Commit b96c309f authored by serg@serg.mysql.com's avatar serg@serg.mysql.com
Browse files

Merge work:/home/bk/mysql-4.0 

into serg.mysql.com:/usr/home/serg/Abk/mysql-4.0
parents 7f5f58dc 59f3a958

include/ft_global.h

+1 −0
Original line number Diff line number Diff line
@@ -49,6 +49,7 @@ extern const char *ft_precompiled_stopwords[]; 
extern ulong ft_min_word_len;

extern ulong ft_max_word_len;

extern ulong ft_max_word_len_for_sort;

extern char *ft_boolean_syntax;



int ft_init_stopwords(const char **);

void ft_free_stopwords(void);

myisam/ft_boolean_search.c

+37 −20
Original line number Diff line number Diff line
@@ -36,7 +36,7 @@ static double _wghts[11]={ 
  3.375000000000000,

  5.062500000000000,

  7.593750000000000};

static double *wghts=_wghts+5; // wghts[i] = 1.5**i

static double *wghts=_wghts+5; /* wghts[i] = 1.5**i */



static double _nwghts[11]={

 -0.065843621399177,
@@ -50,7 +50,7 @@ static double _nwghts[11]={ 
 -1.687500000000000,

 -2.531250000000000,

 -3.796875000000000};

static double *nwghts=_nwghts+5; // nwghts[i] = -0.5*1.5**i

static double *nwghts=_nwghts+5; /* nwghts[i] = -0.5*1.5**i */



typedef struct st_ftb_expr FTB_EXPR;

struct st_ftb_expr {
@@ -114,20 +114,7 @@ void _ftb_parse_query(FTB *ftb, byte **start, byte *end, 
    byte  r=param.plusminus;

    float weight=(param.pmsign ? nwghts : wghts)[(r>5)?5:((r<-5)?-5:r)];

    switch (res) {

      case FTB_LBR:

        ftbe=(FTB_EXPR *)alloc_root(&ftb->mem_root, sizeof(FTB_EXPR));

        ftbe->yesno=param.yesno;

        ftbe->weight=weight;

        ftbe->up=up;

        ftbe->ythresh=0;

        ftbe->docid=HA_POS_ERROR;

        if (ftbe->yesno > 0) up->ythresh++;

        _ftb_parse_query(ftb, start, end, ftbe, depth+1,

                         (param.yesno<0 ? depth+1 : ndepth));

        break;

      case FTB_RBR:

        return;

      case 1:

      case 1: /* word found */

        ftbw=(FTB_WORD *)alloc_root(&ftb->mem_root,

            sizeof(FTB_WORD) + (param.trunc ? MI_MAX_KEY_BUFF : w.len+extra));

        ftbw->len=w.len+1;
@@ -142,6 +129,19 @@ void _ftb_parse_query(FTB *ftb, byte **start, byte *end, 
        if (ftbw->yesno > 0) up->ythresh++;

        queue_insert(& ftb->queue, (byte *)ftbw);

        break;

      case 2: /* left bracket */

        ftbe=(FTB_EXPR *)alloc_root(&ftb->mem_root, sizeof(FTB_EXPR));

        ftbe->yesno=param.yesno;

        ftbe->weight=weight;

        ftbe->up=up;

        ftbe->ythresh=0;

        ftbe->docid=HA_POS_ERROR;

        if (ftbe->yesno > 0) up->ythresh++;

        _ftb_parse_query(ftb, start, end, ftbe, depth+1,

                         (param.yesno<0 ? depth+1 : ndepth));

        break;

      case 3: /* right bracket */

        return;

    }

  }

  return;
@@ -339,7 +339,7 @@ int ft_boolean_read_next(FT_INFO *ftb, char *record) 
      /* curdoc matched ! */

      info->update&= (HA_STATE_CHANGED | HA_STATE_ROW_CHANGED); /* why is this ? */



      /* info->lastpos=curdoc; */ /* do I need this ? */

      info->lastpos=curdoc;

      if (!(*info->read_record)(info,curdoc,record))

      {

        info->update|= HA_STATE_AKTIV;          /* Record is read */
@@ -348,6 +348,7 @@ int ft_boolean_read_next(FT_INFO *ftb, char *record) 
      return my_errno;

    }

  }

  ftb->state=INDEX_DONE;

  return my_errno=HA_ERR_END_OF_FILE;

}
@@ -359,17 +360,33 @@ float ft_boolean_find_relevance(FT_INFO *ftb, my_off_t docid, byte *record) 
  FTB_EXPR *ftbe;

  uint      i;



  if (ftb->state == READY)

  if (ftb->state == READY || ftb->state == INDEX_DONE)

  {

    for (i=1; i<=ftb->queue.elements; i++)

    {

      ftbw=(FTB_WORD *)(ftb->queue.root[i]);

      ftbw->docid=HA_POS_ERROR;

      for (ftbe=ftbw->up; ftbe; ftbe=ftbe->up)

      {

        if (ftbe->docid != HA_POS_ERROR)

        {

          ftbe->cur_weight=ftbe->yesses=ftbe->nos=0;

          ftbe->docid=HA_POS_ERROR;

        }

        else

          break;

      }

    }



    queue_fix(& ftb->queue);

    ftb->state=SCAN;

  }

  else if (ftb->state != SCAN)

    return -1.0;

    return -2.0;



  bzero(&ptree, sizeof(ptree));

  if (_mi_ft_parse(& ptree, ftb->info, ftb->keynr, record))

    return -1.0;

    return -3.0;



  for (i=1; i<=ftb->queue.elements; i++)

  {

myisam/ft_nlq_search.c

+1 −1
Original line number Diff line number Diff line
@@ -253,7 +253,7 @@ float ft_nlq_find_relevance(FT_INFO *handler, my_off_t docid, 
  int a,b,c;

  FT_DOC  *docs=handler->doc;



  // Assuming docs[] is sorted by dpos...

  /* Assuming docs[] is sorted by dpos... */



  for (a=0, b=handler->ndocs, c=(a+b)/2; b-a>1; c=(a+b)/2)

  {

myisam/ft_parser.c

+18 −20
Original line number Diff line number Diff line
@@ -33,9 +33,6 @@ typedef struct st_ft_docstat { 
  double max, nsum, nsum2;

#endif /* EVAL_RUN */



//  MI_INFO *info;

//  uint keynr;

//  byte *keybuf;

} FT_DOCSTAT;



static int FT_WORD_cmp(void* cmp_arg, FT_WORD *w1, FT_WORD *w2)
@@ -63,9 +60,7 @@ static int walk_and_copy(FT_WORD *word,uint32 count,FT_DOCSTAT *docstat) 


/* transforms tree of words into the array, applying normalization */



FT_WORD * ft_linearize(//MI_INFO *info, uint keynr,

    //byte *keybuf,

    TREE *wtree)

FT_WORD * ft_linearize(TREE *wtree)

{

  FT_WORD *wlist,*p;

  FT_DOCSTAT docstat;
@@ -74,9 +69,6 @@ FT_WORD * ft_linearize(//MI_INFO *info, uint keynr, 
  if ((wlist=(FT_WORD *) my_malloc(sizeof(FT_WORD)*

				   (1+wtree->elements_in_tree),MYF(0))))

  {

//    docstat.info=info;

//    docstat.keynr=keynr;

//    docstat.keybuf=keybuf;

    docstat.list=wlist;

    docstat.uniq=wtree->elements_in_tree;

#ifdef EVAL_RUN
@@ -122,12 +114,20 @@ FT_WORD * ft_linearize(//MI_INFO *info, uint keynr, 
#endif

#define word_char(X)		(true_word_char(X) || misc_word_char(X))





/* returns:

 * 0 - eof

 * 1 - word found

 * 2 - left bracket

 * 3 - right bracket

 */

byte ft_get_word(byte **start, byte *end, FT_WORD *word, FTB_PARAM *param)

{

  byte *doc=*start;

  int mwc;



  param->yesno=param->plusminus=param->pmsign=0;

  param->yesno=(FTB_YES==' ')?1:0;

  param->plusminus=param->pmsign=0;



  while (doc<end)

  {
@@ -138,18 +138,16 @@ byte ft_get_word(byte **start, byte *end, FT_WORD *word, FTB_PARAM *param) 
      {

        /* param->prev=' '; */

        *start=doc+1;

        return *doc;

        return (*doc == FTB_RBR)+2;

      }

      if (param->prev == ' ')

      {

        switch (*doc) {

        case FTB_YES: param->yesno=+1; continue;

        case FTB_NO:  param->yesno=-1; continue;

        case FTB_INC: param->plusminus++; continue;

        case FTB_DEC: param->plusminus--; continue;

        case FTB_NEG: param->pmsign=!param->pmsign; continue;

        default: break;

        }

        if (*doc == FTB_YES ) { param->yesno=+1;    continue; } else

        if (*doc == FTB_EGAL) { param->yesno= 0;    continue; } else

        if (*doc == FTB_NO  ) { param->yesno=-1;    continue; } else

        if (*doc == FTB_INC ) { param->plusminus++; continue; } else

        if (*doc == FTB_DEC ) { param->plusminus--; continue; } else

        if (*doc == FTB_NEG ) { param->pmsign=!param->pmsign; continue; }

      }

      param->prev=*doc;

      param->yesno=param->plusminus=param->pmsign=0;
@@ -162,7 +160,7 @@ byte ft_get_word(byte **start, byte *end, FT_WORD *word, FTB_PARAM *param) 
      else if (!misc_word_char(*doc) || mwc++)

        break;



    param->prev='A'; // be sure *prev is true_word_char

    param->prev='A'; /* be sure *prev is true_word_char */

    word->len= (uint)(doc-word->pos) - mwc;

    if ((param->trunc=(doc<end && *doc == FTB_TRUNC)))

      doc++;

myisam/ft_static.c

+1 −0
Original line number Diff line number Diff line
@@ -21,6 +21,7 @@ 
ulong ft_min_word_len=4;

ulong ft_max_word_len=HA_FT_MAXLEN;

ulong ft_max_word_len_for_sort=20;

char *ft_boolean_syntax="+ -><()~*";



const MI_KEYSEG ft_keysegs[FT_SEGS]={

{