![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to sdiction.c CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [Research Unix] / researchv10dc / vol2 / index / tools|
Return to sdiction.c CVS log | Up to [Research Unix] / researchv10dc / vol2 / index / tools |
1.1 root 1: /*
2: * diction -- print all sentences containing one of default phrases
3: *
4: * status returns:
5: * 0 - ok, and some matches
6: * 1 - ok, but no matches
7: * 2 - some error
8: */
9:
10: #include <stdio.h>
11: #include <ctype.h>
12:
13: #define MAXSIZ 6500
14: #define QSIZE 1000
15: #define DICT ""
16: int linemsg;
17: long olcount;
18: long lcount;
19: struct words {
20: char inp;
21: char out;
22: struct words *nst;
23: struct words *link;
24: struct words *fail;
25: } w[MAXSIZ], *smax, *q;
26:
27: char table[128] = {
28: 0, 0, 0, 0, 0, 0, 0, 0,
29: 0, 0, ' ', 0, 0, 0, 0, 0,
30: 0, 0, 0, 0, 0, 0, 0, 0,
31: 0, 0, 0, 0, 0, 0, 0, 0,
32: ' ', '.', ' ', ' ', '$', '%', '&', '\'',
33: '(', ')', '*', '+', ' ', '-', '.', '/',
34: '0', '1', '2', '3', '4', '5', '6', '7',
35: '8', '9', ':', ' ', ' ', '=', ' ', '?',
36: ' ', 'a', 'b', 'c', 'd', 'e', 'f', 'g',
37: 'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o',
38: 'p', 'q', 'r', 's', 't', 'u', 'v', 'w',
39: 'x', 'y', 'z', ' ', '\\', ' ', '^', '_',
40: ' ', 'a', 'b', 'c', 'd', 'e', 'f', 'g',
41: 'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o',
42: 'p', 'q', 'r', 's', 't', 'u', 'v', 'w',
43: 'x', 'y', 'z', '{', ' ', '}', ' ', ' '
44: };
45: int caps = 0;
46: int lineno = 1;
47: int fflag= 0;
48: int nflag = 0; /*use default file*/
49: char *filename;
50: int mflg = 0; /*don't catch output*/
51: int nfile;
52: int nsucc;
53: long nsent = 0;
54: long nhits = 0;
55: char *nlp;
56: char *begp, *endp;
57: int beg, last;
58: char *myst;
59: int myct = 0;
60: int oct = 0;
61: FILE *wordf;
62: FILE *mine;
63: FILE *fl;
64: char *listn;
65: int list = 0;
66: char *argptr;
67: long tl = 0;
68: long th = 0;
69:
70: main(argc, argv)
71: char *argv[];
72: {
73: int sv;
74: char cc;
75: while (--argc > 0 && (++argv)[0][0]=='-')
76: switch (argv[0][1]) {
77:
78: case 'f':
79: fflag++;
80: filename = (++argv)[0];
81: argc--;
82: continue;
83:
84: case 'n':
85: nflag = 0;
86: continue;
87: case 'd':
88: mflg=0;
89: continue;
90: case 'c':
91: caps++;
92: continue;
93: case 'l':
94: lineno++;
95: continue;
96: case 'A': /* for acro */
97: for(cc='A';cc<='Z';cc++)
98: table[cc] = cc;
99: continue;
100: case 'o': /*to put hits to file*/
101: listn = (++argv)[0];
102: argc--;
103: list++;
104: if((fl=fopen(listn,"w"))== NULL){
105: fprintf(stderr,"diction: can't open file %s\n",
106: listn);
107: exit(2);
108: }
109: continue;
110: default:
111: fprintf(stderr, "diction: unknown flag\n");
112: continue;
113: }
114: out:
115: if(nflag){
116: wordf = fopen(DICT,"r");
117: if(wordf == NULL){
118: fprintf(stderr,"diction: can't open default dictionary\n");
119: exit(2);
120: }
121: }
122: else {
123: wordf = fopen(filename,"r");
124: if(wordf == NULL){
125: fprintf(stderr,"diction: can't open %s\n",filename);
126: exit(2);
127: }
128: }
129:
130: #ifdef CATCH
