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
 128     IGNORE(unused);
 129     IGNORE(keywords);
 130     IGNORE(idx);
 131
 132     conf = man_configure(sourceform);
 133
 134     /*
 135      * Open the output file.
 136      */
 137     fp = fopen(conf.filename, "w");
 138     if (!fp) {
 139         error(err_cantopenw, conf.filename);
 140         return;
 141     }
 142
 143     /* Do the version ID */
 144     for (p = sourceform; p; p = p->next)
 145         if (p->type == para_VersionID) {
 146             fprintf(fp, ".\\\" ");
 147             man_text(fp, p->words, TRUE, 0, &conf);
 148         }
 149
 150     /* .TH name-of-program manual-section */
 151     fprintf(fp, ".TH");
 152     if (conf.th && *conf.th) {
 153         char *c;
 154         wchar_t *wp;
 155
 156         for (wp = conf.th; *wp; wp = uadv(wp)) {
 157             fputs(" \"", fp);
 158             man_convert(wp, 0, &c, QUOTE_QUOTES, conf.charset, NULL);
 159             fputs(c, fp);
 160             sfree(c);
 161             fputc('"', fp);
 162         }
 163     }
 164     fputc('\n', fp);
 165
 166     fprintf(fp, ".UC\n");
 167
 168     for (p = sourceform; p; p = p->next) switch (p->type) {
 169         /*
 170          * Things we ignore because we've already processed them or
 171          * aren't going to touch them in this pass.
 172          */
 173       case para_IM:
 174       case para_BR:
 175       case para_Biblio:                /* only touch BiblioCited */
 176       case para_VersionID:
 177       case para_NoCite:
 178       case para_Title:
 179         break;
 180
 181         /*
 182          * Headings.
 183          */
 184       case para_Chapter:
 185       case para_Appendix:
 186       case para_UnnumberedChapter:
 187       case para_Heading:
 188       case para_Subsect:
 189
 190         {
 191             int depth;
 192             if (p->type == para_Subsect)
 193                 depth = p->aux + 2;
 194             else if (p->type == para_Heading)
 195                 depth = 1;
 196             else
 197                 depth = 0;
 198             if (depth >= conf.mindepth) {
 199                 fprintf(fp, ".SH \"");
 200                 if (conf.headnumbers && p->kwtext) {
 201                     man_text(fp, p->kwtext, FALSE, QUOTE_QUOTES, &conf);
 202                     fprintf(fp, " ");
 203                 }
 204                 man_text(fp, p->words, FALSE, QUOTE_QUOTES, &conf);
 205                 fprintf(fp, "\"\n");
 206             }
 207             break;
 208         }
 209
 210         /*
 211          * Code paragraphs.
 212          */
 213       case para_Code:
 214         fprintf(fp, ".PP\n");
 215         man_codepara(fp, p->words, conf.charset);
 216         break;
 217
 218         /*
 219          * Normal paragraphs.
 220          */
 221       case para_Normal:
 222       case para_Copyright:
 223         fprintf(fp, ".PP\n");
 224         man_text(fp, p->words, TRUE, 0, &conf);
 225         break;
 226
 227         /*
 228          * List paragraphs.
 229          */
 230       case para_Description:
 231       case para_BiblioCited:
 232       case para_Bullet:
 233       case para_NumberedList:
 234         if (p->type == para_Bullet) {
 235             char *bullettext;
 236             man_convert(conf.bullet, -1, &bullettext, QUOTE_QUOTES,
 237                         conf.charset, NULL);
 238             fprintf(fp, ".IP \"\\fB%s\\fP\"\n", bullettext);
 239             sfree(bullettext);
 240         } else if (p->type == para_NumberedList) {
 241             fprintf(fp, ".IP \"");
 242             man_text(fp, p->kwtext, FALSE, QUOTE_QUOTES, &conf);
 243             fprintf(fp, "\"\n");
 244         } else if (p->type == para_Description) {
 245             /*
 246              * Do nothing; the .xP for this paragraph is the .IP
 247              * which has come before it in the DescribedThing.
 248              */
 249         } else if (p->type == para_BiblioCited) {
 250             fprintf(fp, ".IP \"");
 251             man_text(fp, p->kwtext, FALSE, QUOTE_QUOTES, &conf);
 252             fprintf(fp, "\"\n");
 253         }
 254         man_text(fp, p->words, TRUE, 0, &conf);
 255         break;
 256
 257       case para_DescribedThing:
 258         fprintf(fp, ".IP \"");
 259         man_text(fp, p->words, FALSE, QUOTE_QUOTES, &conf);
 260         fprintf(fp, "\"\n");
 261         break;
 262
 263       case para_Rule:
 264         /*
 265          * This isn't terribly good. Anyone who wants to do better
 266          * should feel free!
 267          */
 268         fprintf(fp, ".PP\n----------------------------------------\n");
 269         break;
 270
 271       case para_LcontPush:
 272       case para_QuotePush:
 273         fprintf(fp, ".RS\n");
 274         break;
 275       case para_LcontPop:
 276       case para_QuotePop:
 277         fprintf(fp, ".RE\n");
 278         break;
 279     }
 280
 281     /*
 282      * Tidy up.
 283      */
 284     fclose(fp);
 285     man_conf_cleanup(conf);
 286 }
 287
 288 /*
 289  * Convert a wide string into a string of chars; mallocs the
 290  * resulting string and stores a pointer to it in `*result'.
 291  *
 292  * If `state' is non-NULL, updates the charset state pointed to. If
 293  * `state' is NULL, this function uses its own state, initialises
 294  * it from scratch, and cleans it up when finished. If `state' is
 295  * non-NULL but _s_ is NULL, cleans up a provided state.
 296  *
 297  * Return is nonzero if all characters are OK. If not all
 298  * characters are OK but `result' is non-NULL, a result _will_
 299  * still be generated!
 300  *
 301  * This function also does escaping of groff special characters.
 302  */
 303 static int man_convert(wchar_t const *s, int maxlen,
 304                        char **result, int quote_props,
 305                        int charset, charset_state *state) {
 306     charset_state internal_state = CHARSET_INIT_STATE;
 307     int slen, err;
 308     char *p = NULL, *q;
 309     int plen = 0, psize = 0;
 310     rdstringc out = {0, 0, NULL};
 311
 312     if (!state)
 313         state = &internal_state;
 314
 315     slen = (s ? ustrlen(s) : 0);
 316
 317     if (slen > maxlen && maxlen > 0)
 318         slen = maxlen;
 319
 320     psize = 384;
 321     plen = 0;
 322     p = snewn(psize, char);
 323     err = 0;
 324
 325     while (slen > 0) {
 326         int ret = charset_from_unicode(&s, &slen, p+plen, psize-plen,
 327                                    charset, state, (err ? NULL : &err));
 328         if (ret > 0) {
 329             plen += ret;
 330             if (psize - plen < 256) {
 331                 psize = plen + 256;
 332                 p = sresize(p, psize, char);
 333             }
 334         }
 335     }
 336
 337     if (state == &internal_state || s == NULL) {
 338         int ret = charset_from_unicode(NULL, 0, p+plen, psize-plen,
 339                                        charset, state, NULL);
 340         if (ret > 0)
 341             plen += ret;
 342     }
 343
 344     for (q = p; q < p+plen; q++) {
 345         if (q == p && (*q == '.' || *q == '\'') &&
 346             (quote_props & QUOTE_INITCTRL)) {
 347             /*
 348              * Control character (. or ') at the start of a
 349              * line. Quote it by putting \& (troff zero-width
 350              * space) before it.
 351              */
 352             rdaddc(&out, '\\');
 353             rdaddc(&out, '&');
 354         } else if (*q == '\\') {
 355             /*
 356              * Quote backslashes by doubling them, always.
 357              */
 358             rdaddc(&out, '\\');
 359         } else if (*q == '"' && (quote_props & QUOTE_QUOTES)) {
 360             /*
 361              * Double quote within double quotes. Quote it by
 362              * doubling.
 363              */
 364             rdaddc(&out, '"');
 365         }
 366         rdaddc(&out, *q);
 367     }
 368
 369     sfree(p);
 370
 371     if (out.text)
 372         *result = rdtrimc(&out);
 373     else
 374         *result = dupstr("");
 375
 376     return !err;
 377 }
 378
 379 static void man_rdaddwc(rdstringc *rs, word *text, word *end,
 380                         int quote_props, manconfig *conf,
 381                         charset_state *state) {
 382     char *c;
 383
 384     for (; text && text != end; text = text->next) switch (text->type) {
 385       case word_HyperLink:
 386       case word_HyperEnd:
 387       case word_UpperXref:
 388       case word_LowerXref:
 389       case word_XrefEnd:
 390       case word_IndexRef:
 391         break;
 392
 393       case word_Normal:
 394       case word_Emph:
 395       case word_Code:
 396       case word_WeakCode:
 397       case word_WhiteSpace:
 398       case word_EmphSpace:
 399       case word_CodeSpace:
 400       case word_WkCodeSpace:
 401       case word_Quote:
 402       case word_EmphQuote:
 403       case word_CodeQuote:
 404       case word_WkCodeQuote:
 405         assert(text->type != word_CodeQuote &&
 406                text->type != word_WkCodeQuote);
 407
 408         if (towordstyle(text->type) == word_Emph &&
 409             (attraux(text->aux) == attr_First ||
 410              attraux(text->aux) == attr_Only)) {
 411             if (rs->pos > 0)
 412                 quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
 413             man_convert(NULL, 0, &c, quote_props, conf->charset, state);
 414             rdaddsc(rs, c);
 415             sfree(c);
 416             *state = charset_init_state;
 417             rdaddsc(rs, "\\fI");
 418         } else if ((towordstyle(text->type) == word_Code ||
 419                     towordstyle(text->type) == word_WeakCode) &&
 420                    (attraux(text->aux) == attr_First ||
 421                     attraux(text->aux) == attr_Only)) {
 422             if (rs->pos > 0)
 423                 quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
 424             man_convert(NULL, 0, &c, quote_props, conf->charset, state);
 425             rdaddsc(rs, c);
 426             sfree(c);
 427             *state = charset_init_state;
 428             rdaddsc(rs, "\\fB");
 429         }
 430
 431         if (removeattr(text->type) == word_Normal) {
 432             charset_state s2 = *state;
 433
 434             if (rs->pos > 0)
 435                 quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
 436             if (man_convert(text->text, 0, &c, quote_props, conf->charset, &s2) ||
 437                 !text->alt) {
 438                 rdaddsc(rs, c);
 439                 *state = s2;
 440             } else {
 441                 man_rdaddwc(rs, text->alt, NULL, quote_props, conf, state);
 442             }
 443             sfree(c);
 444         } else if (removeattr(text->type) == word_WhiteSpace) {
 445             if (rs->pos > 0)
 446                 quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
 447             man_convert(L" ", 1, &c, quote_props, conf->charset, state);
 448             rdaddsc(rs, c);
 449             sfree(c);
 450         } else if (removeattr(text->type) == word_Quote) {
 451             if (rs->pos > 0)
 452                 quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
 453             man_convert(quoteaux(text->aux) == quote_Open ?
 454                         conf->lquote : conf->rquote, 0,
 455                         &c, quote_props, conf->charset, state);
 456             rdaddsc(rs, c);
 457             sfree(c);
 458         }
 459         if (towordstyle(text->type) != word_Normal &&
 460             (attraux(text->aux) == attr_Last ||
 461              attraux(text->aux) == attr_Only)) {
 462             if (rs->pos > 0)
 463                 quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
 464             man_convert(NULL, 0, &c, quote_props, conf->charset, state);
 465             rdaddsc(rs, c);
 466             sfree(c);
 467             *state = charset_init_state;
 468             rdaddsc(rs, "\\fP");
 469         }
 470         break;
 471     }
 472     man_convert(NULL, 0, &c, quote_props, conf->charset, state);
 473     rdaddsc(rs, c);
 474     sfree(c);
 475 }
 476
 477 static void man_text(FILE *fp, word *text, int newline,
 478                      int quote_props, manconfig *conf) {
 479     rdstringc t = { 0, 0, NULL };
 480     charset_state state = CHARSET_INIT_STATE;
 481
 482     man_rdaddwc(&t, text, NULL, quote_props | QUOTE_INITCTRL, conf, &state);
 483     fprintf(fp, "%s", t.text);
 484     sfree(t.text);
 485     if (newline)
 486         fputc('\n', fp);
 487 }
 488
 489 static void man_codepara(FILE *fp, word *text, int charset) {
 490     fprintf(fp, ".nf\n");
 491     for (; text; text = text->next) if (text->type == word_WeakCode) {
 492         char *c;
 493         wchar_t *t, *e;
 494         int quote_props = QUOTE_INITCTRL;
 495
 496         t = text->text;
 497         if (text->next && text->next->type == word_Emph) {
 498             e = text->next->text;
 499             text = text->next;
 500         } else
 501             e = NULL;
 502
 503         while (e && *e && *t) {
 504             int n;
 505             int ec = *e;
 506
 507             for (n = 0; t[n] && e[n] && e[n] == ec; n++);
 508             if (ec == 'i')
 509                 fprintf(fp, "\\fI");
 510             else if (ec == 'b')
 511                 fprintf(fp, "\\fB");
 512             man_convert(t, n, &c, quote_props, charset, NULL);
 513             quote_props &= ~QUOTE_INITCTRL;
 514             fprintf(fp, "%s", c);
 515             sfree(c);
 516             if (ec == 'i' || ec == 'b')
 517                 fprintf(fp, "\\fP");
 518             t += n;
 519             e += n;
 520         }
 521         man_convert(t, 0, &c, quote_props, charset, NULL);
 522         fprintf(fp, "%s\n", c);
 523         sfree(c);
 524     }
 525     fprintf(fp, ".fi\n");
 526 }