Annotation of mandoc/html.c, Revision 1.252
1.252 ! schwarze 1: /* $Id: html.c,v 1.251 2019/01/11 12:56:42 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},
1.196 schwarze 66: {"h1", HTML_NLAROUND},
67: {"h2", HTML_NLAROUND},
68: {"span", 0},
69: {"link", HTML_NOSTACK | HTML_AUTOCLOSE | HTML_NLALL},
70: {"br", HTML_NOSTACK | HTML_AUTOCLOSE | HTML_NLALL},
71: {"a", 0},
72: {"table", HTML_NLALL | HTML_INDENT},
73: {"tr", HTML_NLALL | HTML_INDENT},
74: {"td", HTML_NLAROUND},
75: {"li", HTML_NLAROUND | HTML_INDENT},
76: {"ul", HTML_NLALL | HTML_INDENT},
77: {"ol", HTML_NLALL | HTML_INDENT},
78: {"dl", HTML_NLALL | HTML_INDENT},
79: {"dt", HTML_NLAROUND},
80: {"dd", HTML_NLAROUND | HTML_INDENT},
1.249 schwarze 81: {"p", HTML_NLAROUND | HTML_INDENT},
1.196 schwarze 82: {"pre", HTML_NLALL | HTML_NOINDENT},
1.207 schwarze 83: {"var", 0},
1.206 schwarze 84: {"cite", 0},
1.196 schwarze 85: {"b", 0},
86: {"i", 0},
87: {"code", 0},
88: {"small", 0},
89: {"style", HTML_NLALL | HTML_INDENT},
90: {"math", HTML_NLALL | HTML_INDENT},
91: {"mrow", 0},
92: {"mi", 0},
1.215 schwarze 93: {"mn", 0},
1.196 schwarze 94: {"mo", 0},
95: {"msup", 0},
96: {"msub", 0},
97: {"msubsup", 0},
98: {"mfrac", 0},
99: {"msqrt", 0},
100: {"mfenced", 0},
101: {"mtable", 0},
102: {"mtr", 0},
103: {"mtd", 0},
104: {"munderover", 0},
105: {"munder", 0},
106: {"mover", 0},
1.90 kristaps 107: };
108:
1.229 schwarze 109: /* Avoid duplicate HTML id= attributes. */
110: static struct ohash id_unique;
111:
1.197 schwarze 112: static void print_byte(struct html *, char);
113: static void print_endword(struct html *);
114: static void print_indent(struct html *);
115: static void print_word(struct html *, const char *);
116:
1.184 schwarze 117: static void print_ctag(struct html *, struct tag *);
1.197 schwarze 118: static int print_escape(struct html *, char);
1.195 schwarze 119: static int print_encode(struct html *, const char *, const char *, int);
120: static void print_href(struct html *, const char *, const char *, int);
1.82 kristaps 121:
1.156 schwarze 122:
1.180 schwarze 123: void *
1.191 schwarze 124: html_alloc(const struct manoutput *outopts)
1.10 kristaps 125: {
1.30 kristaps 126: struct html *h;
127:
1.128 kristaps 128: h = mandoc_calloc(1, sizeof(struct html));
1.10 kristaps 129:
1.204 schwarze 130: h->tag = NULL;
1.186 schwarze 131: h->style = outopts->style;
1.240 schwarze 132: if ((h->base_man1 = outopts->man) == NULL)
133: h->base_man2 = NULL;
134: else if ((h->base_man2 = strchr(h->base_man1, ';')) != NULL)
135: *h->base_man2++ = '\0';
1.186 schwarze 136: h->base_includes = outopts->includes;
137: if (outopts->fragment)
138: h->oflags |= HTML_FRAGMENT;
1.241 schwarze 139: if (outopts->toc)
140: h->oflags |= HTML_TOC;
1.43 kristaps 141:
1.229 schwarze 142: mandoc_ohash_init(&id_unique, 4, 0);
143:
1.188 schwarze 144: return h;
1.29 kristaps 145: }
1.10 kristaps 146:
1.29 kristaps 147: void
148: html_free(void *p)
149: {
1.30 kristaps 150: struct tag *tag;
151: struct html *h;
1.229 schwarze 152: char *cp;
153: unsigned int slot;
1.30 kristaps 154:
155: h = (struct html *)p;
1.204 schwarze 156: while ((tag = h->tag) != NULL) {
157: h->tag = tag->next;
1.30 kristaps 158: free(tag);
159: }
1.229 schwarze 160: free(h);
1.53 kristaps 161:
1.229 schwarze 162: cp = ohash_first(&id_unique, &slot);
163: while (cp != NULL) {
164: free(cp);
165: cp = ohash_next(&id_unique, &slot);
166: }
167: ohash_delete(&id_unique);
1.10 kristaps 168: }
1.2 kristaps 169:
1.51 kristaps 170: void
1.29 kristaps 171: print_gen_head(struct html *h)
172: {
1.165 kristaps 173: struct tag *t;
1.41 kristaps 174:
1.194 schwarze 175: print_otag(h, TAG_META, "?", "charset", "utf-8");
1.222 schwarze 176: if (h->style != NULL) {
177: print_otag(h, TAG_LINK, "?h??", "rel", "stylesheet",
178: h->style, "type", "text/css", "media", "all");
179: return;
180: }
1.165 kristaps 181:
1.168 kristaps 182: /*
1.222 schwarze 183: * Print a minimal embedded style sheet.
