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