Annotation of src/usr.bin/mandoc/term.c, Revision 1.19
1.19 ! schwarze 1: /* $Id: term.c,v 1.18 2009/10/27 21:40:07 schwarze Exp $ */
1.1 kristaps 2: /*
1.2 schwarze 3: * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
1.1 kristaps 4: *
5: * Permission to use, copy, modify, and distribute this software for any
1.2 schwarze 6: * purpose with or without fee is hereby granted, provided that the above
7: * copyright notice and this permission notice appear in all copies.
1.1 kristaps 8: *
1.2 schwarze 9: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
1.1 kristaps 16: */
17: #include <assert.h>
18: #include <stdio.h>
19: #include <stdlib.h>
20: #include <string.h>
1.19 ! schwarze 21: #include <time.h>
1.1 kristaps 22:
1.15 schwarze 23: #include "chars.h"
1.16 schwarze 24: #include "out.h"
1.1 kristaps 25: #include "term.h"
26: #include "man.h"
27: #include "mdoc.h"
1.16 schwarze 28: #include "main.h"
1.1 kristaps 29:
1.16 schwarze 30: /* FIXME: accomodate non-breaking, non-collapsing white-space. */
31: /* FIXME: accomodate non-breaking, collapsing white-space. */
1.1 kristaps 32:
33: static struct termp *term_alloc(enum termenc);
34: static void term_free(struct termp *);
1.11 schwarze 35:
36: static void do_escaped(struct termp *, const char **);
37: static void do_special(struct termp *,
38: const char *, size_t);
39: static void do_reserved(struct termp *,
1.1 kristaps 40: const char *, size_t);
1.11 schwarze 41: static void buffer(struct termp *, char);
42: static void encode(struct termp *, char);
1.1 kristaps 43:
44:
45: void *
46: ascii_alloc(void)
47: {
48:
49: return(term_alloc(TERMENC_ASCII));
50: }
51:
52:
1.13 schwarze 53: void
1.1 kristaps 54: terminal_free(void *arg)
55: {
56:
57: term_free((struct termp *)arg);
58: }
59:
60:
61: static void
62: term_free(struct termp *p)
63: {
64:
65: if (p->buf)
66: free(p->buf);
1.15 schwarze 67: if (p->symtab)
68: chars_free(p->symtab);
1.1 kristaps 69:
70: free(p);
71: }
72:
73:
74: static struct termp *
75: term_alloc(enum termenc enc)
76: {
77: struct termp *p;
78:
1.19 ! schwarze 79: p = calloc(1, sizeof(struct termp));
! 80: if (NULL == p) {
! 81: perror(NULL);
! 82: exit(EXIT_FAILURE);
! 83: }
1.1 kristaps 84: p->maxrmargin = 78;
85: p->enc = enc;
86: return(p);
87: }
88:
89:
90: /*
91: * Flush a line of text. A "line" is loosely defined as being something
92: * that should be followed by a newline, regardless of whether it's
93: * broken apart by newlines getting there. A line can also be a
94: * fragment of a columnar list.
95: *
96: * Specifically, a line is whatever's in p->buf of length p->col, which
97: * is zeroed after this function returns.
98: *
1.6 schwarze 99: * The usage of termp:flags is as follows:
1.1 kristaps 100: *
101: * - TERMP_NOLPAD: when beginning to write the line, don't left-pad the
102: * offset value. This is useful when doing columnar lists where the
103: * prior column has right-padded.
104: *
105: * - TERMP_NOBREAK: this is the most important and is used when making
106: * columns. In short: don't print a newline and instead pad to the
107: * right margin. Used in conjunction with TERMP_NOLPAD.
108: *
1.9 schwarze 109: * - TERMP_TWOSPACE: when padding, make sure there are at least two
110: * space characters of padding. Otherwise, rather break the line.
111: *
1.6 schwarze 112: * - TERMP_DANGLE: don't newline when TERMP_NOBREAK is specified and
113: * the line is overrun, and don't pad-right if it's underrun.
114: *
115: * - TERMP_HANG: like TERMP_DANGLE, but doesn't newline when
116: * overruning, instead save the position and continue at that point
117: * when the next invocation.
1.1 kristaps 118: *
119: * In-line line breaking:
120: *
121: * If TERMP_NOBREAK is specified and the line overruns the right
122: * margin, it will break and pad-right to the right margin after
123: * writing. If maxrmargin is violated, it will break and continue
1.19 ! schwarze 124: * writing from the right-margin, which will lead to the above scenario
! 125: * upon exit. Otherwise, the line will break at the right margin.
