Attachment #510818 for bug #620626

Attachment #510818: Hunspell 1.3.1 patch v2 for bug #620626

View | Details | Raw Unified | Return to bug 620626
Collapse All | Expand All





* use your version of this file under the terms of the MPL, indicate your
* decision by deleting the provisions above and replace them with the notice
* and other provisions required by the GPL or the LGPL. If you do not delete
* the provisions above, a recipient may use your version of this file under
* the terms of any one of the MPL, the GPL or the LGPL.
*
******* END LICENSE BLOCK *******

Hunspell Version:   1.3.1
Additional Patches: 

Hunspell Author: László Németh
MySpell Author: Kevin Hendricks & David Einstein

Hunspell is a spell checker and morphological analyser library. Hunspell
is based on OpenOffice.org's Myspell. Documentation, tests, and examples
are available at http://hunspell.sourceforge.net.






                pos = st; break;
            }
        case '^': { p = nextchar(p); neg = true; break; }
        case ']': { 
                if ((neg && ingroup) || (!neg && !ingroup)) return 0;
                pos = NULL;
                p = nextchar(p);
                // skip the next character
                if (!ingroup && *st) for (st++; (opts & aeUTF8) && (*st & 0xc0) == 0x80; st++);
                if (*st == '\0' && p) return 0; // word <= condition
                break;
            }
         case '.': if (!pos) { // dots are not metacharacters in groups: [.]
                p = nextchar(p);
                // skip the next character
                for (st++; (opts & aeUTF8) && (*st & 0xc0) == 0x80; st++);
                if (*st == '\0' && p) return 0; // word <= condition





  trystring = NULL;
  encoding=NULL;
  csconv=NULL;
  utf8 = 0;
  complexprefixes = 0;
  maptable = NULL;
  nummap = 0;
  breaktable = NULL;
  numbreak = -1;
  reptable = NULL;
  numrep = 0;
  iconvtable = NULL;
  oconvtable = NULL;
  checkcpdtable = NULL;
  // allow simplified compound forms (see 3rd field of CHECKCOMPOUNDPATTERN)
  simplifiedcpd = 0;
  numcheckcpd = 0;

  compoundforbidflag = FLAG_NULL; // compound fordidden flag for suffixed word
  checkcompounddup = 0; // forbid double words in compounds
  checkcompoundrep = 0; // forbid bad compounds (may be non compound word with a REP substitution)
  checkcompoundcase = 0; // forbid upper and lowercase combinations at word bounds
  checkcompoundtriple = 0; // forbid compounds with triple letters
  simplifiedtriple = 0; // allow simplified triple letters in compounds (Schiff+fahrt -> Schiffahrt)
  forbiddenword = FORBIDDENWORD; // forbidden word signing flag
  nosuggest = FLAG_NULL; // don't suggest words signed with NOSUGGEST flag
  nongramsuggest = FLAG_NULL;
  lang = NULL; // language
  langnum = 0; // language code (see http://l10n.openoffice.org/languages.html)
  needaffix = FLAG_NULL; // forbidden root, allowed only with suffixes
  cpdwordmax = -1; // default: unlimited wordcount in compound words
  cpdmin = -1;  // undefined
  cpdmaxsyllable = 0; // default: unlimited syllablecount in compound words
  cpdvowels=NULL; // vowels (for calculating of Hungarian compounding limit, O(n) search! XXX)
  cpdvowels_utf16=NULL; // vowels for UTF-8 encoding (bsearch instead of O(n) search)

  version=NULL; // affix and dictionary file version string
  havecontclass=0; // flags of possible continuing classes (double affix)
  // LEMMA_PRESENT: not put root into the morphological output. Lemma presents
  // in morhological description in dictionary file. It's often combined with PSEUDOROOT.
  lemma_present = FLAG_NULL; 
  circumfix = FLAG_NULL; 
  onlyincompound = FLAG_NULL; 
  maxngramsugs = -1; // undefined
  maxdiff = -1; // undefined
  onlymaxdiff = 0;
  maxcpdsugs = -1; // undefined
  nosplitsugs = 0;
  sugswithdots = 0;
  keepcase = 0;
  forceucase = 0;
  warn = 0;
  forbidwarn = 0;
  checksharps = 0;
  substandard = FLAG_NULL;
  fullstrip = 0;

  sfx = NULL;
  pfx = NULL;

  for (int i=0; i < SETSIZE; i++) {

  FREE_FLAG(compoundbegin);
  FREE_FLAG(compoundmiddle);
  FREE_FLAG(compoundend);
  FREE_FLAG(compoundpermitflag);
  FREE_FLAG(compoundforbidflag);
  FREE_FLAG(compoundroot);
  FREE_FLAG(forbiddenword);
  FREE_FLAG(nosuggest);
  FREE_FLAG(nongramsuggest);
  FREE_FLAG(needaffix);
  FREE_FLAG(lemma_present);
  FREE_FLAG(circumfix);
  FREE_FLAG(onlyincompound);
  
  cpdwordmax = 0;
  pHMgr = NULL;
  cpdmin = 0;


       if (strncmp(line,"NOSUGGEST",9) == 0) {
          if (parse_flag(line, &nosuggest, afflst)) {
             delete afflst;
             return 1;
          }
       }

       if (strncmp(line,"NONGRAMSUGGEST",14) == 0) {
          if (parse_flag(line, &nongramsuggest, afflst)) {
             delete afflst;
             return 1;
          }
       }

       /* parse in the flag used by forbidden words */
       if (strncmp(line,"FORBIDDENWORD",13) == 0) {
          if (parse_flag(line, &forbiddenword, afflst)) {
             delete afflst;
             return 1;
          }
       }



       if (strncmp(line,"MAXNGRAMSUGS",12) == 0) {
          if (parse_num(line, &maxngramsugs, afflst)) {
             delete afflst;
             return 1;
          }
       }

       if (strncmp(line,"ONLYMAXDIFF", 11) == 0)
                   onlymaxdiff = 1;

       if (strncmp(line,"MAXDIFF",7) == 0) {
          if (parse_num(line, &maxdiff, afflst)) {
             delete afflst;
             return 1;
          }
       }

       if (strncmp(line,"MAXCPDSUGS",10) == 0) {
          if (parse_num(line, &maxcpdsugs, afflst)) {
             delete afflst;
             return 1;
          }
       }

       if (strncmp(line,"NOSPLITSUGS",11) == 0) {
                   nosplitsugs=1;
       }

       if (strncmp(line,"FULLSTRIP",9) == 0) {
                   fullstrip=1;
       }


       /* parse in the flag used by forbidden words */
       if (strncmp(line,"KEEPCASE",8) == 0) {
          if (parse_flag(line, &keepcase, afflst)) {
             delete afflst;
             return 1;
          }
       }

       /* parse in the flag used by `forceucase' */
       if (strncmp(line,"FORCEUCASE",10) == 0) {
          if (parse_flag(line, &forceucase, afflst)) {
             delete afflst;
             return 1;
          }
       }

       /* parse in the flag used by `warn' */
       if (strncmp(line,"WARN",4) == 0) {
          if (parse_flag(line, &warn, afflst)) {
             delete afflst;
             return 1;
          }
       }

       if (strncmp(line,"FORBIDWARN",10) == 0) {
                   forbidwarn=1;
       }

       /* parse in the flag used by the affix generator */
       if (strncmp(line,"SUBSTANDARD",11) == 0) {
          if (parse_flag(line, &substandard, afflst)) {
             delete afflst;
             return 1;
          }
       }


                *(expw + strlen(expw)) = (char) i;
        }
    }

    wordchars = mystrdup(expw);
    }

    // default BREAK definition
    if (numbreak == -1) {
        breaktable = (char **) malloc(sizeof(char *) * 3);
        if (!breaktable) return 1;
        breaktable[0] = mystrdup("-");
        breaktable[1] = mystrdup("^-");
        breaktable[2] = mystrdup("-$");
        if (breaktable[0] && breaktable[1] && breaktable[2]) numbreak = 3;
    }
    return 0;

          if (candidate_check(candidate,strlen(candidate))) return 1;
          r++; // search for the next letter
      }
   }
   return 0;
}

// forbid compoundings when there are special patterns at word bound
int AffixMgr::cpdpat_check(const char * word, int pos, hentry * r1, hentry * r2, const char affixed)
{
  int len;
  for (int i = 0; i < numcheckcpd; i++) {
      if (isSubset(checkcpdtable[i].pattern2, word + pos) &&
        (!r1 || !checkcpdtable[i].cond ||
          (r1->astr && TESTAFF(r1->astr, checkcpdtable[i].cond, r1->alen))) &&
        (!r2 || !checkcpdtable[i].cond2 ||
          (r2->astr && TESTAFF(r2->astr, checkcpdtable[i].cond2, r2->alen))) &&
        // zero length pattern => only TESTAFF
        // zero pattern (0/flag) => unmodified stem (zero affixes allowed)
        (!*(checkcpdtable[i].pattern) || (
            (*(checkcpdtable[i].pattern)=='0' && r1->blen <= pos && strncmp(word + pos - r1->blen, r1->word, r1->blen) == 0) ||
            (*(checkcpdtable[i].pattern)!='0' && (len = strlen(checkcpdtable[i].pattern)) &&
                strncmp(word + pos - len, checkcpdtable[i].pattern, len) == 0)))) {
            return 1;
        }
  }
  return 0;
}

// forbid compounding with neighbouring upper and lower case characters at word bounds
int AffixMgr::cpdcase_check(const char * word, int pos)
{
  if (utf8) {

          for ((*cmax)--; (word[*cmax] & 0xc0) == 0x80; (*cmax)--);
        }
    } else {
        *cmin = cpdmin;
        *cmax = len - cpdmin + 1;
    }
}


// check if compound word is correctly spelled
// hu_mov_rule = spec. Hungarian rule (XXX)
struct hentry * AffixMgr::compound_check(const char * word, int len, 
    short wordnum, short numsyllable, short maxwordnum, short wnum, hentry ** words = NULL,
    char hu_mov_rule = 0, char is_sug = 0, int * info = NULL)
{
    int i; 
    short oldnumsyllable, oldnumsyllable2, oldwordnum, oldwordnum2;
    struct hentry * rv = NULL;
    struct hentry * rv_first;
    struct hentry * rwords[MAXWORDLEN]; // buffer for COMPOUND pattern checking
    char st [MAXWORDUTF8LEN + 4];
    char ch = '\0';
    int cmin;
    int cmax;
    int striple = 0;
    int scpd = 0;
    int soldi = 0;
    int oldcmin = 0;
    int oldcmax = 0;
    int oldlen = 0;
    int checkedstriple = 0;
    int onlycpdrule;
    int affixed = 0;
    hentry ** oldwords = words;

    int checked_prefix;

    setcminmax(&cmin, &cmax, word, len);

    strcpy(st, word);

    for (i = cmin; i < cmax; i++) {

        oldnumsyllable = numsyllable;
        oldwordnum = wordnum;
        checked_prefix = 0;

        // go to end of the UTF-8 character
        if (utf8) {
            for (; (st[i] & 0xc0) == 0x80; i++);
            if (i >= cmax) return NULL;
        }

        words = oldwords;
        onlycpdrule = (words) ? 1 : 0;

        do { // onlycpdrule loop

        oldnumsyllable = numsyllable;
        oldwordnum = wordnum;
        checked_prefix = 0;


        do { // simplified checkcompoundpattern loop

        if (scpd > 0) {
          for (; scpd <= numcheckcpd && (!checkcpdtable[scpd-1].pattern3 ||
            strncmp(word + i, checkcpdtable[scpd-1].pattern3, strlen(checkcpdtable[scpd-1].pattern3)) != 0); scpd++);

          if (scpd > numcheckcpd) break; // break simplified checkcompoundpattern loop
          strcpy(st + i, checkcpdtable[scpd-1].pattern);

          len += strlen(checkcpdtable[scpd-1].pattern) + strlen(checkcpdtable[scpd-1].pattern2) - strlen(checkcpdtable[scpd-1].pattern3);
          oldcmin = cmin;
          oldcmax = cmax;
          setcminmax(&cmin, &cmax, st, len);

          cmax = len - cpdmin + 1;
        }


        ch = st[i];
        st[i] = '\0';

        sfx = NULL;
        pfx = NULL;

        // FIRST WORD

        affixed = 1;
        rv = lookup(st); // perhaps without prefix

        // search homonym with compound flag
        while ((rv) && !hu_mov_rule &&
            ((needaffix && TESTAFF(rv->astr, needaffix, rv->alen)) ||
                !((compoundflag && !words && !onlycpdrule && TESTAFF(rv->astr, compoundflag, rv->alen)) ||
                  (compoundbegin && !wordnum && !onlycpdrule && 
                        TESTAFF(rv->astr, compoundbegin, rv->alen)) ||
                  (compoundmiddle && wordnum && !words && !onlycpdrule &&
                    TESTAFF(rv->astr, compoundmiddle, rv->alen)) ||
                  (numdefcpd && onlycpdrule &&
                    ((!words && !wordnum && defcpd_check(&words, wnum, rv, (hentry **) &rwords, 0)) ||
                    (words && defcpd_check(&words, wnum, rv, (hentry **) &rwords, 0))))) ||
                  (scpd != 0 && checkcpdtable[scpd-1].cond != FLAG_NULL &&
                    !TESTAFF(rv->astr, checkcpdtable[scpd-1].cond, rv->alen)))
                  ) {
            rv = rv->next_homonym;
        }

        if (rv) affixed = 0;

        if (!rv) {
            if (onlycpdrule) break;
            if (compoundflag && 
             !(rv = prefix_check(st, i, hu_mov_rule ? IN_CPD_OTHER : IN_CPD_BEGIN, compoundflag))) {
                if ((rv = suffix_check(st, i, 0, NULL, NULL, 0, NULL,
                        FLAG_NULL, compoundflag, hu_mov_rule ? IN_CPD_OTHER : IN_CPD_BEGIN)) && !hu_mov_rule &&
                    sfx->getCont() &&
                        ((compoundforbidflag && TESTAFF(sfx->getCont(), compoundforbidflag, 
                            sfx->getContLen())) || (compoundend &&
                        TESTAFF(sfx->getCont(), compoundend, 

