/* WIDE AREA INFORMATION SERVER SOFTWARE: No guarantees or restrictions. See the readme file for the full standard disclaimer. */ /* Include file for the irhash.c file. Implements the building functions in irext.h */ #ifndef IREXT_H #define IREXT_H /* An interface for adding new server types into the WAIS system. * The idea is to use the parsing and bookkeeping operatios of the serial * indexer, while allowing different invered file and signiture systems * to be added as back ends. * * - Tracy Shen and Brewster 3/91 */ /* * Tracy changes: * - in function "add_word", add two more parameters, source and date * - add a new function "set_query_parameter" * proposed changes by brewster: * replace date_type with time_t: accepted * take out all "unsigned" type modifiers (tracey will concider this) * replace short with long: accepted * replace int with long (we port to 16 bit machines still): accepted * added source to delete_doc_id parameters: accepted * Proposed changes by brewster and tracy: * if routines are successful return 0, otherwise an error code: accepted * Proposed changes by harry: * make the dictionary value be any size. This can be done by * passing in a size arg or by passing in read and write routines. * have a function that says we will not be calling best_hit anymore. * proposed changes by brewster: * took out hash_pos from add word. * change source to a database* db. * added finished_best_hit, finished_search_word(db), finished_add_word * */ #include "cdialect.h" #include "irfiles.h" /* for database */ #ifdef __cplusplus /* declare these as C style functions */ extern "C" { #endif /* def __cplusplus */ /* ============================ * === Building Functions === * ============================*/ long init_add_word _AP ((database *db, long parameter1, long parameter2)); /* * add_word: add this word to the database * return values: 0 if successful, non-0 if error * defined error conditions: * */ long add_word _AP(( char *word, /* the word to be indexed, this could be a word pair. If NULL there are no more words to be indexed */ long char_pos, /* the position of the start of the word */ long line_pos, /* this is passed for the best section calculation */ long weight, /* how important the word looks syntactically (such as is it bold) NOT used by signature system */ long doc_id, /* current document, this will never be 0 */ time_t date, /* display day of this document, 0 if not known */ database* db /* database to insert the document */ )); /* * finished_add_word: states that there are no more words to add * to this database. * * return values: 0 if successful, non-0 if error * defined error conditions: * */ long finished_add_word _AP((database *db)); /* =============================== * === Maintenance Functions === * ===============================*/ /* * delete_doc_id : delete a document * return values: 0, successfull * -1, document not found * */ long delete_doc_id _AP((long doc_id, database *db)); /* ============================= * === Searching Functions === * =============================*/ /* * set_query_parameter : set query parameter * set search attributes such as date factor, document source ids, * and maximum number of documents returned in a search ( the last * one is an important performance factor to signature type system) * The search artributes applies to all comming queries until * they are re-set by next set_query_parameter call. * * return values: none * */ #define SET_MAX_RETRIEVED_MASK 1 #define SET_DATE_FACTOR_MASK 2 #define SET_SELECT_SOURCE 4 /* enum literals for date_factor */ #define DF_INDEPENDENT 1 #define DF_LATER 2 #define DF_EARLIER 3 typedef struct { long max_hit_retrieved; /* max number of hits can be returned by * the search engine. For a signature * type system, the default value is 20 */ long date_factor; /* default is DF_INDEPENDENT */ long num_db; /* value of zero indicating select all, * default is selecting all */ database **db_list; /* list of databases to be searched */ } query_parameter_type; /* * set_query_parameter: set a mode variable for the search engine * return values: 0 if successful, non-0 if error * defined error conditions: * */ long set_query_parameter _AP (( long mask, query_parameter_type *parameters /* fields in the query parameter structure are only * interpreted when the corresponding mask bit * is set in the mask argument. */ )); /* * search_word: searches for a word in the index. it side effects * internal state so that best_hit will return the correct * results. * return values: 0 if successful, non-0 if error * defined error conditions: * */ long search_word _AP ((char *word, /* the word to be searched for */ long char_pos, /* the position of the start of the word */ long line_pos, /* is this needed? not for signature system */ long weight, /* how important the word looks syntactically, such as is it bold */ long doc_id, /* current document, seed words is 0, then it increments into the relevant document */ long dictionary_value, /* this is from the disk dictionary, a signature system would use weight, inverted file systems would put position information */ database *db )); /* * finished_search_word: states that there are no more words that will * be searched for before best_hit will be called. * * return values: 0 if successful, non-0 if error * defined error conditions: * */ long finished_search_word _AP((database *db)); /* * best-hit : * * return values: 0, successfull * -1, no more documents to return * Other values returned to signal future signals. * */ long best_hit _AP ((long *doc_id, long *score)); /* * finished_best_hit: states that there are no more best_hits will be called * before the next set of search_words or add_words. * * return values: 0 if successful, non-0 if error * defined error conditions: * */ long finished_best_hit _AP((void)); #ifdef __cplusplus } #endif /* def __cplusplus */ #endif /* ndef IREXT_H */