affixmgr.cxx

管理项目进度工具的原代码

         }

         // now clean up by adding smart search termination strings:
         // if you are already a superset of the previous prefix
         // but not a subset of the next, search can end here
         // so set NextNE properly

         ptr = (PfxEntry *) pStart[i];
         for (; ptr != NULL; ptr = ptr->getNext()) {
	     PfxEntry * nptr = ptr->getNext();
             PfxEntry * mptr = NULL;
             for (; nptr != NULL; nptr = nptr->getNext()) {
	         if (! isSubset(ptr->getKey(),nptr->getKey())) break;
                 mptr = nptr;
             }
             if (mptr) mptr->setNextNE(NULL);
         }
    }
    return 0;
}



// reinitialize the SfxEntry links NextEQ and NextNE to speed searching
// using the idea of leading subsets this time
int AffixMgr::process_sfx_order()
{
    SfxEntry* ptr;

    // loop through each prefix list starting point
    for (int i=1; i < SETSIZE; i++) {

         ptr = (SfxEntry *) sStart[i];

         // look through the remainder of the list
         //  and find next entry with affix that 
         // the current one is not a subset of
         // mark that as destination for NextNE
         // use next in list that you are a subset
         // of as NextEQ

         for (; ptr != NULL; ptr = ptr->getNext()) {
	     SfxEntry * nptr = ptr->getNext();
             for (; nptr != NULL; nptr = nptr->getNext()) {
	         if (! isSubset(ptr->getKey(),nptr->getKey())) break;
             }
             ptr->setNextNE(nptr);
             ptr->setNextEQ(NULL);
             if ((ptr->getNext()) && isSubset(ptr->getKey(),(ptr->getNext())->getKey())) 
                 ptr->setNextEQ(ptr->getNext());
         }


         // now clean up by adding smart search termination strings:
         // if you are already a superset of the previous suffix
         // but not a subset of the next, search can end here
         // so set NextNE properly

         ptr = (SfxEntry *) sStart[i];
         for (; ptr != NULL; ptr = ptr->getNext()) {
	     SfxEntry * nptr = ptr->getNext();
             SfxEntry * mptr = NULL;
             for (; nptr != NULL; nptr = nptr->getNext()) {
	         if (! isSubset(ptr->getKey(),nptr->getKey())) break;
                 mptr = nptr;
             }
             if (mptr) mptr->setNextNE(NULL);
         }
    }
    return 0;
}



// takes aff file condition string and creates the
// conds array - please see the appendix at the end of the
// file affentry.cxx which describes what is going on here
// in much more detail

void AffixMgr::encodeit(struct affentry * ptr, char * cs)
{
  unsigned char c;
  int i, j, k;
  unsigned char mbr[MAXLNLEN];

  // now clear the conditions array */
  for (i=0;i<SETSIZE;i++) ptr->conds[i] = (unsigned char) 0;

  // now parse the string to create the conds array */
  int nc = strlen(cs);
  int neg = 0;   // complement indicator
  int grp = 0;   // group indicator
  int n = 0;     // number of conditions
  int ec = 0;    // end condition indicator
  int nm = 0;    // number of member in group

  // if no condition just return
  if (strcmp(cs,".")==0) {
    ptr->numconds = 0;
    return;
  }

  i = 0;
  while (i < nc) {
    c = *((unsigned char *)(cs + i));

    // start group indicator
    if (c == '[') {
       grp = 1;
       c = 0;
    }

    // complement flag
    if ((grp == 1) && (c == '^')) {
       neg = 1;
       c = 0;
    }

    // end goup indicator
    if (c == ']') {
       ec = 1;
       c = 0;
    }

    // add character of group to list
    if ((grp == 1) && (c != 0)) {
      *(mbr + nm) = c;
      nm++;
      c = 0;
    }

