Annotation of parser3/src/lib/cord/cordbscs.c, revision 1.2.4.4
1.2 paf 1: /*
2: * Copyright (c) 1993-1994 by Xerox Corporation. All rights reserved.
3: *
4: * THIS MATERIAL IS PROVIDED AS IS, WITH ABSOLUTELY NO WARRANTY EXPRESSED
5: * OR IMPLIED. ANY USE IS AT YOUR OWN RISK.
6: *
7: * Permission is hereby granted to use or copy this program
8: * for any purpose, provided the above notices are retained on all copies.
9: * Permission to modify the code and to distribute modified code is granted,
10: * provided the above notices are retained, and a notice that the code was
11: * modified is included with the above copyright notice.
12: *
13: * Author: Hans-J. Boehm (boehm@parc.xerox.com)
14: */
15: /* Boehm, October 3, 1994 5:19 pm PDT */
16: # include "gc.h"
17: # include "cord.h"
18: # include <stdlib.h>
19: # include <stdio.h>
20: # include <string.h>
21:
22: /* An implementation of the cord primitives. These are the only */
23: /* Functions that understand the representation. We perform only */
24: /* minimal checks on arguments to these functions. Out of bounds */
25: /* arguments to the iteration functions may result in client functions */
26: /* invoked on garbage data. In most cases, client functions should be */
27: /* programmed defensively enough that this does not result in memory */
28: /* smashes. */
29:
30: typedef void (* oom_fn)(void);
31:
32: oom_fn CORD_oom_fn = (oom_fn) 0;
33:
34: # define OUT_OF_MEMORY { if (CORD_oom_fn != (oom_fn) 0) (*CORD_oom_fn)(); \
35: ABORT("Out of memory\n"); }
36: # define ABORT(msg) { fprintf(stderr, "%s\n", msg); abort(); }
37:
38: typedef unsigned long word;
39:
40: typedef union {
41: struct Concatenation {
42: char null;
43: char header;
44: char depth; /* concatenation nesting depth. */
45: unsigned char left_len;
46: /* Length of left child if it is sufficiently */
47: /* short; 0 otherwise. */
48: # define MAX_LEFT_LEN 255
49: word len;
50: CORD left; /* length(left) > 0 */
51: CORD right; /* length(right) > 0 */
52: } concatenation;
53: struct Function {
54: char null;
55: char header;
56: char depth; /* always 0 */
57: char left_len; /* always 0 */
58: word len;
59: CORD_fn fn;
60: void * client_data;
61: } function;
62: struct Generic {
63: char null;
64: char header;
65: char depth;
66: char left_len;
67: word len;
68: } generic;
69: char string[1];
70: } CordRep;
71:
72: # define CONCAT_HDR 1
73:
74: # define FN_HDR 4
75: # define SUBSTR_HDR 6
76: /* Substring nodes are a special case of function nodes. */
77: /* The client_data field is known to point to a substr_args */
78: /* structure, and the function is either CORD_apply_access_fn */
79: /* or CORD_index_access_fn. */
80:
81: /* The following may be applied only to function and concatenation nodes: */
82: #define IS_CONCATENATION(s) (((CordRep *)s)->generic.header == CONCAT_HDR)
83:
84: #define IS_FUNCTION(s) ((((CordRep *)s)->generic.header & FN_HDR) != 0)
85:
86: #define IS_SUBSTR(s) (((CordRep *)s)->generic.header == SUBSTR_HDR)
87:
88: #define LEN(s) (((CordRep *)s) -> generic.len)
89: #define DEPTH(s) (((CordRep *)s) -> generic.depth)
90: #define GEN_LEN(s) (CORD_IS_STRING(s) ? strlen(s) : LEN(s))
91:
92: #define LEFT_LEN(c) ((c) -> left_len != 0? \
93: (c) -> left_len \
94: : (CORD_IS_STRING((c) -> left) ? \
95: (c) -> len - GEN_LEN((c) -> right) \
96: : LEN((c) -> left)))
97:
98: #define SHORT_LIMIT (sizeof(CordRep) - 1)
99: /* Cords shorter than this are C strings */
100:
1.2.4.4 ! paf 101: /* paf: using knowledge of interal structure to speedup */
! 102: char CORD_nul_func(size_t i, void * client_data);
1.2 paf 103:
104: /* Dump the internal representation of x to stdout, with initial */
105: /* indentation level n. */
106: void CORD_dump_inner(CORD x, unsigned n)
107: {
108: register size_t i;
109:
110: for (i = 0; i < (size_t)n; i++) {
111: fputs(" ", stdout);
112: }
113: if (x == 0) {
114: fputs("NIL\n", stdout);
115: } else if (CORD_IS_STRING(x)) {
116: for (i = 0; i <= SHORT_LIMIT; i++) {
117: if (x[i] == '\0') break;
118: putchar(x[i]);
119: }