1.168 kristaps 184: */
1.196 schwarze 185:
1.194 schwarze 186: t = print_otag(h, TAG_STYLE, "");
1.196 schwarze 187: print_text(h, "table.head, table.foot { width: 100%; }");
1.197 schwarze 188: print_endline(h);
1.196 schwarze 189: print_text(h, "td.head-rtitle, td.foot-os { text-align: right; }");
1.197 schwarze 190: print_endline(h);
1.196 schwarze 191: print_text(h, "td.head-vol { text-align: center; }");
1.197 schwarze 192: print_endline(h);
1.198 schwarze 193: print_text(h, "div.Pp { margin: 1ex 0ex; }");
1.225 schwarze 194: print_endline(h);
195: print_text(h, "div.Nd, div.Bf, div.Op { display: inline; }");
1.226 schwarze 196: print_endline(h);
1.227 schwarze 197: print_text(h, "span.Pa, span.Ad { font-style: italic; }");
1.228 schwarze 198: print_endline(h);
199: print_text(h, "span.Ms { font-weight: bold; }");
1.224 schwarze 200: print_endline(h);
201: print_text(h, "dl.Bl-diag ");
202: print_byte(h, '>');
203: print_text(h, " dt { font-weight: bold; }");
1.223 schwarze 204: print_endline(h);
205: print_text(h, "code.Nm, code.Fl, code.Cm, code.Ic, "
206: "code.In, code.Fd, code.Fn,");
207: print_endline(h);
208: print_text(h, "code.Cd { font-weight: bold; "
209: "font-family: inherit; }");
1.165 kristaps 210: print_tagq(h, t);
1.4 kristaps 211: }
212:
1.247 schwarze 213: void
1.132 kristaps 214: print_metaf(struct html *h, enum mandoc_esc deco)
1.88 kristaps 215: {
1.90 kristaps 216: enum htmlfont font;
1.88 kristaps 217:
218: switch (deco) {
1.156 schwarze 219: case ESCAPE_FONTPREV:
1.90 kristaps 220: font = h->metal;
1.88 kristaps 221: break;
1.156 schwarze 222: case ESCAPE_FONTITALIC:
1.90 kristaps 223: font = HTMLFONT_ITALIC;
224: break;
1.156 schwarze 225: case ESCAPE_FONTBOLD:
1.90 kristaps 226: font = HTMLFONT_BOLD;
1.88 kristaps 227: break;
1.156 schwarze 228: case ESCAPE_FONTBI:
1.152 schwarze 229: font = HTMLFONT_BI;
230: break;
1.242 schwarze 231: case ESCAPE_FONTCW:
232: font = HTMLFONT_CW;
233: break;
1.156 schwarze 234: case ESCAPE_FONT:
235: case ESCAPE_FONTROMAN:
1.90 kristaps 236: font = HTMLFONT_NONE;
1.88 kristaps 237: break;
238: default:
1.247 schwarze 239: return;
1.88 kristaps 240: }
241:
1.122 kristaps 242: if (h->metaf) {
243: print_tagq(h, h->metaf);
244: h->metaf = NULL;
245: }
246:
247: h->metal = h->metac;
248: h->metac = font;
249:
1.152 schwarze 250: switch (font) {
1.156 schwarze 251: case HTMLFONT_ITALIC:
1.194 schwarze 252: h->metaf = print_otag(h, TAG_I, "");
1.152 schwarze 253: break;
1.156 schwarze 254: case HTMLFONT_BOLD:
1.194 schwarze 255: h->metaf = print_otag(h, TAG_B, "");
1.152 schwarze 256: break;
1.156 schwarze 257: case HTMLFONT_BI:
1.194 schwarze 258: h->metaf = print_otag(h, TAG_B, "");
259: print_otag(h, TAG_I, "");
1.152 schwarze 260: break;
1.242 schwarze 261: case HTMLFONT_CW:
262: h->metaf = print_otag(h, TAG_SPAN, "c", "Li");
263: break;
1.152 schwarze 264: default:
265: break;
266: }
1.248 schwarze 267: }
268:
1.249 schwarze 269: void
270: html_close_paragraph(struct html *h)
271: {
272: struct tag *t;
273:
1.252 ! schwarze 274: for (t = h->tag; t != NULL && t->closed == 0; t = t->next) {
! 275: switch(t->tag) {
! 276: case TAG_P:
! 277: case TAG_PRE:
1.249 schwarze 278: print_tagq(h, t);
279: break;
1.252 ! schwarze 280: case TAG_A:
! 281: print_tagq(h, t);
! 282: continue;
! 283: default:
! 284: continue;
1.249 schwarze 285: }
1.252 ! schwarze 286: break;
1.249 schwarze 287: }
288: }
289:
1.248 schwarze 290: /*
291: * ROFF_nf switches to no-fill mode, ROFF_fi to fill mode.
