Annotation of mandoc/html.c, Revision 1.255
1.255 ! schwarze 1: /* $Id: html.c,v 1.254 2019/03/03 13:02:11 schwarze Exp $ */
1.1 kristaps 2: /*
1.176 schwarze 3: * Copyright (c) 2008-2011, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
1.249 schwarze 4: * Copyright (c) 2011-2015, 2017-2019 Ingo Schwarze <schwarze@openbsd.org>
1.1 kristaps 5: *
6: * Permission to use, copy, modify, and distribute this software for any
1.29 kristaps 7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
1.1 kristaps 9: *
1.186 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.29 kristaps 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.186 schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.29 kristaps 13: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
1.1 kristaps 17: */
1.92 kristaps 18: #include "config.h"
19:
1.41 kristaps 20: #include <sys/types.h>
1.240 schwarze 21: #include <sys/stat.h>
1.30 kristaps 22:
1.1 kristaps 23: #include <assert.h>
1.68 kristaps 24: #include <ctype.h>
1.76 kristaps 25: #include <stdarg.h>
1.229 schwarze 26: #include <stddef.h>
1.29 kristaps 27: #include <stdio.h>
1.63 kristaps 28: #include <stdint.h>
1.1 kristaps 29: #include <stdlib.h>
1.33 kristaps 30: #include <string.h>
1.45 kristaps 31: #include <unistd.h>
1.1 kristaps 32:
1.210 schwarze 33: #include "mandoc_aux.h"
1.229 schwarze 34: #include "mandoc_ohash.h"
1.100 kristaps 35: #include "mandoc.h"
1.210 schwarze 36: #include "roff.h"
1.58 kristaps 37: #include "out.h"
1.51 kristaps 38: #include "html.h"
1.186 schwarze 39: #include "manconf.h"
1.64 kristaps 40: #include "main.h"
1.63 kristaps 41:
1.29 kristaps 42: struct htmldata {
1.63 kristaps 43: const char *name;
1.29 kristaps 44: int flags;
1.196 schwarze 45: #define HTML_NOSTACK (1 << 0)
46: #define HTML_AUTOCLOSE (1 << 1)
47: #define HTML_NLBEFORE (1 << 2)
48: #define HTML_NLBEGIN (1 << 3)
49: #define HTML_NLEND (1 << 4)
50: #define HTML_NLAFTER (1 << 5)
51: #define HTML_NLAROUND (HTML_NLBEFORE | HTML_NLAFTER)
52: #define HTML_NLINSIDE (HTML_NLBEGIN | HTML_NLEND)
53: #define HTML_NLALL (HTML_NLAROUND | HTML_NLINSIDE)
54: #define HTML_INDENT (1 << 6)
55: #define HTML_NOINDENT (1 << 7)
1.29 kristaps 56: };
1.7 kristaps 57:
1.29 kristaps 58: static const struct htmldata htmltags[TAG_MAX] = {
1.196 schwarze 59: {"html", HTML_NLALL},
60: {"head", HTML_NLALL | HTML_INDENT},
61: {"body", HTML_NLALL},
62: {"meta", HTML_NOSTACK | HTML_AUTOCLOSE | HTML_NLALL},
63: {"title", HTML_NLAROUND},
64: {"div", HTML_NLAROUND},
1.225 schwarze 65: {"div", 0},
1.253 schwarze 66: {"section", HTML_NLALL},
1.196 schwarze 67: {"h1", HTML_NLAROUND},
68: {"h2", HTML_NLAROUND},
69: {"span", 0},
70: {"link", HTML_NOSTACK | HTML_AUTOCLOSE | HTML_NLALL},
71: {"br", HTML_NOSTACK | HTML_AUTOCLOSE | HTML_NLALL},
72: {"a", 0},
73: {"table", HTML_NLALL | HTML_INDENT},
74: {"tr", HTML_NLALL | HTML_INDENT},
75: {"td", HTML_NLAROUND},
76: {"li", HTML_NLAROUND | HTML_INDENT},
77: {"ul", HTML_NLALL | HTML_INDENT},
78: {"ol", HTML_NLALL | HTML_INDENT},
79: {"dl", HTML_NLALL | HTML_INDENT},
80: {"dt", HTML_NLAROUND},
81: {"dd", HTML_NLAROUND | HTML_INDENT},
1.249 schwarze 82: {"p", HTML_NLAROUND | HTML_INDENT},
1.196 schwarze 83: {"pre", HTML_NLALL | HTML_NOINDENT},
1.207 schwarze 84: {"var", 0},
1.206 schwarze 85: {"cite", 0},
1.196 schwarze 86: {"b", 0},
87: {"i", 0},
88: {"code", 0},