1.1 kristaps 126: */
127: void
128: term_flushln(struct termp *p)
129: {
1.19 ! schwarze 130: int i; /* current input position in p->buf */
! 131: size_t vis; /* current visual position on output */
! 132: size_t vbl; /* number of blanks to prepend to output */
! 133: size_t vsz; /* visual characters to write to output */
! 134: size_t bp; /* visual right border position */
! 135: int j; /* temporary loop index */
! 136: size_t maxvis, mmax;
1.9 schwarze 137: static int overstep = 0;
1.1 kristaps 138:
139: /*
140: * First, establish the maximum columns of "visible" content.
141: * This is usually the difference between the right-margin and
142: * an indentation, but can be, for tagged lists or columns, a
1.19 ! schwarze 143: * small set of values.
1.1 kristaps 144: */
145:
146: assert(p->offset < p->rmargin);
1.9 schwarze 147:
1.19 ! schwarze 148: maxvis = (int)(p->rmargin - p->offset) - overstep < 0 ?
! 149: /* LINTED */
! 150: 0 : p->rmargin - p->offset - overstep;
! 151: mmax = (int)(p->maxrmargin - p->offset) - overstep < 0 ?
! 152: /* LINTED */
! 153: 0 : p->maxrmargin - p->offset - overstep;
1.9 schwarze 154:
1.1 kristaps 155: bp = TERMP_NOBREAK & p->flags ? mmax : maxvis;
1.19 ! schwarze 156:
! 157: /*
! 158: * FIXME: if bp is zero, we still output the first word before
! 159: * breaking the line.
! 160: */
! 161:
1.1 kristaps 162: vis = 0;
1.6 schwarze 163:
1.1 kristaps 164: /*
165: * If in the standard case (left-justified), then begin with our
166: * indentation, otherwise (columns, etc.) just start spitting
167: * out text.
168: */
169:
170: if ( ! (p->flags & TERMP_NOLPAD))
171: /* LINTED */
172: for (j = 0; j < (int)p->offset; j++)
173: putchar(' ');
174:
175: for (i = 0; i < (int)p->col; i++) {
176: /*
177: * Count up visible word characters. Control sequences
178: * (starting with the CSI) aren't counted. A space
179: * generates a non-printing word, which is valid (the
180: * space is printed according to regular spacing rules).
181: */
182:
183: /* LINTED */
184: for (j = i, vsz = 0; j < (int)p->col; j++) {
1.10 schwarze 185: if (j && ' ' == p->buf[j])
1.1 kristaps 186: break;
187: else if (8 == p->buf[j])
1.8 schwarze 188: vsz--;
1.1 kristaps 189: else
190: vsz++;
191: }
192:
193: /*
1.5 schwarze 194: * Choose the number of blanks to prepend: no blank at the
195: * beginning of a line, one between words -- but do not
196: * actually write them yet.
1.1 kristaps 197: */
1.5 schwarze 198: vbl = (size_t)(0 == vis ? 0 : 1);
1.1 kristaps 199:
1.5 schwarze 200: /*
201: * Find out whether we would exceed the right margin.
202: * If so, break to the next line. (TODO: hyphenate)
203: * Otherwise, write the chosen number of blanks now.
204: */
205: if (vis && vis + vbl + vsz > bp) {
206: putchar('\n');
207: if (TERMP_NOBREAK & p->flags) {
208: for (j = 0; j < (int)p->rmargin; j++)
209: putchar(' ');
210: vis = p->rmargin - p->offset;
211: } else {
1.1 kristaps 212: for (j = 0; j < (int)p->offset; j++)
213: putchar(' ');
214: vis = 0;
1.5 schwarze 215: }
1.16 schwarze 216: /* Remove the overstep width. */
1.18 schwarze 217: bp += (int)/* LINTED */
218: overstep;
1.17 schwarze 219: overstep = 0;
1.5 schwarze 220: } else {
221: for (j = 0; j < (int)vbl; j++)
1.1 kristaps 222: putchar(' ');
1.5 schwarze 223: vis += vbl;
1.1 kristaps 224: }
225:
1.3 schwarze 226: /*
1.5 schwarze 227: * Finally, write out the word.
