | |
432 | w=halflen-(p2b-p2a)-gap); | |
433 | if(p1b-p1a>w) | |
434 | p1b = p1a; | |
435 | p4a = ltrim(ref,p4b=p2a-(isspace(p2a[-1])!=0), | |
436 | w=halflen-(p3b-p3a)-gap); | |
437 | if(p4b-p4a>w) | |
438 | p4a = p4b; | |
439 | fprintf(outptr,".xx \""); | |
440 | putout(p1a,p1b); | |
441 | /* tilde-1 to account for extra space before TILDE */ | |
442 | if(p1b!=(tilde-1) && p1a!=p1b) | |
443 | fprintf(outptr,"/"); | |
444 | fprintf(outptr,"\" \""); | |
445 | if(p4a==p4b && p2a!=ref && p2a!=p2b) | |
446 | fprintf(outptr,"/"); | |
447 | putout(p2a,p2b); | |
448 | fprintf(outptr,"\" \""); | |
449 | putout(p3a,p3b); | |
450 | /* ++p3b to account for extra blank after TILDE */ | |
451 | /* ++p3b to account for extra space before TILDE */ | |
452 | if(p1a==p1b && ++p3b!=tilde) | |
453 | fprintf(outptr,"/"); | |
454 | fprintf(outptr,"\" \""); | |
455 | if(p1a==p1b && p4a!=ref && p4a!=p4b) | |
456 | fprintf(outptr,"/"); | |
457 | putout(p4a,p4b); | |
458 | if(rflag) | |
459 | fprintf(outptr,"\" %s\n",tilde); | |
460 | else | |
461 | fprintf(outptr,"\"\n"); | |
462 | linep = line; | |
463 | break; | |
464 | ||
465 | case '"': | |
466 | /* put double " for " */ | |
467 | *linep++ = c; | |
468 | default: | |
469 | *linep++ = c; | |
470 | } | |
471 | } | |
472 | } | |
473 | ||
474 | char *rtrim(a,c,d) | |
475 | char *a,*c; | |
476 | { | |
477 | char *b,*x; | |
478 | b = c; | |
479 | for(x=a+1; x<=c&&x-a<=d; x++) | |
480 | if((x==c||isspace(x[0]))&&!isspace(x[-1])) | |
481 | b = x; | |
482 | if(b<c&&!isspace(b[0])) | |
483 | b++; | |
484 | return(b); | |
485 | } | |
486 | ||
487 | char *ltrim(c,b,d) | |
488 | char *c,*b; | |
489 | { | |
490 | char *a,*x; | |
491 | a = c; | |
492 | for(x=b-1; x>=c&&b-x<=d; x--) | |
493 | if(!isspace(x[0])&&(x==c||isspace(x[-1]))) | |
494 | a = x; | |
495 | if(a>c&&!isspace(a[-1])) | |
496 | a--; | |
497 | return(a); | |
498 | } | |
499 | ||
500 | putout(strt,end) | |
501 | char *strt, *end; | |
502 | { | |
503 | char *cp; | |
504 | ||
505 | cp = strt; | |
506 | ||
507 | for(cp=strt; cp<end; cp++) { | |
508 | putc(*cp,outptr); | |
509 | } | |
510 | } | |
511 | ||
bc0c28c1 | 512 | void |
f125e2b0 BJ |
513 | onintr() |
514 | { | |
515 | ||
83583e30 | 516 | unlink(sortfile); |
f125e2b0 BJ |
517 | exit(1); |
518 | } | |
519 | ||
520 | hash(strtp,endp) | |
521 | char *strtp, *endp; | |
522 | { | |
523 | char *cp, c; | |
524 | int i, j, k; | |
525 | ||
526 | /* Return zero hash number for single letter words */ | |
527 | if((endp - strtp) == 1) | |
528 | return(0); | |
529 | ||
530 | cp = strtp; | |
531 | c = *cp++; | |
532 | i = (isupper(c)?tolower(c):c); | |
533 | c = *cp; | |
534 | j = (isupper(c)?tolower(c):c); | |
535 | i = i*j; | |
536 | cp = --endp; | |
537 | c = *cp--; | |
538 | k = (isupper(c)?tolower(c):c); | |
539 | c = *cp; | |
540 | j = (isupper(c)?tolower(c):c); | |
541 | j = k*j; | |
542 | ||
543 | k = (i ^ (j>>2)) & MASK; | |
544 | return(k); | |
545 | } | |
546 | ||
547 | storeh(num,strtp) | |
548 | int num; | |
549 | char *strtp; | |
550 | { | |
551 | int i; | |
552 | ||
553 | for(i=num; i<MAXT; i++) { | |
554 | if(hasht[i] == 0) { | |
555 | hasht[i] = strtp; | |
556 | return(0); | |
557 | } | |
558 | } | |
559 | for(i=0; i<num; i++) { | |
560 | if(hasht[i] == 0) { | |
561 | hasht[i] = strtp; | |
562 | return(0); | |
563 | } | |
564 | } | |
565 | return(1); | |
566 | } |