Annotation of mandoc/html.c, Revision 1.245
1.245 ! schwarze 1: /* $Id: html.c,v 1.244 2018/11/26 01:38:23 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: if (ESCAPE_ERROR == esc)
406: break;
1.82 kristaps 407:
1.132 kristaps 408: switch (esc) {
1.156 schwarze 409: case ESCAPE_FONT:
410: case ESCAPE_FONTPREV:
411: case ESCAPE_FONTBOLD:
412: case ESCAPE_FONTITALIC:
413: case ESCAPE_FONTBI:
1.242 schwarze 414: case ESCAPE_FONTCW:
1.156 schwarze 415: case ESCAPE_FONTROMAN:
1.243 schwarze 416: if (0 == norecurse) {
417: h->flags |= HTML_NOSPACE;
1.151 schwarze 418: print_metaf(h, esc);
1.243 schwarze 419: h->flags &= ~HTML_NOSPACE;
420: }
1.151 schwarze 421: continue;
1.156 schwarze 422: case ESCAPE_SKIPCHAR:
1.151 schwarze 423: h->flags |= HTML_SKIPCHAR;
424: continue;
425: default:
426: break;
427: }
428:
429: if (h->flags & HTML_SKIPCHAR) {
430: h->flags &= ~HTML_SKIPCHAR;
431: continue;
432: }
433:
434: switch (esc) {
1.156 schwarze 435: case ESCAPE_UNICODE:
1.159 schwarze 436: /* Skip past "u" header. */
1.144 kristaps 437: c = mchars_num2uc(seq + 1, len - 1);
438: break;
1.156 schwarze 439: case ESCAPE_NUMBERED:
1.141 kristaps 440: c = mchars_num2char(seq, len);
1.181 schwarze 441: if (c < 0)
442: continue;
1.82 kristaps 443: break;
1.156 schwarze 444: case ESCAPE_SPECIAL:
1.191 schwarze 445: c = mchars_spec2cp(seq, len);
1.181 schwarze 446: if (c <= 0)
447: continue;
1.132 kristaps 448: break;
1.239 schwarze 449: case ESCAPE_DEVICE:
450: print_word(h, "html");
451: continue;
1.214 schwarze 452: case ESCAPE_BREAK:
453: breakline = 1;
454: continue;
1.156 schwarze 455: case ESCAPE_NOSPACE:
1.132 kristaps 456: if ('\0' == *p)
457: nospace = 1;
1.179 schwarze 458: continue;
1.185 schwarze 459: case ESCAPE_OVERSTRIKE:
460: if (len == 0)
461: continue;
462: c = seq[len - 1];
463: break;
1.82 kristaps 464: default:
1.179 schwarze 465: continue;
1.82 kristaps 466: }
1.181 schwarze 467: if ((c < 0x20 && c != 0x09) ||
468: (c > 0x7E && c < 0xA0))
1.179 schwarze 469: c = 0xFFFD;
1.197 schwarze 470: if (c > 0x7E) {
1.216 schwarze 471: (void)snprintf(numbuf, sizeof(numbuf), "&#x%.4X;", c);
1.197 schwarze 472: print_word(h, numbuf);
473: } else if (print_escape(h, c) == 0)
474: print_byte(h, c);
1.32 kristaps 475: }
1.85 kristaps 476:
1.188 schwarze 477: return nospace;
1.14 kristaps 478: }
479:
1.94 kristaps 480: static void
1.195 schwarze 481: print_href(struct html *h, const char *name, const char *sec, int man)
1.94 kristaps 482: {
1.240 schwarze 483: struct stat sb;
1.195 schwarze 484: const char *p, *pp;
1.240 schwarze 485: char *filename;
486:
487: if (man) {
488: pp = h->base_man1;
489: if (h->base_man2 != NULL) {
490: mandoc_asprintf(&filename, "%s.%s", name, sec);
491: if (stat(filename, &sb) == -1)
492: pp = h->base_man2;
493: free(filename);
494: }
495: } else
496: pp = h->base_includes;
1.195 schwarze 497:
498: while ((p = strchr(pp, '%')) != NULL) {
499: print_encode(h, pp, p, 1);
500: if (man && p[1] == 'S') {
501: if (sec == NULL)
1.197 schwarze 502: print_byte(h, '1');
1.195 schwarze 503: else
504: print_encode(h, sec, NULL, 1);
505: } else if ((man && p[1] == 'N') ||
506: (man == 0 && p[1] == 'I'))
507: print_encode(h, name, NULL, 1);
508: else
509: print_encode(h, p, p + 2, 1);
510: pp = p + 2;
511: }
512: if (*pp != '\0')
513: print_encode(h, pp, NULL, 1);
1.94 kristaps 514: }
515:
1.51 kristaps 516: struct tag *
1.194 schwarze 517: print_otag(struct html *h, enum htmltag tag, const char *fmt, ...)
