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 numberstate_Tag numberstate;
  39 typedef struct indexdata_Tag indexdata;
  40 typedef struct indextag_Tag indextag;
  41 typedef struct indexentry_Tag indexentry;
  42 typedef struct macrostack_Tag macrostack;
  43
  44 /*
  45  * Data structure to hold a file name and index, a line and a
  46  * column number, for reporting errors
  47  */
  48 struct filepos_Tag {
  49     char *filename;
  50     int line, col;
  51 };
  52
  53 /*
  54  * Data structure to hold all the file names etc for input
  55  */
  56 typedef struct pushback_Tag {
  57     int chr;
  58     filepos pos;
  59 } pushback;
  60 struct input_Tag {
  61     char **filenames;                  /* complete list of input files */
  62     int nfiles;                        /* how many in the list */
  63     FILE *currfp;                      /* the currently open one */
  64     int currindex;                     /* which one is that in the list */
  65     int wantclose;                     /* does the current file want closing */
  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 isvis(x) ( ( (x) >= word_Normal && (x) <= word_LowerXref ) )
 190 #define isattr(x) ( ( (x) > word_Normal && (x) < word_WhiteSpace ) || \
 191                     ( (x) > word_WhiteSpace && (x) < word_internal_endattrs ) )
 192 #define sameattr(x,y) ( (((x)-(y)) & 3) == 0 )
 193 #define towordstyle(x) ( word_Normal + ((x) & 3) )
 194 #define tospacestyle(x) ( word_WhiteSpace + ((x) & 3) )
 195 #define toquotestyle(x) ( word_Quote + ((x) & 3) )
 196 #define removeattr(x) ( word_Normal + ((x) &~ 3) )
 197
 198 #define attraux(x) ( (x) & attr_mask )
 199 #define quoteaux(x) ( (x) & quote_mask )
 200
 201 /*
 202  * error.c
 203  */
 204 /* out of memory */
 205 void fatalerr_nomemory(void) NORETURN;
 206 /* option `-%s' requires an argument */
 207 void err_optnoarg(const char *sp);
 208 /* unrecognised option `-%s' */
 209 void err_nosuchopt(const char *sp);
 210 /* unrecognised charset %s (cmdline) */
 211 void err_cmdcharset(const char *sp);
 212 /* futile option `-%s'%s */
 213 void err_futileopt(const char *sp, const char *sp2);
 214 /* no input files */
 215 void err_noinput(void);
 216 /* unable to open input file `%s' */
 217 void err_cantopen(const char *sp);
 218 /* no data in input files */
 219 void err_nodata(void);
 220 /* line in codepara didn't begin `\c' */
 221 void err_brokencodepara(const filepos *fpos);
 222 /* expected `}' after keyword */
 223 void err_kwunclosed(const filepos *fpos);
 224 /* paragraph type expects no keyword */
 225 void err_kwexpected(const filepos *fpos);
 226 /* paragraph type expects a keyword */
 227 void err_kwillegal(const filepos *fpos);
 228 /* paragraph type expects only 1 */
 229 void err_kwtoomany(const filepos *fpos);
 230 /* paragraph type expects only kws! */
 231 void err_bodyillegal(const filepos *fpos);
 232 /* invalid command at start of para */
 233 void err_badparatype(const wchar_t *wsp, const filepos *fpos);
 234 /* invalid command in mid-para */
 235 void err_badmidcmd(const wchar_t *wsp, const filepos *fpos);
 236 /* unexpected brace */
 237 void err_unexbrace(const filepos *fpos);
 238 /* expected `{' after command */
 239 void err_explbr(const filepos *fpos);
 240 /* EOF inside braced comment */
 241 void err_commenteof(const filepos *fpos);
 242 /* expected `}' after cross-ref */
 243 void err_kwexprbr(const filepos *fpos);
 244 /* \q within \c is not supported */
 245 void err_codequote(const filepos *fpos);
 246 /* unclosed braces at end of para */
 247 void err_missingrbrace(const filepos *fpos);
 248 /* unclosed braces at end of file */
 249 void err_missingrbrace2(const filepos *fpos);
 250 /* unable to nest text styles */
 251 void err_nestedstyles(const filepos *fpos);
 252 /* unable to nest `\i' thingys */
 253 void err_nestedindex(const filepos *fpos);
 254 /* two \i differing only in case */
 255 void err_indexcase(const filepos *fpos, const wchar_t *wsp,
 256                    const filepos *fpos2, const wchar_t *wsp2);
 257 /* unresolved cross-reference */
 258 void err_nosuchkw(const filepos *fpos, const wchar_t *wsp);
 259 /* multiple \BRs on same keyword */
 260 void err_multiBR(const filepos *fpos, const wchar_t *wsp);
 261 /* \IM on unknown index tag (warning) */