89: {"small", 0},
90: {"style", HTML_NLALL | HTML_INDENT},
91: {"math", HTML_NLALL | HTML_INDENT},
92: {"mrow", 0},
93: {"mi", 0},
1.215 schwarze 94: {"mn", 0},
1.196 schwarze 95: {"mo", 0},
96: {"msup", 0},
97: {"msub", 0},
98: {"msubsup", 0},
99: {"mfrac", 0},
100: {"msqrt", 0},
101: {"mfenced", 0},
102: {"mtable", 0},
103: {"mtr", 0},
104: {"mtd", 0},
105: {"munderover", 0},
106: {"munder", 0},
107: {"mover", 0},
1.90 kristaps 108: };
109:
1.229 schwarze 110: /* Avoid duplicate HTML id= attributes. */
111: static struct ohash id_unique;
112:
1.254 schwarze 113: static void html_reset_internal(struct html *);
1.197 schwarze 114: static void print_byte(struct html *, char);
115: static void print_endword(struct html *);
116: static void print_indent(struct html *);
117: static void print_word(struct html *, const char *);
118:
1.184 schwarze 119: static void print_ctag(struct html *, struct tag *);
1.197 schwarze 120: static int print_escape(struct html *, char);
1.195 schwarze 121: static int print_encode(struct html *, const char *, const char *, int);
122: static void print_href(struct html *, const char *, const char *, int);
1.255 ! schwarze 123: static void print_metaf(struct html *);
1.82 kristaps 124:
1.156 schwarze 125:
1.180 schwarze 126: void *
1.191 schwarze 127: html_alloc(const struct manoutput *outopts)
1.10 kristaps 128: {
1.30 kristaps 129: struct html *h;
130:
1.128 kristaps 131: h = mandoc_calloc(1, sizeof(struct html));
1.10 kristaps 132:
1.204 schwarze 133: h->tag = NULL;
1.186 schwarze 134: h->style = outopts->style;
1.240 schwarze 135: if ((h->base_man1 = outopts->man) == NULL)
136: h->base_man2 = NULL;
137: else if ((h->base_man2 = strchr(h->base_man1, ';')) != NULL)
138: *h->base_man2++ = '\0';
1.186 schwarze 139: h->base_includes = outopts->includes;
140: if (outopts->fragment)
141: h->oflags |= HTML_FRAGMENT;
1.241 schwarze 142: if (outopts->toc)
143: h->oflags |= HTML_TOC;
1.43 kristaps 144:
1.229 schwarze 145: mandoc_ohash_init(&id_unique, 4, 0);
146:
1.188 schwarze 147: return h;
1.29 kristaps 148: }
1.10 kristaps 149:
1.254 schwarze 150: static void
151: html_reset_internal(struct html *h)
1.29 kristaps 152: {
1.30 kristaps 153: struct tag *tag;
1.229 schwarze 154: char *cp;
155: unsigned int slot;
1.30 kristaps 156:
1.204 schwarze 157: while ((tag = h->tag) != NULL) {
158: h->tag = tag->next;
1.30 kristaps 159: free(tag);
160: }
1.229 schwarze 161: cp = ohash_first(&id_unique, &slot);
162: while (cp != NULL) {
163: free(cp);
164: cp = ohash_next(&id_unique, &slot);
165: }
166: ohash_delete(&id_unique);
1.254 schwarze 167: }
168:
169: void
170: html_reset(void *p)
171: {
172: html_reset_internal(p);
173: mandoc_ohash_init(&id_unique, 4, 0);
174: }
175:
176: void
177: html_free(void *p)
178: {
179: html_reset_internal(p);
180: free(p);
1.10 kristaps 181: }
1.2 kristaps 182:
1.51 kristaps 183: void
1.29 kristaps 184: print_gen_head(struct html *h)
185: {
1.165 kristaps 186: struct tag *t;
1.41 kristaps 187:
1.194 schwarze 188: print_otag(h, TAG_META, "?", "charset", "utf-8");
1.222 schwarze 189: if (h->style != NULL) {
190: print_otag(h, TAG_LINK, "?h??", "rel", "stylesheet",
191: h->style, "type", "text/css", "media", "all");
192: return;
193: }
1.165 kristaps 194:
1.168 kristaps 195: /*
1.222 schwarze 196: * Print a minimal embedded style sheet.