1.14 kristaps 518: {
1.194 schwarze 519: va_list ap;
1.30 kristaps 520: struct tag *t;
1.195 schwarze 521: const char *attr;
1.203 schwarze 522: char *arg1, *arg2;
1.244 schwarze 523: int style_written, tflags;
1.196 schwarze 524:
525: tflags = htmltags[tag].flags;
1.30 kristaps 526:
1.204 schwarze 527: /* Push this tag onto the stack of open scopes. */
1.94 kristaps 528:
1.196 schwarze 529: if ((tflags & HTML_NOSTACK) == 0) {
1.128 kristaps 530: t = mandoc_malloc(sizeof(struct tag));
1.30 kristaps 531: t->tag = tag;
1.204 schwarze 532: t->next = h->tag;
533: h->tag = t;
1.30 kristaps 534: } else
535: t = NULL;
1.29 kristaps 536:
1.196 schwarze 537: if (tflags & HTML_NLBEFORE)
1.197 schwarze 538: print_endline(h);
539: if (h->col == 0)
540: print_indent(h);
1.196 schwarze 541: else if ((h->flags & HTML_NOSPACE) == 0) {
542: if (h->flags & HTML_KEEP)
1.216 schwarze 543: print_word(h, " ");
1.196 schwarze 544: else {
545: if (h->flags & HTML_PREKEEP)
546: h->flags |= HTML_KEEP;
1.197 schwarze 547: print_endword(h);
1.105 kristaps 548: }
1.196 schwarze 549: }
1.29 kristaps 550:
1.109 kristaps 551: if ( ! (h->flags & HTML_NONOSPACE))
552: h->flags &= ~HTML_NOSPACE;
1.110 kristaps 553: else
554: h->flags |= HTML_NOSPACE;
1.109 kristaps 555:
1.94 kristaps 556: /* Print out the tag name and attributes. */
557:
1.197 schwarze 558: print_byte(h, '<');
559: print_word(h, htmltags[tag].name);
1.194 schwarze 560:
561: va_start(ap, fmt);
562:
1.244 schwarze 563: while (*fmt != '\0' && *fmt != 's') {
1.203 schwarze 564:
1.238 schwarze 565: /* Parse attributes and arguments. */
1.203 schwarze 566:
567: arg1 = va_arg(ap, char *);
1.238 schwarze 568: arg2 = NULL;
1.194 schwarze 569: switch (*fmt++) {
570: case 'c':
1.195 schwarze 571: attr = "class";
1.194 schwarze 572: break;
573: case 'h':
1.195 schwarze 574: attr = "href";
1.194 schwarze 575: break;
576: case 'i':
1.195 schwarze 577: attr = "id";
1.194 schwarze 578: break;
579: case '?':
1.203 schwarze 580: attr = arg1;
581: arg1 = va_arg(ap, char *);
1.194 schwarze 582: break;
583: default:
584: abort();
585: }
1.203 schwarze 586: if (*fmt == 'M')
587: arg2 = va_arg(ap, char *);
588: if (arg1 == NULL)
589: continue;
590:
1.238 schwarze 591: /* Print the attributes. */
1.203 schwarze 592:
1.197 schwarze 593: print_byte(h, ' ');
594: print_word(h, attr);
595: print_byte(h, '=');
596: print_byte(h, '"');
1.195 schwarze 597: switch (*fmt) {
1.208 schwarze 598: case 'I':
599: print_href(h, arg1, NULL, 0);
600: fmt++;
601: break;
1.195 schwarze 602: case 'M':
1.203 schwarze 603: print_href(h, arg1, arg2, 1);
1.195 schwarze 604: fmt++;
605: break;
1.208 schwarze 606: case 'R':
607: print_byte(h, '#');
608: print_encode(h, arg1, NULL, 1);
1.195 schwarze 609: fmt++;
610: break;
1.208 schwarze 611: case 'T':
612: print_encode(h, arg1, NULL, 1);