120: if (x[i] != '\0') fputs("...", stdout);
121: putchar('\n');
122: } else if (IS_CONCATENATION(x)) {
123: register struct Concatenation * conc =
124: &(((CordRep *)x) -> concatenation);
125: printf("Concatenation: %p (len: %d, depth: %d)\n",
126: x, (int)(conc -> len), (int)(conc -> depth));
127: CORD_dump_inner(conc -> left, n+1);
128: CORD_dump_inner(conc -> right, n+1);
129: } else /* function */{
130: register struct Function * func =
131: &(((CordRep *)x) -> function);
132: if (IS_SUBSTR(x)) printf("(Substring) ");
133: printf("Function: %p (len: %d): ", x, (int)(func -> len));
134: for (i = 0; i < 20 && i < func -> len; i++) {
135: putchar((*(func -> fn))(i, func -> client_data));
136: }
137: if (i < func -> len) fputs("...", stdout);
138: putchar('\n');
139: }
140: }
141:
142: /* Dump the internal representation of x to stdout */
143: void CORD_dump(CORD x)
144: {
145: CORD_dump_inner(x, 0);
146: fflush(stdout);
147: }
148:
149: CORD CORD_cat_char_star(CORD x, const char* y, size_t leny)
150: {
151: register size_t result_len;
152: register size_t lenx;
153: register int depth;
154:
155: if (x == CORD_EMPTY) return(y);
156: //if (leny == 0) leny=strlen(y); // PAF
157: if (y == 0) ABORT("CORD_cat_char_star(,y,) y==0"); // PAF
158: if (*y == 0) ABORT("CORD_cat_char_star(,y,) y==\"\""); // PAF
159: if (leny == 0) ABORT("CORD_cat_char_star(,y,) leny==0"); // PAF
160:
161: if (CORD_IS_STRING(x)) {
162: lenx = strlen(x);
163: result_len = lenx + leny;
164: if (result_len <= SHORT_LIMIT) {
165: register char * result = GC_MALLOC_ATOMIC(result_len+1);
166:
167: if (result == 0) OUT_OF_MEMORY;
168: memcpy(result, x, lenx);
169: memcpy(result + lenx, y, leny);
170: result[result_len] = '\0';
171: return((CORD) result);
172: } else {
173: depth = 1;
174: }
175: } else {
176: register CORD right;
177: register CORD left;
178: register char * new_right;
179: register size_t right_len;
180:
181: lenx = LEN(x);
182:
183: if (leny <= SHORT_LIMIT/2
184: && IS_CONCATENATION(x)
185: && CORD_IS_STRING(right = ((CordRep *)x) -> concatenation.right)) {
186: /* Merge y into right part of x. */
187: if (!CORD_IS_STRING(left = ((CordRep *)x) -> concatenation.left)) {
188: right_len = lenx - LEN(left);
189: } else if (((CordRep *)x) -> concatenation.left_len != 0) {
190: right_len = lenx - ((CordRep *)x) -> concatenation.left_len;
191: } else {
192: right_len = strlen(right);
193: }
194: result_len = right_len + leny; /* length of new_right */
195: if (result_len <= SHORT_LIMIT) {
196: new_right = GC_MALLOC_ATOMIC(result_len + 1);
197: memcpy(new_right, right, right_len);
198: memcpy(new_right + right_len, y, leny);
199: new_right[result_len] = '\0';
200: y = new_right;
201: leny = result_len;
202: x = left;
203: lenx -= right_len;
204: /* Now fall through to concatenate the two pieces: */
205: }
206: if (CORD_IS_STRING(x)) {
207: depth = 1;
208: } else {
209: depth = DEPTH(x) + 1;
210: }
211: } else {
212: depth = DEPTH(x) + 1;
213: }
214: result_len = lenx + leny;
215: }
216: {
217: /* The general case; lenx, result_len is known: */
218: register struct Concatenation * result;
219:
220: result = GC_NEW(struct Concatenation);
221: if (result == 0) OUT_OF_MEMORY;
222: result->header = CONCAT_HDR;
223: result->depth = depth;
224: if (lenx <= MAX_LEFT_LEN) result->left_len = lenx;
225: result->len = result_len;
226: result->left = x;
227: result->right = y;
228: if (depth >= MAX_DEPTH) {
229: return(CORD_balance((CORD)result));
230: } else {
231: return((CORD) result);
232: }
233: }
234: }
235:
236:
237: CORD CORD_cat(CORD x, CORD y)
238: {
239: register size_t result_len;
240: register int depth;
241: register size_t lenx;
242:
243: if (x == CORD_EMPTY) return(y);
244: if (y == CORD_EMPTY) return(x);
245: if (CORD_IS_STRING(y)) {
246: return(CORD_cat_char_star(x, y, strlen(y)));
247: } else if (CORD_IS_STRING(x)) {
248: lenx = strlen(x);
249: depth = DEPTH(y) + 1;
250: } else {
251: register int depthy = DEPTH(y);
252:
253: lenx = LEN(x);
254: depth = DEPTH(x) + 1;
