Annotation of src/usr.bin/mandoc/term.c, Revision 1.30
1.30 ! schwarze 1: /* $Id: term.c,v 1.29 2010/04/13 22:41:48 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: */
1.20 schwarze 17: #include <sys/types.h>
18:
1.1 kristaps 19: #include <assert.h>
1.20 schwarze 20: #include <ctype.h>
1.1 kristaps 21: #include <stdio.h>
22: #include <stdlib.h>
23: #include <string.h>
1.19 schwarze 24: #include <time.h>
1.1 kristaps 25:
1.15 schwarze 26: #include "chars.h"
1.16 schwarze 27: #include "out.h"
1.1 kristaps 28: #include "term.h"
29: #include "man.h"
30: #include "mdoc.h"
1.16 schwarze 31: #include "main.h"
1.1 kristaps 32:
33: static struct termp *term_alloc(enum termenc);
34: static void term_free(struct termp *);
1.20 schwarze 35: static void spec(struct termp *, const char *, size_t);
36: static void res(struct termp *, const char *, size_t);
37: static void buffera(struct termp *, const char *, size_t);
38: static void bufferc(struct termp *, char);
39: static void adjbuf(struct termp *p, size_t);
40: static void encode(struct termp *, const char *, size_t);
1.1 kristaps 41:
42:
43: void *
44: ascii_alloc(void)
45: {
46:
47: return(term_alloc(TERMENC_ASCII));
48: }
49:
50:
1.13 schwarze 51: void
1.1 kristaps 52: terminal_free(void *arg)
53: {
54:
55: term_free((struct termp *)arg);
56: }
57:
58:
59: static void
60: term_free(struct termp *p)
61: {
62:
63: if (p->buf)
64: free(p->buf);
1.15 schwarze 65: if (p->symtab)
66: chars_free(p->symtab);
1.1 kristaps 67:
68: free(p);
69: }
70:
71:
72: static struct termp *
73: term_alloc(enum termenc enc)
74: {
75: struct termp *p;
76:
1.19 schwarze 77: p = calloc(1, sizeof(struct termp));
78: if (NULL == p) {
79: perror(NULL);
80: exit(EXIT_FAILURE);
81: }
1.30 ! schwarze 82: p->tabwidth = 5;
1.1 kristaps 83: p->enc = enc;
84: return(p);
85: }
86:
87:
88: /*
89: * Flush a line of text. A "line" is loosely defined as being something
90: * that should be followed by a newline, regardless of whether it's
91: * broken apart by newlines getting there. A line can also be a
1.27 schwarze 92: * fragment of a columnar list (`Bl -tag' or `Bl -column'), which does
93: * not have a trailing newline.
1.1 kristaps 94: *
1.27 schwarze 95: * The following flags may be specified:
1.1 kristaps 96: *
97: * - TERMP_NOLPAD: when beginning to write the line, don't left-pad the
98: * offset value. This is useful when doing columnar lists where the
99: * prior column has right-padded.
100: *
101: * - TERMP_NOBREAK: this is the most important and is used when making
102: * columns. In short: don't print a newline and instead pad to the
103: * right margin. Used in conjunction with TERMP_NOLPAD.
104: *
1.9 schwarze 105: * - TERMP_TWOSPACE: when padding, make sure there are at least two
106: * space characters of padding. Otherwise, rather break the line.
107: *
1.6 schwarze 108: * - TERMP_DANGLE: don't newline when TERMP_NOBREAK is specified and
109: * the line is overrun, and don't pad-right if it's underrun.
110: *
111: * - TERMP_HANG: like TERMP_DANGLE, but doesn't newline when
112: * overruning, instead save the position and continue at that point
113: * when the next invocation.
1.1 kristaps 114: *
115: * In-line line breaking:
116: *
117: * If TERMP_NOBREAK is specified and the line overruns the right
118: * margin, it will break and pad-right to the right margin after
119: * writing. If maxrmargin is violated, it will break and continue
1.19 schwarze 120: * writing from the right-margin, which will lead to the above scenario
121: * upon exit. Otherwise, the line will break at the right margin.
