/*----------------------------------------------------------------------  File    : tract.c  Contents: item and transaction management  Author  : Christian Borgelt  History : 14.02.1996 file created as apriori.c            24.06.1996 function _get_item optimized            01.07.1996 adapted to modified symtab module            04.01.1998 scan functions moved to module 'tfscan'            09.06.1998 vector enlargement modified            20.06.1998 adapted to changed st_create function            07.08.1998 bug in function _get_tract (is_read) fixed            08.08.1998 item appearances added            17.08.1998 item sorting and recoding added            02.09.1998 several assertions added            05.02.1999 long int changed to int            22.10.1999 bug in item appearances reading fixed            11.11.1999 adapted to name/identifier maps            01.12.1999 check of item appearance added to sort function            15.03.2000 removal of infrequent items added            14.07.2001 adapted to modified module tfscan            27.12.2001 item functions made a separate module            18.11.2001 transaction functions made a separate module            28.12.2001 first version of this module completed            12.01.2002 empty field at end of record reported as error            06.02.2002 item sorting reversed (ascending order)----------------------------------------------------------------------*/#include <stdio.h>#include <stdlib.h>#include <string.h>#include <assert.h>#include "tract.h"#include "vecops.h"#ifdef STORAGE#include "storage.h"#endif/*----------------------------------------------------------------------  Preprocessor Definitions----------------------------------------------------------------------*/#define BLKSIZE  256            /* block size for enlarging vectors *//*----------------------------------------------------------------------  Constants----------------------------------------------------------------------*//* --- item appearance indicators --- */static const char *i_body[] = { /* item to appear in bodies only */  "i",  "in",  "a", "ante", "antecedent", "b", "body", NULL };static const char *i_head[] = { /* item to appear in heads only */  "o",  "out", "c", "cons", "consequent", "h", "head", NULL };static const char *i_both[] = { /* item to appear in both */  "io", "inout", "ac", "bh", "both",                   NULL };static const char *i_ignore[] ={/* item to ignore */  "n", "neither", "none", "ign", "ignore", "-",        NULL };/*----------------------------------------------------------------------  Auxiliary Functions----------------------------------------------------------------------*/static int _appcode (const char *s){                               /* --- get appearance indicator code */  const char **p;               /* to traverse indicator list */  assert(s);                    /* check the function argument */  for (p = i_body;   *p; p++)   /* check 'body' indicators */    if (strcmp(s, *p) == 0) return APP_BODY;  for (p = i_head;   *p; p++)   /* check 'head' indicators */    if (strcmp(s, *p) == 0) return APP_HEAD;  for (p = i_both;   *p; p++)   /* check 'both' indicators */    if (strcmp(s, *p) == 0) return APP_BOTH;  for (p = i_ignore; *p; p++)   /* check 'ignore' indicators */    if (strcmp(s, *p) == 0) return APP_NONE;  return -1;                    /* if none found, return error code */}  /* _appcode() *//*--------------------------------------------------------------------*/static int _get_item (ITEMSET *iset, FILE *file){                               /* --- read an item */  int  d;                       /* delimiter type */  char *buf;                    /* read buffer */  ITEM *item;                   /* pointer to item */  int  *vec;                    /* new item vector */  int  size;                    /* new item vector size */  assert(iset && file);         /* check the function arguments */  d   = tfs_getfld(iset->tfscan, file, NULL, 0);  buf = tfs_buf(iset->tfscan);  /* read the next field (item name) */  if ((d < 0) || (buf[0] == '\0')) return d;  item = nim_byname(iset->nimap, buf);  if (item)                     /* look up the name in name/id map */    item->frq++;                /* if it could be found, count it */  else {                        /* if it could not be found: */    if (iset->app == APP_NONE)  /* if new items are to be ignored, */      return d;                 /* do not register the item */    item = nim_add(iset->nimap, buf, sizeof(ITEM));    if (!item) return E_NOMEM;  /* add the new item to the map, */    item->frq = 1;              /* initialize the frequency counter, */    item->app = iset->app;      /* and set the appearance indicator */  }  size = iset->vsz;             /* get the item vector size */  if (iset->cnt >= size) {      /* if the item vector is full */    size += (size > BLKSIZE) ? (size >> 1) : BLKSIZE;    vec   = (int*)realloc(iset->items, size *sizeof(int));    if (!vec) return E_NOMEM;   /* enlarge the item vector */    iset->items = vec; iset->vsz = size;  }                             /* set the new vector and its size */  iset->items[iset->cnt++] = item->id;  return d;                     /* add the item to the transaction */}  /* _get_item() */            /* and return the delimiter type *//*--------------------------------------------------------------------*/static int _asccmp (const void *p1, const void *p2, void *data){                               /* --- compare item frequencies */  if (((const ITEM*)p1)->app == APP_NONE)    return (((const ITEM*)p2)->app == APP_NONE) ? 