255: if (depthy >= depth) depth = depthy + 1;
256: }
257: result_len = lenx + LEN(y);
258: {
259: register struct Concatenation * result;
260:
261: result = GC_NEW(struct Concatenation);
262: if (result == 0) OUT_OF_MEMORY;
263: result->header = CONCAT_HDR;
264: result->depth = depth;
265: // printf("depth=%d\n", depth);
266: if (lenx <= MAX_LEFT_LEN) result->left_len = lenx;
267: result->len = result_len;
268: result->left = x;
269: result->right = y;
270: // PAF@design.ru bug fix:
271: if (depth >= MAX_DEPTH) {
272: return(CORD_balance((CORD)result));
273: } else {
274: return((CORD) result);
275: }
276: }
277: }
278:
1.2.4.4 ! paf 279: /* paf: assuming both x&y not empty and both of them functions or concatenations of them,
! 280: can modify x! [to speedup simple case]
! 281: */
! 282: CORD CORD_append_blocks(CORD x, CORD y)
! 283: {
! 284: register size_t result_len;
! 285: register int depth;
! 286: register size_t lenx;
! 287:
! 288: if (CORD_IS_STRING(y)) {
! 289: return(CORD_cat_char_star(x, y, strlen(y)));
! 290: } else if (CORD_IS_STRING(x)) {
! 291: lenx = strlen(x);
! 292: depth = DEPTH(y) + 1;
! 293: } else {
! 294: // speedup simple case
! 295: if(IS_FUNCTION(x)
! 296: && IS_FUNCTION(y)
! 297: && ((CordRep *)x) -> function.fn==CORD_nul_func
! 298: && ((CordRep *)y) -> function.fn==CORD_nul_func
! 299: &&
! 300: ((CordRep *)x) -> function.client_data==
! 301: ((CordRep *)y) -> function.client_data) {
! 302: ((CordRep *)x) -> function.len+=
! 303: ((CordRep *)y) -> function.len;
! 304: return x;
! 305: }
! 306: {
! 307: register int depthy = DEPTH(y);
! 308:
! 309: lenx = LEN(x);
! 310: depth = DEPTH(x) + 1;
! 311: if (depthy >= depth) depth = depthy + 1;
! 312: }
! 313: }
! 314: result_len = lenx + LEN(y);
! 315: {
! 316: register struct Concatenation * result;
! 317:
! 318: result = GC_NEW(struct Concatenation);
! 319: if (result == 0) OUT_OF_MEMORY;
! 320: result->header = CONCAT_HDR;
! 321: result->depth = depth;
! 322: // printf("depth=%d\n", depth);
! 323: if (lenx <= MAX_LEFT_LEN) result->left_len = lenx;
! 324: result->len = result_len;
! 325: result->left = x;
! 326: result->right = y;
! 327: // PAF@design.ru bug fix:
! 328: if (depth >= MAX_DEPTH) {
! 329: return(CORD_balance((CORD)result));
! 330: } else {
! 331: return((CORD) result);
! 332: }
! 333: }
! 334: }
! 335:
1.2 paf 336:
337:
338: CORD CORD_from_fn(CORD_fn fn, void * client_data, size_t len)
339: {
340: if (len <= 0) return(0);
341: if (len <= SHORT_LIMIT) {
342: register char * result;
343: register size_t i;
344: char buf[SHORT_LIMIT+1];
345: register char c;
346:
347: for (i = 0; i < len; i++) {
348: c = (*fn)(i, client_data);
349: if (c == '\0') goto gen_case;
350: buf[i] = c;
351: }
352: buf[i] = '\0';
353: result = GC_MALLOC_ATOMIC(len+1);
354: if (result == 0) OUT_OF_MEMORY;
355: strcpy(result, buf);
356: result[len] = '\0';
357: return((CORD) result);
358: }
359: gen_case:
360: {
361: register struct Function * result;
362:
363: result = GC_NEW(struct Function);
364: if (result == 0) OUT_OF_MEMORY;
365: result->header = FN_HDR;
366: /* depth is already 0 */
367: result->len = len;
368: result->fn = fn;
369: result->client_data = client_data;
370: return((CORD) result);
371: }
372: }
373:
1.2.4.2 paf 374: CORD CORD_from_fn_gen(CORD_fn fn, void * client_data, size_t len)
1.2.4.1 paf 375: {
376: register struct Function * result;
1.2.4.2 paf 377: if (len <= 0) return(0);
1.2.4.1 paf 378:
379: result = GC_NEW(struct Function);
380: if (result == 0) OUT_OF_MEMORY;
381: result->header = FN_HDR;
382: /* depth is already 0 */
383: result->len = len;
384: result->fn = fn;
385: result->client_data = client_data;
386: return((CORD) result);
387: }
388:
1.2 paf 389: size_t CORD_len(CORD x)
390: {
391: if (x == 0) {
392: return(0);
393: } else {
394: return(GEN_LEN(x));
395: }
396: }
397:
398: struct substr_args {
399: CordRep * sa_cord;
400: size_t sa_index;
401: };
402:
403: char CORD_index_access_fn(size_t i, void * client_data)
404: {
405: register struct substr_args *descr = (struct substr_args *)client_data;