1.1 kristaps 122: */
123: void
124: term_flushln(struct termp *p)
125: {
1.19 schwarze 126: int i; /* current input position in p->buf */
127: size_t vis; /* current visual position on output */
128: size_t vbl; /* number of blanks to prepend to output */
1.22 schwarze 129: size_t vend; /* end of word visual position on output */
1.19 schwarze 130: size_t bp; /* visual right border position */
131: int j; /* temporary loop index */
1.22 schwarze 132: int jhy; /* last hyphen before line overflow */
1.19 schwarze 133: size_t maxvis, mmax;
1.1 kristaps 134:
135: /*
136: * First, establish the maximum columns of "visible" content.
137: * This is usually the difference between the right-margin and
138: * an indentation, but can be, for tagged lists or columns, a
1.19 schwarze 139: * small set of values.
1.1 kristaps 140: */
141:
142: assert(p->offset < p->rmargin);
1.9 schwarze 143:
1.26 schwarze 144: maxvis = (int)(p->rmargin - p->offset) - p->overstep < 0 ?
1.19 schwarze 145: /* LINTED */
1.26 schwarze 146: 0 : p->rmargin - p->offset - p->overstep;
147: mmax = (int)(p->maxrmargin - p->offset) - p->overstep < 0 ?
1.19 schwarze 148: /* LINTED */
1.26 schwarze 149: 0 : p->maxrmargin - p->offset - p->overstep;
1.9 schwarze 150:
1.1 kristaps 151: bp = TERMP_NOBREAK & p->flags ? mmax : maxvis;
1.19 schwarze 152:
153: /*
154: * FIXME: if bp is zero, we still output the first word before
155: * breaking the line.
156: */
157:
1.30 ! schwarze 158: vis = i = 0;
1.22 schwarze 159: while (i < (int)p->col) {
160:
161: /*
162: * Choose the number of blanks to prepend: no blank at the
163: * beginning of a line, one between words -- but do not
164: * actually write them yet.
165: */
166: vbl = (size_t)(ASCII_EOS == p->buf[i] ? 2 :
167: (0 == vis ? 0 : 1));
168: vis += vbl;
1.30 ! schwarze 169: vend = vis;
! 170:
! 171: /*
! 172: * Handle literal tab characters.
! 173: */
! 174: for (j = i; j < (int)p->col; j++) {
! 175: if ('\t' != p->buf[j])
! 176: break;
! 177: /* Collapse tab with inter-word spacing. */
! 178: if (vis > 0 && j == i)
! 179: vend = vis - 1;
! 180: vend = (vend/p->tabwidth+1)*p->tabwidth;
! 181: vbl += vend - vis;
! 182: vis = vend;
! 183: }
1.22 schwarze 184:
1.1 kristaps 185: /*
186: * Count up visible word characters. Control sequences
187: * (starting with the CSI) aren't counted. A space
188: * generates a non-printing word, which is valid (the
189: * space is printed according to regular spacing rules).
190: */
191:
192: /* LINTED */
1.30 ! schwarze 193: for (jhy = 0; j < (int)p->col; j++) {
! 194: if ((j && ' ' == p->buf[j]) || '\t' == p->buf[j])
1.1 kristaps 195: break;
196: else if (8 == p->buf[j])
1.22 schwarze 197: vend--;
198: else if (ASCII_EOS != p->buf[j]) {
199: if (vend > vis && vend < bp &&
200: '-' == p->buf[j])
201: jhy = j;
202: vend++;
203: }
1.1 kristaps 204: }
205:
206: /*
1.21 schwarze 207: * Skip empty words. This happens due to the ASCII_EOS
208: * after the end of the final sentence of a paragraph.
209: */
1.22 schwarze 210: if (vend == vis && j == (int)p->col)
1.21 schwarze 211: break;
212:
213: /*
1.23 schwarze 214: * Usually, indent the first line of each paragraph.
215: */
1.29 schwarze 216: if (0 == i && ! (p->flags & TERMP_NOLPAD)) {
217: p->viscol += p->offset;
1.23 schwarze 218: /* LINTED */
219: for (j = 0; j < (int)p->offset; j++)
220: putchar(' ');
1.29 schwarze 221: }
1.23 schwarze 222:
223: /*
1.5 schwarze 224: * Find out whether we would exceed the right margin.
225: * If so, break to the next line. (TODO: hyphenate)
226: * Otherwise, write the chosen number of blanks now.
227: */
1.24 schwarze 228: if (vend > bp && 0 == jhy && vis > vbl) {
1.22 schwarze 229: vend -= vis;
1.5 schwarze 230: putchar('\n');
231: if (TERMP_NOBREAK & p->flags) {
1.29 schwarze 232: p->viscol = p->rmargin;
1.5 schwarze 233: for (j = 0; j < (int)p->rmargin; j++)
234: putchar(' ');
1.22 schwarze 235: vend += p->rmargin - p->offset;
1.5 schwarze 236: } else {
1.29 schwarze 237: p->viscol = p->offset;
1.1 kristaps 238: for (j = 0; j < (int)p->offset; j++)
239: putchar(' ');
1.5 schwarze 240: }
1.26 schwarze 241: /* Remove the p->overstep width. */
1.18 schwarze 242: bp += (int)/* LINTED */
1.26 schwarze 243: p->overstep;
244: p->overstep = 0;
1.5 schwarze 245: } else {
1.29 schwarze 246: p->viscol += vbl;
1.5 schwarze 247: for (j = 0; j < (int)vbl; j++)
1.1 kristaps 248: putchar(' ');
249: }
250:
1.3 schwarze 251: /*
1.30 ! schwarze 252: * Skip leading tabs, they were handled above.
! 253: */
! 254: while (i < (int)p->col && '\t' == p->buf[i])
! 255: i++;
! 256:
! 257: /*
1.5 schwarze 258: * Finally, write out the word.
1.1 kristaps 259: */
260: for ( ; i < (int)p->col; i++) {
1.25 schwarze 261: if (vend > bp && jhy > 0 && i > jhy)
1.30 ! schwarze 262: break;
! 263: if ('\t' == p->buf[i])
1.1 kristaps 264: break;
1.22 schwarze 265: if (' ' == p->buf[i]) {
266: i++;
267: break;
268: }
1.21 schwarze 269: if (ASCII_NBRSP == p->buf[i])
1.20 schwarze 270: putchar(' ');
1.21 schwarze 271: else if (ASCII_EOS != p->buf[i])
1.20 schwarze 272: putchar(p->buf[i]);
1.1 kristaps 273: }
1.29 schwarze 274: p->viscol += vend - vis;
1.22 schwarze 275: vis = vend;
1.1 kristaps 276: }
1.18 schwarze 277:
1.9 schwarze 278: p->col = 0;
1.26 schwarze 279: p->overstep = 0;
1.1 kristaps 280:
1.9 schwarze 281: if ( ! (TERMP_NOBREAK & p->flags)) {
1.29 schwarze 282: p->viscol = 0;
1.28 schwarze 283: putchar('\n');
1.1 kristaps 284: return;
285: }
286:
1.9 schwarze 287: if (TERMP_HANG & p->flags) {
288: /* We need one blank after the tag. */
1.26 schwarze 289: p->overstep = /* LINTED */
1.9 schwarze 290: vis - maxvis + 1;
291:
292: /*
293: * Behave exactly the same way as groff:
294: * If we have overstepped the margin, temporarily move
295: * it to the right and flag the rest of the line to be
296: * shorter.
297: * If we landed right at the margin, be happy.
298: * If we are one step before the margin, temporarily
299: * move it one step LEFT and flag the rest of the line
300: * to be longer.