                ((rv = suffix_check(st, i, 0, NULL, NULL, 0, NULL, FLAG_NULL, compoundbegin, hu_mov_rule ? IN_CPD_OTHER : IN_CPD_BEGIN)) ||
                (rv = prefix_check(st, i, hu_mov_rule ? IN_CPD_OTHER : IN_CPD_BEGIN, compoundbegin)))) ||
              ((wordnum > 0) && compoundmiddle &&
                ((rv = suffix_check(st, i, 0, NULL, NULL, 0, NULL, FLAG_NULL, compoundmiddle, hu_mov_rule ? IN_CPD_OTHER : IN_CPD_BEGIN)) ||
                (rv = prefix_check(st, i, hu_mov_rule ? IN_CPD_OTHER : IN_CPD_BEGIN, compoundmiddle)))))
              ) checked_prefix = 1;
        // else check forbiddenwords and needaffix
        } else if (rv->astr && (TESTAFF(rv->astr, forbiddenword, rv->alen) ||
            TESTAFF(rv->astr, needaffix, rv->alen) ||
            TESTAFF(rv->astr, ONLYUPCASEFLAG, rv->alen) ||
            (is_sug && nosuggest && TESTAFF(rv->astr, nosuggest, rv->alen))
             )) {
                st[i] = ch;
                //continue;
                break;
        }

            // check non_compound flag in suffix and prefix
            if ((rv) && !hu_mov_rule &&
                ((pfx && pfx->getCont() &&
                    TESTAFF(pfx->getCont(), compoundforbidflag, 
                        pfx->getContLen())) ||
                (sfx && sfx->getCont() &&

                (sfx && sfx->getCont() &&
                    TESTAFF(sfx->getCont(), compoundmiddle, 
                        sfx->getContLen())))) {
                    rv = NULL;
            }

        // check forbiddenwords
        if ((rv) && (rv->astr) && (TESTAFF(rv->astr, forbiddenword, rv->alen) ||
            TESTAFF(rv->astr, ONLYUPCASEFLAG, rv->alen) ||
            (is_sug && nosuggest && TESTAFF(rv->astr, nosuggest, rv->alen)))) {
                return NULL;
            }

        // increment word number, if the second root has a compoundroot flag
        if ((rv) && compoundroot && 
            (TESTAFF(rv->astr, compoundroot, rv->alen))) {
                wordnum++;

               )
             )
         ) { // first word is ok condition

// LANG_hu section: spec. Hungarian rule
            if (langnum == LANG_hu) {
                // calculate syllable number of the word
                numsyllable += get_syllable(st, i);

                // + 1 word, if syllable number of the prefix > 1 (hungarian convention)
                if (pfx && (get_syllable(pfx->getKey(),strlen(pfx->getKey())) > 1)) wordnum++;
            }
// END of LANG_hu section


            // NEXT WORD(S)
            rv_first = rv;
            st[i] = ch;

        do { // striple loop

            // check simplifiedtriple
            if (simplifiedtriple) { 

                          (compoundend && !words && TESTAFF(rv->astr, compoundend, rv->alen)) ||
                           (numdefcpd && words && defcpd_check(&words, wnum + 1, rv, NULL,1))) ||
                             (scpd != 0 && checkcpdtable[scpd-1].cond2 != FLAG_NULL &&
                                !TESTAFF(rv->astr, checkcpdtable[scpd-1].cond2, rv->alen))
                           )) {
            rv = rv->next_homonym;
        }

            // check FORCEUCASE
            if (rv && forceucase && (rv) &&
                (TESTAFF(rv->astr, forceucase, rv->alen)) && !(info && *info & SPELL_ORIGCAP)) rv = NULL;

            if (rv && words && words[wnum + 1]) return rv_first;

            oldnumsyllable2 = numsyllable;
            oldwordnum2 = wordnum;


// LANG_hu section: spec. Hungarian rule, XXX hardwired dictionary code
            if ((rv) && (langnum == LANG_hu) && (TESTAFF(rv->astr, 'I', rv->alen)) && !(TESTAFF(rv->astr, 'J', rv->alen))) {
                numsyllable--;
            }
// END of LANG_hu section

            // increment word number, if the second root has a compoundroot flag
            if ((rv) && (compoundroot) && 
                (TESTAFF(rv->astr, compoundroot, rv->alen))) {
                    wordnum++;
            }

            // check forbiddenwords
            if ((rv) && (rv->astr) && (TESTAFF(rv->astr, forbiddenword, rv->alen) ||
                TESTAFF(rv->astr, ONLYUPCASEFLAG, rv->alen) ||
               (is_sug && nosuggest && TESTAFF(rv->astr, nosuggest, rv->alen)))) return NULL;

            // second word is acceptable, as a root?
            // hungarian conventions: compounding is acceptable,
            // when compound forms consist of 2 words, or if more,
            // then the syllable number of root words must be 6, or lesser.

            if ((rv) && (

                    )
                && (
                      ((cpdwordmax==-1) || (wordnum+1<cpdwordmax)) || 
                      ((cpdmaxsyllable!=0) && 
                          (numsyllable + get_syllable(HENTRY_WORD(rv), rv->clen)<=cpdmaxsyllable))
                    ) &&
               (
                 // test CHECKCOMPOUNDPATTERN
                 !numcheckcpd || scpd != 0 || !cpdpat_check(word, i, rv_first, rv, 0)
               ) &&
                (
                     (!checkcompounddup || (rv != rv_first))
                   )
            // test CHECKCOMPOUNDPATTERN conditions
                && (scpd == 0 || checkcpdtable[scpd-1].cond2 == FLAG_NULL ||
                      TESTAFF(rv->astr, checkcpdtable[scpd-1].cond2, rv->alen))
                )

            }

            numsyllable = oldnumsyllable2;
            wordnum = oldwordnum2;

            // perhaps second word has prefix or/and suffix
            sfx = NULL;
            sfxflag = FLAG_NULL;
            rv = (compoundflag && !onlycpdrule) ? affix_check((word+i),strlen(word+i), compoundflag, IN_CPD_END) : NULL;
            if (!rv && compoundend && !onlycpdrule) {
                sfx = NULL;
                pfx = NULL;
                rv = affix_check((word+i),strlen(word+i), compoundend, IN_CPD_END);
            }

            if (!rv && numdefcpd && words) {
                rv = affix_check((word+i),strlen(word+i), 0, IN_CPD_END);
                if (rv && defcpd_check(&words, wnum + 1, rv, NULL, 1)) return rv_first;
                rv = NULL;
            }

            // test CHECKCOMPOUNDPATTERN conditions (allowed forms)
            if (rv && !(scpd == 0 || checkcpdtable[scpd-1].cond2 == FLAG_NULL || 
                TESTAFF(rv->astr, checkcpdtable[scpd-1].cond2, rv->alen))) rv = NULL;

            // test CHECKCOMPOUNDPATTERN conditions (forbidden compounds)
            if (rv && numcheckcpd && scpd == 0 && cpdpat_check(word, i, rv_first, rv, affixed)) rv = NULL;

            // check non_compound flag in suffix and prefix
            if ((rv) && 
                ((pfx && pfx->getCont() &&
                    TESTAFF(pfx->getCont(), compoundforbidflag, 
                        pfx->getContLen())) ||
                (sfx && sfx->getCont() &&
                    TESTAFF(sfx->getCont(), compoundforbidflag, 
                        sfx->getContLen())))) {
                    rv = NULL;
            }

            // check FORCEUCASE
            if (rv && forceucase && (rv) &&
                (TESTAFF(rv->astr, forceucase, rv->alen)) && !(info && *info & SPELL_ORIGCAP)) rv = NULL;

            // check forbiddenwords
            if ((rv) && (rv->astr) && (TESTAFF(rv->astr, forbiddenword, rv->alen) ||
                TESTAFF(rv->astr, ONLYUPCASEFLAG, rv->alen) ||
               (is_sug && nosuggest && TESTAFF(rv->astr, nosuggest, rv->alen)))) return NULL;

            // pfxappnd = prefix of word+i, or NULL
            // calculate syllable number of prefix.
            // hungarian convention: when syllable number of prefix is more,
            // than 1, the prefix+word counts as two words.

            if (langnum == LANG_hu) {


                // increment syllable num, if last word has a SYLLABLENUM flag
                // and the suffix is beginning `s'

                if (cpdsyllablenum) {
                    switch (sfxflag) {
                        case 'c': { numsyllable+=2; break; }
                        case 'J': { numsyllable += 1; break; }
                        case 'I': { if (rv && TESTAFF(rv->astr, 'J', rv->alen)) numsyllable += 1; break; }
                    }
                }
            }

            // increment word number, if the second word has a compoundroot flag
            if ((rv) && (compoundroot) && 
                (TESTAFF(rv->astr, compoundroot, rv->alen))) {
                    wordnum++;

            }

            numsyllable = oldnumsyllable2;
            wordnum = oldwordnum2;

            // perhaps second word is a compound word (recursive call)
            if (wordnum < maxwordnum) {
                rv = compound_check((st+i),strlen(st+i), wordnum+1,
                     numsyllable, maxwordnum, wnum + 1, words, 0, is_sug, info);
                
                if (rv && numcheckcpd && ((scpd == 0 && cpdpat_check(word, i, rv_first, rv, affixed)) ||
                   (scpd != 0 && !cpdpat_check(word, i, rv_first, rv, affixed)))) rv = NULL;
            } else {
                rv=NULL;
            }
            if (rv) {
                // forbid compound word, if it is a non compound word with typical fault
                if (checkcompoundrep || forbiddenword) {
                    struct hentry * rv2 = NULL;

                    if (checkcompoundrep && cpdrep_check(word, len)) return NULL;
                    
                    // check first part
                    if (strncmp(rv->word, word + i, rv->blen) == 0) {
                        char r = *(st + i + rv->blen);
                        *(st + i + rv->blen) = '\0';
                        
                        if (checkcompoundrep && cpdrep_check(st, i + rv->blen)) {
                            *(st + i + rv->blen) = r;
                            continue;
                        }

                        if (forbiddenword) {
                            rv2 = lookup(word);
                            if (!rv2) rv2 = affix_check(word, len);
                            if (rv2 && rv2->astr && TESTAFF(rv2->astr, forbiddenword, rv2->alen) && 
                                (strncmp(rv2->word, st, i + rv->blen) == 0)) {
                                    return NULL;
                            }
                        }
                        *(st + i + rv->blen) = r;
                    }
                }
                return rv_first;
            }
          } while (striple && !checkedstriple); // end of striple loop

          if (checkedstriple) {
            i++;
            checkedstriple = 0;
            striple = 0;

          i = soldi;
          soldi = 0;
          len = oldlen;
          cmin = oldcmin;
          cmax = oldcmax;
        }
        scpd++;


        } while (!onlycpdrule && simplifiedcpd && scpd <= numcheckcpd); // end of simplifiedcpd loop

        scpd = 0;
        wordnum = oldwordnum;
        numsyllable = oldnumsyllable;

        if (soldi != 0) {
          i = soldi;
          strcpy(st, word); // XXX add more optim.
          soldi = 0;
        } else st[i] = ch;

        } while (numdefcpd && oldwordnum == 0 && !onlycpdrule && (onlycpdrule = 1)); // end of onlycpd loop

        numsyllable = oldnumsyllable;
    }

    return NULL;
}

// check if compound word is correctly spelled
// hu_mov_rule = spec. Hungarian rule (XXX)
int AffixMgr::compound_check_morph(const char * word, int len, 

    char ch;

    int checked_prefix;
    char presult[MAXLNLEN];

    int cmin;
    int cmax;

    int onlycpdrule;
    int affixed = 0;
    hentry ** oldwords = words;

    setcminmax(&cmin, &cmax, word, len);

    strcpy(st, word);

    for (i = cmin; i < cmax; i++) {
        oldnumsyllable = numsyllable;
        oldwordnum = wordnum;
        checked_prefix = 0;

        // go to end of the UTF-8 character
        if (utf8) {
            for (; (st[i] & 0xc0) == 0x80; i++);
            if (i >= cmax) return 0;
        }

        words = oldwords;
        onlycpdrule = (words) ? 1 : 0;

        do { // onlycpdrule loop

        oldnumsyllable = numsyllable;
        oldwordnum = wordnum;
        checked_prefix = 0;

        ch = st[i];
        st[i] = '\0';
        sfx = NULL;

        // FIRST WORD

        affixed = 1;

        *presult = '\0';
        if (partresult) mystrcat(presult, partresult, MAXLNLEN);

        rv = lookup(st); // perhaps without prefix

        // search homonym with compound flag
        while ((rv) && !hu_mov_rule && 
            ((needaffix && TESTAFF(rv->astr, needaffix, rv->alen)) ||
                !((compoundflag && !words && !onlycpdrule && TESTAFF(rv->astr, compoundflag, rv->alen)) ||
                (compoundbegin && !wordnum && !onlycpdrule &&
                        TESTAFF(rv->astr, compoundbegin, rv->alen)) ||
                (compoundmiddle && wordnum && !words && !onlycpdrule &&
                    TESTAFF(rv->astr, compoundmiddle, rv->alen)) ||
                  (numdefcpd && onlycpdrule &&
                    ((!words && !wordnum && defcpd_check(&words, wnum, rv, (hentry **) &rwords, 0)) ||
                    (words && defcpd_check(&words, wnum, rv, (hentry **) &rwords, 0))))
                  ))) {
            rv = rv->next_homonym;
        }

        if (rv) affixed = 0;

        if (rv)  {
            sprintf(presult + strlen(presult), "%c%s%s", MSEP_FLD, MORPH_PART, st);
            if (!HENTRY_FIND(rv, MORPH_STEM)) {
                sprintf(presult + strlen(presult), "%c%s%s", MSEP_FLD, MORPH_STEM, st);
            }
            // store the pointer of the hash entry
//            sprintf(presult + strlen(presult), "%c%s%p", MSEP_FLD, MORPH_HENTRY, rv);
            if (HENTRY_DATA(rv)) {
                sprintf(presult + strlen(presult), "%c%s", MSEP_FLD, HENTRY_DATA2(rv));
            }
        }        

        if (!rv) {
            if (onlycpdrule) break;
            if (compoundflag &&
             !(rv = prefix_check(st, i, hu_mov_rule ? IN_CPD_OTHER : IN_CPD_BEGIN, compoundflag))) {
                if ((rv = suffix_check(st, i, 0, NULL, NULL, 0, NULL,
                        FLAG_NULL, compoundflag, hu_mov_rule ? IN_CPD_OTHER : IN_CPD_BEGIN)) && !hu_mov_rule &&
                    sfx->getCont() &&
                        ((compoundforbidflag && TESTAFF(sfx->getCont(), compoundforbidflag, 
                            sfx->getContLen())) || (compoundend &&
                        TESTAFF(sfx->getCont(), compoundend, 
                            sfx->getContLen())))) {

                    sprintf(presult + strlen(presult), "%c%s%s%s", MSEP_FLD,
                        MORPH_PART, st, line_uniq_app(&p, MSEP_REC));
                }
                if (p) free(p);
                checked_prefix = 1;
            }
        // else check forbiddenwords
        } else if (rv->astr && (TESTAFF(rv->astr, forbiddenword, rv->alen) ||
            TESTAFF(rv->astr, ONLYUPCASEFLAG, rv->alen) ||
            TESTAFF(rv->astr, needaffix, rv->alen))) {
                st[i] = ch;
                continue;
        }

            // check non_compound flag in suffix and prefix
            if ((rv) && !hu_mov_rule &&
                ((pfx && pfx->getCont() &&

                        pfx->getContLen())) ||
                (sfx && sfx->getCont() &&
                    TESTAFF(sfx->getCont(), compoundmiddle, 
                        sfx->getContLen())))) {
                    rv = NULL;
            }       

        // check forbiddenwords
        if ((rv) && (rv->astr) && (TESTAFF(rv->astr, forbiddenword, rv->alen)
            || TESTAFF(rv->astr, ONLYUPCASEFLAG, rv->alen))) continue;

        // increment word number, if the second root has a compoundroot flag
        if ((rv) && (compoundroot) && 
            (TESTAFF(rv->astr, compoundroot, rv->alen))) {
                wordnum++;
        }

        // first word is acceptable in compound words?

          && ! (( checkcompoundtriple && !words && // test triple letters
                   (word[i-1]==word[i]) && (
                      ((i>1) && (word[i-1]==word[i-2])) || 
                      ((word[i-1]==word[i+1])) // may be word[i+1] == '\0'
                   )
               ) ||
               (
                   // test CHECKCOMPOUNDPATTERN
                   numcheckcpd && !words && cpdpat_check(word, i, rv, NULL, affixed)
               ) ||
               ( 
                 checkcompoundcase && !words && cpdcase_check(word, i)
               ))
         )
// LANG_hu section: spec. Hungarian rule
         || ((!rv) && (langnum == LANG_hu) && hu_mov_rule && (rv = affix_check(st,i)) &&
              (sfx && sfx->getCont() && (

// END of LANG_hu section
            // increment word number, if the second root has a compoundroot flag
            if ((rv) && (compoundroot) && 
                (TESTAFF(rv->astr, compoundroot, rv->alen))) {
                    wordnum++;
            }

            // check forbiddenwords
            if ((rv) && (rv->astr) && (TESTAFF(rv->astr, forbiddenword, rv->alen) ||
                TESTAFF(rv->astr, ONLYUPCASEFLAG, rv->alen))) {
                st[i] = ch;
                continue;
            }

            // second word is acceptable, as a root?
            // hungarian conventions: compounding is acceptable,
            // when compound forms consist of 2 words, or if more,
            // then the syllable number of root words must be 6, or lesser.


            numsyllable = oldnumsyllable2 ;
            wordnum = oldwordnum2;

            // perhaps second word has prefix or/and suffix
            sfx = NULL;
            sfxflag = FLAG_NULL;

            if (compoundflag && !onlycpdrule) rv = affix_check((word+i),strlen(word+i), compoundflag); else rv = NULL;

            if (!rv && compoundend && !onlycpdrule) {
                sfx = NULL;
                pfx = NULL;
                rv = affix_check((word+i),strlen(word+i), compoundend);
            }

            if (!rv && numdefcpd && words) {
                rv = affix_check((word+i),strlen(word+i), 0, IN_CPD_END);
                if (rv && words && defcpd_check(&words, wnum + 1, rv, NULL, 1)) {

                        pfx->getContLen())) ||
                (sfx && sfx->getCont() &&
                    TESTAFF(sfx->getCont(), compoundforbidflag, 
                        sfx->getContLen())))) {
                    rv = NULL;
            }

            // check forbiddenwords
            if ((rv) && (rv->astr) && (TESTAFF(rv->astr,forbiddenword,rv->alen) ||
                    TESTAFF(rv->astr, ONLYUPCASEFLAG, rv->alen))
                    && (! TESTAFF(rv->astr, needaffix, rv->alen))) {
                        st[i] = ch;
                        continue;
                    }

            if (langnum == LANG_hu) {
                // calculate syllable number of the word
                numsyllable += get_syllable(word + i, strlen(word + i));

                             numsyllable, maxwordnum, wnum + 1, words, 0, result, presult);
            } else {
                rv=NULL;
            }
        }
        st[i] = ch;
        wordnum = oldwordnum;
        numsyllable = oldnumsyllable;

        } while (numdefcpd && oldwordnum == 0 && !onlycpdrule && (onlycpdrule = 1)); // end of onlycpd loop

    }
    return 0;
}    

 // return 1 if s1 (reversed) is a leading subset of end of s2
/* inline int AffixMgr::isRevSubset(const char * s1, const char * end_of_s2, int len)
 {
    while ((len > 0) && *s1 && (*s1 == *end_of_s2)) {

                   circumfix, ep->getContLen())) &&
               (!se->getCont() || !(TESTAFF(se->getCont(),circumfix,se->getContLen())))) ||
              // circumfix flag in prefix AND suffix
              ((ppfx && (ep->getCont()) && TESTAFF(ep->getCont(),
                   circumfix, ep->getContLen())) &&
               (se->getCont() && (TESTAFF(se->getCont(),circumfix,se->getContLen())))))  &&
            // fogemorpheme
              (in_compound || 
                 !(se->getCont() && (TESTAFF(se->getCont(), onlyincompound, se->getContLen())))) &&
            // needaffix on prefix or first suffix
              (cclass || 
                   !(se->getCont() && TESTAFF(se->getCont(), needaffix, se->getContLen())) ||
                   (ppfx && !((ep->getCont()) &&
                     TESTAFF(ep->getCont(), needaffix,
                       ep->getContLen())))
              )) {
            ) {
                rv = se->checkword(word,len, sfxopts, ppfx, wlst, maxSug, ns, (FLAG) cclass, 
                    needflag, (in_compound ? 0 : onlyincompound));
                if (rv) {
                    sfx=se; // BUG: sfx not stateless
                    return rv;
                }
            }
        }
       se = se->getNext();
    }

    // now handle the general case
    if (len == 0) return NULL; // FULLSTRIP
    unsigned char sp= *((const unsigned char *)(word + len - 1));
    SfxEntry * sptr = sStart[sp];

    while (sptr) {
        if (isRevSubset(sptr->getKey(), word + len - 1, len)
        ) {
            // suffixes are not allowed in beginning of compounds
            if ((((in_compound != IN_CPD_BEGIN)) || // && !cclass
             // except when signed with compoundpermitflag flag

                 !((sptr->getCont() && (TESTAFF(sptr->getCont(), onlyincompound, sptr->getContLen()))))) &&
            // needaffix on prefix or first suffix
              (cclass || 
                  !(sptr->getCont() && TESTAFF(sptr->getCont(), needaffix, sptr->getContLen())) ||
                  (ppfx && !((ep->getCont()) &&
                     TESTAFF(ep->getCont(), needaffix,
                       ep->getContLen())))
              )
            ) if (in_compound != IN_CPD_END || ppfx || !(sptr->getCont() && TESTAFF(sptr->getCont(), onlyincompound, sptr->getContLen()))) {
                rv = sptr->checkword(word,len, sfxopts, ppfx, wlst,
                    maxSug, ns, cclass, needflag, (in_compound ? 0 : onlyincompound));
                if (rv) {
                    sfx=sptr; // BUG: sfx not stateless
                    sfxflag = sptr->getFlag(); // BUG: sfxflag not stateless
                    if (!sptr->getCont()) sfxappnd=sptr->getKey(); // BUG: sfxappnd not stateless
                    return rv;
                }

    if (rv) return rv;

    // if still not found check all suffixes
    rv = suffix_check(word, len, 0, NULL, NULL, 0, NULL, FLAG_NULL, needflag, in_compound);

    if (havecontclass) {
        sfx = NULL;
        pfx = NULL;

        if (rv) return rv;
        // if still not found check all two-level suffixes
        rv = suffix_check_twosfx(word, len, 0, NULL, needflag);

        if (rv) return rv;
        // if still not found check all two-level suffixes
        rv = prefix_check_twosfx(word, len, IN_CPD_NOT, needflag);
    }

    return rv;
}

// check if word with affixes is correctly spelled
char * AffixMgr::affix_check_morph(const char * word, int len, const FLAG needflag, char in_compound)
{
    char result[MAXLNLEN];
    char * st = NULL;


                int cmp = morphcmp(stemmorph, targetmorph);

                if (cmp == 0) {
                    char * newword = sptr->add(ts, wl);
                    if (newword) {
                        hentry * check = pHMgr->lookup(newword); // XXX extra dic
                        if (!check || !check->astr || 
                            !(TESTAFF(check->astr, forbiddenword, check->alen) || 
                              TESTAFF(check->astr, ONLYUPCASEFLAG, check->alen))) {
                                return newword;
                        }
                        free(newword);
                    }
                }
                
                // recursive call for secondary suffixes
                if ((level == 0) && (cmp == 1) && (sptr->getContLen() > 0) &&

  return fullstrip;
}

FLAG AffixMgr::get_keepcase() const
{
  return keepcase;
}

FLAG AffixMgr::get_forceucase() const
{
  return forceucase;
}

FLAG AffixMgr::get_warn() const
{
  return warn;
}

int AffixMgr::get_forbidwarn() const
{
  return forbidwarn;
}

int AffixMgr::get_checksharps() const
{
  return checksharps;
}

char * AffixMgr::encode_flag(unsigned short aflag) const
{
  return pHMgr->encode_flag(aflag);

}

// return the forbidden words control flag
FLAG AffixMgr::get_nosuggest() const
{
  return nosuggest;
}

// return the forbidden words control flag
FLAG AffixMgr::get_nongramsuggest() const
{
  return nongramsuggest;
}

// return the forbidden words flag modify flag
FLAG AffixMgr::get_needaffix() const
{
  return needaffix;
}

// return the onlyincompound flag
FLAG AffixMgr::get_onlyincompound() const

}

// return utf8
int AffixMgr::get_utf8() const
{
  return utf8;
}

// return nosplitsugs
int AffixMgr::get_maxngramsugs(void) const
{
  return maxngramsugs;
}

int AffixMgr::get_maxcpdsugs(void) const
{
  return maxcpdsugs;
}

int AffixMgr::get_maxdiff(void) const
{
  return maxdiff;
}

int AffixMgr::get_onlymaxdiff(void) const
{
  return onlymaxdiff;
}

// return nosplitsugs
int AffixMgr::get_nosplitsugs(void) const
{
  return nosplitsugs;
}

// return sugswithdots
int AffixMgr::get_sugswithdots(void) const

                  case 0: {
                             if (strncmp(piece,"REP",3) != 0) {
                                 HUNSPELL_WARNING(stderr, "error: line %d: table is corrupt\n", af->getlinenum());
                                 numrep = 0;
                                 return 1;
                             }
                             break;
                          }
                  case 1: {
                            if (*piece == '^') reptable[j].start = true; else reptable[j].start = false;
                            reptable[j].pattern = mystrrep(mystrdup(piece + int(reptable[j].start)),"_"," ");
                            int lr = strlen(reptable[j].pattern) - 1;
                            if (reptable[j].pattern[lr] == '$') {
                                reptable[j].end = true;
                                reptable[j].pattern[lr] = '\0';
                            } else reptable[j].end = false;
                            break;
                          }
                  case 2: { reptable[j].pattern2 = mystrrep(mystrdup(piece),"_"," "); break; }
                  default: break;
               }
               i++;
           }
           piece = mystrsep(&tp, 0);
        }
        if ((!(reptable[j].pattern)) || (!(reptable[j].pattern2))) {

                    break; 
                  }
                  default: break;
               }
               i++;
           }
           piece = mystrsep(&tp, 0);
        }
        if (!pattern || !pattern2) {
            if (pattern)
                free(pattern);
            if (pattern2)
                free(pattern2);
            HUNSPELL_WARNING(stderr, "error: line %d: table is corrupt\n", af->getlinenum());
            return 1;
        }
        (*rl)->add(pattern, pattern2);
   }
   return 0;
}


/* parse in the typical fault correcting table */
int  AffixMgr::parse_phonetable(char * line, FileMgr * af)
{

        }
   }
   return 0;
}

/* parse in the word breakpoint table */
int  AffixMgr::parse_breaktable(char * line, FileMgr * af)
{
   if (numbreak > -1) {
      HUNSPELL_WARNING(stderr, "error: line %d: multiple table definitions\n", af->getlinenum());
      return 1;
   }
   char * tp = line;
   char * piece;
   int i = 0;
   int np = 0;
   piece = mystrsep(&tp, 0);
   while (piece) {
       if (*piece != '\0') {
          switch(i) {
             case 0: { np++; break; }
             case 1: { 
                       numbreak = atoi(piece);
                       if (numbreak < 0) {
                          HUNSPELL_WARNING(stderr, "error: line %d: bad entry number\n", af->getlinenum());
                          return 1;
                       }
                       if (numbreak == 0) return 0;
                       breaktable = (char **) malloc(numbreak * sizeof(char *));
                       if (!breaktable) return 1;
                       np++;
                       break;
                     }
             default: break;
          }
          i++;





  FLAG                compoundpermitflag;
  int                 checkcompounddup;
  int                 checkcompoundrep;
  int                 checkcompoundcase;
  int                 checkcompoundtriple;
  int                 simplifiedtriple;
  FLAG                forbiddenword;
  FLAG                nosuggest;
  FLAG                nongramsuggest;
  FLAG                needaffix;
  int                 cpdmin;
  int                 numrep;
  replentry *         reptable;
  RepList *           iconvtable;
  RepList *           oconvtable;
  int                 nummap;
  mapentry *          maptable;

  char **             breaktable;
  int                 numcheckcpd;
  patentry *          checkcpdtable;
  int                 simplifiedcpd;
  int                 numdefcpd;
  flagentry *         defcpdtable;
  phonetable *        phone;
  int                 maxngramsugs;
  int                 maxcpdsugs;
  int                 maxdiff;
  int                 onlymaxdiff;
  int                 nosplitsugs;
  int                 sugswithdots;
  int                 cpdwordmax;
  int                 cpdmaxsyllable;
  char *              cpdvowels;
  w_char *            cpdvowels_utf16;
  int                 cpdvowels_utf16_len;
  char *              cpdsyllablenum;

  int                 ignorechars_utf16_len;
  char *              version;
  char *              lang;
  int                 langnum;
  FLAG                lemma_present;
  FLAG                circumfix;
  FLAG                onlyincompound;
  FLAG                keepcase;
  FLAG                forceucase;
  FLAG                warn;
  int                 forbidwarn;
  FLAG                substandard;
  int                 checksharps;
  int                 fullstrip;

  int                 havecontclass; // boolean variable
  char                contclasses[CONTSIZE]; // flags of possible continuing classes (twofold affix)

public:

            unsigned short al, char * morph, char * targetmorph, int level);

  int    expand_rootword(struct guessword * wlst, int maxn, const char * ts,
            int wl, const unsigned short * ap, unsigned short al, char * bad,
            int, char *);

  short       get_syllable (const char * word, int wlen);
  int         cpdrep_check(const char * word, int len);
  int         cpdpat_check(const char * word, int len, hentry * r1, hentry * r2,
                    const char affixed);
  int         defcpd_check(hentry *** words, short wnum, hentry * rv,
                    hentry ** rwords, char all);
  int         cpdcase_check(const char * word, int len);
  inline int  candidate_check(const char * word, int len);
  void        setcminmax(int * cmin, int * cmax, const char * word, int len);
  struct hentry * compound_check(const char * word, int len, short wordnum,
            short numsyllable, short maxwordnum, short wnum, hentry ** words,
            char hu_mov_rule, char is_sug, int * info);

  int compound_check_morph(const char * word, int len, short wordnum,
            short numsyllable, short maxwordnum, short wnum, hentry ** words,
            char hu_mov_rule, char ** result, char * partresult);

  struct hentry * lookup(const char * word);
  int                 get_numrep() const;
  struct replentry *  get_reptable() const;

  unsigned short *    get_wordchars_utf16(int * len) const;
  char *              get_ignore() const;
  unsigned short *    get_ignore_utf16(int * len) const;
  int                 get_compound() const;
  FLAG                get_compoundflag() const;
  FLAG                get_compoundbegin() const;
  FLAG                get_forbiddenword() const;
  FLAG                get_nosuggest() const;
  FLAG                get_nongramsuggest() const;
  FLAG                get_needaffix() const;
  FLAG                get_onlyincompound() const;
  FLAG                get_compoundroot() const;
  FLAG                get_lemma_present() const;
  int                 get_checknum() const;
  const char *        get_prefix() const;
  const char *        get_suffix() const;
  const char *        get_derived() const;
  const char *        get_version() const;
  int                 have_contclass() const;
  int                 get_utf8() const;
  int                 get_complexprefixes() const;
  char *              get_suffixed(char ) const;
  int                 get_maxngramsugs() const;
  int                 get_maxcpdsugs() const;
  int                 get_maxdiff() const;
  int                 get_onlymaxdiff() const;
  int                 get_nosplitsugs() const;
  int                 get_sugswithdots(void) const;
  FLAG                get_keepcase(void) const;
  FLAG                get_forceucase(void) const;
  FLAG                get_warn(void) const;
  int                 get_forbidwarn(void) const;
  int                 get_checksharps(void) const;
  char *              encode_flag(unsigned short aflag) const;
  int                 get_fullstrip() const;

private:
  int  parse_file(const char * affpath, const char * key);
  int  parse_flag(char * line, unsigned short * out, FileMgr * af);
  int  parse_num(char * line, int * out, FileMgr * af);





#define aeLONGCOND      (1 << 4)

// compound options
#define IN_CPD_NOT   0
#define IN_CPD_BEGIN 1
#define IN_CPD_END   2
#define IN_CPD_OTHER 3

// info options
#define  SPELL_COMPOUND  (1 << 0)
#define  SPELL_FORBIDDEN (1 << 1)
#define  SPELL_ALLCAP    (1 << 2)
#define  SPELL_NOCAP     (1 << 3)
#define  SPELL_INITCAP   (1 << 4)
#define  SPELL_ORIGCAP   (1 << 5)
#define  SPELL_WARN      (1 << 6)

#define MAXLNLEN        8192

#define MINCPDLEN       3
#define MAXCOMPOUND     10
#define MAXCONDLEN      20
#define MAXCONDLEN_1    (MAXCONDLEN - sizeof(char *))

#define MAXACC          1000





#include <string.h>
#include <stdio.h> 
#include <ctype.h>

#include "csutil.hxx"
#include "atypes.hxx"
#include "langnum.hxx"

// Unicode character encoding information
struct unicode_info {
  unsigned short c;
  unsigned short cupper;
  unsigned short clower;
};

#ifdef OPENOFFICEORG
#  include <unicode/uchar.h>
#else
#  ifndef MOZILLA_CLIENT
#    include "utf_info.cxx"
#    define UTF_LST_LEN (sizeof(utf_lst) / (sizeof(unicode_info)))
#  endif
#endif

#ifdef MOZILLA_CLIENT
#include "nsCOMPtr.h"
#include "nsServiceManagerUtils.h"
#include "nsIUnicodeEncoder.h"
#include "nsIUnicodeDecoder.h"
#include "nsUnicharUtils.h"
#include "nsICharsetConverterManager.h"
#include "nsUnicharUtilCIID.h"
#include "nsUnicharUtils.h"

static NS_DEFINE_CID(kCharsetConverterManagerCID, NS_ICHARSETCONVERTERMANAGER_CID);
static NS_DEFINE_CID(kUnicharUtilCID, NS_UNICHARUTIL_CID);
#endif

struct unicode_info2 {
  char cletter;
  unsigned short cupper;
  unsigned short clower;
};

static struct unicode_info2 * utf_tbl = NULL;
static int utf_tbl_count = 0; // utf_tbl can be used by multiple Hunspell instances

/* only UTF-16 (BMP) implementation */
char * u16_u8(char * dest, int size, const w_char * src, int srclen) {
    signed char * u8 = (signed char *)dest;
    signed char * u8_max = (signed char *)(u8 + size);
    const w_char * u2 = src;

    char * p = strchr(dup, breakchar);
    while (p) {
        linenum++;
        *p = '\0';
        p++;
        p = strchr(p, breakchar);
    }
    linenum++;
//    fprintf(stderr, "LINEN:%d %p %p\n", linenum, lines, *lines);
    *lines = (char **) malloc(linenum * sizeof(char *));
//    fprintf(stderr, "hello\n");
    if (!(*lines)) {
        free(dup);
        return 0;
    }

    p = dup;
    int l = 0;
    for (int i = 0; i < linenum; i++) {

       return dest;
  }
  return NULL;
}

char * mystrrep(char * word, const char * pat, const char * rep) {
    char * pos = strstr(word, pat);
    if (pos) {
      int replen = strlen(rep);
      int patlen = strlen(pat);
      while (pos) {
        if (replen < patlen) {
            char * end = word + strlen(word);
            char * next = pos + replen;
            char * prev = pos + strlen(pat);
            for (; prev < end; *next = *prev, prev++, next++);
            *next = '\0';
        } else if (replen > patlen) {
            char * end = pos + patlen;
            char * next = word + strlen(word) + replen - patlen;
            char * prev = next - replen + patlen;
            for (; prev >= end; *next = *prev, prev--, next--);
        }
        strncpy(pos, rep, replen);
        pos = strstr(word, pat);
      }
    }
    return word;
}

 // reverse word 
 int reverseword(char * word) {
   char r;
   for (char * dest = word + strlen(word) - 1; word < dest; word++, dest--) {

{ 0x00, 0xfa, 0xfa },
{ 0x00, 0xfb, 0xfb },
{ 0x00, 0xfc, 0xfc },
{ 0x00, 0xfd, 0xfd },
{ 0x00, 0xfe, 0xfe },
{ 0x00, 0xff, 0xff }
};

static struct cs_info tis620_tbl[] = {
{ 0x00, 0x00, 0x00 },
{ 0x00, 0x01, 0x01 },
{ 0x00, 0x02, 0x02 },
{ 0x00, 0x03, 0x03 },
{ 0x00, 0x04, 0x04 },
{ 0x00, 0x05, 0x05 },
{ 0x00, 0x06, 0x06 },
{ 0x00, 0x07, 0x07 },
{ 0x00, 0x08, 0x08 },
{ 0x00, 0x09, 0x09 },
{ 0x00, 0x0a, 0x0a },
{ 0x00, 0x0b, 0x0b },
{ 0x00, 0x0c, 0x0c },
{ 0x00, 0x0d, 0x0d },
{ 0x00, 0x0e, 0x0e },
{ 0x00, 0x0f, 0x0f },
{ 0x00, 0x10, 0x10 },
{ 0x00, 0x11, 0x11 },
{ 0x00, 0x12, 0x12 },
{ 0x00, 0x13, 0x13 },
{ 0x00, 0x14, 0x14 },
{ 0x00, 0x15, 0x15 },
{ 0x00, 0x16, 0x16 },
{ 0x00, 0x17, 0x17 },
{ 0x00, 0x18, 0x18 },
{ 0x00, 0x19, 0x19 },
{ 0x00, 0x1a, 0x1a },
{ 0x00, 0x1b, 0x1b },
{ 0x00, 0x1c, 0x1c },
{ 0x00, 0x1d, 0x1d },
{ 0x00, 0x1e, 0x1e },
{ 0x00, 0x1f, 0x1f },
{ 0x00, 0x20, 0x20 },
{ 0x00, 0x21, 0x21 },
{ 0x00, 0x22, 0x22 },
{ 0x00, 0x23, 0x23 },
{ 0x00, 0x24, 0x24 },
{ 0x00, 0x25, 0x25 },
{ 0x00, 0x26, 0x26 },
{ 0x00, 0x27, 0x27 },
{ 0x00, 0x28, 0x28 },
{ 0x00, 0x29, 0x29 },
{ 0x00, 0x2a, 0x2a },
{ 0x00, 0x2b, 0x2b },
{ 0x00, 0x2c, 0x2c },
{ 0x00, 0x2d, 0x2d },
{ 0x00, 0x2e, 0x2e },
{ 0x00, 0x2f, 0x2f },
{ 0x00, 0x30, 0x30 },
{ 0x00, 0x31, 0x31 },
{ 0x00, 0x32, 0x32 },
{ 0x00, 0x33, 0x33 },
{ 0x00, 0x34, 0x34 },
{ 0x00, 0x35, 0x35 },
{ 0x00, 0x36, 0x36 },
{ 0x00, 0x37, 0x37 },
{ 0x00, 0x38, 0x38 },
{ 0x00, 0x39, 0x39 },
{ 0x00, 0x3a, 0x3a },
{ 0x00, 0x3b, 0x3b },
{ 0x00, 0x3c, 0x3c },
{ 0x00, 0x3d, 0x3d },
{ 0x00, 0x3e, 0x3e },
{ 0x00, 0x3f, 0x3f },
{ 0x00, 0x40, 0x40 },
{ 0x01, 0x61, 0x41 },
{ 0x01, 0x62, 0x42 },
{ 0x01, 0x63, 0x43 },
{ 0x01, 0x64, 0x44 },
{ 0x01, 0x65, 0x45 },
{ 0x01, 0x66, 0x46 },
{ 0x01, 0x67, 0x47 },
{ 0x01, 0x68, 0x48 },
{ 0x01, 0x69, 0x49 },
{ 0x01, 0x6a, 0x4a },
{ 0x01, 0x6b, 0x4b },
{ 0x01, 0x6c, 0x4c },
{ 0x01, 0x6d, 0x4d },
{ 0x01, 0x6e, 0x4e },
{ 0x01, 0x6f, 0x4f },
{ 0x01, 0x70, 0x50 },
{ 0x01, 0x71, 0x51 },
{ 0x01, 0x72, 0x52 },
{ 0x01, 0x73, 0x53 },
{ 0x01, 0x74, 0x54 },
{ 0x01, 0x75, 0x55 },
{ 0x01, 0x76, 0x56 },
{ 0x01, 0x77, 0x57 },
{ 0x01, 0x78, 0x58 },
{ 0x01, 0x79, 0x59 },
{ 0x01, 0x7a, 0x5a },
{ 0x00, 0x5b, 0x5b },
{ 0x00, 0x5c, 0x5c },
{ 0x00, 0x5d, 0x5d },
{ 0x00, 0x5e, 0x5e },
{ 0x00, 0x5f, 0x5f },
{ 0x00, 0x60, 0x60 },
{ 0x00, 0x61, 0x41 },
{ 0x00, 0x62, 0x42 },
{ 0x00, 0x63, 0x43 },
{ 0x00, 0x64, 0x44 },
{ 0x00, 0x65, 0x45 },
{ 0x00, 0x66, 0x46 },
{ 0x00, 0x67, 0x47 },
{ 0x00, 0x68, 0x48 },
{ 0x00, 0x69, 0x49 },
{ 0x00, 0x6a, 0x4a },
{ 0x00, 0x6b, 0x4b },
{ 0x00, 0x6c, 0x4c },
{ 0x00, 0x6d, 0x4d },
{ 0x00, 0x6e, 0x4e },
{ 0x00, 0x6f, 0x4f },
{ 0x00, 0x70, 0x50 },
{ 0x00, 0x71, 0x51 },
{ 0x00, 0x72, 0x52 },
{ 0x00, 0x73, 0x53 },
{ 0x00, 0x74, 0x54 },
{ 0x00, 0x75, 0x55 },
{ 0x00, 0x76, 0x56 },
{ 0x00, 0x77, 0x57 },
{ 0x00, 0x78, 0x58 },
{ 0x00, 0x79, 0x59 },
{ 0x00, 0x7a, 0x5a },
{ 0x00, 0x7b, 0x7b },
{ 0x00, 0x7c, 0x7c },
{ 0x00, 0x7d, 0x7d },
{ 0x00, 0x7e, 0x7e },
{ 0x00, 0x7f, 0x7f },
{ 0x00, 0x80, 0x80 },
{ 0x00, 0x81, 0x81 },
{ 0x00, 0x82, 0x82 },
{ 0x00, 0x83, 0x83 },
{ 0x00, 0x84, 0x84 },
{ 0x00, 0x85, 0x85 },
{ 0x00, 0x86, 0x86 },
{ 0x00, 0x87, 0x87 },
{ 0x00, 0x88, 0x88 },
{ 0x00, 0x89, 0x89 },
{ 0x00, 0x8a, 0x8a },
{ 0x00, 0x8b, 0x8b },
{ 0x00, 0x8c, 0x8c },
{ 0x00, 0x8d, 0x8d },
{ 0x00, 0x8e, 0x8e },
{ 0x00, 0x8f, 0x8f },
{ 0x00, 0x90, 0x90 },
{ 0x00, 0x91, 0x91 },
{ 0x00, 0x92, 0x92 },
{ 0x00, 0x93, 0x93 },
{ 0x00, 0x94, 0x94 },
{ 0x00, 0x95, 0x95 },
{ 0x00, 0x96, 0x96 },
{ 0x00, 0x97, 0x97 },
{ 0x00, 0x98, 0x98 },
{ 0x00, 0x99, 0x99 },
{ 0x00, 0x9a, 0x9a },
{ 0x00, 0x9b, 0x9b },
{ 0x00, 0x9c, 0x9c },
{ 0x00, 0x9d, 0x9d },
{ 0x00, 0x9e, 0x9e },
{ 0x00, 0x9f, 0x9f },
{ 0x00, 0xa0, 0xa0 },
{ 0x00, 0xa1, 0xa1 },
{ 0x00, 0xa2, 0xa2 },
{ 0x00, 0xa3, 0xa3 },
{ 0x00, 0xa4, 0xa4 },
{ 0x00, 0xa5, 0xa5 },
{ 0x00, 0xa6, 0xa6 },
{ 0x00, 0xa7, 0xa7 },
{ 0x00, 0xa8, 0xa8 },
{ 0x00, 0xa9, 0xa9 },
{ 0x00, 0xaa, 0xaa },
{ 0x00, 0xab, 0xab },
{ 0x00, 0xac, 0xac },
{ 0x00, 0xad, 0xad },
{ 0x00, 0xae, 0xae },
{ 0x00, 0xaf, 0xaf },
{ 0x00, 0xb0, 0xb0 },
{ 0x00, 0xb1, 0xb1 },
{ 0x00, 0xb2, 0xb2 },
{ 0x00, 0xb3, 0xb3 },
{ 0x00, 0xb4, 0xb4 },
{ 0x00, 0xb5, 0xb5 },
{ 0x00, 0xb6, 0xb6 },
{ 0x00, 0xb7, 0xb7 },
{ 0x00, 0xb8, 0xb8 },
{ 0x00, 0xb9, 0xb9 },
{ 0x00, 0xba, 0xba },
{ 0x00, 0xbb, 0xbb },
{ 0x00, 0xbc, 0xbc },
{ 0x00, 0xbd, 0xbd },
{ 0x00, 0xbe, 0xbe },
{ 0x00, 0xbf, 0xbf },
{ 0x00, 0xc0, 0xc0 },
{ 0x00, 0xc1, 0xc1 },
{ 0x00, 0xc2, 0xc2 },
{ 0x00, 0xc3, 0xc3 },
{ 0x00, 0xc4, 0xc4 },
{ 0x00, 0xc5, 0xc5 },
{ 0x00, 0xc6, 0xc6 },
{ 0x00, 0xc7, 0xc7 },
{ 0x00, 0xc8, 0xc8 },
{ 0x00, 0xc9, 0xc9 },
{ 0x00, 0xca, 0xca },
{ 0x00, 0xcb, 0xcb },
{ 0x00, 0xcc, 0xcc },
{ 0x00, 0xcd, 0xcd },
{ 0x00, 0xce, 0xce },
{ 0x00, 0xcf, 0xcf },
{ 0x00, 0xd0, 0xd0 },
{ 0x00, 0xd1, 0xd1 },
{ 0x00, 0xd2, 0xd2 },
{ 0x00, 0xd3, 0xd3 },
{ 0x00, 0xd4, 0xd4 },
{ 0x00, 0xd5, 0xd5 },
{ 0x00, 0xd6, 0xd6 },
{ 0x00, 0xd7, 0xd7 },
{ 0x00, 0xd8, 0xd8 },
{ 0x00, 0xd9, 0xd9 },
{ 0x00, 0xda, 0xda },
{ 0x00, 0xdb, 0xdb },
{ 0x00, 0xdc, 0xdc },
{ 0x00, 0xdd, 0xdd },
{ 0x00, 0xde, 0xde },
{ 0x00, 0xdf, 0xdf },
{ 0x00, 0xe0, 0xe0 },
{ 0x00, 0xe1, 0xe1 },
{ 0x00, 0xe2, 0xe2 },
{ 0x00, 0xe3, 0xe3 },
{ 0x00, 0xe4, 0xe4 },
{ 0x00, 0xe5, 0xe5 },
{ 0x00, 0xe6, 0xe6 },
{ 0x00, 0xe7, 0xe7 },
{ 0x00, 0xe8, 0xe8 },
{ 0x00, 0xe9, 0xe9 },
{ 0x00, 0xea, 0xea },
{ 0x00, 0xeb, 0xeb },
{ 0x00, 0xec, 0xec },
{ 0x00, 0xed, 0xed },
{ 0x00, 0xee, 0xee },
{ 0x00, 0xef, 0xef },
{ 0x00, 0xf0, 0xf0 },
{ 0x00, 0xf1, 0xf1 },
{ 0x00, 0xf2, 0xf2 },
{ 0x00, 0xf3, 0xf3 },
{ 0x00, 0xf4, 0xf4 },
{ 0x00, 0xf5, 0xf5 },
{ 0x00, 0xf6, 0xf6 },
{ 0x00, 0xf7, 0xf7 },
{ 0x00, 0xf8, 0xf8 },
{ 0x00, 0xf9, 0xf9 },
{ 0x00, 0xfa, 0xfa },
{ 0x00, 0xfb, 0xfb },
{ 0x00, 0xfc, 0xfc },
{ 0x00, 0xfd, 0xfd },
{ 0x00, 0xfe, 0xfe },
{ 0x00, 0xff, 0xff }
};

struct enc_entry {
  const char * enc_name;
  struct cs_info * cs_table;
};

static struct enc_entry encds[] = {
  {"iso88591",iso1_tbl},                     //ISO-8859-1
  {"iso88592",iso2_tbl},                     //ISO-8859-2
  {"iso88593",iso3_tbl},                     //ISO-8859-3
  {"iso88594",iso4_tbl},                     //ISO-8859-4
  {"iso88595",iso5_tbl},                     //ISO-8859-5
  {"iso88596",iso6_tbl},                     //ISO-8859-6
  {"iso88597",iso7_tbl},                     //ISO-8859-7
  {"iso88598",iso8_tbl},                     //ISO-8859-8
  {"iso88599",iso9_tbl},                     //ISO-8859-9
  {"iso885910",iso10_tbl},                   //ISO-8859-10
  {"tis620",tis620_tbl},                     //TIS-620/ISO-8859-11
  {"tis6202533",tis620_tbl},                 //TIS-620/ISO-8859-11
  {"iso885911",tis620_tbl},                  //TIS-620/ISO-8859-11
  {"iso885913", iso13_tbl},                  //ISO-8859-13
  {"iso885914", iso14_tbl},                  //ISO-8859-14
  {"iso885915", iso15_tbl},                  //ISO-8859-15
  {"koi8r",koi8r_tbl},                       //KOI8-R
  {"koi8u",koi8u_tbl},                       //KOI8-U
  {"cp1251",cp1251_tbl},                     //CP-1251
  {"microsoftcp1251",cp1251_tbl},            //microsoft-cp1251
  {"xisciias", iscii_devanagari_tbl},        //x-iscii-as
  {"isciidevanagari", iscii_devanagari_tbl}  //ISCII-DEVANAGARI
};

/* map to lower case and remove non alphanumeric chars */
static void toAsciiLowerAndRemoveNonAlphanumeric( const char* pName, char* pBuf )
{
    while ( *pName )
    {
        /* A-Z */
        if ( (*pName >= 0x41) && (*pName <= 0x5A) )
        {
            *pBuf = (*pName)+0x20;  /* toAsciiLower */
            pBuf++;
        }
        /* a-z, 0-9 */
        else if ( ((*pName >= 0x61) && (*pName <= 0x7A)) ||
                  ((*pName >= 0x30) && (*pName <= 0x39)) )
        {
            *pBuf = *pName;
            pBuf++;
        }

        pName++;
    }

    *pBuf = '\0';
}

struct cs_info * get_current_cs(const char * es) {
  char *normalized_encoding = new char[strlen(es)+1];
  toAsciiLowerAndRemoveNonAlphanumeric(es, normalized_encoding);

  struct cs_info * ccs = NULL;
  int n = sizeof(encds) / sizeof(encds[0]);
  for (int i = 0; i < n; i++) {
    if (strcmp(normalized_encoding,encds[i].enc_name) == 0) {
      ccs = encds[i].cs_table;
      break;
    }
  }

  delete[] normalized_encoding;

  if (!ccs) {
    HUNSPELL_WARNING(stderr, "error: unknown encoding %s: using %s as fallback\n", es, encds[0].enc_name);
    ccs = encds[0].cs_table;
  }

  return ccs;
}
#else
// XXX This function was rewritten for mozilla. Instead of storing the
// conversion tables static in this file, create them when needed
// with help the mozilla backend.
struct cs_info * get_current_cs(const char * es) {
  struct cs_info *ccs;

  if (NS_FAILED(rv))
    return nsnull;
  encoder->SetOutputErrorBehavior(encoder->kOnError_Signal, nsnull, '?');
  rv = ccm->GetUnicodeDecoder(es, getter_AddRefs(decoder));
  if (NS_FAILED(rv))
    return nsnull;
  decoder->SetInputErrorBehavior(decoder->kOnError_Signal);

  if (NS_FAILED(rv))
    return nsnull;

  ccs = new cs_info[256];

  for (unsigned int i = 0; i <= 0xff; ++i) {
    PRBool success = PR_FALSE;
    // We want to find the upper/lowercase equivalents of each byte
    // in this 1-byte character encoding.  Call our encoding/decoding
    // APIs separately for each byte since they may reject some of the
    // bytes, and we want to handle errors separately for each byte.

    }
    *p = '\0';
#ifdef MOZILLA_CLIENT
    delete [] csconv;
#endif
    return mystrdup(expw);
}

// language to encoding default map

struct lang_map {
  const char * lang;
  int num;
};

static struct lang_map lang2enc[] = {
{"ar", LANG_ar},
{"az", LANG_az},
{"az_AZ", LANG_az}, // for back-compatibility
{"bg", LANG_bg},
{"ca", LANG_ca},
{"cs", LANG_cs},
{"da", LANG_da},
{"de", LANG_de},
{"el", LANG_el},
{"en", LANG_en},
{"es", LANG_es},
{"eu", LANG_eu},
{"gl", LANG_gl},
{"fr", LANG_fr},
{"hr", LANG_hr},
{"hu", LANG_hu},
{"hu_HU", LANG_hu}, // for back-compatibility
{"it", LANG_it},
{"la", LANG_la},
{"lv", LANG_lv},
{"nl", LANG_nl},
{"pl", LANG_pl},
{"pt", LANG_pt},
{"sv", LANG_sv},
{"tr", LANG_tr},
{"tr_TR", LANG_tr}, // for back-compatibility
{"ru", LANG_ru},
{"uk", LANG_uk}
};


const char * get_default_enc(const char * lang) {
  int n = sizeof(lang2enc) / sizeof(lang2enc[0]);
  for (int i = 0; i < n; i++) {
    if (strcmp(lang,lang2enc[i].lang) == 0) {
      return lang2enc[i].def_enc;
    }
  }
  return NULL;
}

int get_lang_num(const char * lang) {
  int n = sizeof(lang2enc) / sizeof(lang2enc[0]);
  for (int i = 0; i < n; i++) {
    if (strcmp(lang, lang2enc[i].lang) == 0) {
      return lang2enc[i].num;
    }
  }
  return LANG_xx;
}

#ifndef OPENOFFICEORG
#ifndef MOZILLA_CLIENT






// character encoding information
struct cs_info {
  unsigned char ccase;
  unsigned char clower;
  unsigned char cupper;
};

// Unicode character encoding information
struct unicode_info {
  unsigned short c;
  unsigned short cupper;
  unsigned short clower;
};

struct unicode_info2 {
  char cletter;
  unsigned short cupper;
  unsigned short clower;
};

LIBHUNSPELL_DLL_EXPORTED int initialize_utf_tbl();
LIBHUNSPELL_DLL_EXPORTED void free_utf_tbl();
LIBHUNSPELL_DLL_EXPORTED unsigned short unicodetoupper(unsigned short c, int langnum);
LIBHUNSPELL_DLL_EXPORTED unsigned short unicodetolower(unsigned short c, int langnum);
LIBHUNSPELL_DLL_EXPORTED int unicodeisalpha(unsigned short c);

struct enc_entry {
  const char * enc_name;
  struct cs_info * cs_table;
};

// language to encoding default map

struct lang_map {
  const char * lang;
  const char * def_enc;
  int num;
};

LIBHUNSPELL_DLL_EXPORTED struct cs_info * get_current_cs(const char * es);

LIBHUNSPELL_DLL_EXPORTED const char * get_default_enc(const char * lang);

// get language identifiers of language codes
LIBHUNSPELL_DLL_EXPORTED int get_lang_num(const char * lang);

// get characters of the given 8bit encoding with lower- and uppercase forms
LIBHUNSPELL_DLL_EXPORTED char * get_casechars(const char * enc);

// convert null terminated string to all caps using encoding
LIBHUNSPELL_DLL_EXPORTED void enmkallcap(char * d, const char * p, const char * encoding);


// hash entry macros
LIBHUNSPELL_DLL_EXPORTED inline char* HENTRY_DATA(struct hentry *h)
{
    char *ret;
    if (!h->var)
        ret = NULL;
    else if (h->var & H_OPT_ALIASM)
        ret = get_stored_pointer(HENTRY_WORD(h) + h->blen + 1);
    else 
        ret = HENTRY_WORD(h) + h->blen + 1;
    return ret;
}

// NULL-free version for warning-free OOo build
LIBHUNSPELL_DLL_EXPORTED inline const char* HENTRY_DATA2(const struct hentry *h)
{
    const char *ret;
    if (!h->var)
        ret = "";
    else if (h->var & H_OPT_ALIASM)
        ret = get_stored_pointer(HENTRY_WORD(h) + h->blen + 1);
    else
        ret = HENTRY_WORD(h) + h->blen + 1;
    return ret;
}

LIBHUNSPELL_DLL_EXPORTED inline char* HENTRY_FIND(struct hentry *h, const char *p)
{
    return (HENTRY_DATA(h) ? strstr(HENTRY_DATA(h), p) : NULL);
}






      hv ^= (*word++);
    }
    return (unsigned long) hv % tablesize;
}

int HashMgr::decode_flags(unsigned short ** result, char * flags, FileMgr * af) {
    int len;
    if (*flags == '\0') {
        HUNSPELL_WARNING(stderr, "error: line %d: bad flagvector\n", af->getlinenum());
        *result = NULL;
        return 0;
    }
    switch (flag_mode) {
      case FLAG_LONG: { // two-character flags (1x2yZz -> 1x 2y Zz)
        len = strlen(flags);
        if (len%2 == 1) HUNSPELL_WARNING(stderr, "error: line %d: bad flagvector\n", af->getlinenum());
        len /= 2;





    /* next set up the affix manager */
    /* it needs access to the hash manager lookup methods */
    pAMgr = new AffixMgr(affpath, pHMgr, &maxdic, key);

    /* get the preferred try string and the dictionary */
    /* encoding from the Affix Manager for that dictionary */
    char * try_string = pAMgr->get_try_string();
    encoding = pAMgr->get_encoding();
    csconv = get_current_cs(encoding);
    langnum = pAMgr->get_langnum();
    utf8 = pAMgr->get_utf8();
    if (!utf8)
        csconv = get_current_cs(encoding);
    complexprefixes = pAMgr->get_complexprefixes();
    wordbreak = pAMgr->get_breaktable();

    /* and finally set up the suggestion manager */
    pSMgr = new SuggestMgr(try_string, MAXSUGGESTION, pAMgr);
    if (try_string) free(try_string);
}


    } else break;
  }
  if ((i == wl) && (nstate == NNUM)) return 1;
  if (!info) info = &info2; else *info = 0;

  switch(captype) {
     case HUHCAP:
     case HUHINITCAP:
            *info += SPELL_ORIGCAP;
     case NOCAP: {
            rv = checkword(cw, info, root);
            if ((abbv) && !(rv)) {
                memcpy(wspace,cw,wl);
                *(wspace+wl) = '.';
                *(wspace+wl+1) = '\0';
                rv = checkword(wspace, info, root);
            }
            break;
         }
     case ALLCAP: {
            *info += SPELL_ORIGCAP;
            rv = checkword(cw, info, root);
            if (rv) break;
            if (abbv) {
                memcpy(wspace,cw,wl);
                *(wspace+wl) = '.';
                *(wspace+wl+1) = '\0';
                rv = checkword(wspace, info, root);
                if (rv) break;

                        *(wspace+wl2+1) = '\0';
                        rv = spellsharps(wspace, wspace, 0, 0, tmpword, info, root);
                    }
                }
                if (rv) break;
            }
        }
     case INITCAP: {
             *info += SPELL_ORIGCAP;
             wl = mkallsmall2(cw, unicw, nc);
             memcpy(wspace,cw,(wl+1));
             wl2 = mkinitcap2(cw, unicw, nc);
             if (captype == INITCAP) *info += SPELL_INITCAP;
             rv = checkword(cw, info, root);
             if (captype == INITCAP) *info -= SPELL_INITCAP;
             // forbid bad capitalization
             // (for example, ijs -> Ijs instead of IJs in Dutch)
             // use explicit forms in dic: Ijs/F (F = FORBIDDENWORD flag)
             if (*info & SPELL_FORBIDDEN) {
                rv = NULL;
                break;
             }
             if (rv && is_keepcase(rv) && (captype == ALLCAP)) rv = NULL;

                   // in INITCAP form, too.
                   !(pAMgr->get_checksharps() &&
                      ((utf8 && strstr(wspace, "\xC3\x9F")) ||
                      (!utf8 && strchr(wspace, '\xDF')))))) rv = NULL;
             break;
           }
  }

  if (rv) {
      if (pAMgr && pAMgr->get_warn() && rv->astr &&
          TESTAFF(rv->astr, pAMgr->get_warn(), rv->alen)) {
              *info += SPELL_WARN;
	      if (pAMgr->get_forbidwarn()) return 0;
              return HUNSPELL_OK_WARN;
      }
      return HUNSPELL_OK;
  }

  // recursive breaking at break points
  if (wordbreak) {
    char * s;
    char r;
    int nbr = 0;
    wl = strlen(cw);
    int numbreak = pAMgr ? pAMgr->get_numbreak() : 0;

        unsigned short * ignoredchars_utf16 = pAMgr->get_ignore_utf16(&ignoredchars_utf16_len);
        remove_ignored_chars_utf(w2, ignoredchars_utf16, ignoredchars_utf16_len);
     } else {
        remove_ignored_chars(w2,ignoredchars);
     }
     word = w2;
  } else word = w;

  len = strlen(word);

  if (!len)
      return NULL;

  // word reversing wrapper for complex prefixes
  if (complexprefixes) {
    if (word != w2) {
      strcpy(w2, word);
      word = w2;
    }
    if (utf8) reverseword_utf(w2); else reverseword(w2);
  }

       (pAMgr->get_onlyincompound() && TESTAFF(he->astr, pAMgr->get_onlyincompound(), he->alen)) ||
       (info && (*info & SPELL_INITCAP) && TESTAFF(he->astr, ONLYUPCASEFLAG, he->alen))
    )) he = he->next_homonym;
  }

  // check with affixes
  if (!he && pAMgr) {
     // try stripping off affixes */
     len = strlen(word);
     he = pAMgr->affix_check(word, len, 0);

     // check compound restriction and onlyupcase
     if (he && he->astr && (
        (pAMgr->get_onlyincompound() &&
    	    TESTAFF(he->astr, pAMgr->get_onlyincompound(), he->alen)) ||
        (info && (*info & SPELL_INITCAP) &&
    	    TESTAFF(he->astr, ONLYUPCASEFLAG, he->alen)))) {

        if (root) {
            *root = mystrdup(he->word);
            if (*root && complexprefixes) {
                if (utf8) reverseword_utf(*root); else reverseword(*root);
            }
        }
     // try check compound word
     } else if (pAMgr->get_compound()) {
          he = pAMgr->compound_check(word, len, 0, 0, 100, 0, NULL, 0, 0, info);
          // LANG_hu section: `moving rule' with last dash
          if ((!he) && (langnum == LANG_hu) && (word[len-1] == '-')) {
             char * dup = mystrdup(word);
             if (!dup) return NULL;
             dup[len-1] = '\0';
             he = pAMgr->compound_check(dup, len-1, -5, 0, 100, 0, NULL, 1, 0, info);
             free(dup);
          }
          // end of LANG speficic region
          if (he) {
                if (root) {
                    *root = mystrdup(he->word);
                    if (*root && complexprefixes) {
                        if (utf8) reverseword_utf(*root); else reverseword(*root);

  RepList * rl = (pAMgr) ? pAMgr->get_iconvtable() : NULL;
  if (rl && rl->conv(word, wspace)) wl = cleanword2(cw, wspace, unicw, &nc, &captype, &abbv);
  else wl = cleanword2(cw, word, unicw, &nc, &captype, &abbv);

  if (wl == 0) return 0;
  int ns = 0;
  int capwords = 0;

  // check capitalized form for FORCEUCASE
  if (pAMgr && captype == NOCAP && pAMgr->get_forceucase()) {
    int info = SPELL_ORIGCAP;
    char ** wlst;
    if (checkword(cw, &info, NULL)) {
        if (*slst) {
            wlst = *slst;
        } else {
            wlst = (char **) malloc(MAXSUGGESTION * sizeof(char *));
            if (wlst == NULL) return -1;
            *slst = wlst;
            for (int i = 0; i < MAXSUGGESTION; i++) {
                wlst[i] = NULL;
            }
        }
        wlst[0] = mystrdup(cw);
        mkinitcap(wlst[0]);
        return 1;
    }
  }
 
  switch(captype) {
     case NOCAP:   {
                     ns = pSMgr->suggest(slst, cw, ns, &onlycmpdsug);
                     break;
                   }

     case INITCAP: {
                     capwords = 1;

                                }
                            }
                        }
                     }
                     break;
                   }
  }

 // LANG_hu section: replace '-' with ' ' in Hungarian
  if (langnum == LANG_hu) {
      for (int j=0; j < ns; j++) {
          char * pos = strchr((*slst)[j],'-');
          if (pos) {
              int info;
              char w[MAXWORDUTF8LEN];
              *pos = '\0';
              strcpy(w, (*slst)[j]);

              if ((info & SPELL_COMPOUND) && (info & SPELL_FORBIDDEN)) {
                  *pos = ' ';
              } else *pos = '-';
          }
      }
  }
  // END OF LANG_hu section

  // try ngram approach since found nothing or only compound words
  if (pAMgr && (ns == 0 || onlycmpdsug) && (pAMgr->get_maxngramsugs() != 0) && (*slst)) {
      switch(captype) {
          case NOCAP: {
              ns = pSMgr->ngsuggest(*slst, cw, ns, pHMgr, maxdic);
              break;
          }
	  case HUHINITCAP:
              capwords = 1;
          case HUHCAP: {

  }

  if (*result) {
    // word reversing wrapper for complex prefixes
    if (complexprefixes) {
      if (utf8) reverseword_utf(result); else reverseword(result);
    }
    return line_tok(result, slst, MSEP_REC);

  }

  // compound word with dash (HU) I18n
  char * dash = NULL;
  int nresult = 0;
  // LANG_hu section: set dash information for suggestions
  if (langnum == LANG_hu) dash = (char *) strchr(cw,'-');
  if ((langnum == LANG_hu) && dash) {






#include "hunvisapi.h"

#include "hashmgr.hxx"
#include "affixmgr.hxx"
#include "suggestmgr.hxx"
#include "langnum.hxx"

#define  SPELL_COMPOUND  (1 << 0)
#define  SPELL_FORBIDDEN (1 << 1)
#define  SPELL_ALLCAP    (1 << 2)
#define  SPELL_NOCAP     (1 << 3)
#define  SPELL_INITCAP   (1 << 4)

#define  SPELL_XML "<?xml?>"

#define MAXDIC 20
#define MAXSUGGESTION 15
#define MAXSHARPS 5

#define HUNSPELL_OK       (1 << 0)
#define HUNSPELL_OK_WARN  (1 << 1)

#ifndef _MYSPELLMGR_HXX_
#define _MYSPELLMGR_HXX_

class LIBHUNSPELL_DLL_EXPORTED Hunspell
{
  AffixMgr*       pAMgr;
  HashMgr*        pHMgr[MAXDIC];
  int             maxdic;





    return dat[n];
}

int RepList::near(const char * word) {
    int p1 = 0;
    int p2 = pos;
    while ((p2 - p1) > 1) {
      int m = (p1 + p2) / 2;
//      fprintf(stderr, "m: %d p1: %d p2: %d dat: %s\n", m, p1, p2, dat[m]->pattern);
      int c = strcmp(word, dat[m]->pattern);
      if (c <= 0) {
        if (c < 0) p2 = m; else p1 = p2 = m;
      } else p1 = m;
    }
//    fprintf(stderr, "NEAR: %s (word: %s)\n", dat[p1]->pattern, word);
    return p1;
}

int RepList::match(const char * word, int n) {
    if (strncmp(word, dat[n]->pattern, strlen(dat[n]->pattern)) == 0) return strlen(dat[n]->pattern);
    return 0;
}

int RepList::add(char * pat1, char * pat2) {
    if (pos >= size || pat1 == NULL || pat2 == NULL) return 1;
    replentry * r = (replentry *) malloc(sizeof(replentry));
    if (r == NULL) return 1;
    r->pattern = mystrrep(pat1, "_", " ");
    r->pattern2 = mystrrep(pat2, "_", " ");
    r->start = false;
    r->end = false;
    dat[pos++] = r;
    for (int i = pos - 1; i > 0; i--) {
      r = dat[i];
      if (strcmp(r->pattern, dat[i - 1]->pattern) < 0) {
          dat[i] = dat[i - 1];
          dat[i - 1] = r;
      } else break;
    }
    return 0;
}

int RepList::conv(const char * word, char * dest) {
    int stl = 0;
    int change = 0;
//    for (int i = 0; i < pos; i++) fprintf(stderr, "%d. %s\n", i, dat[i]->pattern);
    for (size_t i = 0; i < strlen(word); i++) {
        int n = near(word + i);
        int l = match(word + i, n);
        if (l) {
          strcpy(dest + stl, dat[n]->pattern2);
          stl += strlen(dat[n]->pattern2);
          i += l - 1;
          change = 1;
        } else dest[stl++] = word[i];
    }
    dest[stl] = '\0';
//    fprintf(stderr, "i: %s o: %s change: %d\n", word, dest, change);
    return change;
}






  utf8 = 0;
  langnum = 0;
  complexprefixes = 0;  
  
  maxSug = maxn;
  nosplitsugs = 0;
  maxngramsugs = MAXNGRAMSUGS;
  maxcpdsugs = MAXCOMPOUNDSUGS;

  if (pAMgr) {
        char * enc = pAMgr->get_encoding();
        csconv = get_current_cs(enc);
        free(enc);
        langnum = pAMgr->get_langnum();
        ckey = pAMgr->get_key_string();
        nosplitsugs = pAMgr->get_nosplitsugs();
        if (pAMgr->get_maxngramsugs() >= 0)
            maxngramsugs = pAMgr->get_maxngramsugs();
        utf8 = pAMgr->get_utf8();
	if (pAMgr->get_maxcpdsugs() >= 0)
	    maxcpdsugs = pAMgr->get_maxcpdsugs();
        if (!utf8)
        {
            char * enc = pAMgr->get_encoding();
            csconv = get_current_cs(enc);
            free(enc);
        }
        complexprefixes = pAMgr->get_complexprefixes();
  }

  if (ckey) {  
    if (utf8) {
        w_char t[MAXSWL];    
        ckeyl = u8_u16(t, MAXSWL, ckey);
        ckey_utf = (w_char *) malloc(ckeyl * sizeof(w_char));


int SuggestMgr::suggest(char*** slst, const char * w, int nsug,
    int * onlycompoundsug)
{
  int nocompoundtwowords = 0;
  char ** wlst;    
  w_char word_utf[MAXSWL];
  int wl = 0;
  int nsugorig = nsug;
  char w2[MAXWORDUTF8LEN];
  const char * word = w;
  int oldSug = 0;

