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 static void man_text(FILE *, word *, int newline, int quote_props);
  11 static void man_codepara(FILE *, word *);
  12 static int man_convert(wchar_t *s, int maxlen,
  13                        char **result, int quote_props);
  14
  15 typedef struct {
  16     wchar_t *th;
  17     int headnumbers;
  18     int mindepth;
  19 } manconfig;
  20
  21 static manconfig man_configure(paragraph *source) {
  22     manconfig ret;
  23
  24     /*
  25      * Defaults.
  26      */
  27     ret.th = NULL;
  28     ret.headnumbers = FALSE;
  29     ret.mindepth = 0;
  30
  31     for (; source; source = source->next) {
  32         if (source->type == para_Config) {
  33             if (!ustricmp(source->keyword, L"man-identity")) {
  34                 wchar_t *wp, *ep;
  35
  36                 wp = uadv(source->keyword);
  37                 ep = wp;
  38                 while (*ep)
  39                     ep = uadv(ep);
  40                 ret.th = mknewa(wchar_t, ep - wp + 1);
  41                 memcpy(ret.th, wp, (ep - wp + 1) * sizeof(wchar_t));
  42             } else if (!ustricmp(source->keyword, L"man-headnumbers")) {
  43                 ret.headnumbers = utob(uadv(source->keyword));
  44             } else if (!ustricmp(source->keyword, L"man-mindepth")) {
  45                 ret.mindepth = utoi(uadv(source->keyword));
  46             }
  47         }
  48     }
  49
  50     return ret;
  51 }
  52
  53 static void man_conf_cleanup(manconfig cf)
  54 {
  55     sfree(cf.th);
  56 }
  57
  58 #define QUOTE_INITCTRL 1 /* quote initial . and ' on a line */
  59 #define QUOTE_QUOTES   2 /* quote double quotes by doubling them */
  60
  61 void man_backend(paragraph *sourceform, keywordlist *keywords,
  62                  indexdata *idx) {
  63     paragraph *p;
  64     FILE *fp;
  65     manconfig conf;
  66     int done_copyright;
  67
  68     IGNORE(keywords);                  /* we don't happen to need this */
  69     IGNORE(idx);                       /* or this */
  70
  71     conf = man_configure(sourceform);
  72
  73     /*
  74      * Determine the output file name, and open the output file
  75      *
  76      * FIXME: want configurable output file names here. For the
  77      * moment, we'll just call it `output.1'.
  78      */
  79     fp = fopen("output.1", "w");
  80     if (!fp) {
  81         error(err_cantopenw, "output.1");
  82         return;
  83     }
  84
  85     /* Do the version ID */
  86     for (p = sourceform; p; p = p->next)
  87         if (p->type == para_VersionID) {
  88             fprintf(fp, ".\\\" ");
  89             man_text(fp, p->words, TRUE, 0);
  90         }
  91
  92     /* .TH name-of-program manual-section */
  93     fprintf(fp, ".TH");
  94     if (conf.th && *conf.th) {
  95         char *c;
  96         wchar_t *wp;
  97
  98         for (wp = conf.th; *wp; wp = uadv(wp)) {
  99             fputs(" \"", fp);
 100             man_convert(wp, 0, &c, QUOTE_QUOTES);
 101             fputs(c, fp);
 102             sfree(c);
 103             fputc('"', fp);
 104         }
 105     }
 106     fputc('\n', fp);
 107
 108     fprintf(fp, ".UC\n");
 109
 110     done_copyright = FALSE;
 111
 112     for (p = sourceform; p; p = p->next) switch (p->type) {
 113         /*
 114          * Things we ignore because we've already processed them or
 115          * aren't going to touch them in this pass.
 116          */
 117       case para_IM:
 118       case para_BR:
 119       case para_Biblio:                /* only touch BiblioCited */
 120       case para_VersionID:
 121       case para_Copyright:
 122       case para_NoCite:
 123       case para_Title:
 124         break;
 125
 126         /*
 127          * Headings.
 128          */
 129       case para_Chapter:
 130       case para_Appendix:
 131       case para_UnnumberedChapter:
 132       case para_Heading:
 133       case para_Subsect:
 134
 135         if (!done_copyright) {
 136             paragraph *p;
 137
 138             /*
 139              * The copyright comes just before the first chapter
 140              * title.
 141              */
 142             for (p = sourceform; p; p = p->next)
 143                 if (p->type == para_Copyright) {
 144                     fprintf(fp, ".PP\n");
 145                     man_text(fp, p->words, TRUE, 0);
 146                 }
 147             done_copyright = TRUE;
 148         }
 149
 150         {
 151             int depth;
 152             if (p->type == para_Subsect)
 153                 depth = p->aux + 2;
 154             else if (p->type == para_Heading)
 155                 depth = 1;
 156             else
 157                 depth = 0;
 158             if (depth >= conf.mindepth) {
 159                 fprintf(fp, ".SH \"");
 160                 if (conf.headnumbers && p->kwtext) {
 161                     man_text(fp, p->kwtext, FALSE, QUOTE_QUOTES);
 162                     fprintf(fp, " ");
 163                 }
 164                 man_text(fp, p->words, FALSE, QUOTE_QUOTES);
 165                 fprintf(fp, "\"\n");
 166             }
 167             break;
 168         }
 169
 170         /*
 171          * Code paragraphs.
