mdw@git.distorted.org.uk Git - sgt/halibut/blob - bk_man.c

   1 /*
   2  * man page backend for Halibut
   3  */
   4
   5 #include <stdio.h>
   6 #include <stdlib.h>
   7 #include <assert.h>
   8 #include "halibut.h"
   9
  10 typedef struct {
  11     wchar_t *th;
  12     int headnumbers;
  13     int mindepth;
  14     char *filename;
  15     int charset;
  16     wchar_t *bullet, *lquote, *rquote;
  17 } manconfig;
  18
  19 static void man_text(FILE *, word *,
  20                      int newline, int quote_props, manconfig *conf);
  21 static void man_codepara(FILE *, word *, int charset);
  22 static int man_convert(wchar_t const *s, int maxlen,
  23                        char **result, int quote_props,
  24                        int charset, charset_state *state);
  25
  26 static manconfig man_configure(paragraph *source) {
  27     paragraph *p;
  28     manconfig ret;
  29
  30     /*
  31      * Defaults.
  32      */
  33     ret.th = NULL;
  34     ret.headnumbers = FALSE;
  35     ret.mindepth = 0;
  36     ret.filename = dupstr("output.1");
  37     ret.charset = CS_ASCII;
  38     ret.bullet = L"\x2022\0o\0\0";
  39     ret.lquote = L"\x2018\0\x2019\0\"\0\"\0\0";
  40     ret.rquote = uadv(ret.lquote);
  41
  42     /*
  43      * Two-pass configuration so that we can pick up global config
  44      * (e.g. `quotes') before having it overridden by specific
  45      * config (`man-quotes'), irrespective of the order in which
  46      * they occur.
  47      */
  48     for (p = source; p; p = p->next) {
  49         if (p->type == para_Config) {
  50             if (!ustricmp(p->keyword, L"quotes")) {
  51                 if (*uadv(p->keyword) && *uadv(uadv(p->keyword))) {
  52                     ret.lquote = uadv(p->keyword);
  53                     ret.rquote = uadv(ret.lquote);
  54                 }
  55             }
  56         }
  57     }
  58
  59     for (p = source; p; p = p->next) {
  60         if (p->type == para_Config) {
  61             if (!ustricmp(p->keyword, L"man-identity")) {
  62                 wchar_t *wp, *ep;
  63
  64                 wp = uadv(p->keyword);
  65                 ep = wp;
  66                 while (*ep)
  67                     ep = uadv(ep);
  68                 sfree(ret.th);
  69                 ret.th = snewn(ep - wp + 1, wchar_t);
  70                 memcpy(ret.th, wp, (ep - wp + 1) * sizeof(wchar_t));
  71             } else if (!ustricmp(p->keyword, L"man-charset")) {
  72                 ret.charset = charset_from_ustr(&p->fpos, uadv(p->keyword));
  73             } else if (!ustricmp(p->keyword, L"man-headnumbers")) {
  74                 ret.headnumbers = utob(uadv(p->keyword));
  75             } else if (!ustricmp(p->keyword, L"man-mindepth")) {
  76                 ret.mindepth = utoi(uadv(p->keyword));
  77             } else if (!ustricmp(p->keyword, L"man-filename")) {
  78                 sfree(ret.filename);
  79                 ret.filename = dupstr(adv(p->origkeyword));
  80             } else if (!ustricmp(p->keyword, L"man-bullet")) {
  81                 ret.bullet = uadv(p->keyword);
  82             } else if (!ustricmp(p->keyword, L"man-quotes")) {
  83                 if (*uadv(p->keyword) && *uadv(uadv(p->keyword))) {
  84                     ret.lquote = uadv(p->keyword);
  85                     ret.rquote = uadv(ret.lquote);
  86                 }
  87             }
  88         }
  89     }
  90
  91     /*
  92      * Now process fallbacks on quote characters and bullets.
  93      */
  94     while (*uadv(ret.rquote) && *uadv(uadv(ret.rquote)) &&
  95            (!cvt_ok(ret.charset, ret.lquote) ||
  96             !cvt_ok(ret.charset, ret.rquote))) {
  97         ret.lquote = uadv(ret.rquote);
  98         ret.rquote = uadv(ret.lquote);
  99     }
 100
 101     while (*ret.bullet && *uadv(ret.bullet) &&
 102            !cvt_ok(ret.charset, ret.bullet))
 103         ret.bullet = uadv(ret.bullet);
 104
 105     return ret;
 106 }
 107
 108 static void man_conf_cleanup(manconfig cf)
 109 {
 110     sfree(cf.th);
 111     sfree(cf.filename);
 112 }
 113
 114 paragraph *man_config_filename(char *filename)
 115 {
 116     return cmdline_cfg_simple("man-filename", filename, NULL);
 117 }
 118
 119 #define QUOTE_INITCTRL 1 /* quote initial . and ' on a line */
 120 #define QUOTE_QUOTES   2 /* quote double quotes by doubling them */
 121
 122 void man_backend(paragraph *sourceform, keywordlist *keywords,
 123                  indexdata *idx, void *unused) {
 124     paragraph *p;
 125     FILE *fp;
 126     manconfig conf;
 127     int had_described_thing;
 128
 129     IGNORE(unused);
 130     IGNORE(keywords);
 131     IGNORE(idx);
 132
 133     conf = man_configure(sourceform);
 134
 135     /*
 136      * Open the output file.
 137      */
 138     fp = fopen(conf.filename, "w");
 139     if (!fp) {
 140         error(err_cantopenw, conf.filename);
 141         return;
 142     }
 143
 144     /* Do the version ID */
 145     for (p = sourceform; p; p = p->next)
 146         if (p->type == para_VersionID) {
 147             fprintf(fp, ".\\\" ");
 148             man_text(fp, p->words, TRUE, 0, &conf);
 149         }
 150
 151     /* .TH name-of-program manual-section */
 152     fprintf(fp, ".TH");
 153     if (conf.th && *conf.th) {
 154         char *c;
 155         wchar_t *wp;
 156
 157         for (wp = conf.th; *wp; wp = uadv(wp)) {
 158             fputs(" \"", fp);
 159             man_convert(wp, 0, &c, QUOTE_QUOTES, conf.charset, NULL);
 160             fputs(c, fp);
 161             sfree(c);
 162             fputc('"', fp);
 163         }
 164     }
 165     fputc('\n', fp);
 166
 167     fprintf(fp, ".UC\n");
 168
 169     had_described_thing = FALSE;
 170 #define cleanup_described_thing do { \
 171     if (had_described_thing) \
 172         fprintf(fp, "\n"); \
 173     had_described_thing = FALSE; \
 174 } while (0)
 175
 176     for (p = sourceform; p; p = p->next) switch (p->type) {
 177         /*
 178          * Things we ignore because we've already processed them or
 179          * aren't going to touch them in this pass.
 180          */
 181       case para_IM:
 182       case para_BR:
 183       case para_Biblio:                /* only touch BiblioCited */
 184       case para_VersionID:
 185       case para_NoCite:
 186       case para_Title:
 187         break;
 188
 189         /*
 190          * Headings.
 191          */
 192       case para_Chapter:
 193       case para_Appendix:
 194       case para_UnnumberedChapter:
 195       case para_Heading:
 196       case para_Subsect:
 197
 198         cleanup_described_thing;
 199         {
 200             int depth;
 201             if (p->type == para_Subsect)
 202                 depth = p->aux + 2;
 203             else if (p->type == para_Heading)
 204                 depth = 1;
 205             else
 206                 depth = 0;
 207             if (depth >= conf.mindepth) {
 208                 fprintf(fp, ".SH \"");
 209                 if (conf.headnumbers && p->kwtext) {
 210                     man_text(fp, p->kwtext, FALSE, QUOTE_QUOTES, &conf);
 211                     fprintf(fp, " ");
 212                 }
 213                 man_text(fp, p->words, FALSE, QUOTE_QUOTES, &conf);
 214                 fprintf(fp, "\"\n");
 215             }
 216             break;
 217         }
 218
 219         /*
 220          * Code paragraphs.
 221          */
 222       case para_Code:
 223         cleanup_described_thing;
 224         fprintf(fp, ".PP\n");
 225         man_codepara(fp, p->words, conf.charset);
 226         break;
 227
 228         /*
 229          * Normal paragraphs.
 230          */
 231       case para_Normal:
 232       case para_Copyright:
 233         cleanup_described_thing;
 234         fprintf(fp, ".PP\n");
 235         man_text(fp, p->words, TRUE, 0, &conf);
 236         break;
 237
 238         /*
 239          * List paragraphs.
 240          */
 241       case para_Description:
 242       case para_BiblioCited:
 243       case para_Bullet:
 244       case para_NumberedList:
 245         if (p->type != para_Description)
 246             cleanup_described_thing;
 247
 248         if (p->type == para_Bullet) {
 249             char *bullettext;
 250             man_convert(conf.bullet, -1, &bullettext, QUOTE_QUOTES,
 251                         conf.charset, NULL);
 252             fprintf(fp, ".IP \"\\fB%s\\fP\"\n", bullettext);
 253             sfree(bullettext);
 254         } else if (p->type == para_NumberedList) {
 255             fprintf(fp, ".IP \"");
 256             man_text(fp, p->kwtext, FALSE, QUOTE_QUOTES, &conf);
 257             fprintf(fp, "\"\n");
 258         } else if (p->type == para_Description) {
 259             if (had_described_thing) {
 260                 /*
 261                  * Do nothing; the .xP for this paragraph is the
 262                  * .IP which has come before it in the
 263                  * DescribedThing.
 264                  */
 265             } else {
 266                 /*
 267                  * A \dd without a preceding \dt is given a blank
 268                  * one.
 269                  */
 270                 fprintf(fp, ".IP \"\"\n");
 271             }
 272         } else if (p->type == para_BiblioCited) {
 273             fprintf(fp, ".IP \"");
 274             man_text(fp, p->kwtext, FALSE, QUOTE_QUOTES, &conf);
 275             fprintf(fp, "\"\n");
 276         }
 277         man_text(fp, p->words, TRUE, 0, &conf);
 278         had_described_thing = FALSE;
 279         break;
 280
 281       case para_DescribedThing:
 282         cleanup_described_thing;
 283         fprintf(fp, ".IP \"");
 284         man_text(fp, p->words, FALSE, QUOTE_QUOTES, &conf);
 285         fprintf(fp, "\"\n");
 286         had_described_thing = TRUE;
 287         break;
 288
 289       case para_Rule:
 290         /*
 291          * This isn't terribly good. Anyone who wants to do better
 292          * should feel free!
 293          */
 294         cleanup_described_thing;
 295         fprintf(fp, ".PP\n----------------------------------------\n");
 296         break;
 297
 298       case para_LcontPush:
 299       case para_QuotePush:
 300         cleanup_described_thing;
 301         fprintf(fp, ".RS\n");
 302         break;
 303       case para_LcontPop:
 304       case para_QuotePop:
 305         cleanup_described_thing;
 306         fprintf(fp, ".RE\n");
 307         break;
 308     }
 309     cleanup_described_thing;
 310
 311     /*
 312      * Tidy up.
 313      */
 314     fclose(fp);
 315     man_conf_cleanup(conf);
 316 }
 317
 318 /*
 319  * Convert a wide string into a string of chars; mallocs the
 320  * resulting string and stores a pointer to it in `*result'.
 321  *
 322  * If `state' is non-NULL, updates the charset state pointed to. If
 323  * `state' is NULL, this function uses its own state, initialises
 324  * it from scratch, and cleans it up when finished. If `state' is
 325  * non-NULL but _s_ is NULL, cleans up a provided state.
 326  *
 327  * Return is nonzero if all characters are OK. If not all
 328  * characters are OK but `result' is non-NULL, a result _will_
 329  * still be generated!
 330  *
 331  * This function also does escaping of groff special characters.
 332  */
 333 static int man_convert(wchar_t const *s, int maxlen,
 334                        char **result, int quote_props,
 335                        int charset, charset_state *state) {
 336     charset_state internal_state = CHARSET_INIT_STATE;
 337     int slen, err;
 338     char *p = NULL, *q;
 339     int plen = 0, psize = 0;
 340     rdstringc out = {0, 0, NULL};
 341
 342     if (!state)
 343         state = &internal_state;
 344
 345     slen = (s ? ustrlen(s) : 0);
 346
 347     if (slen > maxlen && maxlen > 0)
 348         slen = maxlen;
 349
 350     psize = 384;
 351     plen = 0;
 352     p = snewn(psize, char);
 353     err = 0;
 354
 355     while (slen > 0) {
 356         int ret = charset_from_unicode(&s, &slen, p+plen, psize-plen,
 357                                    charset, state, (err ? NULL : &err));
 358         if (ret > 0) {
 359             plen += ret;
 360             if (psize - plen < 256) {
 361                 psize = plen + 256;
 362                 p = sresize(p, psize, char);
 363             }
 364         }
 365     }
 366
 367     if (state == &internal_state || s == NULL) {
 368         int ret = charset_from_unicode(NULL, 0, p+plen, psize-plen,
 369                                        charset, state, NULL);
 370         if (ret > 0)
 371             plen += ret;
 372     }
 373
 374     for (q = p; q < p+plen; q++) {
 375         if (q == p && (*q == '.' || *q == '\'') &&
 376             (quote_props & QUOTE_INITCTRL)) {
 377             /*
 378              * Control character (. or ') at the start of a
 379              * line. Quote it by putting \& (troff zero-width
 380              * space) before it.
 381              */
 382             rdaddc(&out, '\\');
 383             rdaddc(&out, '&');
 384         } else if (*q == '\\') {
 385             /*
 386              * Quote backslashes by doubling them, always.
 387              */
 388             rdaddc(&out, '\\');
 389         } else if (*q == '"' && (quote_props & QUOTE_QUOTES)) {
 390             /*
 391              * Double quote within double quotes. Quote it by
 392              * doubling.
 393              */
 394             rdaddc(&out, '"');
 395         }
 396         rdaddc(&out, *q);
 397     }
 398
 399     sfree(p);
 400
 401     if (out.text)
 402         *result = rdtrimc(&out);
 403     else
 404         *result = dupstr("");
 405
 406     return !err;
 407 }
 408
 409 static void man_rdaddwc(rdstringc *rs, word *text, word *end,
 410                         int quote_props, manconfig *conf,
 411                         charset_state *state) {
 412     char *c;
 413
 414     for (; text && text != end; text = text->next) switch (text->type) {
 415       case word_HyperLink:
 416       case word_HyperEnd:
 417       case word_UpperXref:
 418       case word_LowerXref:
 419       case word_XrefEnd:
 420       case word_IndexRef:
 421         break;
 422
 423       case word_Normal:
 424       case word_Emph:
 425       case word_Code:
 426       case word_WeakCode:
 427       case word_WhiteSpace:
 428       case word_EmphSpace:
 429       case word_CodeSpace:
 430       case word_WkCodeSpace:
 431       case word_Quote:
 432       case word_EmphQuote:
 433       case word_CodeQuote:
 434       case word_WkCodeQuote:
 435         assert(text->type != word_CodeQuote &&
 436                text->type != word_WkCodeQuote);
 437
 438         if (towordstyle(text->type) == word_Emph &&
 439             (attraux(text->aux) == attr_First ||
 440              attraux(text->aux) == attr_Only)) {
 441             if (rs->pos > 0)
 442                 quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
 443             man_convert(NULL, 0, &c, quote_props, conf->charset, state);
 444             rdaddsc(rs, c);
 445             sfree(c);
 446             *state = charset_init_state;
 447             rdaddsc(rs, "\\fI");
 448         } else if ((towordstyle(text->type) == word_Code ||
 449                     towordstyle(text->type) == word_WeakCode) &&
 450                    (attraux(text->aux) == attr_First ||
 451                     attraux(text->aux) == attr_Only)) {
 452             if (rs->pos > 0)
 453                 quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
 454             man_convert(NULL, 0, &c, quote_props, conf->charset, state);
 455             rdaddsc(rs, c);
 456             sfree(c);
 457             *state = charset_init_state;
 458             rdaddsc(rs, "\\fB");
 459         }
 460
 461         if (removeattr(text->type) == word_Normal) {
 462             charset_state s2 = *state;
 463
 464             if (rs->pos > 0)
 465                 quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
 466             if (man_convert(text->text, 0, &c, quote_props, conf->charset, &s2) ||
 467                 !text->alt) {
 468                 rdaddsc(rs, c);
 469                 *state = s2;
 470             } else {
 471                 man_rdaddwc(rs, text->alt, NULL, quote_props, conf, state);
 472             }
 473             sfree(c);
 474         } else if (removeattr(text->type) == word_WhiteSpace) {
 475             if (rs->pos > 0)
 476                 quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
 477             man_convert(L" ", 1, &c, quote_props, conf->charset, state);
 478             rdaddsc(rs, c);
 479             sfree(c);
 480         } else if (removeattr(text->type) == word_Quote) {
 481             if (rs->pos > 0)
 482                 quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
 483             man_convert(quoteaux(text->aux) == quote_Open ?
 484                         conf->lquote : conf->rquote, 0,
 485                         &c, quote_props, conf->charset, state);
 486             rdaddsc(rs, c);
 487             sfree(c);
 488         }
 489         if (towordstyle(text->type) != word_Normal &&
 490             (attraux(text->aux) == attr_Last ||
 491              attraux(text->aux) == attr_Only)) {
 492             if (rs->pos > 0)
 493                 quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
 494             man_convert(NULL, 0, &c, quote_props, conf->charset, state);
 495             rdaddsc(rs, c);
 496             sfree(c);
 497             *state = charset_init_state;
 498             rdaddsc(rs, "\\fP");
 499         }
 500         break;
 501     }
 502     man_convert(NULL, 0, &c, quote_props, conf->charset, state);
 503     rdaddsc(rs, c);
 504     sfree(c);
 505 }
 506
 507 static void man_text(FILE *fp, word *text, int newline,
 508                      int quote_props, manconfig *conf) {
 509     rdstringc t = { 0, 0, NULL };
 510     charset_state state = CHARSET_INIT_STATE;
 511
 512     man_rdaddwc(&t, text, NULL, quote_props | QUOTE_INITCTRL, conf, &state);
 513     fprintf(fp, "%s", t.text);
 514     sfree(t.text);
 515     if (newline)
 516         fputc('\n', fp);
 517 }
 518
 519 static void man_codepara(FILE *fp, word *text, int charset) {
 520     fprintf(fp, ".nf\n");
 521     for (; text; text = text->next) if (text->type == word_WeakCode) {
 522         char *c;
 523         wchar_t *t, *e;
 524         int quote_props = QUOTE_INITCTRL;
 525
 526         t = text->text;
 527         if (text->next && text->next->type == word_Emph) {
 528             e = text->next->text;
 529             text = text->next;
 530         } else
 531             e = NULL;
 532
 533         while (e && *e && *t) {
 534             int n;
 535             int ec = *e;
 536
 537             for (n = 0; t[n] && e[n] && e[n] == ec; n++);
 538             if (ec == 'i')
 539                 fprintf(fp, "\\fI");
 540             else if (ec == 'b')
 541                 fprintf(fp, "\\fB");
 542             man_convert(t, n, &c, quote_props, charset, NULL);
 543             quote_props &= ~QUOTE_INITCTRL;
 544             fprintf(fp, "%s", c);
 545             sfree(c);
 546             if (ec == 'i' || ec == 'b')
 547                 fprintf(fp, "\\fP");
 548             t += n;
 549             e += n;
 550         }
 551         man_convert(t, 0, &c, quote_props, charset, NULL);
 552         fprintf(fp, "%s\n", c);
 553         sfree(c);
 554     }
 555     fprintf(fp, ".fi\n");
 556 }