0 : 1;  if (((const ITEM*)p2)->app == APP_NONE) return -1;  if (((const ITEM*)p1)->frq < (int)data)    return (((const ITEM*)p2)->frq < (int)data) ? 0 : 1;  if (((const ITEM*)p2)->frq < (int)data) return -1;  if (((const ITEM*)p1)->frq > ((const ITEM*)p2)->frq) return  1;  if (((const ITEM*)p1)->frq < ((const ITEM*)p2)->frq) return -1;  return 0;                     /* return sign of frequency diff. */}  /* _asccmp() *//*--------------------------------------------------------------------*/static int _descmp (const void *p1, const void *p2, void *data){                               /* --- compare item frequencies */  if (((const ITEM*)p1)->app == APP_NONE)    return (((const ITEM*)p2)->app == APP_NONE) ? 0 : 1;  if (((const ITEM*)p2)->app == APP_NONE) return -1;  if (((const ITEM*)p1)->frq > ((const ITEM*)p2)->frq) return -1;  if (((const ITEM*)p1)->frq < ((const ITEM*)p2)->frq) return  1;  return 0;                     /* return sign of frequency diff. */}  /* _descmp() *//*----------------------------------------------------------------------  Item Set Functions----------------------------------------------------------------------*/ITEMSET* is_create (void){                               /* --- create an item set */  ITEMSET *iset;                /* created item set */  iset = malloc(sizeof(ITEMSET));  if (!iset) return NULL;       /* create an item set */  iset->tfscan = tfs_create();  /* and its components */  iset->nimap  = nim_create(0, 0, (HASHFN*)0, (SYMFN*)0);  iset->items  = (int*)malloc(BLKSIZE *sizeof(int));  if (!iset->tfscan || !iset->nimap || !iset->items) {    is_delete(iset); return NULL; }  iset->app    = APP_BOTH;      /* initialize the other fields */  iset->vsz    = BLKSIZE;  iset->cnt    = 0;  iset->chars[0] = ' ';  iset->chars[1] = ' ';  iset->chars[2] = '\n'; iset->chars[3] = '\0';  return iset;                  /* return the created item set */}  /* is_create() *//*--------------------------------------------------------------------*/void is_delete (ITEMSET *iset){                               /* --- delete an item set */  assert(iset);                 /* check the function argument */  if (iset->items)  free(iset->items);  if (iset->nimap)  nim_delete(iset->nimap);  if (iset->tfscan) tfs_delete(iset->tfscan);  free(iset);                   /* delete the components */}  /* is_delete() */            /* and the item set body *//*--------------------------------------------------------------------*/void is_chars (ITEMSET *iset, const char *blanks,  const char *fldseps,                              const char *recseps, const char *cominds){                               /* --- set special characters */  assert(iset);                 /* check the function argument */  if (blanks)                   /* set blank characters */    iset->chars[0] = tfs_chars(iset->tfscan, TFS_BLANK,  blanks);  if (fldseps)                  /* set field separators */    iset->chars[1] = tfs_chars(iset->tfscan, TFS_FLDSEP, fldseps);  if (recseps)                  /* set record separators */    iset->chars[2] = tfs_chars(iset->tfscan, TFS_RECSEP, recseps);  if (cominds)                  /* set comment indicators */    tfs_chars(iset->tfscan, TFS_COMMENT, cominds);}  /* is_chars() *//*--------------------------------------------------------------------*/int is_item (ITEMSET *iset, const char *name){                               /* --- get an item identifier */  ITEM *item = nim_byname(iset->nimap, name);  return (item) ? item->id :-1; /* look up the given name */}  /* is_item() */              /* in the name/identifier map *//*--------------------------------------------------------------------*/int is_readapp (ITEMSET *iset, FILE *file){                               /* --- read appearance indicators */  int  d;                       /* delimiter type */  char *buf;                    /* read buffer */  ITEM *item;                   /* to access the item data */  assert(iset && file);         /* check the function arguments */  if (tfs_skip(iset->tfscan, file) < 0)    return E_FREAD;             /* skip leading comments */  buf = tfs_buf(iset->tfscan);  /* read the first record (one field) */  d   = tfs_getfld(iset->tfscan, file, NULL, 0);  if (d <  0)           return E_FREAD;  if (d >= TFS_FLD)     return E_FLDCNT;  iset->app = _appcode(buf);    /* get default appearance code */  if (iset->app < 0)    return E_UNKAPP;  while (d > TFS_EOF) {         /* read item/indicator pairs */    if (tfs_skip(iset->tfscan, file) < 0)      return E_FREAD;           /* skip more comments */    d = tfs_getfld(iset->tfscan, file, NULL, 0);    if (d <= TFS_EOF)           /* read the next item */      return (d < 0) ? E_FREAD : 0;    if (buf[0] == '\0')         /* check for end of file */      return E_ITEMEXP;         /* and for a missing item */    item = nim_add(iset->nimap, buf, sizeof(ITEM));    if (item == EXISTS) return E_DUPITEM;  /* add the new item */    if (item == NULL)   return E_NOMEM;    /* to the name/id map */    item->frq = 0;              /* clear the frequency counter */    if (d < TFS_FLD)    return E_APPEXP;    d = tfs_getfld(iset->tfscan, file, NULL, 0);    if (d <  0)         return E_FREAD;    if (d >= TFS_FLD)   return E_FLDCNT;    item->app = _appcode(buf);  /* get the appearance indicator */    if (item->app <  0)  return E_UNKAPP;  }  return 0;                     /* return 'ok' */}  /* is_readapp() *//*--------------------------------------------------------------------*/