292: * TOKEN_NONE does not switch. The old mode is returned.
293: */
294: enum roff_tok
295: html_fillmode(struct html *h, enum roff_tok want)
296: {
297: struct tag *t;
298: enum roff_tok had;
299:
300: for (t = h->tag; t != NULL; t = t->next)
301: if (t->tag == TAG_PRE)
302: break;
303:
304: had = t == NULL ? ROFF_fi : ROFF_nf;
305:
306: if (want != had) {
307: switch (want) {
308: case ROFF_fi:
309: print_tagq(h, t);
310: break;
311: case ROFF_nf:
1.249 schwarze 312: html_close_paragraph(h);
1.248 schwarze 313: print_otag(h, TAG_PRE, "");
314: break;
315: case TOKEN_NONE:
316: break;
317: default:
318: abort();
319: }
320: }
321: return had;
1.210 schwarze 322: }
323:
324: char *
1.229 schwarze 325: html_make_id(const struct roff_node *n, int unique)
1.210 schwarze 326: {
327: const struct roff_node *nch;
1.229 schwarze 328: char *buf, *bufs, *cp;
329: unsigned int slot;
330: int suffix;
1.210 schwarze 331:
332: for (nch = n->child; nch != NULL; nch = nch->next)
333: if (nch->type != ROFFT_TEXT)
334: return NULL;
335:
336: buf = NULL;
337: deroff(&buf, n);
1.220 schwarze 338: if (buf == NULL)
339: return NULL;
1.210 schwarze 340:
1.230 schwarze 341: /*
342: * In ID attributes, only use ASCII characters that are
343: * permitted in URL-fragment strings according to the
344: * explicit list at:
345: * https://url.spec.whatwg.org/#url-fragment-string
346: */
1.210 schwarze 347:
348: for (cp = buf; *cp != '\0'; cp++)
1.230 schwarze 349: if (isalnum((unsigned char)*cp) == 0 &&
350: strchr("!$&'()*+,-./:;=?@_~", *cp) == NULL)
1.210 schwarze 351: *cp = '_';
352:
1.229 schwarze 353: if (unique == 0)
354: return buf;
355:
356: /* Avoid duplicate HTML id= attributes. */
357:
358: bufs = NULL;
359: suffix = 1;
360: slot = ohash_qlookup(&id_unique, buf);
361: cp = ohash_find(&id_unique, slot);
362: if (cp != NULL) {
363: while (cp != NULL) {
364: free(bufs);
365: if (++suffix > 127) {
366: free(buf);
367: return NULL;
368: }
369: mandoc_asprintf(&bufs, "%s_%d", buf, suffix);
370: slot = ohash_qlookup(&id_unique, bufs);
371: cp = ohash_find(&id_unique, slot);
372: }
373: free(buf);
374: buf = bufs;
375: }
376: ohash_insert(&id_unique, slot, buf);
1.210 schwarze 377: return buf;
1.88 kristaps 378: }
379:
1.85 kristaps 380: static int
1.197 schwarze 381: print_escape(struct html *h, char c)
1.159 schwarze 382: {
383:
384: switch (c) {
385: case '<':
1.197 schwarze 386: print_word(h, "<");
1.159 schwarze 387: break;
388: case '>':
1.197 schwarze 389: print_word(h, ">");
1.159 schwarze 390: break;
391: case '&':
1.197 schwarze 392: print_word(h, "&");
1.159 schwarze 393: break;
394: case '"':
1.197 schwarze 395: print_word(h, """);
1.159 schwarze 396: break;
397: case ASCII_NBRSP:
1.197 schwarze 398: print_word(h, " ");
1.159 schwarze 399: break;
400: case ASCII_HYPH:
1.197 schwarze 401: print_byte(h, '-');
1.189 schwarze 402: break;
1.159 schwarze 403: case ASCII_BREAK:
404: break;
405: default:
1.188 schwarze 406: return 0;
1.159 schwarze 407: }
1.188 schwarze 408: return 1;
1.159 schwarze 409: }
410:
411: static int
1.195 schwarze 412: print_encode(struct html *h, const char *p, const char *pend, int norecurse)
1.29 kristaps 413: {
1.197 schwarze 414: char numbuf[16];
1.214 schwarze 415: const char *seq;
1.77 kristaps 416: size_t sz;
1.214 schwarze 417: int c, len, breakline, nospace;
1.132 kristaps 418: enum mandoc_esc esc;
1.214 schwarze 419: static const char rejs[10] = { ' ', '\\', '<', '>', '&', '"',
1.154 schwarze 420: ASCII_NBRSP, ASCII_HYPH, ASCII_BREAK, '\0' };
1.14 kristaps 421:
1.195 schwarze 422: if (pend == NULL)
423: pend = strchr(p, '\0');
424:
1.214 schwarze 425: breakline = 0;
1.85 kristaps 426: nospace = 0;
427:
1.195 schwarze 428: while (p < pend) {
1.151 schwarze 429: if (HTML_SKIPCHAR & h->flags && '\\' != *p) {
430: h->flags &= ~HTML_SKIPCHAR;
431: p++;
432: continue;
433: }
434:
1.197 schwarze 435: for (sz = strcspn(p, rejs); sz-- && p < pend; p++)
1.214 schwarze 436: print_byte(h, *p);
437:
438: if (breakline &&
439: (p >= pend || *p == ' ' || *p == ASCII_NBRSP)) {
1.245 schwarze 440: print_otag(h, TAG_BR, "");
1.214 schwarze 441: breakline = 0;
442: while (p < pend && (*p == ' ' || *p == ASCII_NBRSP))
443: p++;
444: continue;
445: }
1.77 kristaps 446:
1.195 schwarze 447: if (p >= pend)
1.132 kristaps 448: break;
449:
1.214 schwarze 450: if (*p == ' ') {
451: print_endword(h);
452: p++;
453: continue;
454: }
455:
1.197 schwarze 456: if (print_escape(h, *p++))
1.154 schwarze 457: continue;
1.77 kristaps 458:
1.132 kristaps 459: esc = mandoc_escape(&p, &seq, &len);
460: switch (esc) {
1.156 schwarze 461: case ESCAPE_FONT:
462: case ESCAPE_FONTPREV:
463: case ESCAPE_FONTBOLD:
464: case ESCAPE_FONTITALIC:
465: case ESCAPE_FONTBI:
1.242 schwarze 466: case ESCAPE_FONTCW:
1.156 schwarze 467: case ESCAPE_FONTROMAN:
1.243 schwarze 468: if (0 == norecurse) {
469: h->flags |= HTML_NOSPACE;
1.151 schwarze 470: print_metaf(h, esc);
1.243 schwarze 471: h->flags &= ~HTML_NOSPACE;
472: }
1.151 schwarze 473: continue;
1.156 schwarze 474: case ESCAPE_SKIPCHAR:
1.151 schwarze 475: h->flags |= HTML_SKIPCHAR;
476: continue;
1.246 schwarze 477: case ESCAPE_ERROR:
478: continue;
1.151 schwarze 479: default:
480: break;
481: }
482:
483: if (h->flags & HTML_SKIPCHAR) {
484: h->flags &= ~HTML_SKIPCHAR;
485: continue;
486: }
487:
488: switch (esc) {
1.156 schwarze 489: case ESCAPE_UNICODE:
1.159 schwarze 490: /* Skip past "u" header. */
1.144 kristaps 491: c = mchars_num2uc(seq + 1, len - 1);
492: break;
1.156 schwarze 493: case ESCAPE_NUMBERED:
1.141 kristaps 494: c = mchars_num2char(seq, len);
1.181 schwarze 495: if (c < 0)
496: continue;
1.82 kristaps 497: break;
1.156 schwarze 498: case ESCAPE_SPECIAL:
1.191 schwarze 499: c = mchars_spec2cp(seq, len);
1.181 schwarze 500: if (c <= 0)
501: continue;
1.246 schwarze 502: break;
503: case ESCAPE_UNDEF:
504: c = *seq;
1.132 kristaps 505: break;
1.239 schwarze 506: case ESCAPE_DEVICE:
507: print_word(h, "html");
508: continue;
1.214 schwarze 509: case ESCAPE_BREAK:
510: breakline = 1;
511: continue;
1.156 schwarze 512: case ESCAPE_NOSPACE:
1.132 kristaps 513: if ('\0' == *p)
514: nospace = 1;
1.179 schwarze 515: continue;
1.185 schwarze 516: case ESCAPE_OVERSTRIKE:
517: if (len == 0)
518: continue;
519: c = seq[len - 1];
520: break;
1.82 kristaps 521: default:
1.179 schwarze 522: continue;
1.82 kristaps 523: }
1.181 schwarze 524: if ((c < 0x20 && c != 0x09) ||
525: (c > 0x7E && c < 0xA0))
1.179 schwarze 526: c = 0xFFFD;
1.197 schwarze 527: if (c > 0x7E) {
1.216 schwarze 528: (void)snprintf(numbuf, sizeof(numbuf), "&#x%.4X;", c);
1.197 schwarze 529: print_word(h, numbuf);
530: } else if (print_escape(h, c) == 0)
531: print_byte(h, c);