131: if(fopen(CATCH,"r") != NULL){
132: if((mine=fopen(CATCH,"a"))==NULL)mflg=0;
133: else mflg = 1;
134: }
135: #else
136: mflg = 0;
137: #endif
138: #ifdef MACS
139: if(caps){
140: printf(".so ");
141: printf(MACS);
142: printf("\n");
143: }
144: #endif
145: cgotofn();
146: cfail();
147: nfile = argc;
148: if (argc<=0) {
149: execute((char *)NULL);
150: }
151: else while (--argc >= 0) {
152: execute(*argv);
153: if(lineno){
154: printf("file %s: number of lines %ld number of phrases found %ld\n",
155: *argv, lcount-1, nhits);
156: tl += lcount-1;
157: th += nhits;
158: sv = lcount-1;
159: lcount = nhits = 0;
160: }
161: argv++;
162: }
163: fprintf(stderr,"lcount %ld\n",tl);
164: if(mflg)fprintf(mine,"number of sentences %ld %ld number of hits %ld %ld\n",nsent,tl,nhits,th);
165: if(!caps&& !lineno)printf("number of sentences %ld number of phrases found %ld\n",nsent,nhits);
166: else if(tl != sv)
167: if(!caps)printf("totals: number of lines %ld number of phrases found %ld\n",tl,th);
168: exit(nsucc == 0);
169: }
170:
171: execute(file)
172: char *file;
173: {
174: register char *p;
175: register struct words *c;
176: register ccount;
177: int count1;
178: char *beg1;
179: struct words *savc;
180: char *savp, *seen;
181: int savct;
182: int scr;
183: char buf[1024];
184: int f;
185: int hit;
186: last = 0;
187: if (file) {
188: if ((f = open(file, 0)) < 0) {
189: fprintf(stderr, "diction: can't open %s\n", file);
190: exit(2);
191: }
192: }
193: else f = 0;
194: lcount = olcount = 1;
195: linemsg = 1;
196: ccount = 0;
197: count1 = -1;
198: p = buf;
199: nlp = p;
200: c = w;
201: oct = hit = 0;
202: savc = (struct words *) 0;
203: savp = (char *) 0;
204: for (;;) {
205: if(--ccount <= 0) {
206: if (p == &buf[1024]) p = buf;
207: if (p > &buf[512]) {
208: if ((ccount = read(f, p, &buf[1024] - p)) <= 0) break;
209: }
210: else if ((ccount = read(f, p, 512)) <= 0) break;
211: if(caps && (count1 > 0))
212: fwrite(beg1,sizeof(*beg1),count1,stdout);
213: count1 = ccount;
214: beg1 = p;
215: }
216: if(p == &buf[1024])p=buf;
217: nstate:
218: if (c->inp == table[*p]) {
219: c = c->nst;
220: }
221: else if (c->link != 0) {
222: c = c->link;
223: goto nstate;
224: }
225: else {
226: if(savp != 0){
227: c=savc;
228: p=savp;
229: if(ccount > savct)ccount += savct;
230: else ccount = savct;
231: savc = (struct words *) 0;
232: savp = (char *) 0;
233: goto hadone;
234: }
235: c = c->fail;
236: if (c==0) {
237: c = w;
238: istate:
239: if (c->inp == table[*p]) {
240: c = c->nst;
241: }
242: else if (c->link != 0) {
243: c = c->link;
244: goto istate;
245: }
246: }
247: else goto nstate;
248: }
249: if(c->out){
250: if((c->inp == table[*(p+1)]) && (c->nst != 0)){
251: savp=p;
252: savc=c;
253: savct=ccount;
254: goto cont;
255: }
256: else if(c->link != 0){
257: savc=c;
258: while((savc=savc->link)!= 0){
259: if(savc->inp == table[*(p+1)]){
260: savp=p;
261: savc=c;
262: savct=ccount;
263: goto cont;
264: }
265: }
266: }
267: hadone:
268: savc = (struct words *) 0;
269: savp = (char *) 0;
270: if(c->out == (char)(0377)){
271: c=w;
272: goto nstate;
273: }
274: begp = p - (c->out);