301: */
1.26 schwarze 302: if (p->overstep >= -1) {
303: assert((int)maxvis + p->overstep >= 0);
1.9 schwarze 304: /* LINTED */
1.26 schwarze 305: maxvis += p->overstep;
1.9 schwarze 306: } else
1.26 schwarze 307: p->overstep = 0;
1.9 schwarze 308:
309: } else if (TERMP_DANGLE & p->flags)
310: return;
1.1 kristaps 311:
1.9 schwarze 312: /* Right-pad. */
313: if (maxvis > vis + /* LINTED */
1.29 schwarze 314: ((TERMP_TWOSPACE & p->flags) ? 1 : 0)) {
315: p->viscol += maxvis - vis;
1.9 schwarze 316: for ( ; vis < maxvis; vis++)
317: putchar(' ');
1.29 schwarze 318: } else { /* ...or newline break. */
1.1 kristaps 319: putchar('\n');
1.29 schwarze 320: p->viscol = p->rmargin;
1.9 schwarze 321: for (i = 0; i < (int)p->rmargin; i++)
322: putchar(' ');
323: }
1.1 kristaps 324: }
325:
326:
327: /*
328: * A newline only breaks an existing line; it won't assert vertical
329: * space. All data in the output buffer is flushed prior to the newline
330: * assertion.
331: */
332: void
333: term_newln(struct termp *p)
334: {
335:
336: p->flags |= TERMP_NOSPACE;
1.29 schwarze 337: if (0 == p->col && 0 == p->viscol) {
1.1 kristaps 338: p->flags &= ~TERMP_NOLPAD;
339: return;
340: }
341: term_flushln(p);
342: p->flags &= ~TERMP_NOLPAD;
343: }
344:
345:
346: /*
347: * Asserts a vertical space (a full, empty line-break between lines).
348: * Note that if used twice, this will cause two blank spaces and so on.
349: * All data in the output buffer is flushed prior to the newline
350: * assertion.
351: */
352: void
353: term_vspace(struct termp *p)
354: {
355:
356: term_newln(p);
1.29 schwarze 357: p->viscol = 0;
1.1 kristaps 358: putchar('\n');
359: }
360:
361:
362: static void
1.20 schwarze 363: spec(struct termp *p, const char *word, size_t len)
1.1 kristaps 364: {
365: const char *rhs;
366: size_t sz;
367:
1.15 schwarze 368: rhs = chars_a2ascii(p->symtab, word, len, &sz);
1.20 schwarze 369: if (rhs)
370: encode(p, rhs, sz);
1.11 schwarze 371: }
372:
373:
374: static void
1.20 schwarze 375: res(struct termp *p, const char *word, size_t len)
1.11 schwarze 376: {
377: const char *rhs;
378: size_t sz;
379:
1.15 schwarze 380: rhs = chars_a2res(p->symtab, word, len, &sz);
1.20 schwarze 381: if (rhs)
382: encode(p, rhs, sz);
383: }
384:
385:
386: void
387: term_fontlast(struct termp *p)
388: {
389: enum termfont f;
1.11 schwarze 390:
1.20 schwarze 391: f = p->fontl;
392: p->fontl = p->fontq[p->fonti];
393: p->fontq[p->fonti] = f;
394: }
395:
396:
397: void
398: term_fontrepl(struct termp *p, enum termfont f)
399: {
400:
401: p->fontl = p->fontq[p->fonti];
402: p->fontq[p->fonti] = f;
1.1 kristaps 403: }
404:
405:
1.20 schwarze 406: void
407: term_fontpush(struct termp *p, enum termfont f)
1.1 kristaps 408: {
1.7 schwarze 409:
1.20 schwarze 410: assert(p->fonti + 1 < 10);
411: p->fontl = p->fontq[p->fonti];
412: p->fontq[++p->fonti] = f;
413: }
1.1 kristaps 414:
415:
1.20 schwarze 416: const void *
417: term_fontq(struct termp *p)
418: {
1.1 kristaps 419:
1.20 schwarze 420: return(&p->fontq[p->fonti]);
421: }
1.1 kristaps 422:
423:
1.20 schwarze 424: enum termfont
425: term_fonttop(struct termp *p)
426: {
1.1 kristaps 427:
1.20 schwarze 428: return(p->fontq[p->fonti]);
429: }
1.7 schwarze 430:
431:
1.20 schwarze 432: void
433: term_fontpopq(struct termp *p, const void *key)
434: {
1.1 kristaps 435:
1.20 schwarze 436: while (p->fonti >= 0 && key != &p->fontq[p->fonti])
437: p->fonti--;
438: assert(p->fonti >= 0);
439: }
1.1 kristaps 440:
441:
1.20 schwarze 442: void
443: term_fontpop(struct termp *p)
444: {
1.1 kristaps 445:
1.20 schwarze 446: assert(p->fonti);
447: p->fonti--;
1.1 kristaps 448: }
449:
450:
451: /*
452: * Handle pwords, partial words, which may be either a single word or a
453: * phrase that cannot be broken down (such as a literal string). This
454: * handles word styling.
