mdw@git.distorted.org.uk Git - sgt/halibut/blob - halibut.h

   1 #ifndef HALIBUT_HALIBUT_H
   2 #define HALIBUT_HALIBUT_H
   3
   4 #include <stdio.h>
   5 #include <wchar.h>
   6 #include <time.h>
   7 #include <string.h>
   8
   9 #include "charset.h"
  10
  11 #ifdef __GNUC__
  12 #define NORETURN __attribute__((__noreturn__))
  13 #else
  14 #define NORETURN /* nothing */
  15 #endif
  16
  17 #ifndef TRUE
  18 #define TRUE 1
  19 #endif
  20 #ifndef FALSE
  21 #define FALSE 0
  22 #endif
  23
  24 /* For suppressing unused-parameter warnings */
  25 #define IGNORE(x) ( (x) = (x) )
  26
  27 #include "tree234.h"
  28
  29 /*
  30  * Structure tags
  31  */
  32 typedef struct input_Tag input;
  33 typedef struct filepos_Tag filepos;
  34 typedef struct paragraph_Tag paragraph;
  35 typedef struct word_Tag word;
  36 typedef struct keywordlist_Tag keywordlist;
  37 typedef struct keyword_Tag keyword;
  38 typedef struct userstyle_Tag userstyle;
  39 typedef struct numberstate_Tag numberstate;
  40 typedef struct indexdata_Tag indexdata;
  41 typedef struct indextag_Tag indextag;
  42 typedef struct indexentry_Tag indexentry;
  43 typedef struct macrostack_Tag macrostack;
  44
  45 /*
  46  * Data structure to hold a file name and index, a line and a
  47  * column number, for reporting errors
  48  */
  49 struct filepos_Tag {
  50     char *filename;
  51     int line, col;
  52 };
  53
  54 /*
  55  * Data structure to hold all the file names etc for input
  56  */
  57 typedef struct pushback_Tag {
  58     int chr;
  59     filepos pos;
  60 } pushback;
  61 struct input_Tag {
  62     char **filenames;                  /* complete list of input files */
  63     int nfiles;                        /* how many in the list */
  64     FILE *currfp;                      /* the currently open one */
  65     int currindex;                     /* which one is that in the list */
  66     pushback *pushback;                /* pushed-back input characters */
  67     int npushback, pushbacksize;
  68     filepos pos;
  69     int reportcols;                    /* report column numbers in errors */
  70     macrostack *stack;                 /* macro expansions in force */
  71     int defcharset, charset;           /* character sets for input files */
  72     charset_state csstate;
  73     wchar_t wc[16];                    /* wide chars from input conversion */
  74     int nwc, wcpos;                    /* size of, and position in, wc[] */
  75     char *pushback_chars;              /* used to save input-encoding data */
  76 };
  77
  78 /*
  79  * Data structure to hold the input form of the source, ie a linked
  80  * list of paragraphs
  81  */
  82 struct paragraph_Tag {
  83     paragraph *next;
  84     int type;
  85     wchar_t *keyword;                  /* for most special paragraphs */
  86     char *origkeyword;                 /* same again in original charset */
  87     word *words;                       /* list of words in paragraph */
  88     int aux;                           /* number, in a numbered paragraph
  89                                         * or subsection level
  90                                         */
  91     word *kwtext;                      /* chapter/section indication */
  92     word *kwtext2;                     /* numeric-only form of kwtext */
  93     filepos fpos;
  94
  95     paragraph *parent, *child, *sibling;   /* for hierarchy navigation */
  96
  97     void *private_data;                /* for temp use in backends */
  98 };
  99 enum {
 100     para_IM,                           /* index merge */
 101     para_BR,                           /* bibliography rewrite */
 102     para_Rule,                         /* random horizontal rule */
 103     para_Chapter,
 104     para_Appendix,
 105     para_UnnumberedChapter,
 106     para_Heading,
 107     para_Subsect,
 108     para_Normal,
 109     para_Biblio,                       /* causes no output unless turned ... */
 110     para_BiblioCited,                  /*  ... into this paragraph type */
 111     para_Bullet,
 112     para_NumberedList,
 113     para_DescribedThing,
 114     para_Description,
 115     para_Code,
 116     para_Copyright,
 117     para_NoCite,
 118     para_Title,
 119     para_VersionID,
 120     para_Config,                       /* configuration directive */
 121     para_LcontPush,                    /* begin continuation of list item */
 122     para_LcontPop,                     /* end continuation of list item */
 123     para_QuotePush,                    /* begin block quote */
 124     para_QuotePop,                     /* end block quote */
 125     /*
 126      * Back ends may define their own paragraph types beyond here,
 127      * in case they need to use them internally.
 128      */
 129     para_NotParaType                   /* placeholder value */
 130 };
 131
 132 /*
 133  * Data structure to hold an individual word
 134  */
 135 struct word_Tag {
 136     word *next, *alt;
 137     int type;
 138     int aux;
 139     int breaks;                        /* can a line break after it? */
 140     wchar_t *text;
 141     filepos fpos;
 142
 143     void *private_data;                /* for temp use in backends */
 144 };
 145 enum {
 146     /* ORDERING CONSTRAINT: these normal-word types ... */
 147     word_Normal,
 148     word_Emph,
 149     word_Code,                         /* monospaced; `quoted' in text */
 150     word_WeakCode,                     /* monospaced, normal in text */
 151     /* ... must be in the same order as these space types ... */
 152     word_WhiteSpace,                   /* text is NULL or ignorable */
 153     word_EmphSpace,                    /* WhiteSpace when emphasised */
 154     word_CodeSpace,                    /* WhiteSpace when code */
 155     word_WkCodeSpace,                  /* WhiteSpace when weak code */
 156     /* ... and must be in the same order as these quote types ... */
 157     word_Quote,                        /* text is NULL or ignorable */
 158     word_EmphQuote,                    /* Quote when emphasised */
 159     word_CodeQuote,                    /* (can't happen) */
 160     word_WkCodeQuote,                  /* (can't happen) */
 161     /* END ORDERING CONSTRAINT */
 162     word_internal_endattrs,
 163     word_UpperXref,                    /* \K */
 164     word_LowerXref,                    /* \k */
 165     word_XrefEnd,                      /* (invisible; no text) */
 166     word_IndexRef,                     /* (always an invisible one) */
 167     word_HyperLink,                    /* (invisible) */
 168     word_HyperEnd,                     /* (also invisible; no text) */
 169     /*
 170      * Back ends may define their own word types beyond here, in
 171      * case they need to use them internally.
 172      */
 173     word_NotWordType                   /* placeholder value */
 174 };
 175 /* aux values for attributed words */
 176 enum {
 177     attr_Only   = 0x0000,              /* a lone word with the attribute */
 178     attr_First  = 0x0001,              /* the first of a series */
 179     attr_Last   = 0x0002,              /* the last of a series */
 180     attr_Always = 0x0003,              /* any other part of a series */
 181     attr_mask   = 0x0003,
 182 };
 183 /* aux values for quote-type words */
 184 enum {
 185     quote_Open  = 0x0010,
 186     quote_Close = 0x0020,
 187     quote_mask  = 0x0030,
 188 };
 189 #define isattr(x) ( ( (x) > word_Normal && (x) < word_WhiteSpace ) || \
 190                     ( (x) > word_WhiteSpace && (x) < word_internal_endattrs ) )
 191 #define sameattr(x,y) ( (((x)-(y)) & 3) == 0 )
 192 #define towordstyle(x) ( word_Normal + ((x) & 3) )
 193 #define tospacestyle(x) ( word_WhiteSpace + ((x) & 3) )
 194 #define toquotestyle(x) ( word_Quote + ((x) & 3) )
 195 #define removeattr(x) ( word_Normal + ((x) &~ 3) )
 196
 197 #define attraux(x) ( (x) & attr_mask )
 198 #define quoteaux(x) ( (x) & quote_mask )
 199
 200 /*
 201  * error.c
 202  */
 203 void fatal(int code, ...) NORETURN;
 204 void error(int code, ...);
 205 enum {
 206     err_nomemory,                      /* out of memory */
 207     err_optnoarg,                      /* option `-%s' requires an argument */
 208     err_nosuchopt,                     /* unrecognised option `-%s' */
 209     err_noinput,                       /* no input files */
 210     err_cantopen,                      /* unable to open input file `%s' */
 211     err_nodata,                        /* no data in input files */
 212     err_brokencodepara,                /* line in codepara didn't begin `\c' */
 213     err_kwunclosed,                    /* expected `}' after keyword */
 214     err_kwillegal,                     /* paragraph type expects no keyword */
 215     err_kwexpected,                    /* paragraph type expects a keyword */
 216     err_kwtoomany,                     /* paragraph type expects only 1 */
 217     err_bodyillegal,                   /* paragraph type expects only kws! */
 218     err_badparatype,                   /* invalid command at start of para */
 219     err_badmidcmd,                     /* invalid command in mid-para */
 220     err_unexbrace,                     /* unexpected brace */
 221     err_explbr,                        /* expected `{' after command */
 222     err_commenteof,                    /* EOF inside braced comment */
 223     err_kwexprbr,                      /* expected `}' after cross-ref */
 224     err_codequote,                     /* \q within \c is not supported */
 225     err_missingrbrace,                 /* unclosed braces at end of para */
 226     err_missingrbrace2,                /* unclosed braces at end of file */
 227     err_nestedstyles,                  /* unable to nest text styles */
 228     err_nestedindex,                   /* unable to nest `\i' thingys */
 229     err_indexcase,                     /* two \i differing only in case */
 230     err_nosuchkw,                      /* unresolved cross-reference */
 231     err_multiBR,                       /* multiple \BRs on same keyword */
 232     err_nosuchidxtag,                  /* \IM on unknown index tag (warning) */
 233     err_cantopenw,                     /* can't open output file for write */
 234     err_macroexists,                   /* this macro already exists */
 235     err_sectjump,                      /* jump a heading level, eg \C -> \S */
 236     err_winhelp_ctxclash,              /* WinHelp context ID hash clash */
 237     err_multikw,                       /* keyword clash in sections */
 238     err_misplacedlcont,                /* \lcont not after a list item */
 239     err_sectmarkerinblock,             /* section marker appeared in block */
 240     err_infodirentry,                  /* \cfg{info-dir-entry} missing param */
 241     err_infonodechar,                  /* colon/comma in node name in info */
 242     err_text_codeline,                 /* \c line too long in text backend */
 243     err_whatever                       /* random error of another type */
 244 };
 245
 246 /*
 247  * malloc.c
 248  */
 249 #ifdef LOGALLOC
 250 void *smalloc(char *file, int line, int size);
 251 void *srealloc(char *file, int line, void *p, int size);
 252 void sfree(char *file, int line, void *p);
 253 #define smalloc(x) smalloc(__FILE__, __LINE__, x)
 254 #define srealloc(x, y) srealloc(__FILE__, __LINE__, x, y)
 255 #define sfree(x) sfree(__FILE__, __LINE__, x)
 256 #else
 257 void *smalloc(int size);
 258 void *srealloc(void *p, int size);
 259 void sfree(void *p);
 260 #endif
 261 void free_word_list(word *w);
 262 void free_para_list(paragraph *p);
 263 word *dup_word_list(word *w);
 264 char *dupstr(char *s);
 265
 266 #define snew(type) ( (type *) smalloc (sizeof (type)) )
 267 #define snewn(number, type) ( (type *) smalloc ((number) * sizeof (type)) )
 268 #define sresize(array, number, type) \
 269         ( (type *) srealloc ((array), (number) * sizeof (type)) )
 270 #define lenof(array) ( sizeof(array) / sizeof(*(array)) )
 271
 272 /*
 273  * ustring.c
 274  */
 275 wchar_t *ustrdup(wchar_t const *s);
 276 char *ustrtoa(wchar_t const *s, char *outbuf, int size, int charset);
 277 char *ustrtoa_careful(wchar_t const *s, char *outbuf, int size, int charset);
 278 wchar_t *ustrfroma(char const *s, wchar_t *outbuf, int size, int charset);
 279 char *utoa_dup(wchar_t const *s, int charset);
 280 char *utoa_dup_len(wchar_t const *s, int charset, int *len);
 281 char *utoa_careful_dup(wchar_t const *s, int charset);
 282 wchar_t *ufroma_dup(char const *s, int charset);
 283 char *utoa_locale_dup(wchar_t const *s);
 284 wchar_t *ufroma_locale_dup(char const *s);
 285 int ustrlen(wchar_t const *s);
 286 wchar_t *uadv(wchar_t *s);
 287 wchar_t *ustrcpy(wchar_t *dest, wchar_t const *source);
 288 wchar_t *ustrncpy(wchar_t *dest, wchar_t const *source, int n);
 289 wchar_t utolower(wchar_t);
 290 int uisalpha(wchar_t);
 291 int ustrcmp(wchar_t *lhs, wchar_t *rhs);
 292 int ustricmp(wchar_t const *lhs, wchar_t const *rhs);
 293 int ustrnicmp(wchar_t const *lhs, wchar_t const *rhs, int maxlen);
 294 int utoi(wchar_t const *);
 295 double utof(wchar_t const *);
 296 int utob(wchar_t const *);
 297 int uisdigit(wchar_t);
 298 wchar_t *ustrlow(wchar_t *s);
 299 wchar_t *ustrftime(const wchar_t *wfmt, const struct tm *timespec);
 300 int cvt_ok(int charset, const wchar_t *s);
 301
 302 /*
 303  * wcwidth.c
 304  */
 305 int strwid(char const *s, int charset);
 306 int ustrwid(wchar_t const *s, int charset);
 307
 308 /*
 309  * help.c
 310  */
 311 void help(void);
 312 void usage(void);
 313 void showversion(void);
 314
 315 /*
 316  * licence.c
 317  */
 318 void licence(void);
 319
 320 /*
 321  * version.c
 322  */
 323 const char *const version;
 324
 325 /*
 326  * misc.c
 327  */
 328 char *adv(char *s);
 329
 330 typedef struct stackTag *stack;
 331 stack stk_new(void);
 332 void stk_free(stack);
 333 void stk_push(stack, void *);
 334 void *stk_pop(stack);
 335 void *stk_top(stack);
 336
 337 typedef struct tagRdstring rdstring;
 338 struct tagRdstring {
 339     int pos, size;
 340     wchar_t *text;
 341 };
 342 typedef struct tagRdstringc rdstringc;
 343 struct tagRdstringc {
 344     int pos, size;
 345     char *text;
 346 };
 347 extern const rdstring empty_rdstring;
 348 extern const rdstringc empty_rdstringc;
 349 void rdadd(rdstring *rs, wchar_t c);
 350 void rdadds(rdstring *rs, wchar_t const *p);
 351 wchar_t *rdtrim(rdstring *rs);
 352 void rdaddc(rdstringc *rs, char c);
 353 void rdaddsc(rdstringc *rs, char const *p);
 354 char *rdtrimc(rdstringc *rs);
 355
 356 int compare_wordlists(word *a, word *b);
 357
 358 void mark_attr_ends(word *words);
 359
 360 typedef struct tagWrappedLine wrappedline;
 361 struct tagWrappedLine {
 362     wrappedline *next;
 363     word *begin, *end;                 /* first & last words of line */
 364     int nspaces;                       /* number of whitespaces in line */
 365     int shortfall;                     /* how much shorter than max width */
 366 };
 367 wrappedline *wrap_para(word *, int, int, int (*)(void *, word *), void *, int);
 368 void wrap_free(wrappedline *);
 369 void cmdline_cfg_add(paragraph *cfg, char *string);
 370 paragraph *cmdline_cfg_new(void);
 371 paragraph *cmdline_cfg_simple(char *string, ...);
 372
 373 /*
 374  * input.c
 375  */
 376 paragraph *read_input(input *in, indexdata *idx);
 377
 378 /*
 379  * keywords.c
 380  */
 381 struct keywordlist_Tag {
 382     int nkeywords;
 383     int size;
 384     tree234 *keys;                     /* sorted by `key' field */
 385     word **looseends;                  /* non-keyword list element numbers */
 386     int nlooseends;
 387     int looseendssize;
 388 };
 389 struct keyword_Tag {
 390     wchar_t *key;                      /* the keyword itself */
 391     word *text;                        /* "Chapter 2", "Appendix Q"... */
 392                                        /* (NB: filepos are not set) */
 393     paragraph *para;                   /* the paragraph referenced */
 394 };
 395 keyword *kw_lookup(keywordlist *, wchar_t *);
 396 keywordlist *get_keywords(paragraph *);
 397 void free_keywords(keywordlist *);
 398 void subst_keywords(paragraph *, keywordlist *);
 399
 400 /*
 401  * index.c
 402  */
 403
 404 /*
 405  * Data structure to hold both sides of the index.
 406  */
 407 struct indexdata_Tag {
 408     tree234 *tags;                     /* holds type `indextag' */
 409     tree234 *entries;                  /* holds type `indexentry' */
 410 };
 411
 412 /*
 413  * Data structure to hold an index tag (LHS of index).
 414  */
 415 struct indextag_Tag {
 416     wchar_t *name;
 417     word *implicit_text;
 418     filepos implicit_fpos;
 419     word **explicit_texts;
 420     filepos *explicit_fpos;
 421     int nexplicit, explicit_size;
 422     int nrefs;
 423     indexentry **refs;                 /* array of entries referenced by tag */
 424 };
 425
 426 /*
 427  * Data structure to hold an index entry (RHS of index).
 428  */
 429 struct indexentry_Tag {
 430     word *text;
 431     void *backend_data;                /* private to back end */
 432     filepos fpos;
 433 };
 434
 435 indexdata *make_index(void);
 436 void cleanup_index(indexdata *);
 437 /* index_merge takes responsibility for freeing arg 3 iff implicit; never
 438  * takes responsibility for arg 2 */
 439 void index_merge(indexdata *, int is_explicit, wchar_t *, word *, filepos *);
 440 void build_index(indexdata *);
 441 void index_debug(indexdata *);
 442 indextag *index_findtag(indexdata *idx, wchar_t *name);
 443
 444 /*
 445  * contents.c
 446  */
 447 numberstate *number_init(void);
 448 void number_cfg(numberstate *, paragraph *);
 449 word *number_mktext(numberstate *, paragraph *, wchar_t *, int *, int *);
 450 void number_free(numberstate *);
 451
 452 /*
 453  * biblio.c
 454  */
 455 void gen_citations(paragraph *, keywordlist *);
 456
 457 /*
 458  * style.c
 459  */
 460 struct userstyle_Tag {
 461 };
 462
 463 /*
 464  * bk_text.c
 465  */
 466 void text_backend(paragraph *, keywordlist *, indexdata *, void *);
 467 paragraph *text_config_filename(char *filename);
 468
 469 /*
 470  * bk_html.c
 471  */
 472 void html_backend(paragraph *, keywordlist *, indexdata *, void *);
 473 paragraph *html_config_filename(char *filename);
 474
 475 /*
 476  * bk_whlp.c
 477  */
 478 void whlp_backend(paragraph *, keywordlist *, indexdata *, void *);
 479 paragraph *whlp_config_filename(char *filename);
 480
 481 /*
 482  * bk_man.c
 483  */
 484 void man_backend(paragraph *, keywordlist *, indexdata *, void *);
 485 paragraph *man_config_filename(char *filename);
 486
 487 /*
 488  * bk_info.c
 489  */
 490 void info_backend(paragraph *, keywordlist *, indexdata *, void *);
 491 paragraph *info_config_filename(char *filename);
 492
 493 /*
 494  * bk_paper.c
 495  */
 496 void *paper_pre_backend(paragraph *, keywordlist *, indexdata *);
 497
 498 /*
 499  * bk_ps.c
 500  */
 501 void ps_backend(paragraph *, keywordlist *, indexdata *, void *);
 502 paragraph *ps_config_filename(char *filename);
 503
 504 /*
 505  * bk_pdf.c
 506  */
 507 void pdf_backend(paragraph *, keywordlist *, indexdata *, void *);
 508 paragraph *pdf_config_filename(char *filename);
 509
 510 #endif