 262 void err_nosuchidxtag(const filepos *fpos, const wchar_t *wsp);
 263 /* can't open output file for write */
 264 void err_cantopenw(const char *sp);
 265 /* this macro already exists */
 266 void err_macroexists(const filepos *fpos, const wchar_t *wsp);
 267 /* jump a heading level, eg \C -> \S */
 268 void err_sectjump(const filepos *fpos);
 269 /* WinHelp context ID hash clash */
 270 void err_winhelp_ctxclash(const filepos *fpos, const char *sp, const char *sp2);
 271 /* keyword clash in sections */
 272 void err_multikw(const filepos *fpos, const filepos *fpos2, const wchar_t *wsp);
 273 /* \lcont not after a list item */
 274 void err_misplacedlcont(const filepos *fpos);
 275 /* section marker appeared in block */
 276 void err_sectmarkerinblock(const filepos *fpos, const char *sp);
 277 /* \cfg{%s} insufficient args (<%d) */
 278 void err_cfginsufarg(const filepos *fpos, const char *sp, int i);
 279 /* colon/comma in node name in info */
 280 void err_infonodechar(const filepos *fpos, char c) /* fpos might be NULL */;
 281 /* \c line too long in text backend */
 282 void err_text_codeline(const filepos *fpos, int i, int j);
 283 /* unrecognised HTML version keyword */
 284 void err_htmlver(const filepos *fpos, const wchar_t *wsp);
 285 /* unrecognised character set name */
 286 void err_charset(const filepos *fpos, const wchar_t *wsp);
 287 /* unrecognised font name */
 288 void err_nofont(const filepos *fpos, const wchar_t *wsp);
 289 /* eof in AFM file */
 290 void err_afmeof(const filepos *fpos);
 291 /* missing expected keyword in AFM */
 292 void err_afmkey(const filepos *fpos, const char *sp);
 293 /* unsupported AFM version */
 294 void err_afmvers(const filepos *fpos);
 295 /* missing value(s) for AFM key */
 296 void err_afmval(const filepos *fpos, const char *sp, int i);
 297 /* eof in Type 1 font file */
 298 void err_pfeof(const filepos *fpos);
 299 /* bad Type 1 header line */
 300 void err_pfhead(const filepos *fpos);
 301 /* otherwise invalide Type 1 font */
 302 void err_pfbad(const filepos *fpos);
 303 /* Type 1 font but no AFM */
 304 void err_pfnoafm(const filepos *fpos, const char *sp);
 305 /* need both or neither of hhp+chm */
 306 void err_chmnames(void);
 307 /* required sfnt table missing */
 308 void err_sfntnotable(const filepos *fpos, const char *sp);
 309 /* sfnt has no PostScript name */
 310 void err_sfntnopsname(const filepos *fpos);
 311 /* sfnt table not valid */
 312 void err_sfntbadtable(const filepos *fpos, const char *sp);
 313 /* sfnt has no UCS-2 cmap */
 314 void err_sfntnounicmap(const filepos *fpos);
 315 /* sfnt table version unknown */
 316 void err_sfnttablevers(const filepos *fpos, const char *sp);
 317 /* sfnt has bad header */
 318 void err_sfntbadhdr(const filepos *fpos);
 319 /* sfnt cmap references bad glyph */
 320 void err_sfntbadglyph(const filepos *fpos, unsigned wc);
 321
 322 /*
 323  * malloc.c
 324  */
 325 #ifdef LOGALLOC
 326 void *smalloc(char *file, int line, int size);
 327 void *srealloc(char *file, int line, void *p, int size);
 328 void sfree(char *file, int line, void *p);
 329 #define smalloc(x) smalloc(__FILE__, __LINE__, x)
 330 #define srealloc(x, y) srealloc(__FILE__, __LINE__, x, y)
 331 #define sfree(x) sfree(__FILE__, __LINE__, x)
 332 #else
 333 void *smalloc(int size);
 334 void *srealloc(void *p, int size);
 335 void sfree(void *p);
 336 #endif
 337 void free_word_list(word *w);
 338 void free_para_list(paragraph *p);
 339 word *dup_word_list(word *w);
 340 char *dupstr(char const *s);
 341
 342 #define snew(type) ( (type *) smalloc (sizeof (type)) )
 343 #define snewn(number, type) ( (type *) smalloc ((number) * sizeof (type)) )
 344 #define sresize(array, number, type) \
 345         ( (type *) srealloc ((array), (number) * sizeof (type)) )
 346 #define lenof(array) ( sizeof(array) / sizeof(*(array)) )
 347
 348 /*
 349  * ustring.c
 350  */
 351 wchar_t *ustrdup(wchar_t const *s);
 352 char *ustrtoa(wchar_t const *s, char *outbuf, int size, int charset);
 353 char *ustrtoa_careful(wchar_t const *s, char *outbuf, int size, int charset);
 354 wchar_t *ustrfroma(char const *s, wchar_t *outbuf, int size, int charset);
 355 char *utoa_dup(wchar_t const *s, int charset);
 356 char *utoa_dup_len(wchar_t const *s, int charset, int *len);
 357 char *utoa_careful_dup(wchar_t const *s, int charset);
 358 wchar_t *ufroma_dup(char const *s, int charset);
 359 char *utoa_locale_dup(wchar_t const *s);
 360 wchar_t *ufroma_locale_dup(char const *s);
 361 int ustrlen(wchar_t const *s);
 362 wchar_t *uadv(wchar_t *s);
 363 wchar_t *ustrcpy(wchar_t *dest, wchar_t const *source);
 364 wchar_t *ustrncpy(wchar_t *dest, wchar_t const *source, int n);
 365 wchar_t utolower(wchar_t);
 366 int uisalpha(wchar_t);
 367 int ustrcmp(wchar_t *lhs, wchar_t *rhs);
 368 int ustricmp(wchar_t const *lhs, wchar_t const *rhs);
 369 int ustrnicmp(wchar_t const *lhs, wchar_t const *rhs, int maxlen);
 370 int utoi(wchar_t const *);
 371 double utof(wchar_t const *);
 372 int utob(wchar_t const *);
 373 int uisdigit(wchar_t);
 374 wchar_t *ustrlow(wchar_t *s);
 375 wchar_t *ustrftime(const wchar_t *wfmt, const struct tm *timespec);
 376 int cvt_ok(int charset, const wchar_t *s);
 377 int charset_from_ustr(filepos *fpos, const wchar_t *name);
 378
 379 /*
 380  * wcwidth.c
 381  */
 382 int strwid(char const *s, int charset);
 383 int ustrwid(wchar_t const *s, int charset);
 384
 385 /*
 386  * help.c
 387  */
 388 void help(void);
 389 void usage(void);
 390 void showversion(void);
 391 void listcharsets(void);
 392
 393 /*
 394  * licence.c
 395  */
 396 void licence(void);
 397
 398 /*
 399  * version.c
 400  */
 401 extern const char *const version;
 402
 403 /*
 404  * misc.c
 405  */
 406 char *adv(char *s);
 407
 408 typedef struct stackTag *stack;
 409 stack stk_new(void);
 410 void stk_free(stack);
 411 void stk_push(stack, void *);
 412 void *stk_pop(stack);
 413 void *stk_top(stack);
 414
 415 typedef struct tagRdstring rdstring;
 416 struct tagRdstring {
 417     int pos, size;
 418     wchar_t *text;
 419 };
 420 typedef struct tagRdstringc rdstringc;
 421 struct tagRdstringc {
 422     int pos, size;
 423     char *text;
 424 };
 425 extern const rdstring empty_rdstring;
 426 extern const rdstringc empty_rdstringc;
 427 void rdadd(rdstring *rs, wchar_t c);
 428 void rdadds(rdstring *rs, wchar_t const *p);
 429 wchar_t *rdtrim(rdstring *rs);
 430 void rdaddc(rdstringc *rs, char c);
 431 void rdaddsc(rdstringc *rs, char const *p);
 432 void rdaddsn(rdstringc *rc, char const *p, int len);
 433 char *rdtrimc(rdstringc *rs);
 434
 435 int compare_wordlists(word *a, word *b);
 436
 437 void mark_attr_ends(word *words);
 438
 439 typedef struct tagWrappedLine wrappedline;
 440 struct tagWrappedLine {
 441     wrappedline *next;
 442     word *begin, *end;                 /* first & last words of line */
 443     int nspaces;                       /* number of whitespaces in line */
 444     int shortfall;                     /* how much shorter than max width */
 445 };
 446 wrappedline *wrap_para(word *, int, int, int (*)(void *, word *), void *, int);
 447 void wrap_free(wrappedline *);
 448 void cmdline_cfg_add(paragraph *cfg, char *string);
 449 paragraph *cmdline_cfg_new(void);
 450 paragraph *cmdline_cfg_simple(char *string, ...);
 451
 452 /*
 453  * input.c
 454  */
 455 paragraph *read_input(input *in, indexdata *idx);
 456
 457 /*
 458  * in_afm.c
 459  */
 460 void read_afm_file(input *in);
 461
 462 /*
 463  * in_pf.c
 464  */
 465 void read_pfa_file(input *in);
 466 void read_pfb_file(input *in);
 467
 468 /*
 469  * in_sfnt.c
 470  */
 471 void read_sfnt_file(input *in);
 472
 473 /*
 474  * keywords.c
 475  */
 476 struct keywordlist_Tag {
 477     int nkeywords;
 478     int size;
 479     tree234 *keys;                     /* sorted by `key' field */
 480     word **looseends;                  /* non-keyword list element numbers */
 481     int nlooseends;
 482     int looseendssize;
 483 };
 484 struct keyword_Tag {
 485     wchar_t *key;                      /* the keyword itself */
 486     word *text;                        /* "Chapter 2", "Appendix Q"... */
 487                                        /* (NB: filepos are not set) */
 488     paragraph *para;                   /* the paragraph referenced */
 489 };
 490 keyword *kw_lookup(keywordlist *, wchar_t *);
 491 keywordlist *get_keywords(paragraph *);
 492 void free_keywords(keywordlist *);
 493 void subst_keywords(paragraph *, keywordlist *);
 494
 495 /*
 496  * index.c
 497  */
 498
 499 /*
 500  * Data structure to hold both sides of the index.
 501  */
 502 struct indexdata_Tag {
 503     tree234 *tags;                     /* holds type `indextag' */
 504     tree234 *entries;                  /* holds type `indexentry' */
 505 };
 506
 507 /*
 508  * Data structure to hold an index tag (LHS of index).
 509  */
 510 struct indextag_Tag {
 511     wchar_t *name;
 512     word *implicit_text;
 513     filepos implicit_fpos;
 514     word **explicit_texts;
 515     filepos *explicit_fpos;
 516     int nexplicit, explicit_size;
 517     int nrefs;
 518     indexentry **refs;                 /* array of entries referenced by tag */
 519 };
 520
 521 /*
 522  * Data structure to hold an index entry (RHS of index).
 523  */
 524 struct indexentry_Tag {
 525     word *text;
 526     void *backend_data;                /* private to back end */
 527     filepos fpos;
 528 };
 529
 530 indexdata *make_index(void);
 531 void cleanup_index(indexdata *);
 532 /* index_merge takes responsibility for freeing arg 3 iff implicit; never
 533  * takes responsibility for arg 2 */
 534 void index_merge(indexdata *, int is_explicit, wchar_t *, word *, filepos *);
 535 void build_index(indexdata *);
 536 void index_debug(indexdata *);
 537 indextag *index_findtag(indexdata *idx, wchar_t *name);
 538
 539 /*
 540  * contents.c
 541  */
 542 numberstate *number_init(void);
 543 void number_cfg(numberstate *, paragraph *);
 544 word *number_mktext(numberstate *, paragraph *, wchar_t *, int *, int *);
 545 void number_free(numberstate *);
 546
 547 /*
 548  * biblio.c
 549  */
 550 void gen_citations(paragraph *, keywordlist *);
 551
 552 /*
 553  * bk_text.c
 554  */
 555 void text_backend(paragraph *, keywordlist *, indexdata *, void *);
 556 paragraph *text_config_filename(char *filename);
 557
 558 /*
 559  * bk_html.c
 560  */
 561 void html_backend(paragraph *, keywordlist *, indexdata *, void *);
 562 paragraph *html_config_filename(char *filename);
 563
 564 /*
 565  * bk_whlp.c
 566  */
 567 void whlp_backend(paragraph *, keywordlist *, indexdata *, void *);
 568 paragraph *whlp_config_filename(char *filename);
 569
 570 /*
 571  * bk_man.c
 572  */
 573 void man_backend(paragraph *, keywordlist *, indexdata *, void *);
 574 paragraph *man_config_filename(char *filename);
 575
 576 /*
 577  * bk_info.c
 578  */
 579 void info_backend(paragraph *, keywordlist *, indexdata *, void *);
 580 paragraph *info_config_filename(char *filename);
 581
 582 /*
 583  * bk_paper.c
 584  */
 585 void *paper_pre_backend(paragraph *, keywordlist *, indexdata *);
 586 void listfonts(void);
 587
 588 /*
 589  * bk_ps.c
 590  */
 591 void ps_backend(paragraph *, keywordlist *, indexdata *, void *);
 592 paragraph *ps_config_filename(char *filename);
 593
 594 /*
 595  * bk_pdf.c
 596  */
 597 void pdf_backend(paragraph *, keywordlist *, indexdata *, void *);
 598 paragraph *pdf_config_filename(char *filename);
 599
 600 #endif