275: if(begp < &buf[0])begp = &buf[1024] - (&buf[0]-begp);
276: endp=p;
277: if(mflg){
278: if(begp-20 < &buf[0]){
279: myst = &buf[1024]-20;
280: if(nlp < &buf[512])myst=nlp;
281: }
282: else myst = begp-20;
283: if(myst < nlp)myst = nlp;
284: beg = 0;
285: }
286: hit = 1;
287: nhits++;
288: if(*p == '\n'){lcount++;
289: seen = p;
290: }
291: if (table[*p++] == '.') {
292: linemsg = 1;
293: if (--ccount <= 0) {
294: if (p == &buf[1024]) p = buf;
295: if (p > &buf[512]) {
296: if ((ccount = read(f, p, &buf[1024] - p)) <= 0) break;
297: }
298: else if ((ccount = read(f, p, 512)) <= 0) break;
299: if(caps && (count1 > 0))
300: fwrite(beg1,sizeof(*beg1),count1,stdout);
301: count1=ccount;
302: beg1=p;
303: }
304: }
305: succeed: nsucc = 1;
306: {
307: if (p <= nlp) {
308: outc(&buf[1024],file);
309: nlp = buf;
310: }
311: outc(p,file);
312: }
313: if(mflg)last=1;
314: nomatch:
315: nlp = p;
316: c = w;
317: begp = endp = 0;
318: continue;
319: }
320: cont:
321: if(*p == '\n' && p != seen){lcount++;
322: seen = p;
323: }
324: if (table[*p++] == '.'){
325: if(hit){
326: if(p <= nlp){
327: outc(&buf[1024],file);
328: nlp = buf;
329: }
330: outc(p,file);
331: if(!caps)printf("\n\n");
332: if(mflg && last){putc('\n',mine);myct = 0;}
333: }
334: linemsg = 1;
335: if(*p == '\n')olcount = lcount+1;
336: else
337: olcount=lcount;
338: last = 0;
339: hit = 0;
340: oct = 0;
341: nlp = p;
342: c = w;
343: begp = endp = 0;
344: nsent++;
345: }
346: }
347: if(caps && (count1 > 0))
348: fwrite(beg1,sizeof(*beg1),count1,stdout);
349: close(f);
350: }
351:
352: getargc()
353: {
354: register c;
355: if (wordf){
356: if((c=getc(wordf))==EOF){
357: fclose(wordf);
358: if(nflag && fflag){
359: nflag=0;
360: wordf=fopen(filename,"r");
361: if(wordf == NULL){
362: fprintf(stderr,"diction can't open %s\n",filename);
363: exit(2);
364: }
365: return(getc(wordf));
366: }
367: else return(EOF);
368: }
369: else return(c);
370: }
371: if ((c = *argptr++) == '\0')
372: return(EOF);
373: return(c);
374: }
375:
376: cgotofn() {
377: register c;
378: register struct words *s;
379: register ct;
380: int neg;
381:
382: s = smax = w;
383: neg = ct = 0;
384: nword: for(;;) {
385: c = getargc();
386: if(c == '~'){
387: neg++;
388: c = getargc();
389: }
390: if (c==EOF)
391: return;
392: if (c == '\n') {
393: if(neg)s->out = 0377;
394: else s->out = ct-1;
395: neg = ct = 0;
396: s = w;
397: } else {
398: loop: if (s->inp == c) {
399: s = s->nst;
400: ct++;
401: continue;
402: }
403: if (s->inp == 0) goto enter;
404: if (s->link == 0) {
405: if (smax >= &w[MAXSIZ - 1]) overflo("w1");
406: s->link = ++smax;
407: s = smax;
408: goto enter;
409: }
410: s = s->link;
411: goto loop;
412: }
413: }
414:
415: enter:
416: do {
417: s->inp = c;
418: ct++;
419: if (smax >= &w[MAXSIZ - 1]) overflo("w2");
420: s->nst = ++smax;
421: s = smax;
422: } while ((c = getargc()) != '\n' && c!=EOF);
423: if(neg)smax->out = 0377;
424: else smax->out = ct-1;
425: neg = ct = 0;
426: s = w;
427: if (c != EOF)
428: goto nword;
429: }
430:
431: overflo(s)
432: char *s;
433: {
434: fprintf(stderr, "wordlist too large %s\n",s);
435: exit(2);
436: }
437: cfail() {
438: struct words *queue[QSIZE];
439: struct words **front, **rear;
440: struct words *state;
441: int bstart;
442: register char c;
443: register struct words *s;
444: s = w;
445: front = rear = queue;
446: init: if ((s->inp) != 0) {
447: *rear++ = s->nst;
448: if (rear >= &queue[QSIZE - 1]) overflo("queue1");
449: }
450: if ((s = s->link) != 0) {
451: goto init;
452: }
453:
454: while (rear!=front) {
455: s = *front;
456: if (front == &queue[QSIZE-1])
457: front = queue;
458: else front++;
459: cloop: if ((c = s->inp) != 0) {
460: bstart=0;
461: *rear = (q = s->nst);
462: if (front < rear)
463: if (rear >= &queue[QSIZE-1])
464: if (front == queue) overflo("queue2");
465: else rear = queue;
466: else rear++;
467: else
468: if (++rear == front) overflo("queue3");
469: state = s->fail;
470: floop: if (state == 0){ state = w;bstart=1;}
471: if (state->inp == c) {
472: qloop: q->fail = state->nst;
473: if ((state->nst)->out != 0 && q->out == 0) q->out = (state->nst)->out;
474: if((q=q->link) != 0)goto qloop;
475: }
476: else if((state->link) != 0){
477: state = state->link;
478: goto floop;
479: }
480: else if((state = state->fail) != 0)
481: goto floop;
482: else if(bstart==0){state=0; goto floop;}
483: }
484: if ((s = s->link) != 0)
485: goto cloop;
486: }
487: /* for(s=w;s<=smax;s++)
488: printf("s %d ch %c out %d nst %d link %d fail %d\n",s,
489: s->inp,s->out,s->nst,s->link,s->fail);
490: */
491: }
492: outc(addr,file)
493: char *addr;
494: char *file;
495: {
496: static inside = 0;
497:
498: if(!caps && lineno && linemsg){
499: if(list)fprintf(fl,"%ld ",olcount);
500: printf("beginning line %ld",olcount);
501: if(file != (char *)NULL)printf(" %s\n",file);
502: else printf("\n");
503: linemsg = 0;
504: }
505: while(nlp < addr){
506: if(!caps && oct > 60 && table[*nlp] == ' ' && nlp != begp && nlp != endp){
507: oct=0;
508: putchar('\n');
509: }
510: if(nlp == begp){
511: if(caps)inside++;
512: else {
513: if(list)inside++;
514: if( oct >45){putchar('\n');
515: oct=0;
516: }
517: if( oct==0 || table[*nlp] != ' '){
518: printf("*[");
519: oct+=2;
520: }
521: else {printf(" *[");;
522: oct+=3;
523: }
524: }
525: if(mflg)putc('[',mine);
526: }
527: if(inside && caps){
528: if(islower(*nlp))*nlp = toupper(*nlp);
529: }
530: else {
531: if(inside && list)putc(table[*nlp],fl);
532: if(!caps && *nlp == '\n')*nlp = ' ';
533: if(*nlp == ' ' && oct==0);
534: else if(!caps) {putchar(*nlp); oct++;}
535: }
536: if(nlp == endp){
537: if(caps)
538: inside= 0;
539: else {
540: if(list && inside){
541: inside = 0;
542: putc('\n',fl);
543: }
544: if(*(nlp) != ' '){printf("]*");
545: oct+=2;
546: }
547: else {printf("]* ");
548: oct+=3;
549: }
550: if(oct >60){putchar('\n');
551: oct=0;
552: }
553: }
554: if(mflg)putc(']',mine);
555: beg = 0;
556: }
557: if(mflg){
558: if(nlp == myst)beg = 1;
559: if(beg || last){
560: putc(*nlp,mine);
561: if(myct++ >= 72 || last == 20){
562: putc('\n',mine);
563: if(last == 20)last=myct=0;
564: else myct=0;
565: }
566: if(last)last++;
567: }
568: }
569: nlp++;
570: }
571: }
This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.