1.32 kristaps 532: }
1.85 kristaps 533:
1.188 schwarze 534: return nospace;
1.14 kristaps 535: }
536:
1.94 kristaps 537: static void
1.195 schwarze 538: print_href(struct html *h, const char *name, const char *sec, int man)
1.94 kristaps 539: {
1.240 schwarze 540: struct stat sb;
1.195 schwarze 541: const char *p, *pp;
1.240 schwarze 542: char *filename;
543:
544: if (man) {
545: pp = h->base_man1;
546: if (h->base_man2 != NULL) {
547: mandoc_asprintf(&filename, "%s.%s", name, sec);
548: if (stat(filename, &sb) == -1)
549: pp = h->base_man2;
550: free(filename);
551: }
552: } else
553: pp = h->base_includes;
1.195 schwarze 554:
555: while ((p = strchr(pp, '%')) != NULL) {
556: print_encode(h, pp, p, 1);
557: if (man && p[1] == 'S') {
558: if (sec == NULL)
1.197 schwarze 559: print_byte(h, '1');
1.195 schwarze 560: else
561: print_encode(h, sec, NULL, 1);
562: } else if ((man && p[1] == 'N') ||
563: (man == 0 && p[1] == 'I'))
564: print_encode(h, name, NULL, 1);
565: else
566: print_encode(h, p, p + 2, 1);
567: pp = p + 2;
568: }
569: if (*pp != '\0')
570: print_encode(h, pp, NULL, 1);
1.94 kristaps 571: }
572:
1.51 kristaps 573: struct tag *
1.194 schwarze 574: print_otag(struct html *h, enum htmltag tag, const char *fmt, ...)
1.14 kristaps 575: {
1.194 schwarze 576: va_list ap;
1.30 kristaps 577: struct tag *t;
1.195 schwarze 578: const char *attr;
1.203 schwarze 579: char *arg1, *arg2;
1.244 schwarze 580: int style_written, tflags;
1.196 schwarze 581:
582: tflags = htmltags[tag].flags;
1.30 kristaps 583:
1.204 schwarze 584: /* Push this tag onto the stack of open scopes. */
1.94 kristaps 585:
1.196 schwarze 586: if ((tflags & HTML_NOSTACK) == 0) {
1.128 kristaps 587: t = mandoc_malloc(sizeof(struct tag));
1.30 kristaps 588: t->tag = tag;
1.204 schwarze 589: t->next = h->tag;
1.252 ! schwarze 590: t->refcnt = 0;
! 591: t->closed = 0;
1.204 schwarze 592: h->tag = t;
1.30 kristaps 593: } else
594: t = NULL;
1.29 kristaps 595:
1.196 schwarze 596: if (tflags & HTML_NLBEFORE)
1.197 schwarze 597: print_endline(h);
598: if (h->col == 0)
599: print_indent(h);
1.196 schwarze 600: else if ((h->flags & HTML_NOSPACE) == 0) {
601: if (h->flags & HTML_KEEP)
1.216 schwarze 602: print_word(h, " ");
1.196 schwarze 603: else {
604: if (h->flags & HTML_PREKEEP)
605: h->flags |= HTML_KEEP;
1.197 schwarze 606: print_endword(h);
1.105 kristaps 607: }
1.196 schwarze 608: }
1.29 kristaps 609:
1.109 kristaps 610: if ( ! (h->flags & HTML_NONOSPACE))
611: h->flags &= ~HTML_NOSPACE;
1.110 kristaps 612: else
613: h->flags |= HTML_NOSPACE;
1.109 kristaps 614:
1.94 kristaps 615: /* Print out the tag name and attributes. */
616:
1.197 schwarze 617: print_byte(h, '<');
618: print_word(h, htmltags[tag].name);
1.194 schwarze 619:
620: va_start(ap, fmt);
621:
1.244 schwarze 622: while (*fmt != '\0' && *fmt != 's') {
1.203 schwarze 623:
1.238 schwarze 624: /* Parse attributes and arguments. */
1.203 schwarze 625:
626: arg1 = va_arg(ap, char *);
1.238 schwarze 627: arg2 = NULL;
1.194 schwarze 628: switch (*fmt++) {
629: case 'c':
1.195 schwarze 630: attr = "class";
1.194 schwarze 631: break;
632: case 'h':
1.195 schwarze 633: attr = "href";
1.194 schwarze 634: break;
635: case 'i':
1.195 schwarze 636: attr = "id";
1.194 schwarze 637: break;
638: case '?':
1.203 schwarze 639: attr = arg1;
640: arg1 = va_arg(ap, char *);
1.194 schwarze 641: break;
642: default:
643: abort();
644: }
1.203 schwarze 645: if (*fmt == 'M')
646: arg2 = va_arg(ap, char *);
647: if (arg1 == NULL)
648: continue;
649:
1.238 schwarze 650: /* Print the attributes. */
1.203 schwarze 651:
1.197 schwarze 652: print_byte(h, ' ');
653: print_word(h, attr);
654: print_byte(h, '=');
655: print_byte(h, '"');
1.195 schwarze 656: switch (*fmt) {
1.208 schwarze 657: case 'I':
658: print_href(h, arg1, NULL, 0);
659: fmt++;
660: break;
1.195 schwarze 661: case 'M':
1.203 schwarze 662: print_href(h, arg1, arg2, 1);
1.195 schwarze 663: fmt++;
664: break;
1.208 schwarze 665: case 'R':
666: print_byte(h, '#');
667: print_encode(h, arg1, NULL, 1);
1.195 schwarze 668: fmt++;
1.208 schwarze 669: break;
1.195 schwarze 670: default:
1.244 schwarze 671: print_encode(h, arg1, NULL, 1);
1.195 schwarze 672: break;
673: }
1.197 schwarze 674: print_byte(h, '"');
1.194 schwarze 675: }
1.244 schwarze 676:
677: style_written = 0;
678: while (*fmt++ == 's') {
679: arg1 = va_arg(ap, char *);
680: arg2 = va_arg(ap, char *);
681: if (arg2 == NULL)
682: continue;
683: print_byte(h, ' ');
684: if (style_written == 0) {
685: print_word(h, "style=\"");
686: style_written = 1;
687: }
688: print_word(h, arg1);
689: print_byte(h, ':');
690: print_byte(h, ' ');
691: print_word(h, arg2);
692: print_byte(h, ';');
693: }
694: if (style_written)
695: print_byte(h, '"');
696:
1.194 schwarze 697: va_end(ap);
1.94 kristaps 698:
1.172 kristaps 699: /* Accommodate for "well-formed" singleton escaping. */
1.94 kristaps 700:
1.93 kristaps 701: if (HTML_AUTOCLOSE & htmltags[tag].flags)
1.197 schwarze 702: print_byte(h, '/');
1.93 kristaps 703:
1.197 schwarze 704: print_byte(h, '>');
1.14 kristaps 705:
1.196 schwarze 706: if (tflags & HTML_NLBEGIN)
1.197 schwarze 707: print_endline(h);
1.196 schwarze 708: else
709: h->flags |= HTML_NOSPACE;
1.117 kristaps 710:
1.196 schwarze 711: if (tflags & HTML_INDENT)
712: h->indent++;
713: if (tflags & HTML_NOINDENT)
714: h->noindent++;
1.117 kristaps 715:
1.188 schwarze 716: return t;
1.14 kristaps 717: }
718:
1.29 kristaps 719: static void
1.184 schwarze 720: print_ctag(struct html *h, struct tag *tag)
1.14 kristaps 721: {
1.196 schwarze 722: int tflags;
1.156 schwarze 723:
1.252 ! schwarze 724: if (tag->closed == 0) {
! 725: tag->closed = 1;
! 726: if (tag == h->metaf)
! 727: h->metaf = NULL;
! 728: if (tag == h->tblt)
! 729: h->tblt = NULL;
! 730:
! 731: tflags = htmltags[tag->tag].flags;
! 732: if (tflags & HTML_INDENT)
! 733: h->indent--;
! 734: if (tflags & HTML_NOINDENT)
! 735: h->noindent--;
! 736: if (tflags & HTML_NLEND)
! 737: print_endline(h);
! 738: print_indent(h);
! 739: print_byte(h, '<');
! 740: print_byte(h, '/');
! 741: print_word(h, htmltags[tag->tag].name);
! 742: print_byte(h, '>');
! 743: if (tflags & HTML_NLAFTER)
! 744: print_endline(h);
! 745: }
! 746: if (tag->refcnt == 0) {
! 747: h->tag = tag->next;
! 748: free(tag);
! 749: }
1.14 kristaps 750: }
751:
1.51 kristaps 752: void
1.93 kristaps 753: print_gen_decls(struct html *h)
1.1 kristaps 754: {
1.197 schwarze 755: print_word(h, "<!DOCTYPE html>");
756: print_endline(h);
1.221 schwarze 757: }
758:
759: void
760: print_gen_comment(struct html *h, struct roff_node *n)
761: {
762: int wantblank;
763:
764: print_word(h, "<!-- This is an automatically generated file."
765: " Do not edit.");
766: h->indent = 1;
767: wantblank = 0;
768: while (n != NULL && n->type == ROFFT_COMMENT) {
769: if (strstr(n->string, "-->") == NULL &&
770: (wantblank || *n->string != '\0')) {
771: print_endline(h);
772: print_indent(h);
773: print_word(h, n->string);
774: wantblank = *n->string != '\0';
775: }
776: n = n->next;
777: }
778: if (wantblank)
779: print_endline(h);
780: print_word(h, " -->");
781: print_endline(h);
782: h->indent = 0;
1.1 kristaps 783: }
784:
1.51 kristaps 785: void
1.104 kristaps 786: print_text(struct html *h, const char *word)
1.1 kristaps 787: {
1.197 schwarze 788: if (h->col && (h->flags & HTML_NOSPACE) == 0) {
1.105 kristaps 789: if ( ! (HTML_KEEP & h->flags)) {
790: if (HTML_PREKEEP & h->flags)
791: h->flags |= HTML_KEEP;
1.197 schwarze 792: print_endword(h);
1.105 kristaps 793: } else
1.216 schwarze 794: print_word(h, " ");
1.105 kristaps 795: }
1.30 kristaps 796:
1.122 kristaps 797: assert(NULL == h->metaf);
1.152 schwarze 798: switch (h->metac) {
1.156 schwarze 799: case HTMLFONT_ITALIC:
1.194 schwarze 800: h->metaf = print_otag(h, TAG_I, "");
1.152 schwarze 801: break;
1.156 schwarze 802: case HTMLFONT_BOLD:
1.194 schwarze 803: h->metaf = print_otag(h, TAG_B, "");
1.152 schwarze 804: break;
1.156 schwarze 805: case HTMLFONT_BI:
1.194 schwarze 806: h->metaf = print_otag(h, TAG_B, "");
807: print_otag(h, TAG_I, "");
1.242 schwarze 808: break;
809: case HTMLFONT_CW:
810: h->metaf = print_otag(h, TAG_SPAN, "c", "Li");
1.152 schwarze 811: break;
812: default:
1.197 schwarze 813: print_indent(h);
1.152 schwarze 814: break;
815: }
1.122 kristaps 816:
1.104 kristaps 817: assert(word);
1.195 schwarze 818: if ( ! print_encode(h, word, NULL, 0)) {
1.109 kristaps 819: if ( ! (h->flags & HTML_NONOSPACE))
820: h->flags &= ~HTML_NOSPACE;
1.183 schwarze 821: h->flags &= ~HTML_NONEWLINE;
1.149 kristaps 822: } else
1.183 schwarze 823: h->flags |= HTML_NOSPACE | HTML_NONEWLINE;
1.122 kristaps 824:
825: if (h->metaf) {
826: print_tagq(h, h->metaf);
827: h->metaf = NULL;
828: }
1.113 schwarze 829:
830: h->flags &= ~HTML_IGNDELIM;
1.1 kristaps 831: }
1.30 kristaps 832:
1.51 kristaps 833: void
1.30 kristaps 834: print_tagq(struct html *h, const struct tag *until)
835: {
1.252 ! schwarze 836: struct tag *this, *next;
1.30 kristaps 837:
1.252 ! schwarze 838: for (this = h->tag; this != NULL; this = next) {
! 839: next = this == until ? NULL : this->next;
! 840: print_ctag(h, this);
1.30 kristaps 841: }
842: }
843:
1.250 schwarze 844: /*
845: * Close out all open elements up to but excluding suntil.
846: * Note that a paragraph just inside stays open together with it
847: * because paragraphs include subsequent phrasing content.
848: */
1.51 kristaps 849: void
1.30 kristaps 850: print_stagq(struct html *h, const struct tag *suntil)
851: {
1.252 ! schwarze 852: struct tag *this, *next;
1.30 kristaps 853:
1.252 ! schwarze 854: for (this = h->tag; this != NULL; this = next) {
! 855: next = this->next;
! 856: if (this == suntil || (next == suntil &&
! 857: (this->tag == TAG_P || this->tag == TAG_PRE)))
! 858: break;
! 859: print_ctag(h, this);
1.30 kristaps 860: }
1.171 kristaps 861: }
862:
1.197 schwarze 863:
864: /***********************************************************************
865: * Low level output functions.
866: * They implement line breaking using a short static buffer.
867: ***********************************************************************/
868:
869: /*
870: * Buffer one HTML output byte.
871: * If the buffer is full, flush and deactivate it and start a new line.
872: * If the buffer is inactive, print directly.
873: */
874: static void
875: print_byte(struct html *h, char c)
876: {
877: if ((h->flags & HTML_BUFFER) == 0) {
878: putchar(c);
879: h->col++;
880: return;
881: }
882:
883: if (h->col + h->bufcol < sizeof(h->buf)) {
884: h->buf[h->bufcol++] = c;
885: return;
886: }
887:
888: putchar('\n');
889: h->col = 0;
890: print_indent(h);
891: putchar(' ');
892: putchar(' ');
893: fwrite(h->buf, h->bufcol, 1, stdout);
894: putchar(c);
895: h->col = (h->indent + 1) * 2 + h->bufcol + 1;
896: h->bufcol = 0;
897: h->flags &= ~HTML_BUFFER;
898: }
899:
1.196 schwarze 900: /*
901: * If something was printed on the current output line, end it.
1.197 schwarze 902: * Not to be called right after print_indent().
1.196 schwarze 903: */
1.202 schwarze 904: void
1.197 schwarze 905: print_endline(struct html *h)
1.196 schwarze 906: {
1.197 schwarze 907: if (h->col == 0)
1.196 schwarze 908: return;
909:
1.197 schwarze 910: if (h->bufcol) {
911: putchar(' ');
912: fwrite(h->buf, h->bufcol, 1, stdout);
913: h->bufcol = 0;
914: }
1.196 schwarze 915: putchar('\n');
1.197 schwarze 916: h->col = 0;
917: h->flags |= HTML_NOSPACE;
918: h->flags &= ~HTML_BUFFER;
919: }
920:
921: /*
922: * Flush the HTML output buffer.
923: * If it is inactive, activate it.
924: */
925: static void
926: print_endword(struct html *h)
927: {
928: if (h->noindent) {
929: print_byte(h, ' ');
930: return;
931: }
932:
933: if ((h->flags & HTML_BUFFER) == 0) {
934: h->col++;
935: h->flags |= HTML_BUFFER;
936: } else if (h->bufcol) {
937: putchar(' ');
938: fwrite(h->buf, h->bufcol, 1, stdout);
939: h->col += h->bufcol + 1;
940: }
941: h->bufcol = 0;
1.196 schwarze 942: }
943:
944: /*
945: * If at the beginning of a new output line,
946: * perform indentation and mark the line as containing output.
947: * Make sure to really produce some output right afterwards,
948: * but do not use print_otag() for producing it.
949: */
950: static void
1.197 schwarze 951: print_indent(struct html *h)
1.196 schwarze 952: {
1.197 schwarze 953: size_t i;
1.196 schwarze 954:
1.197 schwarze 955: if (h->col)
1.196 schwarze 956: return;
957:
1.197 schwarze 958: if (h->noindent == 0) {
959: h->col = h->indent * 2;
960: for (i = 0; i < h->col; i++)
1.196 schwarze 961: putchar(' ');
1.197 schwarze 962: }
963: h->flags &= ~HTML_NOSPACE;
964: }
965:
966: /*
967: * Print or buffer some characters
968: * depending on the current HTML output buffer state.
969: */
970: static void
971: print_word(struct html *h, const char *cp)
972: {
973: while (*cp != '\0')
974: print_byte(h, *cp++);
1.68 kristaps 975: }
CVSweb