455: */
1.7 schwarze 456: void
457: term_word(struct termp *p, const char *word)
1.1 kristaps 458: {
1.20 schwarze 459: const char *sv, *seq;
460: int sz;
461: size_t ssz;
462: enum roffdeco deco;
1.1 kristaps 463:
1.14 schwarze 464: sv = word;
465:
1.20 schwarze 466: if (word[0] && '\0' == word[1])
1.14 schwarze 467: switch (word[0]) {
468: case('.'):
469: /* FALLTHROUGH */
470: case(','):
471: /* FALLTHROUGH */
472: case(';'):
473: /* FALLTHROUGH */
474: case(':'):
475: /* FALLTHROUGH */
476: case('?'):
477: /* FALLTHROUGH */
478: case('!'):
479: /* FALLTHROUGH */
480: case(')'):
481: /* FALLTHROUGH */
482: case(']'):
483: if ( ! (TERMP_IGNDELIM & p->flags))
484: p->flags |= TERMP_NOSPACE;
485: break;
486: default:
487: break;
488: }
1.1 kristaps 489:
490: if ( ! (TERMP_NOSPACE & p->flags))
1.20 schwarze 491: bufferc(p, ' ');
1.1 kristaps 492:
493: if ( ! (p->flags & TERMP_NONOSPACE))
494: p->flags &= ~TERMP_NOSPACE;
495:
1.20 schwarze 496: /* FIXME: use strcspn. */
497:
498: while (*word) {
499: if ('\\' != *word) {
500: encode(p, word, 1);
501: word++;
502: continue;
503: }
504:
505: seq = ++word;
506: sz = a2roffdeco(&deco, &seq, &ssz);
507:
508: switch (deco) {
509: case (DECO_RESERVED):
510: res(p, seq, ssz);
511: break;
512: case (DECO_SPECIAL):
513: spec(p, seq, ssz);
514: break;
515: case (DECO_BOLD):
516: term_fontrepl(p, TERMFONT_BOLD);
517: break;
518: case (DECO_ITALIC):
519: term_fontrepl(p, TERMFONT_UNDER);
520: break;
521: case (DECO_ROMAN):
522: term_fontrepl(p, TERMFONT_NONE);
523: break;
524: case (DECO_PREVIOUS):
525: term_fontlast(p);
526: break;
527: default:
528: break;
529: }
530:
531: word += sz;
532: if (DECO_NOSPACE == deco && '\0' == *word)
533: p->flags |= TERMP_NOSPACE;
534: }
1.1 kristaps 535:
1.14 schwarze 536: if (sv[0] && 0 == sv[1])
537: switch (sv[0]) {
538: case('('):
539: /* FALLTHROUGH */
540: case('['):
541: p->flags |= TERMP_NOSPACE;
542: break;
543: default:
544: break;
545: }
1.1 kristaps 546: }
547:
548:
549: static void
1.20 schwarze 550: adjbuf(struct termp *p, size_t sz)
1.1 kristaps 551: {
552:
1.20 schwarze 553: if (0 == p->maxcols)
554: p->maxcols = 1024;
555: while (sz >= p->maxcols)
556: p->maxcols <<= 2;
557:
558: p->buf = realloc(p->buf, p->maxcols);
559: if (NULL == p->buf) {
560: perror(NULL);
561: exit(EXIT_FAILURE);
1.1 kristaps 562: }
563: }
564:
1.4 schwarze 565:
566: static void
1.20 schwarze 567: buffera(struct termp *p, const char *word, size_t sz)
568: {
569:
570: if (p->col + sz >= p->maxcols)
571: adjbuf(p, p->col + sz);
572:
573: memcpy(&p->buf[(int)p->col], word, sz);
574: p->col += sz;
575: }
576:
577:
578: static void
579: bufferc(struct termp *p, char c)
580: {
581:
582: if (p->col + 1 >= p->maxcols)
583: adjbuf(p, p->col + 1);
584:
585: p->buf[(int)p->col++] = c;
586: }
587:
588:
589: static void
590: encode(struct termp *p, const char *word, size_t sz)
1.4 schwarze 591: {
1.20 schwarze 592: enum termfont f;
593: int i;
594:
595: /*
596: * Encode and buffer a string of characters. If the current
597: * font mode is unset, buffer directly, else encode then buffer
598: * character by character.
599: */
600:
601: if (TERMFONT_NONE == (f = term_fonttop(p))) {
602: buffera(p, word, sz);
603: return;
604: }
605:
606: for (i = 0; i < (int)sz; i++) {
607: if ( ! isgraph((u_char)word[i])) {
608: bufferc(p, word[i]);
609: continue;
1.4 schwarze 610: }
1.20 schwarze 611:
612: if (TERMFONT_UNDER == f)
613: bufferc(p, '_');
614: else
615: bufferc(p, word[i]);
616:
617: bufferc(p, 8);
618: bufferc(p, word[i]);
1.4 schwarze 619: }
620: }
1.16 schwarze 621:
622:
623: size_t
624: term_vspan(const struct roffsu *su)
625: {
626: double r;
627:
628: switch (su->unit) {
629: case (SCALE_CM):
630: r = su->scale * 2;
631: break;
632: case (SCALE_IN):
633: r = su->scale * 6;
634: break;
635: case (SCALE_PC):
636: r = su->scale;
637: break;
638: case (SCALE_PT):
639: r = su->scale / 8;
640: break;
641: case (SCALE_MM):
642: r = su->scale / 1000;
643: break;
644: case (SCALE_VS):
645: r = su->scale;
646: break;
647: default:
648: r = su->scale - 1;
649: break;
650: }
651:
652: if (r < 0.0)
653: r = 0.0;
654: return(/* LINTED */(size_t)
655: r);
656: }
657:
658:
659: size_t
660: term_hspan(const struct roffsu *su)
661: {
662: double r;
663:
664: /* XXX: CM, IN, and PT are approximations. */
665:
666: switch (su->unit) {
667: case (SCALE_CM):
668: r = 4 * su->scale;
669: break;
670: case (SCALE_IN):
671: /* XXX: this is an approximation. */
672: r = 10 * su->scale;
673: break;
674: case (SCALE_PC):
675: r = (10 * su->scale) / 6;
676: break;
677: case (SCALE_PT):
678: r = (10 * su->scale) / 72;
679: break;
680: case (SCALE_MM):
681: r = su->scale / 1000; /* FIXME: double-check. */
682: break;
683: case (SCALE_VS):
684: r = su->scale * 2 - 1; /* FIXME: double-check. */
685: break;
686: default:
687: r = su->scale;
688: break;
689: }
690:
691: if (r < 0.0)
692: r = 0.0;
693: return((size_t)/* LINTED */
694: r);
695: }
696:
697: