- Timestamp:
- 2000-07-13T10:21:53+12:00 (24 years ago)
- File:
-
- 1 edited
Legend:
- Unmodified
- Added
- Removed
-
branches/New_Config_Format-branch/gsdl/src/mgpp/text/invf.h
r925 r1279 31 31 32 32 // NOTE: This does not include the magic number 33 // header info for .invf.dict file 33 34 struct invf_dict_header { 34 35 unsigned long lookback; … … 91 92 // this version of the blocked dictionary uses a fixed number 92 93 // of entries per block, not a fixed block size 94 // info for .invf.dict.blocked file 95 // blocked dict has a heap of blocks, some for words, some for tags 96 // and an index into each set of blocks. The index has pointers to 97 // the first entry in each block. Can do a binary search on the index 98 // to find out which block an elemnet is in 93 99 struct block_dict_header : public invf_dict_header { 94 100 // note: word_dict_start and tag_dict_start are undefined 95 101 // for blocked dictionaries 96 102 97 unsigned long entries_per_wblk; 103 unsigned long entries_per_wblk; // word blocks 98 104 unsigned long num_wblks; 99 105 unsigned long max_wblk_size; … … 101 107 unsigned long wblk_idx_start; 102 108 103 unsigned long entries_per_tblk; 109 unsigned long entries_per_tblk; // tag blocks 104 110 unsigned long num_tblks; 105 111 unsigned long max_tblk_size; … … 117 123 struct block_dict_el { 118 124 UCArray el; // word or tag 119 unsigned long frag_occur; // # entries in invf file 120 unsigned long freq; 121 unsigned long invf_ptr; 125 unsigned long frag_occur; // # entries in invf file - if have a 126 // word level index, this is the same as freq, otherwise, its the number 127 // of fragments containing this word 128 unsigned long freq; // # of times this word occurs 129 unsigned long invf_ptr; // pointer into inverted file 122 130 123 131 virtual void Clear (); … … 133 141 134 142 struct word_block_dict_el : public block_dict_el { 135 unsigned long *levelFreqs; 143 unsigned long *levelFreqs; // freq of the word at each level 136 144 137 145 void Clear (); … … 209 217 #define SKIP_MODE_NO_SKIPS 0 210 218 219 // invf file - has a list of frags for each word, but the word is not 220 // stored in the invf file - the dictionaries store the words, along 221 // with num entries, and a pointer into invf file 211 222 struct invf_file_header { 212 223 unsigned long no_of_words;
Note:
See TracChangeset
for help on using the changeset viewer.