  // word reversing wrapper for complex prefixes
  if (complexprefixes) {
    strcpy(w2, w);
    if (utf8) reverseword_utf(w2); else reverseword(w2);
    word = w2;
  }
    

        if (wlst == NULL) return -1;
        for (int i = 0; i < maxSug; i++) {
            wlst[i] = NULL;
        }
    }
    
    if (utf8) {
        wl = u8_u16(word_utf, MAXSWL, word);
	if (wl == -1) {
    		*slst = wlst;
		 return nsug;
	}
    }

    for (int cpdsuggest=0; (cpdsuggest<2) && (nocompoundtwowords==0); cpdsuggest++) {

    // limit compound suggestion
    if (cpdsuggest > 0) oldSug = nsug;

    // suggestions for an uppercase word (html -> HTML)
    if ((nsug < maxSug) && (nsug > -1)) {
        nsug = (utf8) ? capchars_utf(wlst, word_utf, wl, nsug, cpdsuggest) :
                    capchars(wlst, word, nsug, cpdsuggest);
    }

    // perhaps we made a typical fault of spelling
    if ((nsug < maxSug) && (nsug > -1) && (!cpdsuggest || (nsug < oldSug + maxcpdsugs))) {
      nsug = replchars(wlst, word, nsug, cpdsuggest);
    }

    // perhaps we made chose the wrong char from a related set
    if ((nsug < maxSug) && (nsug > -1) && (!cpdsuggest || (nsug < oldSug + maxcpdsugs))) {
      nsug = mapchars(wlst, word, nsug, cpdsuggest);
    }

    // only suggest compound words when no other suggestion
    if ((cpdsuggest == 0) && (nsug > nsugorig)) nocompoundtwowords=1;

    // did we swap the order of chars by mistake
    if ((nsug < maxSug) && (nsug > -1) && (!cpdsuggest || (nsug < oldSug + maxcpdsugs))) {
        nsug = (utf8) ? swapchar_utf(wlst, word_utf, wl, nsug, cpdsuggest) :
                    swapchar(wlst, word, nsug, cpdsuggest);
    }

    // did we swap the order of non adjacent chars by mistake
    if ((nsug < maxSug) && (nsug > -1) && (!cpdsuggest || (nsug < oldSug + maxcpdsugs))) {
        nsug = (utf8) ? longswapchar_utf(wlst, word_utf, wl, nsug, cpdsuggest) :
                    longswapchar(wlst, word, nsug, cpdsuggest);
    }

    // did we just hit the wrong key in place of a good char (case and keyboard)
    if ((nsug < maxSug) && (nsug > -1) && (!cpdsuggest || (nsug < oldSug + maxcpdsugs))) {
        nsug = (utf8) ? badcharkey_utf(wlst, word_utf, wl, nsug, cpdsuggest) :
                    badcharkey(wlst, word, nsug, cpdsuggest);
    }

    // only suggest compound words when no other suggestion
    if ((cpdsuggest == 0) && (nsug > 0)) nocompoundtwowords=1;

    // did we add a char that should not be there
    if ((nsug < maxSug) && (nsug > -1) && (!cpdsuggest || (nsug < oldSug + maxcpdsugs))) {
        nsug = (utf8) ? extrachar_utf(wlst, word_utf, wl, nsug, cpdsuggest) :
                    extrachar(wlst, word, nsug, cpdsuggest);
    }


    // did we forgot a char
    if ((nsug < maxSug) && (nsug > -1) && (!cpdsuggest || (nsug < oldSug + maxcpdsugs))) {
        nsug = (utf8) ? forgotchar_utf(wlst, word_utf, wl, nsug, cpdsuggest) :
                    forgotchar(wlst, word, nsug, cpdsuggest);
    }

    // did we move a char
    if ((nsug < maxSug) && (nsug > -1) && (!cpdsuggest || (nsug < oldSug + maxcpdsugs))) {
        nsug = (utf8) ? movechar_utf(wlst, word_utf, wl, nsug, cpdsuggest) :
                    movechar(wlst, word, nsug, cpdsuggest);
    }

    // did we just hit the wrong key in place of a good char
    if ((nsug < maxSug) && (nsug > -1) && (!cpdsuggest || (nsug < oldSug + maxcpdsugs))) {
        nsug = (utf8) ? badchar_utf(wlst, word_utf, wl, nsug, cpdsuggest) :
                    badchar(wlst, word, nsug, cpdsuggest);
    }

    // did we double two characters
    if ((nsug < maxSug) && (nsug > -1) && (!cpdsuggest || (nsug < oldSug + maxcpdsugs))) {
        nsug = (utf8) ? doubletwochars_utf(wlst, word_utf, wl, nsug, cpdsuggest) :
                    doubletwochars(wlst, word, nsug, cpdsuggest);
    }

    // perhaps we forgot to hit space and two words ran together
    if (!nosplitsugs && (nsug < maxSug) && (nsug > -1) && (!cpdsuggest || (nsug < oldSug + maxcpdsugs))) {
        nsug = twowords(wlst, word, nsug, cpdsuggest);
    }

    } // repeating ``for'' statement compounding support

    if (nsug < 0) {
     // we ran out of memory - we should free up as much as possible
       for (int i = 0; i < maxSug; i++)
         if (wlst[i] != NULL) free(wlst[i]);
       free(wlst);
       wlst = NULL;
    }

    if (!nocompoundtwowords && (nsug > 0) && onlycompoundsug) *onlycompoundsug = 1;

    *slst = wlst;
    return nsug;
}

// generate suggestions for a word with typical mistake
//    pass in address of array of char * pointers
#ifdef HUNSPELL_EXPERIMENTAL
int SuggestMgr::suggest_auto(char*** slst, const char * w, int nsug)
{
    int nocompoundtwowords = 0;
    char ** wlst;
    int oldSug;

  char w2[MAXWORDUTF8LEN];
  const char * word = w;

  // word reversing wrapper for complex prefixes
  if (complexprefixes) {
    strcpy(w2, w);
    if (utf8) reverseword_utf(w2); else reverseword(w2);

        wlst = *slst;
    } else {
        wlst = (char **) malloc(maxSug * sizeof(char *));
        if (wlst == NULL) return -1;
    }

    for (int cpdsuggest=0; (cpdsuggest<2) && (nocompoundtwowords==0); cpdsuggest++) {

    // limit compound suggestion
    if (cpdsuggest > 0) oldSug = nsug;

    // perhaps we made a typical fault of spelling
    if ((nsug < maxSug) && (nsug > -1))
    nsug = replchars(wlst, word, nsug, cpdsuggest);

    // perhaps we made chose the wrong char from a related set
    if ((nsug < maxSug) && (nsug > -1) && (!cpdsuggest || (nsug < oldSug + maxcpdsugs)))
      nsug = mapchars(wlst, word, nsug, cpdsuggest);

    if ((cpdsuggest==0) && (nsug>0)) nocompoundtwowords=1;

    // perhaps we forgot to hit space and two words ran together

    if ((nsug < maxSug) && (nsug > -1) && (!cpdsuggest || (nsug < oldSug + maxcpdsugs)) && check_forbidden(word, strlen(word))) {
                nsug = twowords(wlst, word, nsug, cpdsuggest);
        }
    
    } // repeating ``for'' statement compounding support

    if (nsug < 0) {
       for (int i=0;i<maxSug; i++)
         if (wlst[i] != NULL) free(wlst[i]);

  int numrep = pAMgr->get_numrep();
  struct replentry* reptable = pAMgr->get_reptable();
  if (reptable==NULL) return ns;
  for (int i=0; i < numrep; i++ ) {
      r = word;
      lenr = strlen(reptable[i].pattern2);
      lenp = strlen(reptable[i].pattern);
      // search every occurence of the pattern in the word
      while ((r=strstr(r, reptable[i].pattern)) != NULL && (!reptable[i].end || strlen(r) == strlen(reptable[i].pattern)) &&
        (!reptable[i].start || r == word)) {
          strcpy(candidate, word);
          if (r-word + lenr + strlen(r+lenp) >= MAXSWUTF8L) break;
          strcpy(candidate+(r-word),reptable[i].pattern2);
          strcpy(candidate+(r-word)+lenr, r+lenp);
          ns = testsug(wlst, candidate, wl-lenp+lenr, ns, cpdsuggest, NULL, NULL);
          if (ns == -1) return -1;
          // check REP suggestions with space
          char * sp = strchr(candidate, ' ');
          if (sp) {
            char * prev = candidate;
            while (sp) {
              *sp = '\0';
              if (checkword(prev, strlen(prev), 0, NULL, NULL)) {
                int oldns = ns;
                *sp = ' ';
                ns = testsug(wlst, sp + 1, strlen(sp + 1), ns, cpdsuggest, NULL, NULL);
                if (ns == -1) return -1;
                if (oldns < ns) {
                  free(wlst[ns - 1]);
                  wlst[ns - 1] = mystrdup(candidate);
                  if (!wlst[ns - 1]) return -1;
                }
              }
              *sp = ' ';
              prev = sp + 1;
              sp = strchr(prev, ' ');
            }
                free(wlst[ns - 1]);
                wlst[ns - 1] = mystrdup(candidate);
                if (!wlst[ns - 1]) return -1;
              }
            }            
            *sp = ' ';
          }
          r++; // search for the next letter
      }
   }
   return ns;
}

// perhaps we doubled two characters (pattern aba -> ababa, for example vacation -> vacacation)

  for (i = 0; i < MAX_ROOTS; i++) {
    roots[i] = NULL;
    scores[i] = -100 * i;
    rootsphon[i] = NULL;
    scoresphon[i] = -100 * i;
  }
  lp = MAX_ROOTS - 1;
  lpphon = MAX_ROOTS - 1;
  scphon = -20000;
  int low = NGRAM_LOWERING;
  
  char w2[MAXWORDUTF8LEN];
  char f[MAXSWUTF8L];
  char * word = w;

  // word reversing wrapper for complex prefixes
  if (complexprefixes) {
    strcpy(w2, w);


  char mw[MAXSWUTF8L];
  w_char u8[MAXSWL];
  int nc = strlen(word);
  int n = (utf8) ? u8_u16(u8, MAXSWL, word) : nc;
  
  // set character based ngram suggestion for words with non-BMP Unicode characters
  if (n == -1) {
    utf8 = 0; // XXX not state-free
    n = nc;
    nonbmp = 1;
    low = 0;
  }

  struct hentry* hp = NULL;
  int col = -1;
  phonetable * ph = (pAMgr) ? pAMgr->get_phonetable() : NULL;
  char target[MAXSWUTF8L];
  char candidate[MAXSWUTF8L];
  if (ph) {
    if (utf8) {
      w_char _w[MAXSWL];
      int _wl = u8_u16(_w, MAXSWL, word);
      mkallcap_utf(_w, _wl, langnum);
      u16_u8(candidate, MAXSWUTF8L, _w, _wl);
    } else {
      strcpy(candidate, word);
      if (!nonbmp) mkallcap(candidate, csconv);
    }
    phonet(candidate, target, nc, *ph); // XXX phonet() is 8-bit (nc, not n)
  }

  FLAG forbiddenword = pAMgr ? pAMgr->get_forbiddenword() : FLAG_NULL;
  FLAG nosuggest = pAMgr ? pAMgr->get_nosuggest() : FLAG_NULL;
  FLAG nongramsuggest = pAMgr ? pAMgr->get_nongramsuggest() : FLAG_NULL;
  FLAG onlyincompound = pAMgr ? pAMgr->get_onlyincompound() : FLAG_NULL;

  for (i = 0; i < md; i++) {  
  while (0 != (hp = (pHMgr[i])->walk_hashtable(col, hp))) {
    if ((hp->astr) && (pAMgr) && 
       (TESTAFF(hp->astr, forbiddenword, hp->alen) ||
          TESTAFF(hp->astr, ONLYUPCASEFLAG, hp->alen) ||
          TESTAFF(hp->astr, nosuggest, hp->alen) ||
          TESTAFF(hp->astr, nongramsuggest, hp->alen) ||
          TESTAFF(hp->astr, onlyincompound, hp->alen))) continue;

    sc = ngram(3, word, HENTRY_WORD(hp), NGRAM_LONGER_WORSE + low) +
	leftcommonsubstring(word, HENTRY_WORD(hp));

    // check special pronounciation
    if ((hp->var & H_OPT_PHON) && copy_field(f, HENTRY_DATA(hp), MORPH_PHON)) {
	int sc2 = ngram(3, word, f, NGRAM_LONGER_WORSE + low) +
		+ leftcommonsubstring(word, f);
	if (sc2 > sc) sc = sc2;
    }
    
    scphon = -20000;
    if (ph && (sc > 2) && (abs(n - (int) hp->clen) <= 3)) {
      char target2[MAXSWUTF8L];
      if (utf8) {
        w_char _w[MAXSWL];
        int _wl = u8_u16(_w, MAXSWL, HENTRY_WORD(hp));
        mkallcap_utf(_w, _wl, langnum);
        u16_u8(candidate, MAXSWUTF8L, _w, _wl);
      } else {
        strcpy(candidate, HENTRY_WORD(hp));
        mkallcap(candidate, csconv);
      }
      phonet(candidate, target2, -1, *ph);
      scphon = 2 * ngram(3, target, target2, NGRAM_LONGER_WORSE);
    }

    if (sc > scores[lp]) {
      scores[lp] = sc;  
      roots[lp] = hp;
      lval = sc;
      for (j=0; j < MAX_ROOTS; j++)
        if (scores[j] < lval) {
          lp = j;
          lval = scores[j];
        }
    }


    if (scphon > scoresphon[lpphon]) {
      scoresphon[lpphon] = scphon;
      rootsphon[lpphon] = HENTRY_WORD(hp);
      lval = scphon;
      for (j=0; j < MAX_ROOTS; j++)
        if (scoresphon[j] < lval) {
          lpphon = j;
          lval = scoresphon[j];

