/* Functions for manipulating "document vectors". */ /* Copyright (C) 1997 Andrew McCallum Written by: Andrew Kachites McCallum This file is part of the Bag-Of-Words Library, `libbow'. This library is free software; you can redistribute it and/or modify it under the terms of the GNU Library General Public License as published by the Free Software Foundation, version 2. This library is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Library General Public License for more details. You should have received a copy of the GNU Library General Public License along with this library; if not, write to the Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111, USA */ #include #include unsigned int bow_dv_default_capacity = 2; /* The number of "document vectors" current in existance. */ unsigned int bow_dv_count = 0; /* Create a new, empty "document vector". */ bow_dv * bow_dv_new (int capacity) { bow_dv *ret; if (capacity == 0) capacity = bow_dv_default_capacity; ret = bow_malloc (sizeof (bow_dv) + (sizeof (bow_de) * capacity)); ret->length = 0; ret->idf = 0.0f; ret->size = capacity; bow_dv_count++; return ret; } /* Return the index into (*DV)->entries[] at which the "document entry" for the document with index DI can be found. If necessary, allocate more space, and/or shift other document entries around in order to make room for it. */ static int _bow_dv_index_for_di (bow_dv **dv, int di, int error_on_creation) { int dv_index; /* The "document vector" index at which we are looking for. */ static inline void grow_if_necessary () { if (error_on_creation) bow_error ("Shouldn't be creating new entry for a weight."); if ((*dv)->length == (*dv)->size) { /* The DV must grow to accommodate another entry. */ (*dv)->size *= 3; (*dv)->size /= 2; (*dv) = bow_realloc ((*dv), (sizeof (bow_dv) + sizeof (bow_de) * (*dv)->size)); } } static inline void initialize_dv_index (int dvi) { (*dv)->entry[dvi].di = di; (*dv)->entry[dvi].count = 0; (*dv)->entry[dvi].weight = 0.0f; } assert ((*dv)->length <= (*dv)->size); if ((*dv)->length == 0) { /* The DV is empty. */ assert ((*dv)->size); ((*dv)->length)++; initialize_dv_index (0); return 0; } else if (di == (*dv)->entry[(dv_index = (*dv)->length - 1)].di) { /* An entry already exists for this DI; it's at the end. */ return dv_index; } else if (di > (*dv)->entry[dv_index].di) { /* The entry does not already exist, and the entry belongs at the end of the current DV. */ dv_index = (*dv)->length; ((*dv)->length)++; grow_if_necessary (); initialize_dv_index (dv_index); return dv_index; } else { /* Search for the entry in the middle of the list. */ for (dv_index = 0; (((*dv)->entry[dv_index].di < di) && (dv_index < (*dv)->length)); dv_index++) { if ((*dv)->entry[dv_index].di == di) break; } if ((*dv)->entry[dv_index].di == di) { /* The entry already exists in the middle of the DV. */ return dv_index; } else { /* The entry should be in the middle of the DV, but it isn't there now; we'll have to push some aside to make room. */ int dvi; assert (dv_index < (*dv)->length); ((*dv)->length)++; grow_if_necessary (); /* Scoot some "document entries" up to make room */ for (dvi = (*dv)->length - 2; dvi >= dv_index; dvi--) memcpy (&((*dv)->entry[dvi+1]), &((*dv)->entry[dvi]), sizeof (bow_de)); initialize_dv_index (dv_index); return dv_index; } } } /* Sum the COUNT into the document vector DV at document index DI, creating a new entry in the document vector if necessary. */ void bow_dv_add_di_count_weight (bow_dv **dv, int di, int count, float weight) { int dv_index; /* The "document vector" index at which we are adding COUNT. */ int new_count; dv_index = _bow_dv_index_for_di (dv, di, 0); new_count = (*dv)->entry[dv_index].count + count; if (new_count >= (SHRT_MAX)) { static int already_warned = 0; if (!already_warned) { bow_verbosify (bow_verbose, "bow_dv->entry[].count overflowed short\n"); already_warned = 1; } (*dv)->entry[dv_index].count = (SHRT_MAX) - 1; } else (*dv)->entry[dv_index].count = new_count; assert ((*dv)->entry[dv_index].count > 0); /* If we are recording only binary word absence/presence, force the word count to 0 or 1. */ if (bow_binary_word_counts && (*dv)->entry[dv_index].count > 1) (*dv)->entry[dv_index].count = 1; /* xxx But we don't do anything special with the weight? */ (*dv)->entry[dv_index].weight += weight; } /* Return the number of bytes required for writing the "document vector" DV. */ int bow_dv_write_size (bow_dv *dv) { if (dv == NULL) return sizeof (int); return (sizeof (int) /* length */ + sizeof (float) /* idf */ + (dv->length /* for each entry */ * (sizeof (short) /* di */ + sizeof (short) /* count */ + sizeof (float)))); /* weight */ } /* Write "document vector" DV to the stream FP. */ void bow_dv_write (bow_dv *dv, FILE *fp) { int i; if (dv == NULL) { bow_fwrite_int (0, fp); return; } bow_fwrite_int (dv->length, fp); bow_fwrite_float (dv->idf, fp); assert (dv->idf == dv->idf); /* testing for NaN */ for (i = 0; i < dv->length; i++) { bow_fwrite_short (dv->entry[i].di, fp); bow_fwrite_short (dv->entry[i].count, fp); bow_fwrite_float (dv->entry[i].weight, fp); } } /* Return a new "document vector" read from a pointer into a data file, FP. */ bow_dv * bow_dv_new_from_data_fp (FILE *fp) { int i; int len; bow_dv *ret; assert (feof (fp) == 0); /* Help make sure FP hasn't been closed. */ bow_fread_int (&len, fp); if (len == 0) return NULL; ret = bow_dv_new (len); bow_fread_float (&(ret->idf), fp); assert (ret->idf == ret->idf); /* testing for NaN */ ret->length = len; for (i = 0; i < len; i++) { bow_fread_short (&(ret->entry[i].di), fp); bow_fread_short (&(ret->entry[i].count), fp); bow_fread_float (&(ret->entry[i].weight), fp); } return ret; } void bow_dv_free (bow_dv *dv) { bow_dv_count--; bow_free (dv); }