1.168 kristaps 197: */
1.196 schwarze 198:
1.194 schwarze 199: t = print_otag(h, TAG_STYLE, "");
1.196 schwarze 200: print_text(h, "table.head, table.foot { width: 100%; }");
1.197 schwarze 201: print_endline(h);
1.196 schwarze 202: print_text(h, "td.head-rtitle, td.foot-os { text-align: right; }");
1.197 schwarze 203: print_endline(h);
1.196 schwarze 204: print_text(h, "td.head-vol { text-align: center; }");
1.197 schwarze 205: print_endline(h);
1.198 schwarze 206: print_text(h, "div.Pp { margin: 1ex 0ex; }");
1.225 schwarze 207: print_endline(h);
208: print_text(h, "div.Nd, div.Bf, div.Op { display: inline; }");
1.226 schwarze 209: print_endline(h);
1.227 schwarze 210: print_text(h, "span.Pa, span.Ad { font-style: italic; }");
1.228 schwarze 211: print_endline(h);
212: print_text(h, "span.Ms { font-weight: bold; }");
1.224 schwarze 213: print_endline(h);
214: print_text(h, "dl.Bl-diag ");
215: print_byte(h, '>');
216: print_text(h, " dt { font-weight: bold; }");
1.223 schwarze 217: print_endline(h);
218: print_text(h, "code.Nm, code.Fl, code.Cm, code.Ic, "
219: "code.In, code.Fd, code.Fn,");
220: print_endline(h);
221: print_text(h, "code.Cd { font-weight: bold; "
222: "font-family: inherit; }");
1.165 kristaps 223: print_tagq(h, t);
1.4 kristaps 224: }
225:
1.255 ! schwarze 226: int
! 227: html_setfont(struct html *h, enum mandoc_esc font)
1.88 kristaps 228: {
1.255 ! schwarze 229: switch (font) {
1.156 schwarze 230: case ESCAPE_FONTPREV:
1.90 kristaps 231: font = h->metal;
1.88 kristaps 232: break;
1.156 schwarze 233: case ESCAPE_FONTITALIC:
234: case ESCAPE_FONTBOLD:
235: case ESCAPE_FONTBI:
1.242 schwarze 236: case ESCAPE_FONTCW:
1.255 ! schwarze 237: case ESCAPE_FONTROMAN:
1.242 schwarze 238: break;
1.156 schwarze 239: case ESCAPE_FONT:
1.255 ! schwarze 240: font = ESCAPE_FONTROMAN;
1.88 kristaps 241: break;
242: default:
1.255 ! schwarze 243: return 0;
1.88 kristaps 244: }
1.255 ! schwarze 245: h->metal = h->metac;
! 246: h->metac = font;
! 247: return 1;
! 248: }
1.88 kristaps 249:
1.255 ! schwarze 250: static void
! 251: print_metaf(struct html *h)
! 252: {
1.122 kristaps 253: if (h->metaf) {
254: print_tagq(h, h->metaf);
255: h->metaf = NULL;
256: }
1.255 ! schwarze 257: switch (h->metac) {
! 258: case ESCAPE_FONTITALIC:
1.194 schwarze 259: h->metaf = print_otag(h, TAG_I, "");
1.152 schwarze 260: break;
1.255 ! schwarze 261: case ESCAPE_FONTBOLD:
1.194 schwarze 262: h->metaf = print_otag(h, TAG_B, "");
1.152 schwarze 263: break;
1.255 ! schwarze 264: case ESCAPE_FONTBI:
1.194 schwarze 265: h->metaf = print_otag(h, TAG_B, "");
266: print_otag(h, TAG_I, "");
1.152 schwarze 267: break;
1.255 ! schwarze 268: case ESCAPE_FONTCW:
1.242 schwarze 269: h->metaf = print_otag(h, TAG_SPAN, "c", "Li");
270: break;
1.152 schwarze 271: default:
272: break;
273: }
1.248 schwarze 274: }
275:
1.249 schwarze 276: void
277: html_close_paragraph(struct html *h)
278: {
279: struct tag *t;
280:
1.252 schwarze 281: for (t = h->tag; t != NULL && t->closed == 0; t = t->next) {
282: switch(t->tag) {
283: case TAG_P:
284: case TAG_PRE:
1.249 schwarze 285: print_tagq(h, t);
286: break;
1.252 schwarze 287: case TAG_A:
288: print_tagq(h, t);
289: continue;
290: default:
291: continue;
1.249 schwarze 292: }
1.252 schwarze 293: break;
1.249 schwarze 294: }
295: }
296:
1.248 schwarze 297: /*
298: * ROFF_nf switches to no-fill mode, ROFF_fi to fill mode.
299: * TOKEN_NONE does not switch. The old mode is returned.
