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