    // end of condition 
    if (c != 0) {
       ec = 1;
    }

    
    if (ec) {
      if (grp == 1) {
        if (neg == 0) {
          // set the proper bits in the condition array vals for those chars
	  for (j=0;j<nm;j++) {
	     k = (unsigned int) mbr[j];
             ptr->conds[k] = ptr->conds[k] | (1 << n);
          }
	} else {
	  // complement so set all of them and then unset indicated ones
	   for (j=0;j<SETSIZE;j++) ptr->conds[j] = ptr->conds[j] | (1 << n);
	   for (j=0;j<nm;j++) {
	     k = (unsigned int) mbr[j];
             ptr->conds[k] = ptr->conds[k] & ~(1 << n);
	   }
        }
        neg = 0;
        grp = 0;   
        nm = 0;
      } else {
         // not a group so just set the proper bit for this char
         // but first handle special case of . inside condition
         if (c == '.') {
	    // wild card character so set them all
            for (j=0;j<SETSIZE;j++) ptr->conds[j] = ptr->conds[j] | (1 << n);
         } else {  
	    ptr->conds[(unsigned int) c] = ptr->conds[(unsigned int)c] | (1 << n);
         }
      }
      n++;
      ec = 0;
    }


    i++;
  }
  ptr->numconds = n;
  return;
}


// check word for prefixes
struct hentry * AffixMgr::prefix_check (const char * word, int len)
{
    struct hentry * rv= NULL;
 
    // first handle the special case of 0 length prefixes
    PfxEntry * pe = (PfxEntry *) pStart[0];
    while (pe) {
       rv = pe->check(word,len);
       if (rv) return rv;
       pe = pe->getNext();
    }
  
    // now handle the general case
    unsigned char sp = *((const unsigned char *)word);
    PfxEntry * pptr = (PfxEntry *)pStart[sp];

    while (pptr) {
        if (isSubset(pptr->getKey(),word)) {
	     rv = pptr->check(word,len);
             if (rv) return rv;
             pptr = pptr->getNextEQ();
        } else {
	     pptr = pptr->getNextNE();
        }
    }
    
    return NULL;
}

// check if compound word is correctly spelled
struct hentry * AffixMgr::compound_check (const char * word, int len, char compound_flag)
{
    int i;
    struct hentry * rv= NULL;
    char * st;
    char ch;
    
    // handle case of string too short to be a piece of a compound word 
    if (len < cpdmin) return NULL;

    st = mystrdup(word);
    
    for (i=cpdmin; i < (len - (cpdmin-1)); i++) {

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

	rv = lookup(st);
        if (!rv) rv = affix_check(st,i);

	if ((rv) && (TESTAFF(rv->astr, compound_flag, rv->alen))) {
	    rv = lookup((word+i));
	    if ((rv) && (TESTAFF(rv->astr, compound_flag, rv->alen))) {
		free(st);
		return rv;
	    }
	    rv = affix_check((word+i),strlen(word+i));
	    if ((rv) && (TESTAFF(rv->astr, compound_flag, rv->alen))) {
		free(st);
		return rv;
	    }
	    rv = compound_check((word+i),strlen(word+i),compound_flag); 
	    if (rv) {
		free(st);
		return rv;
	    }
	    
	}
        st[i] = ch;
    }
    free(st);
    return NULL;
}    



// check word for suffixes
struct hentry * AffixMgr::suffix_check (const char * word, int len, 
                       int sfxopts, AffEntry * ppfx)
{
    struct hentry * rv = NULL;

    // first handle the special case of 0 length suffixes
    SfxEntry * se = (SfxEntry *) sStart[0];
    while (se) {
       rv = se->check(word,len, sfxopts, ppfx);
       if (rv) return rv;
       se = se->getNext();
    }
  
    // now handle the general case
    unsigned char sp = *((const unsigned char *)(word + len - 1));