  // find minimum threshold for a passable suggestion
  // mangle original word three differnt ways
  // and score them to generate a minimum acceptable score
  int thresh = 0;
  for (int sp = 1; sp < 4; sp++) {
     if (utf8) {
       for (int k=sp; k < n; k+=4) *((unsigned short *) u8 + k) = '*';
       u16_u8(mw, MAXSWUTF8L, u8, n);
       thresh = thresh + ngram(n, word, mw, NGRAM_ANY_MISMATCH + low);
     } else {
       strcpy(mw, word);
       for (int k=sp; k < n; k+=4) *(mw + k) = '*';
       thresh = thresh + ngram(n, word, mw, NGRAM_ANY_MISMATCH + low);
     }
  }
  thresh = thresh / 3;
  thresh--;

 // now expand affixes on each of these root words and
  // and use length adjusted ngram scores to select
  // possible suggestions
  char * guess[MAX_GUESS];
  char * guessorig[MAX_GUESS];
  int gscore[MAX_GUESS];
  for(i=0;i<MAX_GUESS;i++) {
     guess[i] = NULL;
     guessorig[i] = NULL;

  for (i = 0; i < MAX_ROOTS; i++) {
      if (roots[i]) {
        struct hentry * rp = roots[i];
        int nw = pAMgr->expand_rootword(glst, MAX_WORDS, HENTRY_WORD(rp), rp->blen,
            	    rp->astr, rp->alen, word, nc, 
                    ((rp->var & H_OPT_PHON) ? copy_field(f, HENTRY_DATA(rp), MORPH_PHON) : NULL));

        for (int k = 0; k < nw ; k++) {
           sc = ngram(n, word, glst[k].word, NGRAM_ANY_MISMATCH + low) +
               leftcommonsubstring(word, glst[k].word);

           if (sc > thresh) {
              if (sc > gscore[lp]) {
                 if (guess[lp]) {
                    free (guess[lp]);
                    if (guessorig[lp]) {
                	free(guessorig[lp]);
                	guessorig[lp] = NULL;
            	    }
                 }

                       lp = j;
                       lval = gscore[j];
                    }
              } else { 
                free(glst[k].word);
                if (glst[k].orig) free(glst[k].orig);
              }
           } else {
                free(glst[k].word);
                if (glst[k].orig) free(glst[k].orig);
           }
        }
      }
  }
  free(glst);

  // now we are done generating guesses

  
  
  bubblesort(&guess[0], &guessorig[0], &gscore[0], MAX_GUESS);
  if (ph) bubblesort(&rootsphon[0], NULL, &scoresphon[0], MAX_ROOTS);

  // weight suggestions with a similarity index, based on
  // the longest common subsequent algorithm and resort

  int is_swap = 0;
  int re = 0;
  double fact = 1.0;
  if (pAMgr) {
	int maxd = pAMgr->get_maxdiff();
	if (maxd >= 0) fact = (10.0 - maxd)/5.0;
  }

  for (i=0; i < MAX_GUESS; i++) {
      if (guess[i]) {
        // lowering guess[i]
        char gl[MAXSWUTF8L];
        int len;
        if (utf8) {
          w_char _w[MAXSWL];
          len = u8_u16(_w, MAXSWL, guess[i]);
          mkallsmall_utf(_w, len, langnum);
          u16_u8(gl, MAXSWUTF8L, _w, len);
        } else {
          strcpy(gl, guess[i]);
          if (!nonbmp) mkallsmall(gl, csconv);
          len = strlen(guess[i]);
        }

        int _lcs = lcslen(word, gl);

        // same characters with different casing
        if ((n == len) && (n == _lcs)) {
            gscore[i] += 2000;
            break;
        }
        // using 2-gram instead of 3, and other weightening
        gscore[i] =
        gscore[i] +=
          // length of longest common subsequent minus length difference
          2 * _lcs - abs((int) (n - len)) +
          // weight length of the left common substring
          leftcommonsubstring(word, gl) +
          // weight equal character positions
          (!nonbmp && commoncharacterpositions(word, gl, &is_swap) ? 1: 0) +
          // swap character (not neighboring)
          ((is_swap) ? 10 : 0) +
          // ngram
          ngram(4, word, gl, NGRAM_ANY_MISMATCH + low) +
          // weighted ngrams
          (re = ngram(2, word, gl, NGRAM_ANY_MISMATCH + low + NGRAM_WEIGHTED)) +
          (re += ngram(2, gl, word, NGRAM_ANY_MISMATCH + low + NGRAM_WEIGHTED)) +
          // different limit for dictionaries with PHONE rules
          (ph ? (re < len * fact ? -1000 : 0) : (re < (n + len)*fact? -1000 : 0));
      }
  }

  bubblesort(&guess[0], &guessorig[0], &gscore[0], MAX_GUESS);

// phonetic version
  if (ph) for (i=0; i < MAX_ROOTS; i++) {
      if (rootsphon[i]) {

        int len;
        if (utf8) {
          w_char _w[MAXSWL];
          len = u8_u16(_w, MAXSWL, rootsphon[i]);
          mkallsmall_utf(_w, len, langnum);
          u16_u8(gl, MAXSWUTF8L, _w, len);
        } else {
          strcpy(gl, rootsphon[i]);
          if (!nonbmp) mkallsmall(gl, csconv);
          len = strlen(rootsphon[i]);
        }

        // heuristic weigthing of ngram scores
        scoresphon[i] += 2 * lcslen(word, gl) - abs((int) (n - len)) +
          // weight length of the left common substring
          leftcommonsubstring(word, gl);
      }

  int oldns = ns;

  int same = 0;
  for (i=0; i < MAX_GUESS; i++) {
    if (guess[i]) {
      if ((ns < oldns + maxngramsugs) && (ns < maxSug) && (!same || (gscore[i] > 1000))) {
        int unique = 1;
        // leave only excellent suggestions, if exists
        if (gscore[i] > 1000) same = 1; else if (gscore[i] < -100) {
            same = 1;
	    // keep the best ngram suggestions, unless in ONLYMAXDIFF mode
            if (ns > oldns || (pAMgr && pAMgr->get_onlymaxdiff())) {
    	        free(guess[i]);
    	        if (guessorig[i]) free(guessorig[i]);
                continue;
            }
        }
        for (j = 0; j < ns; j++) {
          // don't suggest previous suggestions or a previous suggestion with prefixes or affixes
          if ((!guessorig[i] && strstr(guess[i], wlst[j])) ||
	     (guessorig[i] && strstr(guessorig[i], wlst[j])) ||
            // check forbidden words
            !checkword(guess[i], strlen(guess[i]), 0, NULL, NULL)) unique = 0;
        }
        if (unique) {

      }
    }
  }

  oldns = ns;
  if (ph) for (i=0; i < MAX_ROOTS; i++) {
    if (rootsphon[i]) {
      if ((ns < oldns + MAXPHONSUGS) && (ns < maxSug)) {
	int unique = 1;
        for (j = 0; j < ns; j++) {
          // don't suggest previous suggestions or a previous suggestion with prefixes or affixes
          if (strstr(rootsphon[i], wlst[j]) || 
            // check forbidden words
            !checkword(rootsphon[i], strlen(rootsphon[i]), 0, NULL, NULL)) unique = 0;
        }
        if (unique) {
            wlst[ns++] = mystrdup(rootsphon[i]);

// needs to check both root words and words with affixes

// obsolote MySpell-HU modifications:
// return value 2 and 3 marks compounding with hyphen (-)
// `3' marks roots without suffix
int SuggestMgr::checkword(const char * word, int len, int cpdsuggest, int * timer, clock_t * timelimit)
{
  struct hentry * rv=NULL;
  struct hentry * rv2=NULL;
  int nosuffix = 0;

  // check time limit
  if (timer) {
    (*timer)--;
    if (!(*timer) && timelimit) {
      if ((clock() - *timelimit) > TIMELIMIT) return 0;
      *timer = MAXPLUSTIMER;
    }
  }
  
  if (pAMgr) { 
    if (cpdsuggest==1) {
      if (pAMgr->get_compound()) {
        rv = pAMgr->compound_check(word, len, 0, 0, 100, 0, NULL, 0, 1, 0); //EXT
        if (rv && (!(rv2 = pAMgr->lookup(word)) || !rv2->astr || 
            !(TESTAFF(rv2->astr,pAMgr->get_forbiddenword(),rv2->alen) ||
            TESTAFF(rv2->astr,pAMgr->get_nosuggest(),rv2->alen)))) return 3; // XXX obsolote categorisation + only ICONV needs affix flag check?
        }
        return 0;
    }

    rv = pAMgr->lookup(word);

    if (rv) {
        if ((rv->astr) && (TESTAFF(rv->astr,pAMgr->get_forbiddenword(),rv->alen)


// generate an n-gram score comparing s1 and s2
int SuggestMgr::ngram(int n, char * s1, const char * s2, int opt)
{
  int nscore = 0;
  int ns;
  int l1;
  int l2;
  int test = 0;

  if (utf8) {
    w_char su1[MAXSWL];
    w_char su2[MAXSWL];
    l1 = u8_u16(su1, MAXSWL, s1);
    l2 = u8_u16(su2, MAXSWL, s2);
    if ((l2 <= 0) || (l1 == -1)) return 0;
    // lowering dictionary word
    if (opt & NGRAM_LOWERING) mkallsmall_utf(su2, l2, langnum);
    for (int j = 1; j <= n; j++) {
      ns = 0;
      for (int i = 0; i <= (l1-j); i++) {
	int k = 0;
        for (int l = 0; l <= (l2-j); l++) {
            for (k = 0; k < j; k++) {
            for (k = 0; (k < j); k++) {
              w_char * c1 = su1 + i + k;
              w_char * c2 = su2 + l + k;
              if ((c1->l != c2->l) || (c1->h != c2->h)) break;
            }
            if (k == j) {
		ns++;
                break;
            } 
	}
	if (k != j && opt & NGRAM_WEIGHTED) {
	  ns--;
	  test++;
	  if (i == 0 || i == l1-j) ns--; // side weight
	}
      }
      nscore = nscore + ns;
      if (ns < 2 && !(opt & NGRAM_WEIGHTED)) break;
    }
  } else {  
    l2 = strlen(s2);
    if (l2 == 0) return 0;
    l1 = strlen(s1);
    char *t = mystrdup(s2);
    if (opt & NGRAM_LOWERING) mkallsmall(t, csconv);
    for (int j = 1; j <= n; j++) {
      ns = 0;
      for (int i = 0; i <= (l1-j); i++) {
        char c = *(s1 + i + j);
        *(s1 + i + j) = '\0';
        if (strstr(t,(s1+i))) {
	  ns++;
	} else if (opt & NGRAM_WEIGHTED) {
	  ns--;
test++;
	  if (i == 0 || i == l1-j) ns--; // side weight
	}
        *(s1 + i + j ) = c;
      }
      nscore = nscore + ns;
      if (ns < 2 && !(opt & NGRAM_WEIGHTED)) break;
    }
    free(t);
  }
  
  ns = 0;
  if (opt & NGRAM_LONGER_WORSE) ns = (l2-l1)-2;
  if (opt & NGRAM_ANY_MISMATCH) ns = abs(l2-l1)-2;
  ns = (nscore - ((ns > 0) ? ns : 0));
  return ns;
}

// length of the left common substring of s1 and (decapitalised) s2
int SuggestMgr::leftcommonsubstring(char * s1, const char * s2) {
  if (utf8) {
    w_char su1[MAXSWL];
    w_char su2[MAXSWL];
    su1[0].l = su2[0].l = su1[0].h = su2[0].h = 0;
    // decapitalize dictionary word
    if (complexprefixes) {
      int l1 = u8_u16(su1, MAXSWL, s1);
      int l2 = u8_u16(su2, MAXSWL, s2);
      if (*((short *)su1+l1-1) == *((short *)su2+l2-1)) return 1;
    } else {
      int i;
      u8_u16(su1, 1, s1);






#define MAXSWL 100
#define MAXSWUTF8L (MAXSWL * 4)
#define MAX_ROOTS 100
#define MAX_WORDS 100
#define MAX_GUESS 200
#define MAXNGRAMSUGS 4
#define MAXPHONSUGS 2
#define MAXCOMPOUNDSUGS 3

// timelimit: max ~1/4 sec (process time on Linux) for a time consuming function
#define TIMELIMIT (CLOCKS_PER_SEC >> 2)
#define MINTIMER 100
#define MAXPLUSTIMER 100

#define NGRAM_LONGER_WORSE  (1 << 0)
#define NGRAM_ANY_MISMATCH  (1 << 1)
#define NGRAM_LOWERING      (1 << 2)
#define NGRAM_WEIGHTED      (1 << 3)

#include "hunvisapi.h"

#include "atypes.hxx"
#include "affixmgr.hxx"
#include "hashmgr.hxx"
#include "langnum.hxx"
#include <time.h>


  AffixMgr*       pAMgr;
  int             maxSug;
  struct cs_info * csconv;
  int             utf8;
  int             langnum;
  int             nosplitsugs;
  int             maxngramsugs;
  int             maxcpdsugs;
  int             complexprefixes;


public:
  SuggestMgr(const char * tryme, int maxn, AffixMgr *aptr);
  ~SuggestMgr();

  int suggest(char*** slst, const char * word, int nsug, int * onlycmpdsug);

(-)a/extensions/spellcheck/hunspell/src/w_char.hxx (+2 lines)

Line	Link Here

Lines 42-52 typedef struct __attribute__ ((packed))	Link Here

    unsigned char l;

    unsigned char l;

    unsigned char h;

    unsigned char h;

} w_char;

} w_char;

// two character arrays

// two character arrays

struct replentry {

struct replentry {

  char * pattern;

  char * pattern;

  char * pattern2;

  char * pattern2;

  bool start;

  bool end;

};

};

#endif

#endif

Return to bug 620626