 172          */
 173       case para_Code:
 174         fprintf(fp, ".PP\n");
 175         man_codepara(fp, p->words);
 176         break;
 177
 178         /*
 179          * Normal paragraphs.
 180          */
 181       case para_Normal:
 182         fprintf(fp, ".PP\n");
 183         man_text(fp, p->words, TRUE, 0);
 184         break;
 185
 186         /*
 187          * List paragraphs.
 188          */
 189       case para_Description:
 190       case para_BiblioCited:
 191       case para_Bullet:
 192       case para_NumberedList:
 193         if (p->type == para_Bullet) {
 194             fprintf(fp, ".IP \"\\fBo\\fP\"\n");   /* FIXME: configurable? */
 195         } else if (p->type == para_NumberedList) {
 196             fprintf(fp, ".IP \"");
 197             man_text(fp, p->kwtext, FALSE, QUOTE_QUOTES);
 198             fprintf(fp, "\"\n");
 199         } else if (p->type == para_Description) {
 200             /*
 201              * Do nothing; the .xP for this paragraph is the .IP
 202              * which has come before it in the DescribedThing.
 203              */
 204         } else if (p->type == para_BiblioCited) {
 205             fprintf(fp, ".IP \"");
 206             man_text(fp, p->kwtext, FALSE, QUOTE_QUOTES);
 207             fprintf(fp, "\"\n");
 208         }
 209         man_text(fp, p->words, TRUE, 0);
 210         break;
 211
 212       case para_DescribedThing:
 213         fprintf(fp, ".IP \"");
 214         man_text(fp, p->words, FALSE, QUOTE_QUOTES);
 215         fprintf(fp, "\"\n");
 216         break;
 217
 218       case para_Rule:
 219         /*
 220          * This isn't terribly good. Anyone who wants to do better
 221          * should feel free!
 222          */
 223         fprintf(fp, ".PP\n----------------------------------------\n");
 224         break;
 225
 226       case para_LcontPush:
 227       case para_QuotePush:
 228         fprintf(fp, ".RS\n");
 229         break;
 230       case para_LcontPop:
 231       case para_QuotePop:
 232         fprintf(fp, ".RE\n");
 233         break;
 234     }
 235
 236     /*
 237      * Tidy up.
 238      */
 239     fclose(fp);
 240     man_conf_cleanup(conf);
 241 }
 242
 243 /*
 244  * Convert a wide string into a string of chars. If `result' is
 245  * non-NULL, mallocs the resulting string and stores a pointer to
 246  * it in `*result'. If `result' is NULL, merely checks whether all
 247  * characters in the string are feasible for the output character
 248  * set.
 249  *
 250  * Return is nonzero if all characters are OK. If not all
 251  * characters are OK but `result' is non-NULL, a result _will_
 252  * still be generated!
 253  *
 254  * FIXME: Here is probably also a good place to do escaping sorts
 255  * of things. I know I at least need to escape backslash, and full
 256  * stops at the starts of words are probably trouble as well.
 257  */
 258 static int man_convert(wchar_t *s, int maxlen,
 259                        char **result, int quote_props) {
 260     /*
 261      * FIXME. Currently this is ISO8859-1 only.
 262      */
 263     int doing = (result != 0);
 264     int ok = TRUE;
 265     char *p = NULL;
 266     int plen = 0, psize = 0;
 267
 268     if (maxlen <= 0)
 269         maxlen = -1;
 270
 271     for (; *s && maxlen != 0; s++, maxlen--) {
 272         wchar_t c = *s;
 273         char outc;
 274
 275         if ((c >= 32 && c <= 126) ||
 276             (c >= 160 && c <= 255)) {
 277             /* Char is OK. */
 278             outc = (char)c;
 279         } else {
 280             /* Char is not OK. */
 281             ok = FALSE;
 282             outc = 0xBF;               /* approximate the good old DEC `uh?' */
 283         }
 284         if (doing) {
 285             if (plen+3 >= psize) {
 286                 psize = plen + 256;
 287                 p = resize(p, psize);
 288             }
 289             if (plen == 0 && (outc == '.' || outc == '\'') &&
 290                 (quote_props & QUOTE_INITCTRL)) {
 291                 /*
 292                  * Control character (. or ') at the start of a
 293                  * line. Quote it by putting \& (troff zero-width
 294                  * space) before it.
 295                  */
 296                 p[plen++] = '\\';
 297                 p[plen++] = '&';
 298             } else if (outc == '\\') {
 299                 /*
 300                  * Quote backslashes by doubling them, always.
 301                  */
 302                 p[plen++] = '\\';
 303             } else if (outc == '"' && (quote_props & QUOTE_QUOTES)) {
 304                 /*
 305                  * Double quote within double quotes. Quote it by
 306                  * doubling.
 307                  */
 308                 p[plen++] = '"';
 309             }
 310             p[plen++] = outc;
 311         }
 312     }
 313     if (doing) {
 314         p = resize(p, plen+1);
 315         p[plen] = '\0';
 316         *result = p;
 317     }
 318     return ok;
 319 }
 320
 321 static void man_rdaddwc(rdstringc *rs, word *text, word *end,
 322                         int quote_props) {
 323     char *c;
 324
 325     for (; text && text != end; text = text->next) switch (text->type) {
 326       case word_HyperLink:
 327       case word_HyperEnd:
 328       case word_UpperXref:
 329       case word_LowerXref:
 330       case word_XrefEnd:
 331       case word_IndexRef:
 332         break;
 333
 334       case word_Normal:
 335       case word_Emph:
 336       case word_Code:
 337       case word_WeakCode:
 338       case word_WhiteSpace:
 339       case word_EmphSpace:
 340       case word_CodeSpace:
 341       case word_WkCodeSpace:
 342       case word_Quote:
 343       case word_EmphQuote:
 344       case word_CodeQuote:
 345       case word_WkCodeQuote:
 346         assert(text->type != word_CodeQuote &&
 347                text->type != word_WkCodeQuote);
 348         if (towordstyle(text->type) == word_Emph &&
 349             (attraux(text->aux) == attr_First ||
 350              attraux(text->aux) == attr_Only))
 351             rdaddsc(rs, "\\fI");
 352         else if ((towordstyle(text->type) == word_Code ||
 353                   towordstyle(text->type) == word_WeakCode) &&
 354                  (attraux(text->aux) == attr_First ||
 355                   attraux(text->aux) == attr_Only))
 356             rdaddsc(rs, "\\fB");
 357         if (removeattr(text->type) == word_Normal) {
 358             if (rs->pos > 0)
 359                 quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
 360             if (man_convert(text->text, 0, &c, quote_props))
 361                 rdaddsc(rs, c);
 362             else
 363                 man_rdaddwc(rs, text->alt, NULL, quote_props);
 364             sfree(c);
 365         } else if (removeattr(text->type) == word_WhiteSpace) {
 366             rdaddc(rs, ' ');
 367         } else if (removeattr(text->type) == word_Quote) {
 368             rdaddc(rs, '"');
 369             if (quote_props & QUOTE_QUOTES)
 370                 rdaddc(rs, '"');
 371         }
 372         if (towordstyle(text->type) == word_Emph &&
 373             (attraux(text->aux) == attr_Last ||
 374              attraux(text->aux) == attr_Only))
 375             rdaddsc(rs, "\\fP");
 376         else if ((towordstyle(text->type) == word_Code ||
 377                   towordstyle(text->type) == word_WeakCode) &&
 378                  (attraux(text->aux) == attr_Last ||
 379                   attraux(text->aux) == attr_Only))
 380             rdaddsc(rs, "\\fP");
 381         break;
 382     }
 383 }
 384
 385 static void man_text(FILE *fp, word *text, int newline, int quote_props) {
 386     rdstringc t = { 0, 0, NULL };
 387
 388     man_rdaddwc(&t, text, NULL, quote_props | QUOTE_INITCTRL);
 389     fprintf(fp, "%s", t.text);
 390     sfree(t.text);
 391     if (newline)
 392         fputc('\n', fp);
 393 }
 394
 395 static void man_codepara(FILE *fp, word *text) {
 396     fprintf(fp, ".nf\n");
 397     for (; text; text = text->next) if (text->type == word_WeakCode) {
 398         char *c;
 399         wchar_t *t, *e;
 400         int quote_props = QUOTE_INITCTRL;
 401
 402         t = text->text;
 403         if (text->next && text->next->type == word_Emph) {
 404             e = text->next->text;
 405             text = text->next;
 406         } else
 407             e = NULL;
 408
 409         while (e && *e && *t) {
 410             int n;
 411             int ec = *e;
 412
 413             for (n = 0; t[n] && e[n] && e[n] == ec; n++);
 414             if (ec == 'i')
 415                 fprintf(fp, "\\fI");
 416             else if (ec == 'b')
 417                 fprintf(fp, "\\fB");
 418             man_convert(t, n, &c, quote_props);
 419             quote_props &= ~QUOTE_INITCTRL;
 420             fprintf(fp, "%s", c);
 421             sfree(c);
 422             if (ec == 'i' || ec == 'b')
 423                 fprintf(fp, "\\fP");
 424             t += n;
 425             e += n;
 426         }
 427         man_convert(t, 0, &c, quote_props);
 428         fprintf(fp, "%s\n", c);
 429         sfree(c);
 430     }
 431     fprintf(fp, ".fi\n");
 432 }