406:
407: return(((char *)(descr->sa_cord))[i + descr->sa_index]);
408: }
409:
410: char CORD_apply_access_fn(size_t i, void * client_data)
411: {
412: register struct substr_args *descr = (struct substr_args *)client_data;
413: register struct Function * fn_cord = &(descr->sa_cord->function);
414:
415: return((*(fn_cord->fn))(i + descr->sa_index, fn_cord->client_data));
416: }
417:
418: /* A version of CORD_substr that simply returns a function node, thus */
419: /* postponing its work. The fourth argument is a function that may */
420: /* be used for efficient access to the ith character. */
421: /* Assumes i >= 0 and i + n < length(x). */
422: CORD CORD_substr_closure(CORD x, size_t i, size_t n, CORD_fn f)
423: {
424: register struct substr_args * sa = GC_NEW(struct substr_args);
425: CORD result;
426:
427: if (sa == 0) OUT_OF_MEMORY;
428: sa->sa_cord = (CordRep *)x;
429: sa->sa_index = i;
430: result = CORD_from_fn(f, (void *)sa, n);
431: ((CordRep *)result) -> function.header = SUBSTR_HDR;
432: return (result);
433: }
434:
435: # define SUBSTR_LIMIT (10 * SHORT_LIMIT)
436: /* Substrings of function nodes and flat strings shorter than */
437: /* this are flat strings. Othewise we use a functional */
438: /* representation, which is significantly slower to access. */
439:
440: /* A version of CORD_substr that assumes i >= 0, n > 0, and i + n < length(x).*/
441: CORD CORD_substr_checked(CORD x, size_t i, size_t n)
442: {
443: if (CORD_IS_STRING(x)) {
444: if (n > SUBSTR_LIMIT) {
445: return(CORD_substr_closure(x, i, n, CORD_index_access_fn));
446: } else {
447: register char * result = GC_MALLOC_ATOMIC(n+1);
448:
449: if (result == 0) OUT_OF_MEMORY;
450: strncpy(result, x+i, n);
451: result[n] = '\0';
452: return(result);
453: }
454: } else if (IS_CONCATENATION(x)) {
455: register struct Concatenation * conc
456: = &(((CordRep *)x) -> concatenation);
457: register size_t left_len;
458: register size_t right_len;
459:
460: left_len = LEFT_LEN(conc);
461: right_len = conc -> len - left_len;
462: if (i >= left_len) {
463: if (n == right_len) return(conc -> right);
464: return(CORD_substr_checked(conc -> right, i - left_len, n));
465: } else if (i+n <= left_len) {
466: if (n == left_len) return(conc -> left);
467: return(CORD_substr_checked(conc -> left, i, n));
468: } else {
469: /* Need at least one character from each side. */
470: register CORD left_part;
471: register CORD right_part;
472: register size_t left_part_len = left_len - i;
473:
474: if (i == 0) {
475: left_part = conc -> left;
476: } else {
477: left_part = CORD_substr_checked(conc -> left, i, left_part_len);
478: }
479: if (i + n == right_len + left_len) {
480: right_part = conc -> right;
481: } else {
482: right_part = CORD_substr_checked(conc -> right, 0,
483: n - left_part_len);
484: }
485: return(CORD_cat(left_part, right_part));
486: }
487: } else /* function */ {
488: if (n > SUBSTR_LIMIT) {
489: if (IS_SUBSTR(x)) {
490: /* Avoid nesting substring nodes. */
491: register struct Function * f = &(((CordRep *)x) -> function);
492: register struct substr_args *descr =
493: (struct substr_args *)(f -> client_data);
494:
495: return(CORD_substr_closure((CORD)descr->sa_cord,
496: i + descr->sa_index,
497: n, f -> fn));
498: } else {
499: return(CORD_substr_closure(x, i, n, CORD_apply_access_fn));
500: }
501: } else {
502: char * result;
503: register struct Function * f = &(((CordRep *)x) -> function);
504: char buf[SUBSTR_LIMIT+1];
505: register char * p = buf;
506: register char c;
507: register int j;
508: register int lim = i + n;
509:
510: for (j = i; j < lim; j++) {
511: c = (*(f -> fn))(j, f -> client_data);
512: if (c == '\0') {
513: return(CORD_substr_closure(x, i, n, CORD_apply_access_fn));
514: }
515: *p++ = c;
516: }
517: *p = '\0';
518: result = GC_MALLOC_ATOMIC(n+1);
519: if (result == 0) OUT_OF_MEMORY;
520: strcpy(result, buf);
521: return(result);
522: }
523: }
524: }
525:
526: CORD CORD_substr(CORD x, size_t i, size_t n)
527: {
528: register size_t len = CORD_len(x);
529:
530: if (i >= len || n <= 0) return(0);
531: /* n < 0 is impossible in a correct C implementation, but */
532: /* quite possible under SunOS 4.X. */
533: if (i + n > len) n = len - i;
534: # ifndef __STDC__
535: if (i < 0) ABORT("CORD_substr: second arg. negative");
536: /* Possible only if both client and C implementation are buggy. */
537: /* But empirically this happens frequently. */
538: # endif
539: return(CORD_substr_checked(x, i, n));
540: }
541:
542: /* See cord.h for definition. We assume i is in range. */
543: int CORD_iter5(CORD x, size_t i, CORD_iter_fn f1,
544: CORD_batched_iter_fn f2, void * client_data)
545: {
1.2.4.3 paf 546: int result;
1.2 paf 547: if (x == 0) return(0);
548: if (CORD_IS_STRING(x)) {
549: register const char* p = x+i;
550:
551: if (*p == '\0') ABORT("2nd arg to CORD_iter5 too big");
552: if (f2 != CORD_NO_FN) {
553: return((*f2)(p, client_data));
554: } else {
555: while (*p) {
1.2.4.3 paf 556: if (result=(*f1)(*p, client_data))
557: return result;
1.2 paf 558: p++;
559: }
560: return(0);
561: }
562: } else if (IS_CONCATENATION(x)) {
1.2.4.3 paf 563: register struct Concatenation * conc = &(((CordRep *)x) -> concatenation);
1.2 paf 564: if (i > 0) {
565: register size_t left_len = LEFT_LEN(conc);
566:
567: if (i >= left_len) {
568: return(CORD_iter5(conc -> right, i - left_len, f1, f2,
569: client_data));
570: }
571: }
1.2.4.3 paf 572: result=CORD_iter5(conc -> left, i, f1, f2, client_data);
573: if (result) return result;
1.2 paf 574: return(CORD_iter5(conc -> right, 0, f1, f2, client_data));
575: } else /* function */ {
576: register struct Function * f = &(((CordRep *)x) -> function);
577: register size_t j;
578: register size_t lim = f -> len;
579:
580: for (j = i; j < lim; j++) {
1.2.4.3 paf 581: if (result=(*f1)((*(f -> fn))(j, f -> client_data), client_data))
582: return result;
1.2 paf 583: }
584: return(0);
1.2.4.1 paf 585: }
586: }
587:
588: /* See cord.h for definition. We assume i is in range. */
1.2.4.3 paf 589: int CORD_block_iter(CORD x, size_t i, CORD_block_iter_fn fb, void * client_data)
1.2.4.1 paf 590: {
1.2.4.3 paf 591: int result;
1.2.4.1 paf 592: if (x == 0) return(0);
1.2.4.3 paf 593: if (CORD_IS_STRING(x)) {
594: register const char* p = x+i;
595: const char *b=p;
596: char bc=*b;
597: int pc;
1.2.4.1 paf 598:
1.2.4.3 paf 599: if (bc == '\0') ABORT("2nd arg to CORD_iter5 too big");
600: do {
601: pc=*++p;
602: if(pc!=bc) {
603: if (result=fb(bc, p-b, client_data))
604: return result;
605: b=p; bc=pc;
606: }
607: } while (pc);
608: return(0);
609: } else if (IS_CONCATENATION(x)) {
610: register struct Concatenation * conc= &(((CordRep *)x) -> concatenation);
1.2.4.1 paf 611: if (i > 0) {
1.2.4.3 paf 612: register size_t left_len = LEFT_LEN(conc);
1.2.4.1 paf 613:
1.2.4.3 paf 614: if (i >= left_len) {
615: return(CORD_block_iter(conc -> right, i - left_len, fb, client_data));
616: }
1.2.4.1 paf 617: }
1.2.4.3 paf 618: result=CORD_block_iter(conc -> left, i, fb, client_data);
619: if (result) return result;
620: return(CORD_block_iter(conc -> right, 0, fb, client_data));
621: } else /* function */ {
1.2.4.1 paf 622: register struct Function * f = &(((CordRep *)x) -> function);
1.2.4.3 paf 623: register size_t lim = f -> len;
1.2.4.1 paf 624:
1.2.4.3 paf 625: if(f->fn == CORD_nul_func ) {
626: if (result=fb((char)(unsigned long)f -> client_data, f -> len-i, client_data)) return result;
627: } else if(f->fn == CORD_apply_access_fn) {
628: register struct substr_args *descr = (struct substr_args *)f->client_data;
629: register struct Function * fn_cord = &(descr->sa_cord->function);
630:
631: if(fn_cord->fn == CORD_nul_func ) {
632: if (result=fb((char)(unsigned long)fn_cord->client_data, f -> len-i, client_data))
633: return result;
634: } else
635: ABORT("CORD_block_iter:CORD_apply_access_fn:unknown_fn should not happen");
636: } else {
637: if(f->fn == CORD_index_access_fn)
638: ABORT("CORD_block_iter:CORD_index_access_fn should not happen");
639: ABORT("CORD_block_iter:unknown_fn should not happen");
640: }
1.2 paf 641: }
1.2.4.3 paf 642: return(0);
1.2 paf 643: }
644:
1.2.4.3 paf 645:
1.2 paf 646: #undef CORD_iter
647: int CORD_iter(CORD x, CORD_iter_fn f1, void * client_data)
648: {
649: return(CORD_iter5(x, 0, f1, CORD_NO_FN, client_data));
650: }
651:
652: int CORD_riter4(CORD x, size_t i, CORD_iter_fn f1, void * client_data)
653: {
654: if (x == 0) return(0);
655: if (CORD_IS_STRING(x)) {
656: register const char* p = x + i;
657: register char c;
658:
659: for(;;) {
660: c = *p;
661: if (c == '\0') ABORT("2nd arg to CORD_riter4 too big");
662: if ((*f1)(c, client_data)) return(1);
663: if (p == x) break;
664: p--;
665: }
666: return(0);
667: } else if (IS_CONCATENATION(x)) {
668: register struct Concatenation * conc
669: = &(((CordRep *)x) -> concatenation);
670: register CORD left_part = conc -> left;
671: register size_t left_len;
672:
673: left_len = LEFT_LEN(conc);
674: if (i >= left_len) {
675: if (CORD_riter4(conc -> right, i - left_len, f1, client_data)) {
676: return(1);
677: }
678: return(CORD_riter4(left_part, left_len - 1, f1, client_data));
679: } else {
680: return(CORD_riter4(left_part, i, f1, client_data));
681: }
682: } else /* function */ {
683: register struct Function * f = &(((CordRep *)x) -> function);
684: register size_t j;
685:
686: for (j = i; ; j--) {
687: if ((*f1)((*(f -> fn))(j, f -> client_data), client_data)) {
688: return(1);
689: }
690: if (j == 0) return(0);
691: }
692: }
693: }
694:
695: int CORD_riter(CORD x, CORD_iter_fn f1, void * client_data)
696: {
697: return(CORD_riter4(x, CORD_len(x) - 1, f1, client_data));
698: }
699:
700: /*
701: * The following functions are concerned with balancing cords.
702: * Strategy:
703: * Scan the cord from left to right, keeping the cord scanned so far
704: * as a forest of balanced trees of exponentialy decreasing length.
705: * When a new subtree needs to be added to the forest, we concatenate all
706: * shorter ones to the new tree in the appropriate order, and then insert
707: * the result into the forest.
708: * Crucial invariants:
709: * 1. The concatenation of the forest (in decreasing order) with the
710: * unscanned part of the rope is equal to the rope being balanced.
711: * 2. All trees in the forest are balanced.
712: * 3. forest[i] has depth at most i.
713: */
714:
715: typedef struct {
716: CORD c;
717: size_t len; /* Actual length of c */
718: } ForestElement;
719:
720: static size_t min_len [ MAX_DEPTH ];
721:
722: static int min_len_init = 0;
723:
724: int CORD_max_len;
725:
726: typedef ForestElement Forest [ MAX_DEPTH ];
727: /* forest[i].len >= fib(i+1) */
728: /* The string is the concatenation */
729: /* of the forest in order of DECREASING */
730: /* indices. */
731:
732: void CORD_init_min_len()
733: {
734: register int i;
735: register size_t last, previous, current;
736:
737: min_len[0] = previous = 1;
738: min_len[1] = last = 2;
739: for (i = 2; i < MAX_DEPTH; i++) {
740: current = last + previous;
741: if (current < last) /* overflow */ current = last;
742: min_len[i] = current;
743: previous = last;
744: last = current;
745: }
746: CORD_max_len = last - 1;
747: min_len_init = 1;
748: }
749:
750:
751: void CORD_init_forest(ForestElement * forest, size_t max_len)
752: {
753: register int i;
754:
755: for (i = 0; i < MAX_DEPTH; i++) {
756: forest[i].c = 0;
757: if (min_len[i] > max_len) return;
758: }
759: ABORT("Cord too long");
760: }
761:
762: /* Add a leaf to the appropriate level in the forest, cleaning */
763: /* out lower levels as necessary. */
764: /* Also works if x is a balanced tree of concatenations; however */
765: /* in this case an extra concatenation node may be inserted above x; */
766: /* This node should not be counted in the statement of the invariants. */
767: void CORD_add_forest(ForestElement * forest, CORD x, size_t len)
768: {
769: register int i = 0;
770: register CORD sum = CORD_EMPTY;
771: register size_t sum_len = 0;
772:
773: while (len > min_len[i + 1]) {
774: if (forest[i].c != 0) {
775: sum = CORD_cat(forest[i].c, sum);
776: sum_len += forest[i].len;
777: forest[i].c = 0;
778: }
779: i++;
780: }
781: /* Sum has depth at most 1 greter than what would be required */
782: /* for balance. */
783: sum = CORD_cat(sum, x);
784: sum_len += len;
785: /* If x was a leaf, then sum is now balanced. To see this */
786: /* consider the two cases in which forest[i-1] either is or is */
787: /* not empty. */
788: while (sum_len >= min_len[i]) {
789: if (forest[i].c != 0) {
790: sum = CORD_cat(forest[i].c, sum);
791: sum_len += forest[i].len;
792: /* This is again balanced, since sum was balanced, and has */
793: /* allowable depth that differs from i by at most 1. */
794: forest[i].c = 0;
795: }
796: i++;
797: }
798: i--;
799: forest[i].c = sum;
800: forest[i].len = sum_len;
801: }
802:
803: CORD CORD_concat_forest(ForestElement * forest, size_t expected_len)
804: {
805: register int i = 0;
806: CORD sum = 0;
807: size_t sum_len = 0;
808:
809: while (sum_len != expected_len) {
810: if (forest[i].c != 0) {
811: sum = CORD_cat(forest[i].c, sum);
812: sum_len += forest[i].len;
813: }
814: i++;
815: }
816: return(sum);
817: }
818:
819: /* Insert the frontier of x into forest. Balanced subtrees are */
820: /* treated as leaves. This potentially adds one to the depth */
821: /* of the final tree. */
822: void CORD_balance_insert(CORD x, size_t len, ForestElement * forest)
823: {
824: register int depth;
825:
826: if (CORD_IS_STRING(x)) {
827: CORD_add_forest(forest, x, len);
828: } else if (IS_CONCATENATION(x)
829: && ((depth = DEPTH(x)) >= MAX_DEPTH
830: || len < min_len[depth])) {
831: register struct Concatenation * conc
832: = &(((CordRep *)x) -> concatenation);
833: size_t left_len = LEFT_LEN(conc);
834:
835: CORD_balance_insert(conc -> left, left_len, forest);
836: CORD_balance_insert(conc -> right, len - left_len, forest);
837: } else /* function or balanced */ {
838: CORD_add_forest(forest, x, len);
839: }
840: }
841:
842:
843: CORD CORD_balance(CORD x)
844: {
845: Forest forest;
846: register size_t len;
847:
848: if (x == 0) return(0);
849: if (CORD_IS_STRING(x)) return(x);
850: if (!min_len_init) CORD_init_min_len();
851: len = LEN(x);
852: CORD_init_forest(forest, len);
853: CORD_balance_insert(x, len, forest);
854: return(CORD_concat_forest(forest, len));
855: }
856:
857:
858: /* Position primitives */
859:
860: /* Private routines to deal with the hard cases only: */
861:
862: /* P contains a prefix of the path to cur_pos. Extend it to a full */
863: /* path and set up leaf info. */
864: /* Return 0 if past the end of cord, 1 o.w. */
865: void CORD__extend_path(register CORD_pos p)
866: {
867: register struct CORD_pe * current_pe = &(p[0].path[p[0].path_len]);
868: register CORD top = current_pe -> pe_cord;
869: register size_t pos = p[0].cur_pos;
870: register size_t top_pos = current_pe -> pe_start_pos;
871: register size_t top_len = GEN_LEN(top);
872:
873: /* Fill in the rest of the path. */
874: while(!CORD_IS_STRING(top) && IS_CONCATENATION(top)) {
875: register struct Concatenation * conc =
876: &(((CordRep *)top) -> concatenation);
877: register size_t left_len;
878:
879: left_len = LEFT_LEN(conc);
880: current_pe++;
881: if (pos >= top_pos + left_len) {
882: current_pe -> pe_cord = top = conc -> right;
883: current_pe -> pe_start_pos = top_pos = top_pos + left_len;
884: top_len -= left_len;
885: } else {
886: current_pe -> pe_cord = top = conc -> left;
887: current_pe -> pe_start_pos = top_pos;
888: top_len = left_len;
889: }
890: p[0].path_len++;
891: }
892: /* Fill in leaf description for fast access. */
893: if (CORD_IS_STRING(top)) {
894: p[0].cur_leaf = top;
895: p[0].cur_start = top_pos;
896: p[0].cur_end = top_pos + top_len;
897: } else {
898: p[0].cur_end = 0;
899: }
900: if (pos >= top_pos + top_len) p[0].path_len = CORD_POS_INVALID;
901: }
902:
903: char CORD__pos_fetch(register CORD_pos p)
904: {
905: /* Leaf is a function node */
906: struct CORD_pe * pe = &((p)[0].path[(p)[0].path_len]);
907: CORD leaf = pe -> pe_cord;
908: register struct Function * f = &(((CordRep *)leaf) -> function);
909:
910: if (!IS_FUNCTION(leaf)) ABORT("CORD_pos_fetch: bad leaf");
911: return ((*(f -> fn))(p[0].cur_pos - pe -> pe_start_pos, f -> client_data));
912: }
913:
914: void CORD__next(register CORD_pos p)
915: {
916: register size_t cur_pos = p[0].cur_pos + 1;
917: register struct CORD_pe * current_pe = &((p)[0].path[(p)[0].path_len]);
918: register CORD leaf = current_pe -> pe_cord;
919:
920: /* Leaf is not a string or we're at end of leaf */
921: p[0].cur_pos = cur_pos;
922: if (!CORD_IS_STRING(leaf)) {
923: /* Function leaf */
924: register struct Function * f = &(((CordRep *)leaf) -> function);
925: register size_t start_pos = current_pe -> pe_start_pos;
926: register size_t end_pos = start_pos + f -> len;
927:
928: if (cur_pos < end_pos) {
929: /* Fill cache and return. */
930: register size_t i;
931: register size_t limit = cur_pos + FUNCTION_BUF_SZ;
932: register CORD_fn fn = f -> fn;
933: register void * client_data = f -> client_data;
934:
935: if (limit > end_pos) {
936: limit = end_pos;
937: }
938: for (i = cur_pos; i < limit; i++) {
939: p[0].function_buf[i - cur_pos] =
940: (*fn)(i - start_pos, client_data);
941: }
942: p[0].cur_start = cur_pos;
943: p[0].cur_leaf = p[0].function_buf;
944: p[0].cur_end = limit;
945: return;
946: }
947: }
948: /* End of leaf */
949: /* Pop the stack until we find two concatenation nodes with the */
950: /* same start position: this implies we were in left part. */
951: {
952: while (p[0].path_len > 0
953: && current_pe[0].pe_start_pos != current_pe[-1].pe_start_pos) {
954: p[0].path_len--;
955: current_pe--;
956: }
957: if (p[0].path_len == 0) {
958: p[0].path_len = CORD_POS_INVALID;
959: return;
960: }
961: }
962: p[0].path_len--;
963: CORD__extend_path(p);
964: }
965:
966: void CORD__prev(register CORD_pos p)
967: {
968: register struct CORD_pe * pe = &(p[0].path[p[0].path_len]);
969:
970: if (p[0].cur_pos == 0) {
971: p[0].path_len = CORD_POS_INVALID;
972: return;
973: }
974: p[0].cur_pos--;
975: if (p[0].cur_pos >= pe -> pe_start_pos) return;
976:
977: /* Beginning of leaf */
978:
979: /* Pop the stack until we find two concatenation nodes with the */
980: /* different start position: this implies we were in right part. */
981: {
982: register struct CORD_pe * current_pe = &((p)[0].path[(p)[0].path_len]);
983:
984: while (p[0].path_len > 0
985: && current_pe[0].pe_start_pos == current_pe[-1].pe_start_pos) {
986: p[0].path_len--;
987: current_pe--;
988: }
989: }
990: p[0].path_len--;
991: CORD__extend_path(p);
992: }
993:
994: #undef CORD_pos_fetch
995: #undef CORD_next
996: #undef CORD_prev
997: #undef CORD_pos_to_index
998: #undef CORD_pos_to_cord
999: #undef CORD_pos_valid
1000:
1001: char CORD_pos_fetch(register CORD_pos p)
1002: {
1003: if (p[0].cur_start <= p[0].cur_pos && p[0].cur_pos < p[0].cur_end) {
1004: return(p[0].cur_leaf[p[0].cur_pos - p[0].cur_start]);
1005: } else {
1006: return(CORD__pos_fetch(p));
1007: }
1008: }
1009:
1010: void CORD_next(CORD_pos p)
1011: {
1012: if (p[0].cur_pos < p[0].cur_end - 1) {
1013: p[0].cur_pos++;
1014: } else {
1015: CORD__next(p);
1016: }
1017: }
1018:
1019: void CORD_prev(CORD_pos p)
1020: {
1021: if (p[0].cur_end != 0 && p[0].cur_pos > p[0].cur_start) {
1022: p[0].cur_pos--;
1023: } else {
1024: CORD__prev(p);
1025: }
1026: }
1027:
1028: size_t CORD_pos_to_index(CORD_pos p)
1029: {
1030: return(p[0].cur_pos);
1031: }
1032:
1033: CORD CORD_pos_to_cord(CORD_pos p)
1034: {
1035: return(p[0].path[0].pe_cord);
1036: }
1037:
1038: int CORD_pos_valid(CORD_pos p)
1039: {
1040: return(p[0].path_len != CORD_POS_INVALID);
1041: }
1042:
1043: void CORD_set_pos(CORD_pos p, CORD x, size_t i)
1044: {
1045: if (x == CORD_EMPTY) {
1046: p[0].path_len = CORD_POS_INVALID;
1047: return;
1048: }
1049: p[0].path[0].pe_cord = x;
1050: p[0].path[0].pe_start_pos = 0;
1051: p[0].path_len = 0;
1052: p[0].cur_pos = i;
1053: CORD__extend_path(p);
1054: }
E-mail:
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to cordbscs.c CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [parser3project] / parser3 / src / lib / cord