300: */
301: enum roff_tok
302: html_fillmode(struct html *h, enum roff_tok want)
303: {
304: struct tag *t;
305: enum roff_tok had;
306:
307: for (t = h->tag; t != NULL; t = t->next)
308: if (t->tag == TAG_PRE)
309: break;
310:
311: had = t == NULL ? ROFF_fi : ROFF_nf;
312:
313: if (want != had) {
314: switch (want) {
315: case ROFF_fi:
316: print_tagq(h, t);
317: break;
318: case ROFF_nf:
1.249 schwarze 319: html_close_paragraph(h);
1.248 schwarze 320: print_otag(h, TAG_PRE, "");
321: break;
322: case TOKEN_NONE:
323: break;
324: default:
325: abort();
326: }
327: }
328: return had;
1.210 schwarze 329: }
330:
331: char *
1.229 schwarze 332: html_make_id(const struct roff_node *n, int unique)
1.210 schwarze 333: {
334: const struct roff_node *nch;
1.229 schwarze 335: char *buf, *bufs, *cp;
336: unsigned int slot;
337: int suffix;
1.210 schwarze 338:
339: for (nch = n->child; nch != NULL; nch = nch->next)
340: if (nch->type != ROFFT_TEXT)
341: return NULL;
342:
343: buf = NULL;
344: deroff(&buf, n);
1.220 schwarze 345: if (buf == NULL)
346: return NULL;
1.210 schwarze 347:
1.230 schwarze 348: /*
349: * In ID attributes, only use ASCII characters that are
350: * permitted in URL-fragment strings according to the
351: * explicit list at:
352: * https://url.spec.whatwg.org/#url-fragment-string
353: */
1.210 schwarze 354:
355: for (cp = buf; *cp != '\0'; cp++)
1.230 schwarze 356: if (isalnum((unsigned char)*cp) == 0 &&
357: strchr("!$&'()*+,-./:;=?@_~", *cp) == NULL)
1.210 schwarze 358: *cp = '_';
359:
1.229 schwarze 360: if (unique == 0)
361: return buf;
362:
363: /* Avoid duplicate HTML id= attributes. */
364:
365: bufs = NULL;
366: suffix = 1;
367: slot = ohash_qlookup(&id_unique, buf);
368: cp = ohash_find(&id_unique, slot);
369: if (cp != NULL) {
370: while (cp != NULL) {
371: free(bufs);
372: if (++suffix > 127) {
373: free(buf);
374: return NULL;
375: }
376: mandoc_asprintf(&bufs, "%s_%d", buf, suffix);
377: slot = ohash_qlookup(&id_unique, bufs);
378: cp = ohash_find(&id_unique, slot);
379: }
380: free(buf);
381: buf = bufs;
382: }
383: ohash_insert(&id_unique, slot, buf);
1.210 schwarze 384: return buf;
1.88 kristaps 385: }
386:
1.85 kristaps 387: static int
1.197 schwarze 388: print_escape(struct html *h, char c)
1.159 schwarze 389: {
390:
391: switch (c) {
392: case '<':
1.197 schwarze 393: print_word(h, "<");
1.159 schwarze 394: break;
395: case '>':
1.197 schwarze 396: print_word(h, ">");
1.159 schwarze 397: break;
398: case '&':
1.197 schwarze 399: print_word(h, "&");
1.159 schwarze 400: break;
401: case '"':
1.197 schwarze 402: print_word(h, """);
1.159 schwarze 403: break;
404: case ASCII_NBRSP:
1.197 schwarze 405: print_word(h, " ");
1.159 schwarze 406: break;
407: case ASCII_HYPH:
1.197 schwarze 408: print_byte(h, '-');
1.189 schwarze 409: break;
1.159 schwarze 410: case ASCII_BREAK:
411: break;
412: default:
1.188 schwarze 413: return 0;
1.159 schwarze 414: }
1.188 schwarze 415: return 1;
1.159 schwarze 416: }
417:
418: static int
1.195 schwarze 419: print_encode(struct html *h, const char *p, const char *pend, int norecurse)
1.29 kristaps 420: {
1.197 schwarze 421: char numbuf[16];
1.214 schwarze 422: const char *seq;
1.77 kristaps 423: size_t sz;
1.214 schwarze 424: int c, len, breakline, nospace;
1.132 kristaps 425: enum mandoc_esc esc;
1.214 schwarze 426: static const char rejs[10] = { ' ', '\\', '<', '>', '&', '"',
1.154 schwarze 427: ASCII_NBRSP, ASCII_HYPH, ASCII_BREAK, '\0' };
1.14 kristaps 428:
1.195 schwarze 429: if (pend == NULL)
430: pend = strchr(p, '\0');
431:
1.214 schwarze 432: breakline = 0;
1.85 kristaps 433: nospace = 0;
434:
1.195 schwarze 435: while (p < pend) {
1.151 schwarze 436: if (HTML_SKIPCHAR & h->flags && '\\' != *p) {
437: h->flags &= ~HTML_SKIPCHAR;
438: p++;
439: continue;
440: }
441:
1.197 schwarze 442: for (sz = strcspn(p, rejs); sz-- && p < pend; p++)
1.214 schwarze 443: print_byte(h, *p);
444:
445: if (breakline &&
