| /* vi:set ts=8 sts=4 sw=4: |
| * |
| * VIM - Vi IMproved by Bram Moolenaar |
| * |
| * Do ":help uganda" in Vim to read copying and usage conditions. |
| * Do ":help credits" in Vim to see a list of people who contributed. |
| * See README.txt for an overview of the Vim source code. |
| */ |
| |
| /* |
| * spell.h: common code for spell checking, used by spell.c and spellfile.c. |
| */ |
| |
| /* Use SPELL_PRINTTREE for debugging: dump the word tree after adding a word. |
| * Only use it for small word lists! */ |
| #if 0 |
| # define SPELL_PRINTTREE |
| #endif |
| |
| /* Use SPELL_COMPRESS_ALLWAYS for debugging: compress the word tree after |
| * adding a word. Only use it for small word lists! */ |
| #if 0 |
| # define SPELL_COMPRESS_ALLWAYS |
| #endif |
| |
| /* Use DEBUG_TRIEWALK to print the changes made in suggest_trie_walk() for a |
| * specific word. */ |
| #if 0 |
| # define DEBUG_TRIEWALK |
| #endif |
| |
| #define MAXWLEN 254 /* Assume max. word len is this many bytes. |
| Some places assume a word length fits in a |
| byte, thus it can't be above 255. |
| Must be >= PFD_NOTSPECIAL. */ |
| |
| /* Type used for indexes in the word tree need to be at least 4 bytes. If int |
| * is 8 bytes we could use something smaller, but what? */ |
| #if VIM_SIZEOF_INT > 3 |
| typedef int idx_T; |
| #else |
| typedef long idx_T; |
| #endif |
| |
| #ifdef FEAT_MBYTE |
| typedef int salfirst_T; |
| #else |
| typedef short salfirst_T; |
| #endif |
| |
| /* |
| * Structure used to store words and other info for one language, loaded from |
| * a .spl file. |
| * The main access is through the tree in "sl_fbyts/sl_fidxs", storing the |
| * case-folded words. "sl_kbyts/sl_kidxs" is for keep-case words. |
| * |
| * The "byts" array stores the possible bytes in each tree node, preceded by |
| * the number of possible bytes, sorted on byte value: |
| * <len> <byte1> <byte2> ... |
| * The "idxs" array stores the index of the child node corresponding to the |
| * byte in "byts". |
| * Exception: when the byte is zero, the word may end here and "idxs" holds |
| * the flags, region mask and affixID for the word. There may be several |
| * zeros in sequence for alternative flag/region/affixID combinations. |
| */ |
| typedef struct slang_S slang_T; |
| struct slang_S |
| { |
| slang_T *sl_next; /* next language */ |
| char_u *sl_name; /* language name "en", "en.rare", "nl", etc. */ |
| char_u *sl_fname; /* name of .spl file */ |
| int sl_add; /* TRUE if it's a .add file. */ |
| |
| char_u *sl_fbyts; /* case-folded word bytes */ |
| idx_T *sl_fidxs; /* case-folded word indexes */ |
| char_u *sl_kbyts; /* keep-case word bytes */ |
| idx_T *sl_kidxs; /* keep-case word indexes */ |
| char_u *sl_pbyts; /* prefix tree word bytes */ |
| idx_T *sl_pidxs; /* prefix tree word indexes */ |
| |
| char_u *sl_info; /* infotext string or NULL */ |
| |
| char_u sl_regions[17]; /* table with up to 8 region names plus NUL */ |
| |
| char_u *sl_midword; /* MIDWORD string or NULL */ |
| |
| hashtab_T sl_wordcount; /* hashtable with word count, wordcount_T */ |
| |
| int sl_compmax; /* COMPOUNDWORDMAX (default: MAXWLEN) */ |
