Annotation of mandoc/html.c, Revision 1.76
1.76 ! kristaps 1: /* $Id: html.c,v 1.75 2009/10/31 06:10:57 kristaps Exp $ */
1.1 kristaps 2: /*
1.29 kristaps 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
1.29 kristaps 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: *
1.29 kristaps 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: */
1.41 kristaps 17: #include <sys/types.h>
1.30 kristaps 18:
1.1 kristaps 19: #include <assert.h>
1.68 kristaps 20: #include <ctype.h>
1.76 ! kristaps 21: #include <stdarg.h>
1.29 kristaps 22: #include <stdio.h>
1.63 kristaps 23: #include <stdint.h>
1.1 kristaps 24: #include <stdlib.h>
1.33 kristaps 25: #include <string.h>
1.45 kristaps 26: #include <unistd.h>
1.1 kristaps 27:
1.58 kristaps 28: #include "out.h"
1.32 kristaps 29: #include "chars.h"
1.51 kristaps 30: #include "html.h"
1.64 kristaps 31: #include "main.h"
1.2 kristaps 32:
1.63 kristaps 33: #define UNCONST(a) ((void *)(uintptr_t)(const void *)(a))
34:
1.29 kristaps 35: #define DOCTYPE "-//W3C//DTD HTML 4.01//EN"
36: #define DTD "http://www.w3.org/TR/html4/strict.dtd"
1.8 kristaps 37:
1.29 kristaps 38: struct htmldata {
1.63 kristaps 39: const char *name;
1.29 kristaps 40: int flags;
1.30 kristaps 41: #define HTML_CLRLINE (1 << 0)
42: #define HTML_NOSTACK (1 << 1)
1.29 kristaps 43: };
1.7 kristaps 44:
1.29 kristaps 45: static const struct htmldata htmltags[TAG_MAX] = {
1.30 kristaps 46: {"html", HTML_CLRLINE}, /* TAG_HTML */
47: {"head", HTML_CLRLINE}, /* TAG_HEAD */
48: {"body", HTML_CLRLINE}, /* TAG_BODY */
49: {"meta", HTML_CLRLINE | HTML_NOSTACK}, /* TAG_META */
1.33 kristaps 50: {"title", HTML_CLRLINE}, /* TAG_TITLE */
1.30 kristaps 51: {"div", HTML_CLRLINE}, /* TAG_DIV */
1.29 kristaps 52: {"h1", 0}, /* TAG_H1 */
53: {"h2", 0}, /* TAG_H2 */
1.30 kristaps 54: {"p", HTML_CLRLINE}, /* TAG_P */
1.29 kristaps 55: {"span", 0}, /* TAG_SPAN */
1.30 kristaps 56: {"link", HTML_CLRLINE | HTML_NOSTACK}, /* TAG_LINK */
57: {"br", HTML_CLRLINE | HTML_NOSTACK}, /* TAG_LINK */
58: {"a", 0}, /* TAG_A */
1.33 kristaps 59: {"table", HTML_CLRLINE}, /* TAG_TABLE */
60: {"col", HTML_CLRLINE | HTML_NOSTACK}, /* TAG_COL */
61: {"tr", HTML_CLRLINE}, /* TAG_TR */
62: {"td", HTML_CLRLINE}, /* TAG_TD */
1.34 kristaps 63: {"li", HTML_CLRLINE}, /* TAG_LI */
64: {"ul", HTML_CLRLINE}, /* TAG_UL */
65: {"ol", HTML_CLRLINE}, /* TAG_OL */
1.41 kristaps 66: {"base", HTML_CLRLINE | HTML_NOSTACK}, /* TAG_BASE */
1.29 kristaps 67: };
1.10 kristaps 68:
1.29 kristaps 69: static const char *const htmlattrs[ATTR_MAX] = {
70: "http-equiv",
71: "content",
72: "name",
73: "rel",
74: "href",
75: "type",
76: "media",
1.33 kristaps 77: "class",
78: "style",
79: "width",
80: "valign",
1.54 kristaps 81: "target",
1.57 kristaps 82: "id",
1.67 kristaps 83: "summary",
1.29 kristaps 84: };
1.10 kristaps 85:
1.33 kristaps 86: #ifdef __linux__
1.43 kristaps 87: extern int getsubopt(char **, char * const *, char **);
1.33 kristaps 88: #endif
1.29 kristaps 89:
90: void *
1.43 kristaps 91: html_alloc(char *outopts)
1.10 kristaps 92: {
1.30 kristaps 93: struct html *h;
1.63 kristaps 94: const char *toks[4];
95: char *v;
1.43 kristaps 96:
97: toks[0] = "style";
1.53 kristaps 98: toks[1] = "man";
1.54 kristaps 99: toks[2] = "includes";
100: toks[3] = NULL;
1.30 kristaps 101:
1.72 kristaps 102: h = calloc(1, sizeof(struct html));
103: if (NULL == h) {
1.75 kristaps 104: perror(NULL);
1.72 kristaps 105: exit(EXIT_FAILURE);
106: }
1.10 kristaps 107:
1.66 kristaps 108: h->tags.head = NULL;
109: h->ords.head = NULL;
1.72 kristaps 110: h->symtab = chars_init(CHARS_HTML);
1.41 kristaps 111:
1.47 kristaps 112: while (outopts && *outopts)
1.63 kristaps 113: switch (getsubopt(&outopts, UNCONST(toks), &v)) {
1.43 kristaps 114: case (0):
115: h->style = v;
116: break;
117: case (1):
1.53 kristaps 118: h->base_man = v;
1.43 kristaps 119: break;
1.54 kristaps 120: case (2):
121: h->base_includes = v;
122: break;
1.43 kristaps 123: default:
124: break;
125: }
126:
1.30 kristaps 127: return(h);
1.29 kristaps 128: }
1.10 kristaps 129:
1.33 kristaps 130:
1.29 kristaps 131: void
132: html_free(void *p)
133: {
1.30 kristaps 134: struct tag *tag;
1.37 kristaps 135: struct ord *ord;
1.30 kristaps 136: struct html *h;
137:
138: h = (struct html *)p;
1.10 kristaps 139:
1.66 kristaps 140: while ((ord = h->ords.head) != NULL) {
141: h->ords.head = ord->next;
1.37 kristaps 142: free(ord);
143: }
144:
1.66 kristaps 145: while ((tag = h->tags.head) != NULL) {
146: h->tags.head = tag->next;
1.30 kristaps 147: free(tag);
148: }
1.36 kristaps 149:
150: if (h->symtab)
151: chars_free(h->symtab);
1.53 kristaps 152:
1.30 kristaps 153: free(h);
1.10 kristaps 154: }
1.2 kristaps 155:
1.33 kristaps 156:
1.51 kristaps 157: void
1.29 kristaps 158: print_gen_head(struct html *h)
159: {
1.41 kristaps 160: struct htmlpair tag[4];
161:
162: tag[0].key = ATTR_HTTPEQUIV;
163: tag[0].val = "Content-Type";
164: tag[1].key = ATTR_CONTENT;
165: tag[1].val = "text/html; charset=utf-8";
166: print_otag(h, TAG_META, 2, tag);
167:
168: tag[0].key = ATTR_NAME;
169: tag[0].val = "resource-type";
170: tag[1].key = ATTR_CONTENT;
171: tag[1].val = "document";
172: print_otag(h, TAG_META, 2, tag);
173:
174: if (h->style) {
175: tag[0].key = ATTR_REL;
176: tag[0].val = "stylesheet";
177: tag[1].key = ATTR_HREF;
178: tag[1].val = h->style;
179: tag[2].key = ATTR_TYPE;
180: tag[2].val = "text/css";
181: tag[3].key = ATTR_MEDIA;
182: tag[3].val = "all";
183: print_otag(h, TAG_LINK, 4, tag);
184: }
1.4 kristaps 185: }
186:
1.33 kristaps 187:
1.29 kristaps 188: static void
1.32 kristaps 189: print_spec(struct html *h, const char *p, int len)
190: {
191: const char *rhs;
192: size_t sz;
193:
194: rhs = chars_a2ascii(h->symtab, p, (size_t)len, &sz);
195:
196: if (NULL == rhs)
197: return;
1.76 ! kristaps 198: fwrite(rhs, 1, sz, stdout);
1.32 kristaps 199: }
200:
1.33 kristaps 201:
1.32 kristaps 202: static void
203: print_res(struct html *h, const char *p, int len)
204: {
205: const char *rhs;
206: size_t sz;
207:
208: rhs = chars_a2res(h->symtab, p, (size_t)len, &sz);
209:
210: if (NULL == rhs)
211: return;
1.76 ! kristaps 212: fwrite(rhs, 1, sz, stdout);
1.32 kristaps 213: }
214:
1.33 kristaps 215:
1.32 kristaps 216: static void
217: print_escape(struct html *h, const char **p)
218: {
219: int j, type;
220: const char *wp;
221:
222: wp = *p;
223: type = 1;
224:
225: if (0 == *(++wp)) {
226: *p = wp;
227: return;
228: }
229:
230: if ('(' == *wp) {
231: wp++;
232: if (0 == *wp || 0 == *(wp + 1)) {
233: *p = 0 == *wp ? wp : wp + 1;
234: return;
235: }
236:
237: print_spec(h, wp, 2);
238: *p = ++wp;
239: return;
240:
241: } else if ('*' == *wp) {
242: if (0 == *(++wp)) {
243: *p = wp;
244: return;
245: }
246:
247: switch (*wp) {
248: case ('('):
249: wp++;
250: if (0 == *wp || 0 == *(wp + 1)) {
251: *p = 0 == *wp ? wp : wp + 1;
252: return;
253: }
254:
255: print_res(h, wp, 2);
256: *p = ++wp;
257: return;
258: case ('['):
259: type = 0;
260: break;
261: default:
262: print_res(h, wp, 1);
263: *p = wp;
264: return;
265: }
266:
267: } else if ('f' == *wp) {
268: if (0 == *(++wp)) {
269: *p = wp;
270: return;
271: }
272:
273: switch (*wp) {
274: case ('B'):
275: /* TODO */
276: break;
277: case ('I'):
278: /* TODO */
279: break;
280: case ('P'):
281: /* FALLTHROUGH */
282: case ('R'):
283: /* TODO */
284: break;
285: default:
286: break;
287: }
288:
289: *p = wp;
290: return;
291:
292: } else if ('[' != *wp) {
293: print_spec(h, wp, 1);
294: *p = wp;
295: return;
296: }
297:
298: wp++;
299: for (j = 0; *wp && ']' != *wp; wp++, j++)
300: /* Loop... */ ;
301:
302: if (0 == *wp) {
303: *p = wp;
304: return;
305: }
306:
307: if (type)
308: print_spec(h, wp - j, j);
309: else
310: print_res(h, wp - j, j);
311:
312: *p = wp;
313: }
314:
1.9 kristaps 315:
1.29 kristaps 316: static void
1.32 kristaps 317: print_encode(struct html *h, const char *p)
1.29 kristaps 318: {
1.14 kristaps 319:
1.32 kristaps 320: for (; *p; p++) {
1.34 kristaps 321: if ('\\' == *p) {
322: print_escape(h, &p);
323: continue;
324: }
325: switch (*p) {
326: case ('<'):
327: printf("<");
328: break;
329: case ('>'):
330: printf(">");
331: break;
332: case ('&'):
333: printf("&");
334: break;
335: default:
1.32 kristaps 336: putchar(*p);
1.34 kristaps 337: break;
1.32 kristaps 338: }
339: }
1.14 kristaps 340: }
341:
342:
1.51 kristaps 343: struct tag *
1.29 kristaps 344: print_otag(struct html *h, enum htmltag tag,
345: int sz, const struct htmlpair *p)
1.14 kristaps 346: {
1.29 kristaps 347: int i;
1.30 kristaps 348: struct tag *t;
349:
350: if ( ! (HTML_NOSTACK & htmltags[tag].flags)) {
1.72 kristaps 351: t = malloc(sizeof(struct tag));
352: if (NULL == t) {
1.75 kristaps 353: perror(NULL);
1.72 kristaps 354: exit(EXIT_FAILURE);
355: }
1.30 kristaps 356: t->tag = tag;
1.66 kristaps 357: t->next = h->tags.head;
358: h->tags.head = t;
1.30 kristaps 359: } else
360: t = NULL;
1.29 kristaps 361:
362: if ( ! (HTML_NOSPACE & h->flags))
1.30 kristaps 363: if ( ! (HTML_CLRLINE & htmltags[tag].flags))
1.29 kristaps 364: printf(" ");
365:
366: printf("<%s", htmltags[tag].name);
367: for (i = 0; i < sz; i++) {
368: printf(" %s=\"", htmlattrs[p[i].key]);
369: assert(p->val);
1.32 kristaps 370: print_encode(h, p[i].val);
1.29 kristaps 371: printf("\"");
372: }
373: printf(">");
1.14 kristaps 374:
1.29 kristaps 375: h->flags |= HTML_NOSPACE;
1.30 kristaps 376: if (HTML_CLRLINE & htmltags[tag].flags)
377: h->flags |= HTML_NEWLINE;
378: else
379: h->flags &= ~HTML_NEWLINE;
1.14 kristaps 380:
1.30 kristaps 381: return(t);
1.14 kristaps 382: }
383:
384:
385: /* ARGSUSED */
1.29 kristaps 386: static void
387: print_ctag(struct html *h, enum htmltag tag)
1.14 kristaps 388: {
389:
1.29 kristaps 390: printf("</%s>", htmltags[tag].name);
1.71 kristaps 391: if (HTML_CLRLINE & htmltags[tag].flags) {
1.29 kristaps 392: h->flags |= HTML_NOSPACE;
1.30 kristaps 393: h->flags |= HTML_NEWLINE;
1.71 kristaps 394: printf("\n");
395: } else
1.30 kristaps 396: h->flags &= ~HTML_NEWLINE;
1.14 kristaps 397: }
398:
399:
1.29 kristaps 400: /* ARGSUSED */
1.51 kristaps 401: void
1.29 kristaps 402: print_gen_doctype(struct html *h)
1.1 kristaps 403: {
1.29 kristaps 404:
1.46 kristaps 405: printf("<!DOCTYPE HTML PUBLIC \"%s\" \"%s\">", DOCTYPE, DTD);
1.1 kristaps 406: }
407:
408:
1.51 kristaps 409: void
1.29 kristaps 410: print_text(struct html *h, const char *p)
1.1 kristaps 411: {
412:
1.29 kristaps 413: if (*p && 0 == *(p + 1))
414: switch (*p) {
415: case('.'):
416: /* FALLTHROUGH */
417: case(','):
418: /* FALLTHROUGH */
419: case(';'):
420: /* FALLTHROUGH */
421: case(':'):
422: /* FALLTHROUGH */
423: case('?'):
424: /* FALLTHROUGH */
425: case('!'):
426: /* FALLTHROUGH */
427: case(')'):
428: /* FALLTHROUGH */
429: case(']'):
430: /* FALLTHROUGH */
431: case('}'):
1.52 kristaps 432: if ( ! (HTML_IGNDELIM & h->flags))
433: h->flags |= HTML_NOSPACE;
1.30 kristaps 434: break;
1.29 kristaps 435: default:
436: break;
437: }
1.1 kristaps 438:
1.29 kristaps 439: if ( ! (h->flags & HTML_NOSPACE))
440: printf(" ");
1.30 kristaps 441:
1.29 kristaps 442: h->flags &= ~HTML_NOSPACE;
1.30 kristaps 443: h->flags &= ~HTML_NEWLINE;
1.1 kristaps 444:
1.29 kristaps 445: if (p)
1.32 kristaps 446: print_encode(h, p);
1.8 kristaps 447:
1.29 kristaps 448: if (*p && 0 == *(p + 1))
449: switch (*p) {
450: case('('):
451: /* FALLTHROUGH */
452: case('['):
453: /* FALLTHROUGH */
454: case('{'):
455: h->flags |= HTML_NOSPACE;
1.30 kristaps 456: break;
1.29 kristaps 457: default:
458: break;
459: }
1.1 kristaps 460: }
1.30 kristaps 461:
462:
1.51 kristaps 463: void
1.30 kristaps 464: print_tagq(struct html *h, const struct tag *until)
465: {
466: struct tag *tag;
467:
1.66 kristaps 468: while ((tag = h->tags.head) != NULL) {
1.30 kristaps 469: print_ctag(h, tag->tag);
1.66 kristaps 470: h->tags.head = tag->next;
1.30 kristaps 471: free(tag);
472: if (until && tag == until)
473: return;
474: }
475: }
476:
477:
1.51 kristaps 478: void
1.30 kristaps 479: print_stagq(struct html *h, const struct tag *suntil)
480: {
481: struct tag *tag;
482:
1.66 kristaps 483: while ((tag = h->tags.head) != NULL) {
1.30 kristaps 484: if (suntil && tag == suntil)
485: return;
486: print_ctag(h, tag->tag);
1.66 kristaps 487: h->tags.head = tag->next;
1.30 kristaps 488: free(tag);
489: }
490: }
1.55 kristaps 491:
492:
493: void
494: bufinit(struct html *h)
495: {
496:
497: h->buf[0] = '\0';
498: h->buflen = 0;
499: }
500:
501:
502: void
1.58 kristaps 503: bufcat_style(struct html *h, const char *key, const char *val)
504: {
505:
506: bufcat(h, key);
507: bufncat(h, ":", 1);
508: bufcat(h, val);
509: bufncat(h, ";", 1);
510: }
511:
512:
513: void
1.55 kristaps 514: bufcat(struct html *h, const char *p)
515: {
516:
517: bufncat(h, p, strlen(p));
518: }
519:
520:
521: void
522: buffmt(struct html *h, const char *fmt, ...)
523: {
524: va_list ap;
525:
526: va_start(ap, fmt);
1.56 kristaps 527: (void)vsnprintf(h->buf + (int)h->buflen,
1.55 kristaps 528: BUFSIZ - h->buflen - 1, fmt, ap);
529: va_end(ap);
530: h->buflen = strlen(h->buf);
531: }
532:
533:
534: void
535: bufncat(struct html *h, const char *p, size_t sz)
536: {
537:
538: if (h->buflen + sz > BUFSIZ - 1)
539: sz = BUFSIZ - 1 - h->buflen;
540:
541: (void)strncat(h->buf, p, sz);
542: h->buflen += sz;
543: }
544:
545:
546: void
547: buffmt_includes(struct html *h, const char *name)
548: {
549: const char *p, *pp;
550:
551: pp = h->base_includes;
1.61 kristaps 552:
553: while (NULL != (p = strchr(pp, '%'))) {
1.56 kristaps 554: bufncat(h, pp, (size_t)(p - pp));
1.55 kristaps 555: switch (*(p + 1)) {
556: case('I'):
557: bufcat(h, name);
558: break;
559: default:
560: bufncat(h, p, 2);
561: break;
562: }
563: pp = p + 2;
564: }
565: if (pp)
566: bufcat(h, pp);
567: }
568:
569:
570: void
571: buffmt_man(struct html *h,
572: const char *name, const char *sec)
573: {
574: const char *p, *pp;
575:
576: pp = h->base_man;
1.61 kristaps 577:
578: /* LINTED */
579: while (NULL != (p = strchr(pp, '%'))) {
1.56 kristaps 580: bufncat(h, pp, (size_t)(p - pp));
1.55 kristaps 581: switch (*(p + 1)) {
582: case('S'):
1.58 kristaps 583: bufcat(h, sec ? sec : "1");
1.55 kristaps 584: break;
585: case('N'):
1.58 kristaps 586: buffmt(h, name);
1.55 kristaps 587: break;
588: default:
589: bufncat(h, p, 2);
590: break;
591: }
592: pp = p + 2;
593: }
594: if (pp)
595: bufcat(h, pp);
596: }
1.58 kristaps 597:
598:
599: void
600: bufcat_su(struct html *h, const char *p, const struct roffsu *su)
601: {
1.62 kristaps 602: double v;
1.63 kristaps 603: const char *u;
1.58 kristaps 604:
605: v = su->scale;
606:
607: switch (su->unit) {
608: case (SCALE_CM):
609: u = "cm";
610: break;
611: case (SCALE_IN):
612: u = "in";
613: break;
614: case (SCALE_PC):
615: u = "pc";
616: break;
617: case (SCALE_PT):
618: u = "pt";
619: break;
1.59 kristaps 620: case (SCALE_EM):
621: u = "em";
622: break;
1.58 kristaps 623: case (SCALE_MM):
624: if (0 == (v /= 100))
625: v = 1;
626: u = "em";
627: break;
1.59 kristaps 628: case (SCALE_EN):
629: u = "ex";
630: break;
631: case (SCALE_BU):
632: u = "ex";
633: break;
1.58 kristaps 634: case (SCALE_VS):
635: u = "em";
636: break;
637: default:
638: u = "ex";
639: break;
640: }
641:
1.62 kristaps 642: if (su->pt)
643: buffmt(h, "%s: %f%s;", p, v, u);
644: else
645: /* LINTED */
646: buffmt(h, "%s: %d%s;", p, (int)v, u);
1.58 kristaps 647: }
1.65 kristaps 648:
1.68 kristaps 649:
650: void
1.70 kristaps 651: html_idcat(char *dst, const char *src, int sz)
1.68 kristaps 652: {
1.70 kristaps 653: int ssz;
1.68 kristaps 654:
655: assert(sz);
656:
657: /* Cf. <http://www.w3.org/TR/html4/types.html#h-6.2>. */
658:
1.70 kristaps 659: for ( ; *dst != '\0' && sz; dst++, sz--)
1.68 kristaps 660: /* Jump to end. */ ;
661:
1.70 kristaps 662: assert(sz > 2);
1.68 kristaps 663:
1.70 kristaps 664: /* We can't start with a number (bah). */
1.68 kristaps 665:
1.70 kristaps 666: *dst++ = 'x';
1.68 kristaps 667: *dst = '\0';
1.70 kristaps 668: sz--;
669:
670: for ( ; *src != '\0' && sz > 1; src++) {
1.73 kristaps 671: ssz = snprintf(dst, (size_t)sz, "%.2x", *src);
1.70 kristaps 672: sz -= ssz;
673: dst += ssz;
674: }
1.68 kristaps 675: }
CVSweb