1.1 kristaps 228: */
229: for ( ; i < (int)p->col; i++) {
230: if (' ' == p->buf[i])
231: break;
232: putchar(p->buf[i]);
233: }
234: vis += vsz;
235: }
1.18 schwarze 236:
1.9 schwarze 237: p->col = 0;
1.18 schwarze 238: overstep = 0;
1.1 kristaps 239:
1.9 schwarze 240: if ( ! (TERMP_NOBREAK & p->flags)) {
241: putchar('\n');
1.1 kristaps 242: return;
243: }
244:
1.9 schwarze 245: if (TERMP_HANG & p->flags) {
246: /* We need one blank after the tag. */
247: overstep = /* LINTED */
248: vis - maxvis + 1;
249:
250: /*
251: * Behave exactly the same way as groff:
252: * If we have overstepped the margin, temporarily move
253: * it to the right and flag the rest of the line to be
254: * shorter.
255: * If we landed right at the margin, be happy.
256: * If we are one step before the margin, temporarily
257: * move it one step LEFT and flag the rest of the line
258: * to be longer.
259: */
260: if (overstep >= -1) {
261: assert((int)maxvis + overstep >= 0);
262: /* LINTED */
263: maxvis += overstep;
264: } else
265: overstep = 0;
266:
267: } else if (TERMP_DANGLE & p->flags)
268: return;
1.1 kristaps 269:
1.9 schwarze 270: /* Right-pad. */
271: if (maxvis > vis + /* LINTED */
272: ((TERMP_TWOSPACE & p->flags) ? 1 : 0))
273: for ( ; vis < maxvis; vis++)
274: putchar(' ');
275: else { /* ...or newline break. */
1.1 kristaps 276: putchar('\n');
1.9 schwarze 277: for (i = 0; i < (int)p->rmargin; i++)
278: putchar(' ');
279: }
1.1 kristaps 280: }
281:
282:
283: /*
284: * A newline only breaks an existing line; it won't assert vertical
285: * space. All data in the output buffer is flushed prior to the newline
286: * assertion.
287: */
288: void
289: term_newln(struct termp *p)
290: {
291:
292: p->flags |= TERMP_NOSPACE;
293: if (0 == p->col) {
294: p->flags &= ~TERMP_NOLPAD;
295: return;
296: }
297: term_flushln(p);
298: p->flags &= ~TERMP_NOLPAD;
299: }
300:
301:
302: /*
303: * Asserts a vertical space (a full, empty line-break between lines).
304: * Note that if used twice, this will cause two blank spaces and so on.
305: * All data in the output buffer is flushed prior to the newline
306: * assertion.
307: */
308: void
309: term_vspace(struct termp *p)
310: {
311:
312: term_newln(p);
313: putchar('\n');
314: }
315:
316:
317: static void
1.11 schwarze 318: do_special(struct termp *p, const char *word, size_t len)
1.1 kristaps 319: {
320: const char *rhs;
321: size_t sz;
1.4 schwarze 322: int i;
1.1 kristaps 323:
1.15 schwarze 324: rhs = chars_a2ascii(p->symtab, word, len, &sz);
1.7 schwarze 325:
1.11 schwarze 326: if (NULL == rhs) {
327: #if 0
328: fputs("Unknown special character: ", stderr);
329: for (i = 0; i < (int)len; i++)
330: fputc(word[i], stderr);
331: fputc('\n', stderr);
332: #endif
333: return;
334: }
335: for (i = 0; i < (int)sz; i++)
336: encode(p, rhs[i]);
337: }
338:
339:
340: static void
341: do_reserved(struct termp *p, const char *word, size_t len)
342: {
343: const char *rhs;
344: size_t sz;
345: int i;
346:
1.15 schwarze 347: rhs = chars_a2res(p->symtab, word, len, &sz);
1.11 schwarze 348:
349: if (NULL == rhs) {
350: #if 0
351: fputs("Unknown reserved word: ", stderr);
352: for (i = 0; i < (int)len; i++)
353: fputc(word[i], stderr);
354: fputc('\n', stderr);
355: #endif
356: return;
357: }
358: for (i = 0; i < (int)sz; i++)
359: encode(p, rhs[i]);
1.1 kristaps 360: }
361:
362:
363: /*
364: * Handle an escape sequence: determine its length and pass it to the
365: * escape-symbol look table. Note that we assume mdoc(3) has validated
366: * the escape sequence (we assert upon badly-formed escape sequences).
367: */
368: static void
1.11 schwarze 369: do_escaped(struct termp *p, const char **word)
1.1 kristaps 370: {
1.11 schwarze 371: int j, type;
1.7 schwarze 372: const char *wp;
373:
374: wp = *word;
1.11 schwarze 375: type = 1;
1.1 kristaps 376:
1.7 schwarze 377: if (0 == *(++wp)) {
378: *word = wp;
1.1 kristaps 379: return;
1.7 schwarze 380: }
1.1 kristaps 381:
1.7 schwarze 382: if ('(' == *wp) {
383: wp++;
384: if (0 == *wp || 0 == *(wp + 1)) {
385: *word = 0 == *wp ? wp : wp + 1;
1.1 kristaps 386: return;
1.7 schwarze 387: }
1.1 kristaps 388:
1.11 schwarze 389: do_special(p, wp, 2);
1.7 schwarze 390: *word = ++wp;
1.1 kristaps 391: return;
392:
1.7 schwarze 393: } else if ('*' == *wp) {
394: if (0 == *(++wp)) {
395: *word = wp;
1.1 kristaps 396: return;
1.7 schwarze 397: }
1.1 kristaps 398:
1.7 schwarze 399: switch (*wp) {
1.1 kristaps 400: case ('('):
1.7 schwarze 401: wp++;
402: if (0 == *wp || 0 == *(wp + 1)) {
403: *word = 0 == *wp ? wp : wp + 1;
1.1 kristaps 404: return;
1.7 schwarze 405: }
1.1 kristaps 406:
1.11 schwarze 407: do_reserved(p, wp, 2);
1.7 schwarze 408: *word = ++wp;
1.1 kristaps 409: return;
410: case ('['):
1.11 schwarze 411: type = 0;
1.1 kristaps 412: break;
413: default:
1.11 schwarze 414: do_reserved(p, wp, 1);
1.7 schwarze 415: *word = wp;
1.1 kristaps 416: return;
417: }
418:
1.7 schwarze 419: } else if ('f' == *wp) {
420: if (0 == *(++wp)) {
421: *word = wp;
1.1 kristaps 422: return;
1.7 schwarze 423: }
424:
425: switch (*wp) {
1.1 kristaps 426: case ('B'):
1.12 schwarze 427: p->bold++;
1.1 kristaps 428: break;
429: case ('I'):
1.12 schwarze 430: p->under++;
1.1 kristaps 431: break;
432: case ('P'):
433: /* FALLTHROUGH */
434: case ('R'):
1.12 schwarze 435: p->bold = p->under = 0;
1.1 kristaps 436: break;
437: default:
438: break;
439: }
1.7 schwarze 440:
441: *word = wp;
1.1 kristaps 442: return;
443:
1.7 schwarze 444: } else if ('[' != *wp) {
1.11 schwarze 445: do_special(p, wp, 1);
1.7 schwarze 446: *word = wp;
1.1 kristaps 447: return;
448: }
449:
1.7 schwarze 450: wp++;
451: for (j = 0; *wp && ']' != *wp; wp++, j++)
1.1 kristaps 452: /* Loop... */ ;
453:
1.7 schwarze 454: if (0 == *wp) {
455: *word = wp;
1.1 kristaps 456: return;
1.7 schwarze 457: }
1.1 kristaps 458:
1.11 schwarze 459: if (type)
460: do_special(p, wp - j, (size_t)j);
461: else
462: do_reserved(p, wp - j, (size_t)j);
1.7 schwarze 463: *word = wp;
1.1 kristaps 464: }
465:
466:
467: /*
468: * Handle pwords, partial words, which may be either a single word or a
469: * phrase that cannot be broken down (such as a literal string). This
470: * handles word styling.
471: */
1.7 schwarze 472: void
473: term_word(struct termp *p, const char *word)
1.1 kristaps 474: {
1.7 schwarze 475: const char *sv;
1.1 kristaps 476:
1.14 schwarze 477: sv = word;
478:
479: if (word[0] && 0 == word[1])
480: switch (word[0]) {
481: case('.'):
482: /* FALLTHROUGH */
483: case(','):
484: /* FALLTHROUGH */
485: case(';'):
486: /* FALLTHROUGH */
487: case(':'):
488: /* FALLTHROUGH */
489: case('?'):
490: /* FALLTHROUGH */
491: case('!'):
492: /* FALLTHROUGH */
493: case(')'):
494: /* FALLTHROUGH */
495: case(']'):
496: /* FALLTHROUGH */
497: case('}'):
498: if ( ! (TERMP_IGNDELIM & p->flags))
499: p->flags |= TERMP_NOSPACE;
500: break;
501: default:
502: break;
503: }
1.1 kristaps 504:
505: if ( ! (TERMP_NOSPACE & p->flags))
1.11 schwarze 506: buffer(p, ' ');
1.1 kristaps 507:
508: if ( ! (p->flags & TERMP_NONOSPACE))
509: p->flags &= ~TERMP_NOSPACE;
510:
1.14 schwarze 511: for ( ; *word; word++)
1.7 schwarze 512: if ('\\' != *word)
1.11 schwarze 513: encode(p, *word);
1.4 schwarze 514: else
1.11 schwarze 515: do_escaped(p, &word);
1.1 kristaps 516:
1.14 schwarze 517: if (sv[0] && 0 == sv[1])
518: switch (sv[0]) {
519: case('('):
520: /* FALLTHROUGH */
521: case('['):
522: /* FALLTHROUGH */
523: case('{'):
524: p->flags |= TERMP_NOSPACE;
525: break;
526: default:
527: break;
528: }
1.1 kristaps 529: }
530:
531:
532: /*
533: * Insert a single character into the line-buffer. If the buffer's
534: * space is exceeded, then allocate more space by doubling the buffer
535: * size.
536: */
537: static void
1.11 schwarze 538: buffer(struct termp *p, char c)
1.1 kristaps 539: {
540: size_t s;
541:
542: if (p->col + 1 >= p->maxcols) {
543: if (0 == p->maxcols)
544: p->maxcols = 256;
545: s = p->maxcols * 2;
546: p->buf = realloc(p->buf, s);
1.19 ! schwarze 547: if (NULL == p->buf) {
! 548: perror(NULL);
! 549: exit(EXIT_FAILURE);
! 550: }
1.1 kristaps 551: p->maxcols = s;
552: }
553: p->buf[(int)(p->col)++] = c;
554: }
555:
1.4 schwarze 556:
557: static void
1.11 schwarze 558: encode(struct termp *p, char c)
1.4 schwarze 559: {
1.7 schwarze 560:
1.12 schwarze 561: if (' ' != c) {
1.18 schwarze 562: if (p->under) {
563: buffer(p, '_');
564: buffer(p, 8);
565: }
1.12 schwarze 566: if (p->bold) {
1.11 schwarze 567: buffer(p, c);
568: buffer(p, 8);
1.4 schwarze 569: }
570: }
1.11 schwarze 571: buffer(p, c);
1.4 schwarze 572: }
1.16 schwarze 573:
574:
575: size_t
576: term_vspan(const struct roffsu *su)
577: {
578: double r;
579:
580: switch (su->unit) {
581: case (SCALE_CM):
582: r = su->scale * 2;
583: break;
584: case (SCALE_IN):
585: r = su->scale * 6;
586: break;
587: case (SCALE_PC):
588: r = su->scale;
589: break;
590: case (SCALE_PT):
591: r = su->scale / 8;
592: break;
593: case (SCALE_MM):
594: r = su->scale / 1000;
595: break;
596: case (SCALE_VS):
597: r = su->scale;
598: break;
599: default:
600: r = su->scale - 1;
601: break;
602: }
603:
604: if (r < 0.0)
605: r = 0.0;
606: return(/* LINTED */(size_t)
607: r);
608: }
609:
610:
611: size_t
612: term_hspan(const struct roffsu *su)
613: {
614: double r;
615:
616: /* XXX: CM, IN, and PT are approximations. */
617:
618: switch (su->unit) {
619: case (SCALE_CM):
620: r = 4 * su->scale;
621: break;
622: case (SCALE_IN):
623: /* XXX: this is an approximation. */
624: r = 10 * su->scale;
625: break;
626: case (SCALE_PC):
627: r = (10 * su->scale) / 6;
628: break;
629: case (SCALE_PT):
630: r = (10 * su->scale) / 72;
631: break;
632: case (SCALE_MM):
633: r = su->scale / 1000; /* FIXME: double-check. */
634: break;
635: case (SCALE_VS):
636: r = su->scale * 2 - 1; /* FIXME: double-check. */
637: break;
638: default:
639: r = su->scale;
640: break;
641: }
642:
643: if (r < 0.0)
644: r = 0.0;
645: return((size_t)/* LINTED */
646: r);
647: }
648:
649: