Annotation of mandoc/html.c, Revision 1.246
1.246 ! schwarze 1: /* $Id: html.c,v 1.245 2018/12/04 18:29:38 schwarze Exp $ */
1.1 kristaps 2: /*
1.176 schwarze 3: * Copyright (c) 2008-2011, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
1.221 schwarze 4: * Copyright (c) 2011-2015, 2017, 2018 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},
81: {"pre", HTML_NLALL | HTML_NOINDENT},
1.207 schwarze 82: {"var", 0},
1.206 schwarze 83: {"cite", 0},
1.196 schwarze 84: {"b", 0},
85: {"i", 0},
86: {"code", 0},
87: {"small", 0},
88: {"style", HTML_NLALL | HTML_INDENT},
89: {"math", HTML_NLALL | HTML_INDENT},
90: {"mrow", 0},
91: {"mi", 0},
1.215 schwarze 92: {"mn", 0},
1.196 schwarze 93: {"mo", 0},
94: {"msup", 0},
95: {"msub", 0},
96: {"msubsup", 0},
97: {"mfrac", 0},
98: {"msqrt", 0},
99: {"mfenced", 0},
100: {"mtable", 0},
101: {"mtr", 0},
102: {"mtd", 0},
103: {"munderover", 0},
104: {"munder", 0},
105: {"mover", 0},
1.90 kristaps 106: };
107:
1.229 schwarze 108: /* Avoid duplicate HTML id= attributes. */
109: static struct ohash id_unique;
110:
1.197 schwarze 111: static void print_byte(struct html *, char);
112: static void print_endword(struct html *);
113: static void print_indent(struct html *);
114: static void print_word(struct html *, const char *);
115:
1.184 schwarze 116: static void print_ctag(struct html *, struct tag *);
1.197 schwarze 117: static int print_escape(struct html *, char);
1.195 schwarze 118: static int print_encode(struct html *, const char *, const char *, int);
119: static void print_href(struct html *, const char *, const char *, int);
1.141 kristaps 120: static void print_metaf(struct html *, enum mandoc_esc);
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:
1.126 schwarze 213: static 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:
239: abort();
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: }
1.210 schwarze 267: }
268:
269: char *
1.229 schwarze 270: html_make_id(const struct roff_node *n, int unique)
1.210 schwarze 271: {
272: const struct roff_node *nch;
1.229 schwarze 273: char *buf, *bufs, *cp;
274: unsigned int slot;
275: int suffix;
1.210 schwarze 276:
277: for (nch = n->child; nch != NULL; nch = nch->next)
278: if (nch->type != ROFFT_TEXT)
279: return NULL;
280:
281: buf = NULL;
282: deroff(&buf, n);
1.220 schwarze 283: if (buf == NULL)
284: return NULL;
1.210 schwarze 285:
1.230 schwarze 286: /*
287: * In ID attributes, only use ASCII characters that are
288: * permitted in URL-fragment strings according to the
289: * explicit list at:
290: * https://url.spec.whatwg.org/#url-fragment-string
291: */
1.210 schwarze 292:
293: for (cp = buf; *cp != '\0'; cp++)
1.230 schwarze 294: if (isalnum((unsigned char)*cp) == 0 &&
295: strchr("!$&'()*+,-./:;=?@_~", *cp) == NULL)
1.210 schwarze 296: *cp = '_';
297:
1.229 schwarze 298: if (unique == 0)
299: return buf;
300:
301: /* Avoid duplicate HTML id= attributes. */
302:
303: bufs = NULL;
304: suffix = 1;
305: slot = ohash_qlookup(&id_unique, buf);
306: cp = ohash_find(&id_unique, slot);
307: if (cp != NULL) {
308: while (cp != NULL) {
309: free(bufs);
310: if (++suffix > 127) {
311: free(buf);
312: return NULL;
313: }
314: mandoc_asprintf(&bufs, "%s_%d", buf, suffix);
315: slot = ohash_qlookup(&id_unique, bufs);
316: cp = ohash_find(&id_unique, slot);
317: }
318: free(buf);
319: buf = bufs;
320: }
321: ohash_insert(&id_unique, slot, buf);
1.210 schwarze 322: return buf;
1.88 kristaps 323: }
324:
1.85 kristaps 325: static int
1.197 schwarze 326: print_escape(struct html *h, char c)
1.159 schwarze 327: {
328:
329: switch (c) {
330: case '<':
1.197 schwarze 331: print_word(h, "<");
1.159 schwarze 332: break;
333: case '>':
1.197 schwarze 334: print_word(h, ">");
1.159 schwarze 335: break;
336: case '&':
1.197 schwarze 337: print_word(h, "&");
1.159 schwarze 338: break;
339: case '"':
1.197 schwarze 340: print_word(h, """);
1.159 schwarze 341: break;
342: case ASCII_NBRSP:
1.197 schwarze 343: print_word(h, " ");
1.159 schwarze 344: break;
345: case ASCII_HYPH:
1.197 schwarze 346: print_byte(h, '-');
1.189 schwarze 347: break;
1.159 schwarze 348: case ASCII_BREAK:
349: break;
350: default:
1.188 schwarze 351: return 0;
1.159 schwarze 352: }
1.188 schwarze 353: return 1;
1.159 schwarze 354: }
355:
356: static int
1.195 schwarze 357: print_encode(struct html *h, const char *p, const char *pend, int norecurse)
1.29 kristaps 358: {
1.197 schwarze 359: char numbuf[16];
1.214 schwarze 360: const char *seq;
1.77 kristaps 361: size_t sz;
1.214 schwarze 362: int c, len, breakline, nospace;
1.132 kristaps 363: enum mandoc_esc esc;
1.214 schwarze 364: static const char rejs[10] = { ' ', '\\', '<', '>', '&', '"',
1.154 schwarze 365: ASCII_NBRSP, ASCII_HYPH, ASCII_BREAK, '\0' };
1.14 kristaps 366:
1.195 schwarze 367: if (pend == NULL)
368: pend = strchr(p, '\0');
369:
1.214 schwarze 370: breakline = 0;
1.85 kristaps 371: nospace = 0;
372:
1.195 schwarze 373: while (p < pend) {
1.151 schwarze 374: if (HTML_SKIPCHAR & h->flags && '\\' != *p) {
375: h->flags &= ~HTML_SKIPCHAR;
376: p++;
377: continue;
378: }
379:
1.197 schwarze 380: for (sz = strcspn(p, rejs); sz-- && p < pend; p++)
1.214 schwarze 381: print_byte(h, *p);
382:
383: if (breakline &&
384: (p >= pend || *p == ' ' || *p == ASCII_NBRSP)) {
1.245 schwarze 385: print_otag(h, TAG_BR, "");
1.214 schwarze 386: breakline = 0;
387: while (p < pend && (*p == ' ' || *p == ASCII_NBRSP))
388: p++;
389: continue;
390: }
1.77 kristaps 391:
1.195 schwarze 392: if (p >= pend)
1.132 kristaps 393: break;
394:
1.214 schwarze 395: if (*p == ' ') {
396: print_endword(h);
397: p++;
398: continue;
399: }
400:
1.197 schwarze 401: if (print_escape(h, *p++))
1.154 schwarze 402: continue;
1.77 kristaps 403:
1.132 kristaps 404: esc = mandoc_escape(&p, &seq, &len);
405: switch (esc) {
1.156 schwarze 406: case ESCAPE_FONT:
407: case ESCAPE_FONTPREV:
408: case ESCAPE_FONTBOLD:
409: case ESCAPE_FONTITALIC:
410: case ESCAPE_FONTBI:
1.242 schwarze 411: case ESCAPE_FONTCW:
1.156 schwarze 412: case ESCAPE_FONTROMAN:
1.243 schwarze 413: if (0 == norecurse) {
414: h->flags |= HTML_NOSPACE;
1.151 schwarze 415: print_metaf(h, esc);
1.243 schwarze 416: h->flags &= ~HTML_NOSPACE;
417: }
1.151 schwarze 418: continue;
1.156 schwarze 419: case ESCAPE_SKIPCHAR:
1.151 schwarze 420: h->flags |= HTML_SKIPCHAR;
421: continue;
1.246 ! schwarze 422: case ESCAPE_ERROR:
! 423: continue;
1.151 schwarze 424: default:
425: break;
426: }
427:
428: if (h->flags & HTML_SKIPCHAR) {
429: h->flags &= ~HTML_SKIPCHAR;
430: continue;
431: }
432:
433: switch (esc) {
1.156 schwarze 434: case ESCAPE_UNICODE:
1.159 schwarze 435: /* Skip past "u" header. */
1.144 kristaps 436: c = mchars_num2uc(seq + 1, len - 1);
437: break;
1.156 schwarze 438: case ESCAPE_NUMBERED:
1.141 kristaps 439: c = mchars_num2char(seq, len);
1.181 schwarze 440: if (c < 0)
441: continue;
1.82 kristaps 442: break;
1.156 schwarze 443: case ESCAPE_SPECIAL:
1.191 schwarze 444: c = mchars_spec2cp(seq, len);
1.181 schwarze 445: if (c <= 0)
446: continue;
1.246 ! schwarze 447: break;
! 448: case ESCAPE_UNDEF:
! 449: c = *seq;
1.132 kristaps 450: break;
1.239 schwarze 451: case ESCAPE_DEVICE:
452: print_word(h, "html");
453: continue;
1.214 schwarze 454: case ESCAPE_BREAK:
455: breakline = 1;
456: continue;
1.156 schwarze 457: case ESCAPE_NOSPACE:
1.132 kristaps 458: if ('\0' == *p)
459: nospace = 1;
1.179 schwarze 460: continue;
1.185 schwarze 461: case ESCAPE_OVERSTRIKE:
462: if (len == 0)
463: continue;
464: c = seq[len - 1];
465: break;
1.82 kristaps 466: default:
1.179 schwarze 467: continue;
1.82 kristaps 468: }
1.181 schwarze 469: if ((c < 0x20 && c != 0x09) ||
470: (c > 0x7E && c < 0xA0))
1.179 schwarze 471: c = 0xFFFD;
1.197 schwarze 472: if (c > 0x7E) {
1.216 schwarze 473: (void)snprintf(numbuf, sizeof(numbuf), "&#x%.4X;", c);
1.197 schwarze 474: print_word(h, numbuf);
475: } else if (print_escape(h, c) == 0)
476: print_byte(h, c);
1.32 kristaps 477: }
1.85 kristaps 478:
1.188 schwarze 479: return nospace;
1.14 kristaps 480: }
481:
1.94 kristaps 482: static void
1.195 schwarze 483: print_href(struct html *h, const char *name, const char *sec, int man)
1.94 kristaps 484: {
1.240 schwarze 485: struct stat sb;
1.195 schwarze 486: const char *p, *pp;
1.240 schwarze 487: char *filename;
488:
489: if (man) {
490: pp = h->base_man1;
491: if (h->base_man2 != NULL) {
492: mandoc_asprintf(&filename, "%s.%s", name, sec);
493: if (stat(filename, &sb) == -1)
494: pp = h->base_man2;
495: free(filename);
496: }
497: } else
498: pp = h->base_includes;
1.195 schwarze 499:
500: while ((p = strchr(pp, '%')) != NULL) {
501: print_encode(h, pp, p, 1);
502: if (man && p[1] == 'S') {
503: if (sec == NULL)
1.197 schwarze 504: print_byte(h, '1');
1.195 schwarze 505: else
506: print_encode(h, sec, NULL, 1);
507: } else if ((man && p[1] == 'N') ||
508: (man == 0 && p[1] == 'I'))
509: print_encode(h, name, NULL, 1);
510: else
511: print_encode(h, p, p + 2, 1);
512: pp = p + 2;
513: }
514: if (*pp != '\0')
515: print_encode(h, pp, NULL, 1);
1.94 kristaps 516: }
517:
1.51 kristaps 518: struct tag *
1.194 schwarze 519: print_otag(struct html *h, enum htmltag tag, const char *fmt, ...)
1.14 kristaps 520: {
1.194 schwarze 521: va_list ap;
1.30 kristaps 522: struct tag *t;
1.195 schwarze 523: const char *attr;
1.203 schwarze 524: char *arg1, *arg2;
1.244 schwarze 525: int style_written, tflags;
1.196 schwarze 526:
527: tflags = htmltags[tag].flags;
1.30 kristaps 528:
1.204 schwarze 529: /* Push this tag onto the stack of open scopes. */
1.94 kristaps 530:
1.196 schwarze 531: if ((tflags & HTML_NOSTACK) == 0) {
1.128 kristaps 532: t = mandoc_malloc(sizeof(struct tag));
1.30 kristaps 533: t->tag = tag;
1.204 schwarze 534: t->next = h->tag;
535: h->tag = t;
1.30 kristaps 536: } else
537: t = NULL;
1.29 kristaps 538:
1.196 schwarze 539: if (tflags & HTML_NLBEFORE)
1.197 schwarze 540: print_endline(h);
541: if (h->col == 0)
542: print_indent(h);
1.196 schwarze 543: else if ((h->flags & HTML_NOSPACE) == 0) {
544: if (h->flags & HTML_KEEP)
1.216 schwarze 545: print_word(h, " ");
1.196 schwarze 546: else {
547: if (h->flags & HTML_PREKEEP)
548: h->flags |= HTML_KEEP;
1.197 schwarze 549: print_endword(h);
1.105 kristaps 550: }
1.196 schwarze 551: }
1.29 kristaps 552:
1.109 kristaps 553: if ( ! (h->flags & HTML_NONOSPACE))
554: h->flags &= ~HTML_NOSPACE;
1.110 kristaps 555: else
556: h->flags |= HTML_NOSPACE;
1.109 kristaps 557:
1.94 kristaps 558: /* Print out the tag name and attributes. */
559:
1.197 schwarze 560: print_byte(h, '<');
561: print_word(h, htmltags[tag].name);
1.194 schwarze 562:
563: va_start(ap, fmt);
564:
1.244 schwarze 565: while (*fmt != '\0' && *fmt != 's') {
1.203 schwarze 566:
1.238 schwarze 567: /* Parse attributes and arguments. */
1.203 schwarze 568:
569: arg1 = va_arg(ap, char *);
1.238 schwarze 570: arg2 = NULL;
1.194 schwarze 571: switch (*fmt++) {
572: case 'c':
1.195 schwarze 573: attr = "class";
1.194 schwarze 574: break;
575: case 'h':
1.195 schwarze 576: attr = "href";
1.194 schwarze 577: break;
578: case 'i':
1.195 schwarze 579: attr = "id";
1.194 schwarze 580: break;
581: case '?':
1.203 schwarze 582: attr = arg1;
583: arg1 = va_arg(ap, char *);
1.194 schwarze 584: break;
585: default:
586: abort();
587: }
1.203 schwarze 588: if (*fmt == 'M')
589: arg2 = va_arg(ap, char *);
590: if (arg1 == NULL)
591: continue;
592:
1.238 schwarze 593: /* Print the attributes. */
1.203 schwarze 594:
1.197 schwarze 595: print_byte(h, ' ');
596: print_word(h, attr);
597: print_byte(h, '=');
598: print_byte(h, '"');
1.195 schwarze 599: switch (*fmt) {
1.208 schwarze 600: case 'I':
601: print_href(h, arg1, NULL, 0);
602: fmt++;
603: break;
1.195 schwarze 604: case 'M':
1.203 schwarze 605: print_href(h, arg1, arg2, 1);
1.195 schwarze 606: fmt++;
607: break;
1.208 schwarze 608: case 'R':
609: print_byte(h, '#');
610: print_encode(h, arg1, NULL, 1);
1.195 schwarze 611: fmt++;
612: break;
1.208 schwarze 613: case 'T':
614: print_encode(h, arg1, NULL, 1);
615: print_word(h, "\" title=\"");
616: print_encode(h, arg1, NULL, 1);
1.195 schwarze 617: fmt++;
1.208 schwarze 618: break;
1.195 schwarze 619: default:
1.244 schwarze 620: print_encode(h, arg1, NULL, 1);
1.195 schwarze 621: break;
622: }
1.197 schwarze 623: print_byte(h, '"');
1.194 schwarze 624: }
1.244 schwarze 625:
626: style_written = 0;
627: while (*fmt++ == 's') {
628: arg1 = va_arg(ap, char *);
629: arg2 = va_arg(ap, char *);
630: if (arg2 == NULL)
631: continue;
632: print_byte(h, ' ');
633: if (style_written == 0) {
634: print_word(h, "style=\"");
635: style_written = 1;
636: }
637: print_word(h, arg1);
638: print_byte(h, ':');
639: print_byte(h, ' ');
640: print_word(h, arg2);
641: print_byte(h, ';');
642: }
643: if (style_written)
644: print_byte(h, '"');
645:
1.194 schwarze 646: va_end(ap);
1.94 kristaps 647:
1.172 kristaps 648: /* Accommodate for "well-formed" singleton escaping. */
1.94 kristaps 649:
1.93 kristaps 650: if (HTML_AUTOCLOSE & htmltags[tag].flags)
1.197 schwarze 651: print_byte(h, '/');
1.93 kristaps 652:
1.197 schwarze 653: print_byte(h, '>');
1.14 kristaps 654:
1.196 schwarze 655: if (tflags & HTML_NLBEGIN)
1.197 schwarze 656: print_endline(h);
1.196 schwarze 657: else
658: h->flags |= HTML_NOSPACE;
1.117 kristaps 659:
1.196 schwarze 660: if (tflags & HTML_INDENT)
661: h->indent++;
662: if (tflags & HTML_NOINDENT)
663: h->noindent++;
1.117 kristaps 664:
1.188 schwarze 665: return t;
1.14 kristaps 666: }
667:
1.29 kristaps 668: static void
1.184 schwarze 669: print_ctag(struct html *h, struct tag *tag)
1.14 kristaps 670: {
1.196 schwarze 671: int tflags;
1.156 schwarze 672:
1.184 schwarze 673: /*
674: * Remember to close out and nullify the current
675: * meta-font and table, if applicable.
676: */
677: if (tag == h->metaf)
678: h->metaf = NULL;
679: if (tag == h->tblt)
680: h->tblt = NULL;
681:
1.196 schwarze 682: tflags = htmltags[tag->tag].flags;
683:
684: if (tflags & HTML_INDENT)
685: h->indent--;
686: if (tflags & HTML_NOINDENT)
687: h->noindent--;
688: if (tflags & HTML_NLEND)
1.197 schwarze 689: print_endline(h);
690: print_indent(h);
691: print_byte(h, '<');
692: print_byte(h, '/');
693: print_word(h, htmltags[tag->tag].name);
694: print_byte(h, '>');
1.196 schwarze 695: if (tflags & HTML_NLAFTER)
1.197 schwarze 696: print_endline(h);
1.184 schwarze 697:
1.204 schwarze 698: h->tag = tag->next;
1.184 schwarze 699: free(tag);
1.14 kristaps 700: }
701:
1.51 kristaps 702: void
1.93 kristaps 703: print_gen_decls(struct html *h)
1.1 kristaps 704: {
1.197 schwarze 705: print_word(h, "<!DOCTYPE html>");
706: print_endline(h);
1.221 schwarze 707: }
708:
709: void
710: print_gen_comment(struct html *h, struct roff_node *n)
711: {
712: int wantblank;
713:
714: print_word(h, "<!-- This is an automatically generated file."
715: " Do not edit.");
716: h->indent = 1;
717: wantblank = 0;
718: while (n != NULL && n->type == ROFFT_COMMENT) {
719: if (strstr(n->string, "-->") == NULL &&
720: (wantblank || *n->string != '\0')) {
721: print_endline(h);
722: print_indent(h);
723: print_word(h, n->string);
724: wantblank = *n->string != '\0';
725: }
726: n = n->next;
727: }
728: if (wantblank)
729: print_endline(h);
730: print_word(h, " -->");
731: print_endline(h);
732: h->indent = 0;
1.1 kristaps 733: }
734:
1.51 kristaps 735: void
1.104 kristaps 736: print_text(struct html *h, const char *word)
1.1 kristaps 737: {
1.197 schwarze 738: if (h->col && (h->flags & HTML_NOSPACE) == 0) {
1.105 kristaps 739: if ( ! (HTML_KEEP & h->flags)) {
740: if (HTML_PREKEEP & h->flags)
741: h->flags |= HTML_KEEP;
1.197 schwarze 742: print_endword(h);
1.105 kristaps 743: } else
1.216 schwarze 744: print_word(h, " ");
1.105 kristaps 745: }
1.30 kristaps 746:
1.122 kristaps 747: assert(NULL == h->metaf);
1.152 schwarze 748: switch (h->metac) {
1.156 schwarze 749: case HTMLFONT_ITALIC:
1.194 schwarze 750: h->metaf = print_otag(h, TAG_I, "");
1.152 schwarze 751: break;
1.156 schwarze 752: case HTMLFONT_BOLD:
1.194 schwarze 753: h->metaf = print_otag(h, TAG_B, "");
1.152 schwarze 754: break;
1.156 schwarze 755: case HTMLFONT_BI:
1.194 schwarze 756: h->metaf = print_otag(h, TAG_B, "");
757: print_otag(h, TAG_I, "");
1.242 schwarze 758: break;
759: case HTMLFONT_CW:
760: h->metaf = print_otag(h, TAG_SPAN, "c", "Li");
1.152 schwarze 761: break;
762: default:
1.197 schwarze 763: print_indent(h);
1.152 schwarze 764: break;
765: }
1.122 kristaps 766:
1.104 kristaps 767: assert(word);
1.195 schwarze 768: if ( ! print_encode(h, word, NULL, 0)) {
1.109 kristaps 769: if ( ! (h->flags & HTML_NONOSPACE))
770: h->flags &= ~HTML_NOSPACE;
1.183 schwarze 771: h->flags &= ~HTML_NONEWLINE;
1.149 kristaps 772: } else
1.183 schwarze 773: h->flags |= HTML_NOSPACE | HTML_NONEWLINE;
1.122 kristaps 774:
775: if (h->metaf) {
776: print_tagq(h, h->metaf);
777: h->metaf = NULL;
778: }
1.113 schwarze 779:
780: h->flags &= ~HTML_IGNDELIM;
1.1 kristaps 781: }
1.30 kristaps 782:
1.51 kristaps 783: void
1.30 kristaps 784: print_tagq(struct html *h, const struct tag *until)
785: {
786: struct tag *tag;
787:
1.204 schwarze 788: while ((tag = h->tag) != NULL) {
1.184 schwarze 789: print_ctag(h, tag);
1.30 kristaps 790: if (until && tag == until)
791: return;
792: }
793: }
794:
1.51 kristaps 795: void
1.30 kristaps 796: print_stagq(struct html *h, const struct tag *suntil)
797: {
798: struct tag *tag;
799:
1.204 schwarze 800: while ((tag = h->tag) != NULL) {
1.30 kristaps 801: if (suntil && tag == suntil)
802: return;
1.184 schwarze 803: print_ctag(h, tag);
1.30 kristaps 804: }
805: }
1.171 kristaps 806:
807: void
808: print_paragraph(struct html *h)
809: {
810: struct tag *t;
811:
1.198 schwarze 812: t = print_otag(h, TAG_DIV, "c", "Pp");
1.171 kristaps 813: print_tagq(h, t);
814: }
815:
1.197 schwarze 816:
817: /***********************************************************************
818: * Low level output functions.
819: * They implement line breaking using a short static buffer.
820: ***********************************************************************/
821:
822: /*
823: * Buffer one HTML output byte.
824: * If the buffer is full, flush and deactivate it and start a new line.
825: * If the buffer is inactive, print directly.
826: */
827: static void
828: print_byte(struct html *h, char c)
829: {
830: if ((h->flags & HTML_BUFFER) == 0) {
831: putchar(c);
832: h->col++;
833: return;
834: }
835:
836: if (h->col + h->bufcol < sizeof(h->buf)) {
837: h->buf[h->bufcol++] = c;
838: return;
839: }
840:
841: putchar('\n');
842: h->col = 0;
843: print_indent(h);
844: putchar(' ');
845: putchar(' ');
846: fwrite(h->buf, h->bufcol, 1, stdout);
847: putchar(c);
848: h->col = (h->indent + 1) * 2 + h->bufcol + 1;
849: h->bufcol = 0;
850: h->flags &= ~HTML_BUFFER;
851: }
852:
1.196 schwarze 853: /*
854: * If something was printed on the current output line, end it.
1.197 schwarze 855: * Not to be called right after print_indent().
1.196 schwarze 856: */
1.202 schwarze 857: void
1.197 schwarze 858: print_endline(struct html *h)
1.196 schwarze 859: {
1.197 schwarze 860: if (h->col == 0)
1.196 schwarze 861: return;
862:
1.197 schwarze 863: if (h->bufcol) {
864: putchar(' ');
865: fwrite(h->buf, h->bufcol, 1, stdout);
866: h->bufcol = 0;
867: }
1.196 schwarze 868: putchar('\n');
1.197 schwarze 869: h->col = 0;
870: h->flags |= HTML_NOSPACE;
871: h->flags &= ~HTML_BUFFER;
872: }
873:
874: /*
875: * Flush the HTML output buffer.
876: * If it is inactive, activate it.
877: */
878: static void
879: print_endword(struct html *h)
880: {
881: if (h->noindent) {
882: print_byte(h, ' ');
883: return;
884: }
885:
886: if ((h->flags & HTML_BUFFER) == 0) {
887: h->col++;
888: h->flags |= HTML_BUFFER;
889: } else if (h->bufcol) {
890: putchar(' ');
891: fwrite(h->buf, h->bufcol, 1, stdout);
892: h->col += h->bufcol + 1;
893: }
894: h->bufcol = 0;
1.196 schwarze 895: }
896:
897: /*
898: * If at the beginning of a new output line,
899: * perform indentation and mark the line as containing output.
900: * Make sure to really produce some output right afterwards,
901: * but do not use print_otag() for producing it.
902: */
903: static void
1.197 schwarze 904: print_indent(struct html *h)
1.196 schwarze 905: {
1.197 schwarze 906: size_t i;
1.196 schwarze 907:
1.197 schwarze 908: if (h->col)
1.196 schwarze 909: return;
910:
1.197 schwarze 911: if (h->noindent == 0) {
912: h->col = h->indent * 2;
913: for (i = 0; i < h->col; i++)
1.196 schwarze 914: putchar(' ');
1.197 schwarze 915: }
916: h->flags &= ~HTML_NOSPACE;
917: }
918:
919: /*
920: * Print or buffer some characters
921: * depending on the current HTML output buffer state.
922: */
923: static void
924: print_word(struct html *h, const char *cp)
925: {
926: while (*cp != '\0')
927: print_byte(h, *cp++);
1.68 kristaps 928: }
CVSweb