613: print_word(h, "\" title=\"");
614: print_encode(h, arg1, NULL, 1);
1.195 schwarze 615: fmt++;
1.208 schwarze 616: break;
1.195 schwarze 617: default:
1.244 schwarze 618: print_encode(h, arg1, NULL, 1);
1.195 schwarze 619: break;
620: }
1.197 schwarze 621: print_byte(h, '"');
1.194 schwarze 622: }
1.244 schwarze 623:
624: style_written = 0;
625: while (*fmt++ == 's') {
626: arg1 = va_arg(ap, char *);
627: arg2 = va_arg(ap, char *);
628: if (arg2 == NULL)
629: continue;
630: print_byte(h, ' ');
631: if (style_written == 0) {
632: print_word(h, "style=\"");
633: style_written = 1;
634: }
635: print_word(h, arg1);
636: print_byte(h, ':');
637: print_byte(h, ' ');
638: print_word(h, arg2);
639: print_byte(h, ';');
640: }
641: if (style_written)
642: print_byte(h, '"');
643:
1.194 schwarze 644: va_end(ap);
1.94 kristaps 645:
1.172 kristaps 646: /* Accommodate for "well-formed" singleton escaping. */
1.94 kristaps 647:
1.93 kristaps 648: if (HTML_AUTOCLOSE & htmltags[tag].flags)
1.197 schwarze 649: print_byte(h, '/');
1.93 kristaps 650:
1.197 schwarze 651: print_byte(h, '>');
1.14 kristaps 652:
1.196 schwarze 653: if (tflags & HTML_NLBEGIN)
1.197 schwarze 654: print_endline(h);
1.196 schwarze 655: else
656: h->flags |= HTML_NOSPACE;
1.117 kristaps 657:
1.196 schwarze 658: if (tflags & HTML_INDENT)
659: h->indent++;
660: if (tflags & HTML_NOINDENT)
661: h->noindent++;
1.117 kristaps 662:
1.188 schwarze 663: return t;
1.14 kristaps 664: }
665:
1.29 kristaps 666: static void
1.184 schwarze 667: print_ctag(struct html *h, struct tag *tag)
1.14 kristaps 668: {
1.196 schwarze 669: int tflags;
1.156 schwarze 670:
1.184 schwarze 671: /*
672: * Remember to close out and nullify the current
673: * meta-font and table, if applicable.
674: */
675: if (tag == h->metaf)
676: h->metaf = NULL;
677: if (tag == h->tblt)
678: h->tblt = NULL;
679:
1.196 schwarze 680: tflags = htmltags[tag->tag].flags;
681:
682: if (tflags & HTML_INDENT)
683: h->indent--;
684: if (tflags & HTML_NOINDENT)
685: h->noindent--;
686: if (tflags & HTML_NLEND)
1.197 schwarze 687: print_endline(h);
688: print_indent(h);
689: print_byte(h, '<');
690: print_byte(h, '/');
691: print_word(h, htmltags[tag->tag].name);
692: print_byte(h, '>');
1.196 schwarze 693: if (tflags & HTML_NLAFTER)
1.197 schwarze 694: print_endline(h);
1.184 schwarze 695:
1.204 schwarze 696: h->tag = tag->next;
1.184 schwarze 697: free(tag);
1.14 kristaps 698: }
699:
1.51 kristaps 700: void
1.93 kristaps 701: print_gen_decls(struct html *h)
1.1 kristaps 702: {
1.197 schwarze 703: print_word(h, "<!DOCTYPE html>");
704: print_endline(h);
1.221 schwarze 705: }
706:
707: void
708: print_gen_comment(struct html *h, struct roff_node *n)
709: {
710: int wantblank;
711:
712: print_word(h, "<!-- This is an automatically generated file."
713: " Do not edit.");
714: h->indent = 1;
715: wantblank = 0;
716: while (n != NULL && n->type == ROFFT_COMMENT) {
717: if (strstr(n->string, "-->") == NULL &&
718: (wantblank || *n->string != '\0')) {
719: print_endline(h);
720: print_indent(h);
721: print_word(h, n->string);
722: wantblank = *n->string != '\0';
723: }
724: n = n->next;
725: }
726: if (wantblank)
727: print_endline(h);
728: print_word(h, " -->");
729: print_endline(h);
730: h->indent = 0;
1.1 kristaps 731: }
732:
1.51 kristaps 733: void
1.104 kristaps 734: print_text(struct html *h, const char *word)
1.1 kristaps 735: {
1.197 schwarze 736: if (h->col && (h->flags & HTML_NOSPACE) == 0) {
1.105 kristaps 737: if ( ! (HTML_KEEP & h->flags)) {
738: if (HTML_PREKEEP & h->flags)
739: h->flags |= HTML_KEEP;
1.197 schwarze 740: print_endword(h);
1.105 kristaps 741: } else
1.216 schwarze 742: print_word(h, " ");
1.105 kristaps 743: }
1.30 kristaps 744:
1.122 kristaps 745: assert(NULL == h->metaf);
1.152 schwarze 746: switch (h->metac) {
1.156 schwarze 747: case HTMLFONT_ITALIC:
1.194 schwarze 748: h->metaf = print_otag(h, TAG_I, "");
1.152 schwarze 749: break;
1.156 schwarze 750: case HTMLFONT_BOLD:
1.194 schwarze 751: h->metaf = print_otag(h, TAG_B, "");
1.152 schwarze 752: break;
1.156 schwarze 753: case HTMLFONT_BI:
1.194 schwarze 754: h->metaf = print_otag(h, TAG_B, "");
755: print_otag(h, TAG_I, "");
1.242 schwarze 756: break;
757: case HTMLFONT_CW:
758: h->metaf = print_otag(h, TAG_SPAN, "c", "Li");
1.152 schwarze 759: break;
760: default:
1.197 schwarze 761: print_indent(h);
1.152 schwarze 762: break;
763: }
1.122 kristaps 764:
1.104 kristaps 765: assert(word);
1.195 schwarze 766: if ( ! print_encode(h, word, NULL, 0)) {
1.109 kristaps 767: if ( ! (h->flags & HTML_NONOSPACE))
768: h->flags &= ~HTML_NOSPACE;
1.183 schwarze 769: h->flags &= ~HTML_NONEWLINE;
1.149 kristaps 770: } else
1.183 schwarze 771: h->flags |= HTML_NOSPACE | HTML_NONEWLINE;
1.122 kristaps 772:
773: if (h->metaf) {
774: print_tagq(h, h->metaf);
775: h->metaf = NULL;
776: }
1.113 schwarze 777:
778: h->flags &= ~HTML_IGNDELIM;
1.1 kristaps 779: }
1.30 kristaps 780:
1.51 kristaps 781: void
1.30 kristaps 782: print_tagq(struct html *h, const struct tag *until)
783: {
784: struct tag *tag;
785:
1.204 schwarze 786: while ((tag = h->tag) != NULL) {
1.184 schwarze 787: print_ctag(h, tag);
1.30 kristaps 788: if (until && tag == until)
789: return;
790: }
791: }
792:
1.51 kristaps 793: void
1.30 kristaps 794: print_stagq(struct html *h, const struct tag *suntil)
795: {
796: struct tag *tag;
797:
1.204 schwarze 798: while ((tag = h->tag) != NULL) {
1.30 kristaps 799: if (suntil && tag == suntil)
800: return;
1.184 schwarze 801: print_ctag(h, tag);
1.30 kristaps 802: }
803: }
1.171 kristaps 804:
805: void
806: print_paragraph(struct html *h)
807: {
808: struct tag *t;
809:
1.198 schwarze 810: t = print_otag(h, TAG_DIV, "c", "Pp");
1.171 kristaps 811: print_tagq(h, t);
812: }
813:
1.197 schwarze 814:
815: /***********************************************************************
816: * Low level output functions.
817: * They implement line breaking using a short static buffer.
818: ***********************************************************************/
819:
820: /*
821: * Buffer one HTML output byte.
822: * If the buffer is full, flush and deactivate it and start a new line.
823: * If the buffer is inactive, print directly.
824: */
825: static void
826: print_byte(struct html *h, char c)
827: {
828: if ((h->flags & HTML_BUFFER) == 0) {
829: putchar(c);
830: h->col++;
831: return;
832: }
833:
834: if (h->col + h->bufcol < sizeof(h->buf)) {
835: h->buf[h->bufcol++] = c;
836: return;
837: }
838:
839: putchar('\n');
840: h->col = 0;
841: print_indent(h);
842: putchar(' ');
843: putchar(' ');
844: fwrite(h->buf, h->bufcol, 1, stdout);
845: putchar(c);
846: h->col = (h->indent + 1) * 2 + h->bufcol + 1;
847: h->bufcol = 0;
848: h->flags &= ~HTML_BUFFER;
849: }
850:
1.196 schwarze 851: /*
852: * If something was printed on the current output line, end it.
1.197 schwarze 853: * Not to be called right after print_indent().
1.196 schwarze 854: */
1.202 schwarze 855: void
1.197 schwarze 856: print_endline(struct html *h)
1.196 schwarze 857: {
1.197 schwarze 858: if (h->col == 0)
1.196 schwarze 859: return;
860:
1.197 schwarze 861: if (h->bufcol) {
862: putchar(' ');
863: fwrite(h->buf, h->bufcol, 1, stdout);
864: h->bufcol = 0;
865: }
1.196 schwarze 866: putchar('\n');
1.197 schwarze 867: h->col = 0;
868: h->flags |= HTML_NOSPACE;
869: h->flags &= ~HTML_BUFFER;
870: }
871:
872: /*
873: * Flush the HTML output buffer.
874: * If it is inactive, activate it.
875: */
876: static void
877: print_endword(struct html *h)
878: {
879: if (h->noindent) {
880: print_byte(h, ' ');
881: return;
882: }
883:
884: if ((h->flags & HTML_BUFFER) == 0) {
885: h->col++;
886: h->flags |= HTML_BUFFER;
887: } else if (h->bufcol) {
888: putchar(' ');
889: fwrite(h->buf, h->bufcol, 1, stdout);
890: h->col += h->bufcol + 1;
891: }
892: h->bufcol = 0;
1.196 schwarze 893: }
894:
895: /*
896: * If at the beginning of a new output line,
897: * perform indentation and mark the line as containing output.
898: * Make sure to really produce some output right afterwards,
899: * but do not use print_otag() for producing it.
900: */
901: static void
1.197 schwarze 902: print_indent(struct html *h)
1.196 schwarze 903: {
1.197 schwarze 904: size_t i;
1.196 schwarze 905:
1.197 schwarze 906: if (h->col)
1.196 schwarze 907: return;
908:
1.197 schwarze 909: if (h->noindent == 0) {
910: h->col = h->indent * 2;
911: for (i = 0; i < h->col; i++)
1.196 schwarze 912: putchar(' ');
1.197 schwarze 913: }
914: h->flags &= ~HTML_NOSPACE;
915: }
916:
917: /*
918: * Print or buffer some characters
919: * depending on the current HTML output buffer state.
920: */
921: static void
922: print_word(struct html *h, const char *cp)
923: {
924: while (*cp != '\0')
925: print_byte(h, *cp++);
1.68 kristaps 926: }
CVSweb