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     char const *sep;
  66     manconfig conf;
  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     /* Do the preamble and copyright */
 111     sep = "";
 112     for (p = sourceform; p; p = p->next)
 113         if (p->type == para_Preamble) {
 114             fprintf(fp, "%s", sep);
 115             man_text(fp, p->words, TRUE, 0);
 116             sep = "\n";
 117         }
 118     for (p = sourceform; p; p = p->next)
 119         if (p->type == para_Copyright) {
 120             fprintf(fp, "%s", sep);
 121             man_text(fp, p->words, TRUE, 0);
 122             sep = "\n";
 123         }
 124
 125     for (p = sourceform; p; p = p->next) switch (p->type) {
 126         /*
 127          * Things we ignore because we've already processed them or
 128          * aren't going to touch them in this pass.
 129          */
 130       case para_IM:
 131       case para_BR:
 132       case para_Biblio:                /* only touch BiblioCited */
 133       case para_VersionID:
 134       case para_Copyright:
 135       case para_Preamble:
 136       case para_NoCite:
 137       case para_Title:
 138         break;
 139
 140         /*
 141          * Headings.
 142          */
 143       case para_Chapter:
 144       case para_Appendix:
 145       case para_UnnumberedChapter:
 146       case para_Heading:
 147       case para_Subsect:
 148         {
 149             int depth;
 150             if (p->type == para_Subsect)
 151                 depth = p->aux + 2;
 152             else if (p->type == para_Heading)
 153                 depth = 1;
 154             else
 155                 depth = 0;
 156             if (depth >= conf.mindepth) {
 157                 fprintf(fp, ".SH \"");
 158                 if (conf.headnumbers && p->kwtext) {
 159                     man_text(fp, p->kwtext, FALSE, QUOTE_QUOTES);
 160                     fprintf(fp, " ");
 161                 }
 162                 man_text(fp, p->words, FALSE, QUOTE_QUOTES);
 163                 fprintf(fp, "\"\n");
 164             }
 165             break;
 166         }
 167
 168         /*
 169          * Code paragraphs.
 170          */
 171       case para_Code:
 172         fprintf(fp, ".PP\n");
 173         man_codepara(fp, p->words);
 174         break;
 175
 176         /*
 177          * Normal paragraphs.
 178          */
 179       case para_Normal:
 180         fprintf(fp, ".PP\n");
 181         man_text(fp, p->words, TRUE, 0);
 182         break;
 183
 184         /*
 185          * List paragraphs.
 186          */
 187       case para_Description:
 188       case para_BiblioCited:
 189       case para_Bullet:
 190       case para_NumberedList:
 191         if (p->type == para_Bullet) {
 192             fprintf(fp, ".IP \"\\fBo\\fP\"\n");   /* FIXME: configurable? */
 193         } else if (p->type == para_NumberedList) {
 194             fprintf(fp, ".IP \"");
 195             man_text(fp, p->kwtext, FALSE, QUOTE_QUOTES);
 196             fprintf(fp, "\"\n");
 197         } else if (p->type == para_Description) {
 198             /*
 199              * Do nothing; the .xP for this paragraph is the .IP
 200              * which has come before it in the DescribedThing.
 201              */
 202         } else if (p->type == para_BiblioCited) {
 203             fprintf(fp, ".IP \"");
 204             man_text(fp, p->kwtext, FALSE, QUOTE_QUOTES);
 205             fprintf(fp, "\"\n");
 206         }
 207         man_text(fp, p->words, TRUE, 0);
 208         break;
 209
 210       case para_DescribedThing:
 211         fprintf(fp, ".IP \"");
 212         man_text(fp, p->words, FALSE, QUOTE_QUOTES);
 213         fprintf(fp, "\"\n");
 214         break;
 215
 216       case para_Rule:
 217         /*
 218          * This isn't terribly good. Anyone who wants to do better
 219          * should feel free!
 220          */
 221         fprintf(fp, ".PP\n----------------------------------------\n");
 222         break;
 223
 224       case para_LcontPush:
 225         fprintf(fp, ".RS\n");
 226         break;
 227       case para_LcontPop:
 228         fprintf(fp, ".RE\n");
 229         break;
 230     }
 231
 232     /*
 233      * Tidy up.
 234      */
 235     fclose(fp);
 236     man_conf_cleanup(conf);
 237 }
 238
 239 /*
 240  * Convert a wide string into a string of chars. If `result' is
 241  * non-NULL, mallocs the resulting string and stores a pointer to
 242  * it in `*result'. If `result' is NULL, merely checks whether all
 243  * characters in the string are feasible for the output character
 244  * set.
 245  *
 246  * Return is nonzero if all characters are OK. If not all
 247  * characters are OK but `result' is non-NULL, a result _will_
 248  * still be generated!
 249  *
 250  * FIXME: Here is probably also a good place to do escaping sorts
 251  * of things. I know I at least need to escape backslash, and full
 252  * stops at the starts of words are probably trouble as well.
 253  */
 254 static int man_convert(wchar_t *s, int maxlen,
 255                        char **result, int quote_props) {
 256     /*
 257      * FIXME. Currently this is ISO8859-1 only.
 258      */
 259     int doing = (result != 0);
 260     int ok = TRUE;
 261     char *p = NULL;
 262     int plen = 0, psize = 0;
 263
 264     if (maxlen <= 0)
 265         maxlen = -1;
 266
 267     for (; *s && maxlen != 0; s++, maxlen--) {
 268         wchar_t c = *s;
 269         char outc;
 270
 271         if ((c >= 32 && c <= 126) ||
 272             (c >= 160 && c <= 255)) {
 273             /* Char is OK. */
 274             outc = (char)c;
 275         } else {
 276             /* Char is not OK. */
 277             ok = FALSE;
 278             outc = 0xBF;               /* approximate the good old DEC `uh?' */
 279         }
 280         if (doing) {
 281             if (plen+3 >= psize) {
 282                 psize = plen + 256;
 283                 p = resize(p, psize);
 284             }
 285             if (plen == 0 && (outc == '.' || outc == '\'') &&
 286                 (quote_props & QUOTE_INITCTRL)) {
 287                 /*
 288                  * Control character (. or ') at the start of a
 289                  * line. Quote it by putting \& (troff zero-width
 290                  * space) before it.
 291                  */
 292                 p[plen++] = '\\';
 293                 p[plen++] = '&';
 294             } else if (outc == '\\') {
 295                 /*
 296                  * Quote backslashes by doubling them, always.
 297                  */
 298                 p[plen++] = '\\';
 299             } else if (outc == '"' && (quote_props & QUOTE_QUOTES)) {
 300                 /*
 301                  * Double quote within double quotes. Quote it by
 302                  * doubling.
 303                  */
 304                 p[plen++] = '"';
 305             }
 306             p[plen++] = outc;
 307         }
 308     }
 309     if (doing) {
 310         p = resize(p, plen+1);
 311         p[plen] = '\0';
 312         *result = p;
 313     }
 314     return ok;
 315 }
 316
 317 static void man_rdaddwc(rdstringc *rs, word *text, word *end,
 318                         int quote_props) {
 319     char *c;
 320
 321     for (; text && text != end; text = text->next) switch (text->type) {
 322       case word_HyperLink:
 323       case word_HyperEnd:
 324       case word_UpperXref:
 325       case word_LowerXref:
 326       case word_XrefEnd:
 327       case word_IndexRef:
 328         break;
 329
 330       case word_Normal:
 331       case word_Emph:
 332       case word_Code:
 333       case word_WeakCode:
 334       case word_WhiteSpace:
 335       case word_EmphSpace:
 336       case word_CodeSpace:
 337       case word_WkCodeSpace:
 338       case word_Quote:
 339       case word_EmphQuote:
 340       case word_CodeQuote:
 341       case word_WkCodeQuote:
 342         assert(text->type != word_CodeQuote &&
 343                text->type != word_WkCodeQuote);
 344         if (towordstyle(text->type) == word_Emph &&
 345             (attraux(text->aux) == attr_First ||
 346              attraux(text->aux) == attr_Only))
 347             rdaddsc(rs, "\\fI");
 348         else if ((towordstyle(text->type) == word_Code ||
 349                   towordstyle(text->type) == word_WeakCode) &&
 350                  (attraux(text->aux) == attr_First ||
 351                   attraux(text->aux) == attr_Only))
 352             rdaddsc(rs, "\\fB");
 353         if (removeattr(text->type) == word_Normal) {
 354             if (rs->pos > 0)
 355                 quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
 356             if (man_convert(text->text, 0, &c, quote_props))
 357                 rdaddsc(rs, c);
 358             else
 359                 man_rdaddwc(rs, text->alt, NULL, quote_props);
 360             sfree(c);
 361         } else if (removeattr(text->type) == word_WhiteSpace) {
 362             rdaddc(rs, ' ');
 363         } else if (removeattr(text->type) == word_Quote) {
 364             rdaddc(rs, '"');
 365             if (quote_props & QUOTE_QUOTES)
 366                 rdaddc(rs, '"');
 367         }
 368         if (towordstyle(text->type) == word_Emph &&
 369             (attraux(text->aux) == attr_Last ||
 370              attraux(text->aux) == attr_Only))
 371             rdaddsc(rs, "\\fP");
 372         else if ((towordstyle(text->type) == word_Code ||
 373                   towordstyle(text->type) == word_WeakCode) &&
 374                  (attraux(text->aux) == attr_Last ||
 375                   attraux(text->aux) == attr_Only))
 376             rdaddsc(rs, "\\fP");
 377         break;
 378     }
 379 }
 380
 381 static void man_text(FILE *fp, word *text, int newline, int quote_props) {
 382     rdstringc t = { 0, 0, NULL };
 383
 384     man_rdaddwc(&t, text, NULL, quote_props | QUOTE_INITCTRL);
 385     fprintf(fp, "%s", t.text);
 386     sfree(t.text);
 387     if (newline)
 388         fputc('\n', fp);
 389 }
 390
 391 static void man_codepara(FILE *fp, word *text) {
 392     fprintf(fp, ".nf\n");
 393     for (; text; text = text->next) if (text->type == word_WeakCode) {
 394         char *c;
 395         wchar_t *t, *e;
 396         int quote_props = QUOTE_INITCTRL;
 397
 398         t = text->text;
 399         if (text->next && text->next->type == word_Emph) {
 400             e = text->next->text;
 401             text = text->next;
 402         } else
 403             e = NULL;
 404
 405         while (e && *e && *t) {
 406             int n;
 407             int ec = *e;
 408
 409             for (n = 0; t[n] && e[n] && e[n] == ec; n++);
 410             if (ec == 'i')
 411                 fprintf(fp, "\\fI");
 412             else if (ec == 'b')
 413                 fprintf(fp, "\\fB");
 414             man_convert(t, n, &c, quote_props);
 415             quote_props &= ~QUOTE_INITCTRL;
 416             fprintf(fp, "%s", c);
 417             sfree(c);
 418             if (ec == 'i' || ec == 'b')
 419                 fprintf(fp, "\\fP");
 420             t += n;
 421             e += n;
 422         }
 423         man_convert(t, 0, &c, quote_props);
 424         fprintf(fp, "%s\n", c);
 425         sfree(c);
 426     }
 427     fprintf(fp, ".fi\n");
 428 }