446: (p >= pend || *p == ' ' || *p == ASCII_NBRSP)) {
1.245 schwarze 447: print_otag(h, TAG_BR, "");
1.214 schwarze 448: breakline = 0;
449: while (p < pend && (*p == ' ' || *p == ASCII_NBRSP))
450: p++;
451: continue;
452: }
1.77 kristaps 453:
1.195 schwarze 454: if (p >= pend)
1.132 kristaps 455: break;
456:
1.214 schwarze 457: if (*p == ' ') {
458: print_endword(h);
459: p++;
460: continue;
461: }
462:
1.197 schwarze 463: if (print_escape(h, *p++))
1.154 schwarze 464: continue;
1.77 kristaps 465:
1.132 kristaps 466: esc = mandoc_escape(&p, &seq, &len);
467: switch (esc) {
1.156 schwarze 468: case ESCAPE_FONT:
469: case ESCAPE_FONTPREV:
470: case ESCAPE_FONTBOLD:
471: case ESCAPE_FONTITALIC:
472: case ESCAPE_FONTBI:
1.242 schwarze 473: case ESCAPE_FONTCW:
1.156 schwarze 474: case ESCAPE_FONTROMAN:
1.243 schwarze 475: if (0 == norecurse) {
476: h->flags |= HTML_NOSPACE;
1.255 ! schwarze 477: if (html_setfont(h, esc))
! 478: print_metaf(h);
1.243 schwarze 479: h->flags &= ~HTML_NOSPACE;
480: }
1.151 schwarze 481: continue;
1.156 schwarze 482: case ESCAPE_SKIPCHAR:
1.151 schwarze 483: h->flags |= HTML_SKIPCHAR;
484: continue;
1.246 schwarze 485: case ESCAPE_ERROR:
486: continue;
1.151 schwarze 487: default:
488: break;
489: }
490:
491: if (h->flags & HTML_SKIPCHAR) {
492: h->flags &= ~HTML_SKIPCHAR;
493: continue;
494: }
495:
496: switch (esc) {
1.156 schwarze 497: case ESCAPE_UNICODE:
1.159 schwarze 498: /* Skip past "u" header. */
1.144 kristaps 499: c = mchars_num2uc(seq + 1, len - 1);
500: break;
1.156 schwarze 501: case ESCAPE_NUMBERED:
1.141 kristaps 502: c = mchars_num2char(seq, len);
1.181 schwarze 503: if (c < 0)
504: continue;
1.82 kristaps 505: break;
1.156 schwarze 506: case ESCAPE_SPECIAL:
1.191 schwarze 507: c = mchars_spec2cp(seq, len);
1.181 schwarze 508: if (c <= 0)
509: continue;
1.246 schwarze 510: break;
511: case ESCAPE_UNDEF:
512: c = *seq;
1.132 kristaps 513: break;
1.239 schwarze 514: case ESCAPE_DEVICE:
515: print_word(h, "html");
516: continue;
1.214 schwarze 517: case ESCAPE_BREAK:
518: breakline = 1;
519: continue;
1.156 schwarze 520: case ESCAPE_NOSPACE:
1.132 kristaps 521: if ('\0' == *p)
522: nospace = 1;
1.179 schwarze 523: continue;
1.185 schwarze 524: case ESCAPE_OVERSTRIKE:
525: if (len == 0)
526: continue;
527: c = seq[len - 1];
528: break;
1.82 kristaps 529: default:
1.179 schwarze 530: continue;
1.82 kristaps 531: }
1.181 schwarze 532: if ((c < 0x20 && c != 0x09) ||
533: (c > 0x7E && c < 0xA0))
1.179 schwarze 534: c = 0xFFFD;
1.197 schwarze 535: if (c > 0x7E) {
1.216 schwarze 536: (void)snprintf(numbuf, sizeof(numbuf), "&#x%.4X;", c);
1.197 schwarze 537: print_word(h, numbuf);
538: } else if (print_escape(h, c) == 0)
539: print_byte(h, c);
1.32 kristaps 540: }
1.85 kristaps 541:
1.188 schwarze 542: return nospace;
1.14 kristaps 543: }
544:
1.94 kristaps 545: static void
1.195 schwarze 546: print_href(struct html *h, const char *name, const char *sec, int man)
1.94 kristaps 547: {
1.240 schwarze 548: struct stat sb;
1.195 schwarze 549: const char *p, *pp;
1.240 schwarze 550: char *filename;
551:
552: if (man) {
553: pp = h->base_man1;
554: if (h->base_man2 != NULL) {
555: mandoc_asprintf(&filename, "%s.%s", name, sec);
556: if (stat(filename, &sb) == -1)
557: pp = h->base_man2;
558: free(filename);
559: }
560: } else
561: pp = h->base_includes;
1.195 schwarze 562:
563: while ((p = strchr(pp, '%')) != NULL) {
564: print_encode(h, pp, p, 1);
565: if (man && p[1] == 'S') {
566: if (sec == NULL)
1.197 schwarze 567: print_byte(h, '1');
1.195 schwarze 568: else
569: print_encode(h, sec, NULL, 1);
570: } else if ((man && p[1] == 'N') ||
571: (man == 0 && p[1] == 'I'))
572: print_encode(h, name, NULL, 1);
573: else
574: print_encode(h, p, p + 2, 1);
575: pp = p + 2;
576: }
577: if (*pp != '\0')
578: print_encode(h, pp, NULL, 1);
1.94 kristaps 579: }
580:
1.51 kristaps 581: struct tag *
1.194 schwarze 582: print_otag(struct html *h, enum htmltag tag, const char *fmt, ...)