    SfxEntry * sptr = (SfxEntry *) sStart[sp];

    while (sptr) {
        if (isRevSubset(sptr->getKey(),(word+len-1), len)) {
	     rv = sptr->check(word,len, sfxopts, ppfx);
             if (rv) {
                  return rv;
             }
             sptr = sptr->getNextEQ();
        } else {
	     sptr = sptr->getNextNE();
        }
    }
    return NULL;
}



// check if word with affixes is correctly spelled
struct hentry * AffixMgr::affix_check (const char * word, int len)
{
    struct hentry * rv= NULL;

    // check all prefixes (also crossed with suffixes if allowed) 
    rv = prefix_check(word, len);
    if (rv) return rv;

    // if still not found check all suffixes
    rv = suffix_check(word, len, 0, NULL);
    return rv;
}


int AffixMgr::expand_rootword(struct guessword * wlst, int maxn, 
                       const char * ts, int wl, const char * ap, int al)
{

    int nh=0;

    // first add root word to list

    if (nh < maxn) {
       wlst[nh].word = mystrdup(ts);
       wlst[nh].allow = (1 == 0);
       nh++;
    }

    // handle suffixes
    for (int i = 0; i < al; i++) {
       unsigned char c = (unsigned char) ap[i];
       SfxEntry * sptr = (SfxEntry *)sFlag[c];
       while (sptr) {
	 char * newword = sptr->add(ts, wl);
         if (newword) {
           if (nh < maxn) {
	      wlst[nh].word = newword;
              wlst[nh].allow = sptr->allowCross();
              nh++;
	   } else {
	      free(newword);
           }
	 }
         sptr = (SfxEntry *)sptr ->getFlgNxt();
       }
    }

    int n = nh;

    // handle cross products of prefixes and suffixes
    for (int j=1;j<n ;j++)
       if (wlst[j].allow) {
          for (int k = 0; k < al; k++) {
             unsigned char c = (unsigned char) ap[k];
             PfxEntry * cptr = (PfxEntry *) pFlag[c];
             while (cptr) {
                if (cptr->allowCross()) {
	            int l1 = strlen(wlst[j].word);
	            char * newword = cptr->add(wlst[j].word, l1);
                    if (newword) {
		       if (nh < maxn) {
	                  wlst[nh].word = newword;
                          wlst[nh].allow = cptr->allowCross();
                          nh++;
		       } else {
			  free(newword);
                       }
	            }
                }
                cptr = (PfxEntry *)cptr ->getFlgNxt();
             }
	  }
       }


    // now handle pure prefixes
    for (int m = 0; m < al; m ++) {
       unsigned char c = (unsigned char) ap[m];
       PfxEntry * ptr = (PfxEntry *) pFlag[c];
       while (ptr) {
	 char * newword = ptr->add(ts, wl);
         if (newword) {
	     if (nh < maxn) {
	        wlst[nh].word = newword;
                wlst[nh].allow = ptr->allowCross();
                nh++;
             } else {
	        free(newword);
	     } 
	 }
         ptr = (PfxEntry *)ptr ->getFlgNxt();
       }
    }

    return nh;
}


// return length of replacing table
int AffixMgr::get_numrep()
{
  return numrep;
}

// return replacing table
struct replentry * AffixMgr::get_reptable()
{
  if (! reptable ) return NULL;
  return reptable;
}


// return length of character map table
int AffixMgr::get_nummap()
{
  return nummap;
}

// return character map table
struct mapentry * AffixMgr::get_maptable()
{
  if (! maptable ) return NULL;
  return maptable;
}

// return text encoding of dictionary
char * AffixMgr::get_encoding()
{
  if (! encoding ) {
      encoding = mystrdup("ISO8859-1");
  }
  return mystrdup(encoding);
}


// return the preferred try string for suggestions
char * AffixMgr::get_try_string()
{
  if (! trystring ) return NULL;
  return mystrdup(trystring);
}

// return the compound words control flag
char * AffixMgr::get_compound()