Support the locale-supplied character set where appropriate. It's
[sgt/halibut] / halibut.h
CommitLineData
d7482997 1#ifndef HALIBUT_HALIBUT_H
2#define HALIBUT_HALIBUT_H
3
4#include <stdio.h>
5#include <wchar.h>
6#include <time.h>
9c1cf191 7#include <string.h>
d7482997 8
e34ba5c3 9#include "charset.h"
10
d7482997 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 */
32typedef struct input_Tag input;
33typedef struct filepos_Tag filepos;
34typedef struct paragraph_Tag paragraph;
35typedef struct word_Tag word;
36typedef struct keywordlist_Tag keywordlist;
37typedef struct keyword_Tag keyword;
38typedef struct userstyle_Tag userstyle;
39typedef struct numberstate_Tag numberstate;
40typedef struct indexdata_Tag indexdata;
41typedef struct indextag_Tag indextag;
42typedef struct indexentry_Tag indexentry;
43typedef 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 */
49struct 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 */
57typedef struct pushback_Tag {
58 int chr;
59 filepos pos;
60} pushback;
61struct 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 */
e34ba5c3 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[] */
e4ea58f8 75 char *pushback_chars; /* used to save input-encoding data */
d7482997 76};
77
78/*
79 * Data structure to hold the input form of the source, ie a linked
80 * list of paragraphs
81 */
82struct paragraph_Tag {
83 paragraph *next;
84 int type;
85 wchar_t *keyword; /* for most special paragraphs */
e4ea58f8 86 char *origkeyword; /* same again in original charset */
d7482997 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};
99enum {
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,
7136a6c7 113 para_DescribedThing,
114 para_Description,
d7482997 115 para_Code,
116 para_Copyright,
d7482997 117 para_NoCite,
118 para_Title,
119 para_VersionID,
120 para_Config, /* configuration directive */
7136a6c7 121 para_LcontPush, /* begin continuation of list item */
122 para_LcontPop, /* end continuation of list item */
2614b01d 123 para_QuotePush, /* begin block quote */
124 para_QuotePop, /* end block quote */
3f3d1acc 125 /*
126 * Back ends may define their own paragraph types beyond here,
127 * in case they need to use them internally.
128 */
d7482997 129 para_NotParaType /* placeholder value */
130};
131
132/*
133 * Data structure to hold an individual word
134 */
135struct 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;
5dd44dce 142
143 void *private_data; /* for temp use in backends */
d7482997 144};
145enum {
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) */
3f3d1acc 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 */
d7482997 174};
175/* aux values for attributed words */
176enum {
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 */
184enum {
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 */
203void fatal(int code, ...) NORETURN;
204void error(int code, ...);
205enum {
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_missingrbrace, /* unclosed braces at end of para */
7136a6c7 225 err_missingrbrace2, /* unclosed braces at end of file */
d7482997 226 err_nestedstyles, /* unable to nest text styles */
227 err_nestedindex, /* unable to nest `\i' thingys */
228 err_nosuchkw, /* unresolved cross-reference */
229 err_multiBR, /* multiple \BRs on same keyword */
230 err_nosuchidxtag, /* \IM on unknown index tag (warning) */
231 err_cantopenw, /* can't open output file for write */
232 err_macroexists, /* this macro already exists */
233 err_sectjump, /* jump a heading level, eg \C -> \S */
234 err_winhelp_ctxclash, /* WinHelp context ID hash clash */
235 err_multikw, /* keyword clash in sections */
7136a6c7 236 err_misplacedlcont, /* \lcont not after a list item */
2614b01d 237 err_sectmarkerinblock, /* section marker appeared in block */
d4c7e130 238 err_infodirentry, /* \cfg{info-dir-entry} missing param */
f4551933 239 err_infonodechar, /* colon/comma in node name in info */
d7482997 240 err_whatever /* random error of another type */
241};
242
243/*
244 * malloc.c
245 */
246#ifdef LOGALLOC
247void *smalloc(char *file, int line, int size);
248void *srealloc(char *file, int line, void *p, int size);
249void sfree(char *file, int line, void *p);
250#define smalloc(x) smalloc(__FILE__, __LINE__, x)
251#define srealloc(x, y) srealloc(__FILE__, __LINE__, x, y)
252#define sfree(x) sfree(__FILE__, __LINE__, x)
253#else
254void *smalloc(int size);
255void *srealloc(void *p, int size);
256void sfree(void *p);
257#endif
258void free_word_list(word *w);
259void free_para_list(paragraph *p);
260word *dup_word_list(word *w);
261char *dupstr(char *s);
262
263#define mknew(type) ( (type *) smalloc (sizeof (type)) )
264#define mknewa(type, number) ( (type *) smalloc ((number) * sizeof (type)) )
265#define resize(array, len) ( srealloc ((array), (len) * sizeof (*(array))) )
266#define lenof(array) ( sizeof(array) / sizeof(*(array)) )
267
268/*
269 * ustring.c
270 */
e4ea58f8 271wchar_t *ustrdup(wchar_t const *s);
272char *ustrtoa(wchar_t const *s, char *outbuf, int size, int charset);
273char *ustrtoa_careful(wchar_t const *s, char *outbuf, int size, int charset);
274wchar_t *ustrfroma(char const *s, wchar_t *outbuf, int size, int charset);
275char *utoa_dup(wchar_t const *s, int charset);
276char *utoa_dup_len(wchar_t const *s, int charset, int *len);
277char *utoa_careful_dup(wchar_t const *s, int charset);
278wchar_t *ufroma_dup(char const *s, int charset);
7e976207 279char *utoa_locale_dup(wchar_t const *s);
280wchar_t *ufroma_locale_dup(char const *s);
5dd44dce 281int ustrlen(wchar_t const *s);
d7482997 282wchar_t *uadv(wchar_t *s);
5dd44dce 283wchar_t *ustrcpy(wchar_t *dest, wchar_t const *source);
d7482997 284wchar_t utolower(wchar_t);
831da32e 285int uisalpha(wchar_t);
d7482997 286int ustrcmp(wchar_t *lhs, wchar_t *rhs);
287int ustricmp(wchar_t *lhs, wchar_t *rhs);
288int utoi(wchar_t *);
289int utob(wchar_t *);
290int uisdigit(wchar_t);
291wchar_t *ustrlow(wchar_t *s);
292wchar_t *ustrftime(wchar_t *fmt, struct tm *timespec);
91f93b94 293int cvt_ok(int charset, const wchar_t *s);
d7482997 294
295/*
296 * help.c
297 */
298void help(void);
299void usage(void);
300void showversion(void);
301
302/*
303 * licence.c
304 */
305void licence(void);
306
307/*
308 * version.c
309 */
310const char *const version;
311
312/*
313 * misc.c
314 */
e4ea58f8 315char *adv(char *s);
316
d7482997 317typedef struct stackTag *stack;
318stack stk_new(void);
319void stk_free(stack);
320void stk_push(stack, void *);
321void *stk_pop(stack);
7136a6c7 322void *stk_top(stack);
d7482997 323
324typedef struct tagRdstring rdstring;
325struct tagRdstring {
326 int pos, size;
327 wchar_t *text;
328};
329typedef struct tagRdstringc rdstringc;
330struct tagRdstringc {
331 int pos, size;
332 char *text;
333};
334extern const rdstring empty_rdstring;
335extern const rdstringc empty_rdstringc;
336void rdadd(rdstring *rs, wchar_t c);
5dd44dce 337void rdadds(rdstring *rs, wchar_t const *p);
d7482997 338wchar_t *rdtrim(rdstring *rs);
339void rdaddc(rdstringc *rs, char c);
5dd44dce 340void rdaddsc(rdstringc *rs, char const *p);
d7482997 341char *rdtrimc(rdstringc *rs);
342
343int compare_wordlists(word *a, word *b);
344
345void mark_attr_ends(paragraph *sourceform);
346
347typedef struct tagWrappedLine wrappedline;
348struct tagWrappedLine {
349 wrappedline *next;
350 word *begin, *end; /* first & last words of line */
351 int nspaces; /* number of whitespaces in line */
352 int shortfall; /* how much shorter than max width */
353};
43341922 354wrappedline *wrap_para(word *, int, int, int (*)(void *, word *), void *, int);
d7482997 355void wrap_free(wrappedline *);
e4ea58f8 356void cmdline_cfg_add(paragraph *cfg, char *string);
357paragraph *cmdline_cfg_new(void);
358paragraph *cmdline_cfg_simple(char *string, ...);
d7482997 359
360/*
361 * input.c
362 */
363paragraph *read_input(input *in, indexdata *idx);
364
365/*
366 * keywords.c
367 */
368struct keywordlist_Tag {
369 int nkeywords;
370 int size;
371 tree234 *keys; /* sorted by `key' field */
372 word **looseends; /* non-keyword list element numbers */
373 int nlooseends;
374 int looseendssize;
375};
376struct keyword_Tag {
377 wchar_t *key; /* the keyword itself */
378 word *text; /* "Chapter 2", "Appendix Q"... */
379 /* (NB: filepos are not set) */
380 paragraph *para; /* the paragraph referenced */
381};
382keyword *kw_lookup(keywordlist *, wchar_t *);
383keywordlist *get_keywords(paragraph *);
384void free_keywords(keywordlist *);
385void subst_keywords(paragraph *, keywordlist *);
386
387/*
388 * index.c
389 */
390
391/*
392 * Data structure to hold both sides of the index.
393 */
394struct indexdata_Tag {
395 tree234 *tags; /* holds type `indextag' */
396 tree234 *entries; /* holds type `indexentry' */
397};
398
399/*
400 * Data structure to hold an index tag (LHS of index).
401 */
402struct indextag_Tag {
403 wchar_t *name;
404 word *implicit_text;
f4551933 405 filepos implicit_fpos;
d7482997 406 word **explicit_texts;
f4551933 407 filepos *explicit_fpos;
d7482997 408 int nexplicit, explicit_size;
409 int nrefs;
410 indexentry **refs; /* array of entries referenced by tag */
411};
412
413/*
414 * Data structure to hold an index entry (RHS of index).
415 */
416struct indexentry_Tag {
417 word *text;
418 void *backend_data; /* private to back end */
f4551933 419 filepos fpos;
d7482997 420};
421
422indexdata *make_index(void);
423void cleanup_index(indexdata *);
424/* index_merge takes responsibility for freeing arg 3 iff implicit; never
425 * takes responsibility for arg 2 */
f4551933 426void index_merge(indexdata *, int is_explicit, wchar_t *, word *, filepos *);
d7482997 427void build_index(indexdata *);
428void index_debug(indexdata *);
429indextag *index_findtag(indexdata *idx, wchar_t *name);
430
431/*
432 * contents.c
433 */
434numberstate *number_init(void);
435void number_cfg(numberstate *, paragraph *);
96f3af16 436word *number_mktext(numberstate *, paragraph *, wchar_t *, int *, int *);
d7482997 437void number_free(numberstate *);
438
439/*
440 * biblio.c
441 */
442void gen_citations(paragraph *, keywordlist *);
443
444/*
445 * style.c
446 */
447struct userstyle_Tag {
448};
449
450/*
451 * bk_text.c
452 */
43341922 453void text_backend(paragraph *, keywordlist *, indexdata *, void *);
ba9c1487 454paragraph *text_config_filename(char *filename);
d7482997 455
456/*
457 * bk_xhtml.c
458 */
43341922 459void xhtml_backend(paragraph *, keywordlist *, indexdata *, void *);
ba9c1487 460paragraph *xhtml_config_filename(char *filename);
d7482997 461
462/*
463 * bk_whlp.c
464 */
43341922 465void whlp_backend(paragraph *, keywordlist *, indexdata *, void *);
ba9c1487 466paragraph *whlp_config_filename(char *filename);
d7482997 467
7136a6c7 468/*
469 * bk_man.c
470 */
43341922 471void man_backend(paragraph *, keywordlist *, indexdata *, void *);
ba9c1487 472paragraph *man_config_filename(char *filename);
7136a6c7 473
5dd44dce 474/*
475 * bk_info.c
476 */
43341922 477void info_backend(paragraph *, keywordlist *, indexdata *, void *);
5dd44dce 478paragraph *info_config_filename(char *filename);
479
43341922 480/*
481 * bk_paper.c
482 */
483void *paper_pre_backend(paragraph *, keywordlist *, indexdata *);
484
485/*
486 * bk_ps.c
487 */
488void ps_backend(paragraph *, keywordlist *, indexdata *, void *);
489paragraph *ps_config_filename(char *filename);
490
491/*
492 * bk_pdf.c
493 */
494void pdf_backend(paragraph *, keywordlist *, indexdata *, void *);
495paragraph *pdf_config_filename(char *filename);
496
d7482997 497#endif