| int sl_compminlen; /* COMPOUNDMIN (default: 0) */ |
| int sl_compsylmax; /* COMPOUNDSYLMAX (default: MAXWLEN) */ |
| int sl_compoptions; /* COMP_* flags */ |
| garray_T sl_comppat; /* CHECKCOMPOUNDPATTERN items */ |
| regprog_T *sl_compprog; /* COMPOUNDRULE turned into a regexp progrm |
| * (NULL when no compounding) */ |
| char_u *sl_comprules; /* all COMPOUNDRULE concatenated (or NULL) */ |
| char_u *sl_compstartflags; /* flags for first compound word */ |
| char_u *sl_compallflags; /* all flags for compound words */ |
| char_u sl_nobreak; /* When TRUE: no spaces between words */ |
| char_u *sl_syllable; /* SYLLABLE repeatable chars or NULL */ |
| garray_T sl_syl_items; /* syllable items */ |
| |
| int sl_prefixcnt; /* number of items in "sl_prefprog" */ |
| regprog_T **sl_prefprog; /* table with regprogs for prefixes */ |
| |
| garray_T sl_rep; /* list of fromto_T entries from REP lines */ |
| short sl_rep_first[256]; /* indexes where byte first appears, -1 if |
| there is none */ |
| garray_T sl_sal; /* list of salitem_T entries from SAL lines */ |
| salfirst_T sl_sal_first[256]; /* indexes where byte first appears, -1 if |
| there is none */ |
| int sl_followup; /* SAL followup */ |
| int sl_collapse; /* SAL collapse_result */ |
| int sl_rem_accents; /* SAL remove_accents */ |
| int sl_sofo; /* SOFOFROM and SOFOTO instead of SAL items: |
| * "sl_sal_first" maps chars, when has_mbyte |
| * "sl_sal" is a list of wide char lists. */ |
| garray_T sl_repsal; /* list of fromto_T entries from REPSAL lines */ |
| short sl_repsal_first[256]; /* sl_rep_first for REPSAL lines */ |
| int sl_nosplitsugs; /* don't suggest splitting a word */ |
| int sl_nocompoundsugs; /* don't suggest compounding */ |
| |
| /* Info from the .sug file. Loaded on demand. */ |
| time_t sl_sugtime; /* timestamp for .sug file */ |
| char_u *sl_sbyts; /* soundfolded word bytes */ |
| idx_T *sl_sidxs; /* soundfolded word indexes */ |
| buf_T *sl_sugbuf; /* buffer with word number table */ |
| int sl_sugloaded; /* TRUE when .sug file was loaded or failed to |
| load */ |
| |
| int sl_has_map; /* TRUE if there is a MAP line */ |
| #ifdef FEAT_MBYTE |
| hashtab_T sl_map_hash; /* MAP for multi-byte chars */ |
| int sl_map_array[256]; /* MAP for first 256 chars */ |
| #else |
| char_u sl_map_array[256]; /* MAP for first 256 chars */ |
| #endif |
| hashtab_T sl_sounddone; /* table with soundfolded words that have |
| handled, see add_sound_suggest() */ |
| }; |
| |
| #ifdef VMS |
| # define SPL_FNAME_TMPL "%s_%s.spl" |
| # define SPL_FNAME_ADD "_add." |
| # define SPL_FNAME_ASCII "_ascii." |
| #else |
| # define SPL_FNAME_TMPL "%s.%s.spl" |
| # define SPL_FNAME_ADD ".add." |
| # define SPL_FNAME_ASCII ".ascii." |
| #endif |
| |
| /* Flags used for a word. Only the lowest byte can be used, the region byte |
| * comes above it. */ |
| #define WF_REGION 0x01 /* region byte follows */ |
| #define WF_ONECAP 0x02 /* word with one capital (or all capitals) */ |
| #define WF_ALLCAP 0x04 /* word must be all capitals */ |
| #define WF_RARE 0x08 /* rare word */ |
| #define WF_BANNED 0x10 /* bad word */ |
| #define WF_AFX 0x20 /* affix ID follows */ |
| #define WF_FIXCAP 0x40 /* keep-case word, allcap not allowed */ |
| #define WF_KEEPCAP 0x80 /* keep-case word */ |
| |
| /* for <flags2>, shifted up one byte to be used in wn_flags */ |
| #define WF_HAS_AFF 0x0100 /* word includes affix */ |
| #define WF_NEEDCOMP 0x0200 /* word only valid in compound */ |
| #define WF_NOSUGGEST 0x0400 /* word not to be suggested */ |
| #define WF_COMPROOT 0x0800 /* already compounded word, COMPOUNDROOT */ |
| #define WF_NOCOMPBEF 0x1000 /* no compounding before this word */ |
| #define WF_NOCOMPAFT 0x2000 /* no compounding after this word */ |
| |
| /* flags for <pflags> */ |
| #define WFP_RARE 0x01 /* rare prefix */ |
| #define WFP_NC 0x02 /* prefix is not combining */ |
| #define WFP_UP 0x04 /* to-upper prefix */ |
| #define WFP_COMPPERMIT 0x08 /* prefix with COMPOUNDPERMITFLAG */ |
| #define WFP_COMPFORBID 0x10 /* prefix with COMPOUNDFORBIDFLAG */ |
| |
| /* Flags for postponed prefixes in "sl_pidxs". Must be above affixID (one |
| * byte) and prefcondnr (two bytes). */ |
| #define WF_RAREPFX (WFP_RARE << 24) /* rare postponed prefix */ |
| #define WF_PFX_NC (WFP_NC << 24) /* non-combining postponed prefix */ |
| #define WF_PFX_UP (WFP_UP << 24) /* to-upper postponed prefix */ |
| #define WF_PFX_COMPPERMIT (WFP_COMPPERMIT << 24) /* postponed prefix with |
| * COMPOUNDPERMITFLAG */ |
| #define WF_PFX_COMPFORBID (WFP_COMPFORBID << 24) /* postponed prefix with |
| * COMPOUNDFORBIDFLAG */ |
| |
| /* flags for <compoptions> */ |
| #define COMP_CHECKDUP 1 /* CHECKCOMPOUNDDUP */ |
| #define COMP_CHECKREP 2 /* CHECKCOMPOUNDREP */ |
| #define COMP_CHECKCASE 4 /* CHECKCOMPOUNDCASE */ |
| #define COMP_CHECKTRIPLE 8 /* CHECKCOMPOUNDTRIPLE */ |
| |
| /* Info from "REP", "REPSAL" and "SAL" entries in ".aff" file used in si_rep, |
| * si_repsal, sl_rep, and si_sal. Not for sl_sal! |
| * One replacement: from "ft_from" to "ft_to". */ |
| typedef struct fromto_S |
| { |
| char_u *ft_from; |
| char_u *ft_to; |
| } fromto_T; |
| |
| /* Info from "SAL" entries in ".aff" file used in sl_sal. |
| * The info is split for quick processing by spell_soundfold(). |
| * Note that "sm_oneof" and "sm_rules" point into sm_lead. */ |
| typedef struct salitem_S |
| { |
| char_u *sm_lead; /* leading letters */ |
| int sm_leadlen; /* length of "sm_lead" */ |
| char_u *sm_oneof; /* letters from () or NULL */ |
| char_u *sm_rules; /* rules like ^, $, priority */ |
| char_u *sm_to; /* replacement. */ |
| #ifdef FEAT_MBYTE |
| int *sm_lead_w; /* wide character copy of "sm_lead" */ |
| int *sm_oneof_w; /* wide character copy of "sm_oneof" */ |
| int *sm_to_w; /* wide character copy of "sm_to" */ |
| #endif |
| } salitem_T; |
| |
| /* Values for SP_*ERROR are negative, positive values are used by |
| * read_cnt_string(). */ |
| #define SP_TRUNCERROR -1 /* spell file truncated error */ |
| #define SP_FORMERROR -2 /* format error in spell file */ |
| #define SP_OTHERERROR -3 /* other error while reading spell file */ |
| |
| /* |
| * Structure used in "b_langp", filled from 'spelllang'. |
| */ |
| typedef struct langp_S |
| { |
| slang_T *lp_slang; /* info for this language */ |
| slang_T *lp_sallang; /* language used for sound folding or NULL */ |
| slang_T *lp_replang; /* language used for REP items or NULL */ |
| int lp_region; /* bitmask for region or REGION_ALL */ |
| } langp_T; |
| |
| #define LANGP_ENTRY(ga, i) (((langp_T *)(ga).ga_data) + (i)) |
| |
| #define VIMSUGMAGIC "VIMsug" /* string at start of Vim .sug file */ |
| #define VIMSUGMAGICL 6 |
| #define VIMSUGVERSION 1 |
| |
| /* |
| * The tables used for recognizing word characters according to spelling. |
| * These are only used for the first 256 characters of 'encoding'. |
| */ |
| typedef struct spelltab_S |
| { |
| char_u st_isw[256]; /* flags: is word char */ |
| char_u st_isu[256]; /* flags: is uppercase char */ |
| char_u st_fold[256]; /* chars: folded case */ |
| char_u st_upper[256]; /* chars: upper case */ |
| } spelltab_T; |
| |
| /* |
| * Use our own character-case definitions, because the current locale may |
| * differ from what the .spl file uses. |
| * These must not be called with negative number! |
| */ |
| #ifndef FEAT_MBYTE |
| /* Non-multi-byte implementation. */ |
| # define SPELL_TOFOLD(c) ((c) < 256 ? (int)spelltab.st_fold[c] : (c)) |
| # define SPELL_TOUPPER(c) ((c) < 256 ? (int)spelltab.st_upper[c] : (c)) |
| # define SPELL_ISUPPER(c) ((c) < 256 ? spelltab.st_isu[c] : FALSE) |
| #else |
| # if defined(HAVE_WCHAR_H) |
| # include <wchar.h> /* for towupper() and towlower() */ |
| # endif |
| /* Multi-byte implementation. For Unicode we can call utf_*(), but don't do |
| * that for ASCII, because we don't want to use 'casemap' here. Otherwise use |
| * the "w" library function for characters above 255 if available. */ |
| # ifdef HAVE_TOWLOWER |
| # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \ |
| : (c) < 256 ? (int)spelltab.st_fold[c] : (int)towlower(c)) |
| # else |
| # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \ |
| : (c) < 256 ? (int)spelltab.st_fold[c] : (c)) |
| # endif |
| |
| # ifdef HAVE_TOWUPPER |
| # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \ |
| : (c) < 256 ? (int)spelltab.st_upper[c] : (int)towupper(c)) |
| # else |
| # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \ |
| : (c) < 256 ? (int)spelltab.st_upper[c] : (c)) |
| # endif |
| |
| # ifdef HAVE_ISWUPPER |
| # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \ |
| : (c) < 256 ? spelltab.st_isu[c] : iswupper(c)) |
| # else |
| # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \ |
| : (c) < 256 ? spelltab.st_isu[c] : (FALSE)) |
| # endif |
| #endif |
| |
| #ifdef FEAT_SPELL |
| /* First language that is loaded, start of the linked list of loaded |
| * languages. */ |
| # ifdef IN_SPELL_C |
| # define SPELL_EXTERN |
| # define SPELL_INIT(x) x |
| # else |
| # define SPELL_EXTERN extern |
| # define SPELL_INIT(x) |
| # endif |
| |
| SPELL_EXTERN slang_T *first_lang SPELL_INIT(= NULL); |
| |
| /* file used for "zG" and "zW" */ |
| SPELL_EXTERN char_u *int_wordlist SPELL_INIT(= NULL); |
| |
| |
| SPELL_EXTERN char e_format[] SPELL_INIT(= N_("E759: Format error in spell file")); |
| |
| SPELL_EXTERN spelltab_T spelltab; |
| SPELL_EXTERN int did_set_spelltab; |
| |
| #endif |