1.14 kristaps 583: {
1.194 schwarze 584: va_list ap;
1.30 kristaps 585: struct tag *t;
1.195 schwarze 586: const char *attr;
1.203 schwarze 587: char *arg1, *arg2;
1.244 schwarze 588: int style_written, tflags;
1.196 schwarze 589:
590: tflags = htmltags[tag].flags;
1.30 kristaps 591:
1.204 schwarze 592: /* Push this tag onto the stack of open scopes. */
1.94 kristaps 593:
1.196 schwarze 594: if ((tflags & HTML_NOSTACK) == 0) {
1.128 kristaps 595: t = mandoc_malloc(sizeof(struct tag));
1.30 kristaps 596: t->tag = tag;
1.204 schwarze 597: t->next = h->tag;
1.252 schwarze 598: t->refcnt = 0;
599: t->closed = 0;
1.204 schwarze 600: h->tag = t;
1.30 kristaps 601: } else
602: t = NULL;
1.29 kristaps 603:
1.196 schwarze 604: if (tflags & HTML_NLBEFORE)
1.197 schwarze 605: print_endline(h);
606: if (h->col == 0)
607: print_indent(h);
1.196 schwarze 608: else if ((h->flags & HTML_NOSPACE) == 0) {
609: if (h->flags & HTML_KEEP)
1.216 schwarze 610: print_word(h, " ");
1.196 schwarze 611: else {
612: if (h->flags & HTML_PREKEEP)
613: h->flags |= HTML_KEEP;
1.197 schwarze 614: print_endword(h);
1.105 kristaps 615: }
1.196 schwarze 616: }
1.29 kristaps 617:
1.109 kristaps 618: if ( ! (h->flags & HTML_NONOSPACE))
619: h->flags &= ~HTML_NOSPACE;
1.110 kristaps 620: else
621: h->flags |= HTML_NOSPACE;
1.109 kristaps 622:
1.94 kristaps 623: /* Print out the tag name and attributes. */
624:
1.197 schwarze 625: print_byte(h, '<');
626: print_word(h, htmltags[tag].name);
1.194 schwarze 627:
628: va_start(ap, fmt);
629:
1.244 schwarze 630: while (*fmt != '\0' && *fmt != 's') {
1.203 schwarze 631:
1.238 schwarze 632: /* Parse attributes and arguments. */
1.203 schwarze 633:
634: arg1 = va_arg(ap, char *);
1.238 schwarze 635: arg2 = NULL;
1.194 schwarze 636: switch (*fmt++) {
637: case 'c':
1.195 schwarze 638: attr = "class";
1.194 schwarze 639: break;
640: case 'h':
1.195 schwarze 641: attr = "href";
1.194 schwarze 642: break;
643: case 'i':
1.195 schwarze 644: attr = "id";
1.194 schwarze 645: break;
646: case '?':
1.203 schwarze 647: attr = arg1;
648: arg1 = va_arg(ap, char *);
1.194 schwarze 649: break;
650: default:
651: abort();
652: }
1.203 schwarze 653: if (*fmt == 'M')
654: arg2 = va_arg(ap, char *);
655: if (arg1 == NULL)
656: continue;
657:
1.238 schwarze 658: /* Print the attributes. */
1.203 schwarze 659:
1.197 schwarze 660: print_byte(h, ' ');
661: print_word(h, attr);
662: print_byte(h, '=');
663: print_byte(h, '"');
1.195 schwarze 664: switch (*fmt) {
1.208 schwarze 665: case 'I':
666: print_href(h, arg1, NULL, 0);
667: fmt++;
668: break;
1.195 schwarze 669: case 'M':
1.203 schwarze 670: print_href(h, arg1, arg2, 1);
1.195 schwarze 671: fmt++;
672: break;
1.208 schwarze 673: case 'R':
674: print_byte(h, '#');
675: print_encode(h, arg1, NULL, 1);
1.195 schwarze 676: fmt++;
1.208 schwarze 677: break;
1.195 schwarze 678: default:
1.244 schwarze 679: print_encode(h, arg1, NULL, 1);
1.195 schwarze 680: break;
681: }
1.197 schwarze 682: print_byte(h, '"');
1.194 schwarze 683: }
1.244 schwarze 684:
685: style_written = 0;
686: while (*fmt++ == 's') {
687: arg1 = va_arg(ap, char *);
688: arg2 = va_arg(ap, char *);
689: if (arg2 == NULL)
690: continue;
691: print_byte(h, ' ');
692: if (style_written == 0) {
693: print_word(h, "style=\"");
694: style_written = 1;
695: }
696: print_word(h, arg1);
697: print_byte(h, ':');
698: print_byte(h, ' ');
699: print_word(h, arg2);
700: print_byte(h, ';');
701: }
702: if (style_written)
703: print_byte(h, '"');
704:
1.194 schwarze 705: va_end(ap);
1.94 kristaps 706:
1.172 kristaps 707: /* Accommodate for "well-formed" singleton escaping. */
1.94 kristaps 708:
1.93 kristaps 709: if (HTML_AUTOCLOSE & htmltags[tag].flags)
1.197 schwarze 710: print_byte(h, '/');
1.93 kristaps 711:
1.197 schwarze 712: print_byte(h, '>');
1.14 kristaps 713:
1.196 schwarze 714: if (tflags & HTML_NLBEGIN)
1.197 schwarze 715: print_endline(h);
1.196 schwarze 716: else
717: h->flags |= HTML_NOSPACE;
1.117 kristaps 718:
1.196 schwarze 719: if (tflags & HTML_INDENT)
720: h->indent++;
721: if (tflags & HTML_NOINDENT)
722: h->noindent++;
1.117 kristaps 723:
1.188 schwarze 724: return t;
1.14 kristaps 725: }
726:
1.29 kristaps 727: static void
1.184 schwarze 728: print_ctag(struct html *h, struct tag *tag)
1.14 kristaps 729: {
1.196 schwarze 730: int tflags;
1.156 schwarze 731:
1.252 schwarze 732: if (tag->closed == 0) {
733: tag->closed = 1;
734: if (tag == h->metaf)
735: h->metaf = NULL;
736: if (tag == h->tblt)
737: h->tblt = NULL;
738:
739: tflags = htmltags[tag->tag].flags;
740: if (tflags & HTML_INDENT)
741: h->indent--;
742: if (tflags & HTML_NOINDENT)
743: h->noindent--;
744: if (tflags & HTML_NLEND)
745: print_endline(h);
746: print_indent(h);
747: print_byte(h, '<');
748: print_byte(h, '/');
749: print_word(h, htmltags[tag->tag].name);
750: print_byte(h, '>');
751: if (tflags & HTML_NLAFTER)
752: print_endline(h);
753: }
754: if (tag->refcnt == 0) {
755: h->tag = tag->next;
756: free(tag);
757: }
1.14 kristaps 758: }
759:
1.51 kristaps 760: void
1.93 kristaps 761: print_gen_decls(struct html *h)
1.1 kristaps 762: {
1.197 schwarze 763: print_word(h, "<!DOCTYPE html>");
764: print_endline(h);
1.221 schwarze 765: }
766:
767: void
768: print_gen_comment(struct html *h, struct roff_node *n)
769: {
770: int wantblank;
771:
772: print_word(h, "<!-- This is an automatically generated file."
773: " Do not edit.");
774: h->indent = 1;
775: wantblank = 0;
776: while (n != NULL && n->type == ROFFT_COMMENT) {
777: if (strstr(n->string, "-->") == NULL &&
778: (wantblank || *n->string != '\0')) {
779: print_endline(h);
780: print_indent(h);
781: print_word(h, n->string);
782: wantblank = *n->string != '\0';
783: }
784: n = n->next;
785: }
786: if (wantblank)
787: print_endline(h);
788: print_word(h, " -->");
789: print_endline(h);
790: h->indent = 0;
1.1 kristaps 791: }
792:
1.51 kristaps 793: void
1.104 kristaps 794: print_text(struct html *h, const char *word)
1.1 kristaps 795: {
1.197 schwarze 796: if (h->col && (h->flags & HTML_NOSPACE) == 0) {
1.105 kristaps 797: if ( ! (HTML_KEEP & h->flags)) {
798: if (HTML_PREKEEP & h->flags)
799: h->flags |= HTML_KEEP;
1.197 schwarze 800: print_endword(h);
1.105 kristaps 801: } else
1.216 schwarze 802: print_word(h, " ");
1.105 kristaps 803: }
1.30 kristaps 804:
1.255 ! schwarze 805: assert(h->metaf == NULL);
! 806: print_metaf(h);
! 807: print_indent(h);
1.195 schwarze 808: if ( ! print_encode(h, word, NULL, 0)) {
1.109 kristaps 809: if ( ! (h->flags & HTML_NONOSPACE))
810: h->flags &= ~HTML_NOSPACE;
1.183 schwarze 811: h->flags &= ~HTML_NONEWLINE;
1.149 kristaps 812: } else
1.183 schwarze 813: h->flags |= HTML_NOSPACE | HTML_NONEWLINE;
1.122 kristaps 814:
1.255 ! schwarze 815: if (h->metaf != NULL) {
1.122 kristaps 816: print_tagq(h, h->metaf);
817: h->metaf = NULL;
818: }
1.113 schwarze 819:
820: h->flags &= ~HTML_IGNDELIM;
1.1 kristaps 821: }
1.30 kristaps 822:
1.51 kristaps 823: void
1.30 kristaps 824: print_tagq(struct html *h, const struct tag *until)
825: {
1.252 schwarze 826: struct tag *this, *next;
1.30 kristaps 827:
1.252 schwarze 828: for (this = h->tag; this != NULL; this = next) {
829: next = this == until ? NULL : this->next;
830: print_ctag(h, this);
1.30 kristaps 831: }
832: }
833:
1.250 schwarze 834: /*
835: * Close out all open elements up to but excluding suntil.
