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                 if (depth > conf.mindepth)
 209                     fprintf(fp, ".SS \"");
 210                 else
 211                     fprintf(fp, ".SH \"");
 212                 if (conf.headnumbers && p->kwtext) {
 213                     man_text(fp, p->kwtext, FALSE, QUOTE_QUOTES, &conf);
 214                     fprintf(fp, " ");
 215                 }
 216                 man_text(fp, p->words, FALSE, QUOTE_QUOTES, &conf);
 217                 fprintf(fp, "\"\n");
 218             }
 219             break;
 220         }
 221
 222         /*
 223          * Code paragraphs.
 224          */
 225       case para_Code:
 226         cleanup_described_thing;
 227         fprintf(fp, ".PP\n");
 228         man_codepara(fp, p->words, conf.charset);
 229         break;
 230
 231         /*
 232          * Normal paragraphs.
 233          */
 234       case para_Normal:
 235       case para_Copyright:
 236         cleanup_described_thing;
 237         fprintf(fp, ".PP\n");
 238         man_text(fp, p->words, TRUE, 0, &conf);
 239         break;
 240
 241         /*
 242          * List paragraphs.
 243          */
 244       case para_Description:
 245       case para_BiblioCited:
 246       case para_Bullet:
 247       case para_NumberedList:
 248         if (p->type != para_Description)
 249             cleanup_described_thing;
 250
 251         if (p->type == para_Bullet) {
 252             char *bullettext;
 253             man_convert(conf.bullet, -1, &bullettext, QUOTE_QUOTES,
 254                         conf.charset, NULL);
 255             fprintf(fp, ".IP \"\\fB%s\\fP\"\n", bullettext);
 256             sfree(bullettext);
 257         } else if (p->type == para_NumberedList) {
 258             fprintf(fp, ".IP \"");
 259             man_text(fp, p->kwtext, FALSE, QUOTE_QUOTES, &conf);
 260             fprintf(fp, "\"\n");
 261         } else if (p->type == para_Description) {
 262             if (had_described_thing) {
 263                 /*
 264                  * Do nothing; the .xP for this paragraph is the
 265                  * .IP which has come before it in the
 266                  * DescribedThing.
 267                  */
 268             } else {
 269                 /*
 270                  * A \dd without a preceding \dt is given a blank
 271                  * one.
 272                  */
 273                 fprintf(fp, ".IP \"\"\n");
 274             }
 275         } else if (p->type == para_BiblioCited) {
 276             fprintf(fp, ".IP \"");
 277             man_text(fp, p->kwtext, FALSE, QUOTE_QUOTES, &conf);
 278             fprintf(fp, "\"\n");
 279         }
 280         man_text(fp, p->words, TRUE, 0, &conf);
 281         had_described_thing = FALSE;
 282         break;
 283
 284       case para_DescribedThing:
 285         cleanup_described_thing;
 286         fprintf(fp, ".IP \"");
 287         man_text(fp, p->words, FALSE, QUOTE_QUOTES, &conf);
 288         fprintf(fp, "\"\n");
 289         had_described_thing = TRUE;
 290         break;
 291
 292       case para_Rule:
 293         /*
 294          * This isn't terribly good. Anyone who wants to do better
 295          * should feel free!
 296          */
 297         cleanup_described_thing;
 298         fprintf(fp, ".PP\n----------------------------------------\n");
 299         break;
 300
 301       case para_LcontPush:
 302       case para_QuotePush:
 303         cleanup_described_thing;
 304         fprintf(fp, ".RS\n");
 305         break;
 306       case para_LcontPop:
 307       case para_QuotePop:
 308         cleanup_described_thing;
 309         fprintf(fp, ".RE\n");
 310         break;
 311     }
 312     cleanup_described_thing;
 313
 314     /*
 315      * Tidy up.
 316      */
 317     fclose(fp);
 318     man_conf_cleanup(conf);
 319 }
 320
 321 /*
 322  * Convert a wide string into a string of chars; mallocs the
 323  * resulting string and stores a pointer to it in `*result'.
 324  *
 325  * If `state' is non-NULL, updates the charset state pointed to. If
 326  * `state' is NULL, this function uses its own state, initialises
 327  * it from scratch, and cleans it up when finished. If `state' is
 328  * non-NULL but _s_ is NULL, cleans up a provided state.
 329  *
 330  * Return is nonzero if all characters are OK. If not all
 331  * characters are OK but `result' is non-NULL, a result _will_
 332  * still be generated!
 333  *
 334  * This function also does escaping of groff special characters.
 335  */
 336 static int man_convert(wchar_t const *s, int maxlen,
 337                        char **result, int quote_props,
 338                        int charset, charset_state *state) {
 339     charset_state internal_state = CHARSET_INIT_STATE;
 340     int slen, err;
 341     char *p = NULL, *q;
 342     int plen = 0, psize = 0;
 343     rdstringc out = {0, 0, NULL};
 344
 345     if (!state)
 346         state = &internal_state;
 347
 348     slen = (s ? ustrlen(s) : 0);
 349
 350     if (slen > maxlen && maxlen > 0)
 351         slen = maxlen;
 352
 353     psize = 384;
 354     plen = 0;
 355     p = snewn(psize, char);
 356     err = 0;
 357
 358     while (slen > 0) {
 359         int ret = charset_from_unicode(&s, &slen, p+plen, psize-plen,
 360                                    charset, state, (err ? NULL : &err));
 361         if (ret > 0) {
 362             plen += ret;
 363             if (psize - plen < 256) {
 364                 psize = plen + 256;
 365                 p = sresize(p, psize, char);
 366             }
 367         }
 368     }
 369
 370     if (state == &internal_state || s == NULL) {
 371         int ret = charset_from_unicode(NULL, 0, p+plen, psize-plen,
 372                                        charset, state, NULL);
 373         if (ret > 0)
 374             plen += ret;
 375     }
 376
 377     for (q = p; q < p+plen; q++) {
 378         if (q == p && (*q == '.' || *q == '\'') &&
 379             (quote_props & QUOTE_INITCTRL)) {
 380             /*
 381              * Control character (. or ') at the start of a
 382              * line. Quote it by putting \& (troff zero-width
 383              * space) before it.
 384              */
 385             rdaddc(&out, '\\');
 386             rdaddc(&out, '&');
 387         } else if (*q == '\\') {
 388             /*
 389              * Quote backslashes by doubling them, always.
 390              */
 391             rdaddc(&out, '\\');
 392         } else if (*q == '"' && (quote_props & QUOTE_QUOTES)) {
 393             /*
 394              * Double quote within double quotes. Quote it by
 395              * doubling.
 396              */
 397             rdaddc(&out, '"');
 398         }
 399         rdaddc(&out, *q);
 400     }
 401
 402     sfree(p);
 403
 404     if (out.text)
 405         *result = rdtrimc(&out);
 406     else
 407         *result = dupstr("");
 408
 409     return !err;
 410 }
 411
 412 static int man_rdaddwc(rdstringc *rs, word *text, word *end,
 413                        int quote_props, manconfig *conf,
 414                        charset_state *state) {
 415     char *c;
 416
 417     for (; text && text != end; text = text->next) switch (text->type) {
 418       case word_HyperLink:
 419       case word_HyperEnd:
 420       case word_UpperXref:
 421       case word_LowerXref:
 422       case word_XrefEnd:
 423       case word_IndexRef:
 424         break;
 425
 426       case word_Normal:
 427       case word_Emph:
 428       case word_Code:
 429       case word_WeakCode:
 430       case word_WhiteSpace:
 431       case word_EmphSpace:
 432       case word_CodeSpace:
 433       case word_WkCodeSpace:
 434       case word_Quote:
 435       case word_EmphQuote:
 436       case word_CodeQuote:
 437       case word_WkCodeQuote:
 438         assert(text->type != word_CodeQuote &&
 439                text->type != word_WkCodeQuote);
 440
 441         if (towordstyle(text->type) == word_Emph &&
 442             (attraux(text->aux) == attr_First ||
 443              attraux(text->aux) == attr_Only)) {
 444             man_convert(NULL, 0, &c, quote_props, conf->charset, state);
 445             rdaddsc(rs, c);
 446             if (*c)
 447                 quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
 448             sfree(c);
 449             *state = charset_init_state;
 450             rdaddsc(rs, "\\fI");
 451         } else if ((towordstyle(text->type) == word_Code ||
 452                     towordstyle(text->type) == word_WeakCode) &&
 453                    (attraux(text->aux) == attr_First ||
 454                     attraux(text->aux) == attr_Only)) {
 455             man_convert(NULL, 0, &c, quote_props, conf->charset, state);
 456             rdaddsc(rs, c);
 457             if (*c)
 458                 quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
 459             sfree(c);
 460             *state = charset_init_state;
 461             rdaddsc(rs, "\\fB");
 462         }
 463
 464         if (removeattr(text->type) == word_Normal) {
 465             charset_state s2 = *state;
 466
 467             if (man_convert(text->text, 0, &c, quote_props, conf->charset, &s2) ||
 468                 !text->alt) {
 469                 rdaddsc(rs, c);
 470                 if (*c)
 471                     quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
 472                 *state = s2;
 473             } else {
 474                 quote_props = man_rdaddwc(rs, text->alt, NULL,
 475                                           quote_props, conf, state);
 476             }
 477             sfree(c);
 478         } else if (removeattr(text->type) == word_WhiteSpace) {
 479             man_convert(L" ", 1, &c, quote_props, conf->charset, state);
 480             rdaddsc(rs, c);
 481             if (*c)
 482                 quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
 483             sfree(c);
 484         } else if (removeattr(text->type) == word_Quote) {
 485             man_convert(quoteaux(text->aux) == quote_Open ?
 486                         conf->lquote : conf->rquote, 0,
 487                         &c, quote_props, conf->charset, state);
 488             rdaddsc(rs, c);
 489             if (*c)
 490                 quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
 491             sfree(c);
 492         }
 493         if (towordstyle(text->type) != word_Normal &&
 494             (attraux(text->aux) == attr_Last ||
 495              attraux(text->aux) == attr_Only)) {
 496             man_convert(NULL, 0, &c, quote_props, conf->charset, state);
 497             rdaddsc(rs, c);
 498             if (*c)
 499                 quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
 500             sfree(c);
 501             *state = charset_init_state;
 502             rdaddsc(rs, "\\fP");
 503         }
 504         break;
 505     }
 506     man_convert(NULL, 0, &c, quote_props, conf->charset, state);
 507     rdaddsc(rs, c);
 508     if (*c)
 509         quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
 510     sfree(c);
 511
 512     return quote_props;
 513 }
 514
 515 static void man_text(FILE *fp, word *text, int newline,
 516                      int quote_props, manconfig *conf) {
 517     rdstringc t = { 0, 0, NULL };
 518     charset_state state = CHARSET_INIT_STATE;
 519
 520     man_rdaddwc(&t, text, NULL, quote_props | QUOTE_INITCTRL, conf, &state);
 521     fprintf(fp, "%s", t.text);
 522     sfree(t.text);
 523     if (newline)
 524         fputc('\n', fp);
 525 }
 526
 527 static void man_codepara(FILE *fp, word *text, int charset) {
 528     fprintf(fp, ".nf\n");
 529     for (; text; text = text->next) if (text->type == word_WeakCode) {
 530         char *c;
 531         wchar_t *t, *e;
 532         int quote_props = QUOTE_INITCTRL;
 533
 534         t = text->text;
 535         if (text->next && text->next->type == word_Emph) {
 536             e = text->next->text;
 537             text = text->next;
 538         } else
 539             e = NULL;
 540
 541         while (e && *e && *t) {
 542             int n;
 543             int ec = *e;
 544
 545             for (n = 0; t[n] && e[n] && e[n] == ec; n++);
 546             if (ec == 'i')
 547                 fprintf(fp, "\\fI");
 548             else if (ec == 'b')
 549                 fprintf(fp, "\\fB");
 550             man_convert(t, n, &c, quote_props, charset, NULL);
 551             quote_props &= ~QUOTE_INITCTRL;
 552             fprintf(fp, "%s", c);
 553             sfree(c);
 554             if (ec == 'i' || ec == 'b')
 555                 fprintf(fp, "\\fP");
 556             t += n;
 557             e += n;
 558         }
 559         man_convert(t, 0, &c, quote_props, charset, NULL);
 560         fprintf(fp, "%s\n", c);
 561         sfree(c);
 562     }
 563     fprintf(fp, ".fi\n");
 564 }