836: * Note that a paragraph just inside stays open together with it
837: * because paragraphs include subsequent phrasing content.
838: */
1.51 kristaps 839: void
1.30 kristaps 840: print_stagq(struct html *h, const struct tag *suntil)
841: {
1.252 schwarze 842: struct tag *this, *next;
1.30 kristaps 843:
1.252 schwarze 844: for (this = h->tag; this != NULL; this = next) {
845: next = this->next;
846: if (this == suntil || (next == suntil &&
847: (this->tag == TAG_P || this->tag == TAG_PRE)))
848: break;
849: print_ctag(h, this);
1.30 kristaps 850: }
1.171 kristaps 851: }
852:
1.197 schwarze 853:
854: /***********************************************************************
855: * Low level output functions.
856: * They implement line breaking using a short static buffer.
857: ***********************************************************************/
858:
859: /*
860: * Buffer one HTML output byte.
861: * If the buffer is full, flush and deactivate it and start a new line.
862: * If the buffer is inactive, print directly.
863: */
864: static void
865: print_byte(struct html *h, char c)
866: {
867: if ((h->flags & HTML_BUFFER) == 0) {
868: putchar(c);
869: h->col++;
870: return;
871: }
872:
873: if (h->col + h->bufcol < sizeof(h->buf)) {
874: h->buf[h->bufcol++] = c;
875: return;
876: }
877:
878: putchar('\n');
879: h->col = 0;
880: print_indent(h);
881: putchar(' ');
882: putchar(' ');
883: fwrite(h->buf, h->bufcol, 1, stdout);
884: putchar(c);
885: h->col = (h->indent + 1) * 2 + h->bufcol + 1;
886: h->bufcol = 0;
887: h->flags &= ~HTML_BUFFER;
888: }
889:
1.196 schwarze 890: /*
891: * If something was printed on the current output line, end it.
1.197 schwarze 892: * Not to be called right after print_indent().
1.196 schwarze 893: */
1.202 schwarze 894: void
1.197 schwarze 895: print_endline(struct html *h)
1.196 schwarze 896: {
1.197 schwarze 897: if (h->col == 0)
1.196 schwarze 898: return;
899:
1.197 schwarze 900: if (h->bufcol) {
901: putchar(' ');
902: fwrite(h->buf, h->bufcol, 1, stdout);
903: h->bufcol = 0;
904: }
1.196 schwarze 905: putchar('\n');
1.197 schwarze 906: h->col = 0;
907: h->flags |= HTML_NOSPACE;
908: h->flags &= ~HTML_BUFFER;
909: }
910:
911: /*
912: * Flush the HTML output buffer.
913: * If it is inactive, activate it.
914: */
915: static void
916: print_endword(struct html *h)
917: {
918: if (h->noindent) {
919: print_byte(h, ' ');
920: return;
921: }
922:
923: if ((h->flags & HTML_BUFFER) == 0) {
924: h->col++;
925: h->flags |= HTML_BUFFER;
926: } else if (h->bufcol) {
927: putchar(' ');
928: fwrite(h->buf, h->bufcol, 1, stdout);
929: h->col += h->bufcol + 1;
930: }
931: h->bufcol = 0;
1.196 schwarze 932: }
933:
934: /*
935: * If at the beginning of a new output line,
936: * perform indentation and mark the line as containing output.
937: * Make sure to really produce some output right afterwards,
938: * but do not use print_otag() for producing it.
939: */
940: static void
1.197 schwarze 941: print_indent(struct html *h)
1.196 schwarze 942: {
1.197 schwarze 943: size_t i;
1.196 schwarze 944:
1.197 schwarze 945: if (h->col)
1.196 schwarze 946: return;
947:
1.197 schwarze 948: if (h->noindent == 0) {
949: h->col = h->indent * 2;
950: for (i = 0; i < h->col; i++)
1.196 schwarze 951: putchar(' ');
1.197 schwarze 952: }
953: h->flags &= ~HTML_NOSPACE;
954: }
955:
956: /*
957: * Print or buffer some characters
958: * depending on the current HTML output buffer state.
959: */
960: static void
961: print_word(struct html *h, const char *cp)
962: {
963: while (*cp != '\0')
964: print_byte(h, *cp++);
1.68 kristaps 965: }
CVSweb