]]> DtSrAPI library call DtSrAPI Describes overview, constants, and structures for DtSearch online API The DtSearch API provides programmatic access to the DtSearch search and retrieval engine. The API functions are located in the library libDtSr, and are directly linked into user written search programs. Search and retrieval of DtSearch databases is available through three essential API functions: DtSearchInit Opens databases and other files, and generally initializes the search engine for subsequent requests. DtSearchQuery Is passed a user query and some search options, performs the requested search, and returns a linked list of structures, called a results list, representing the objects satisfying the search. The results list contains abstracted information about the documents suitable for display to an end user, as well as private information used for subsequent retrievals. DtSearchRetrieve Retrieves an object given data from a results list node. When a results list contains all the information an application needs, retrieval by DtSearch may not be required. For example when the documents themselves are not stored in DtSearch databases and the document references are available from the results list, the calling program may access the objects directly. All functions can potentially return multiple messages on a global linked list of messages called the MessageList. Most unsuccessful return codes append at least one message to the MessageList, but even successful returns may append messages, and multiple messages are always possible. Messages are standard C text strings terminated by a zero byte, and were designed to be displayed directly to users. Several API utility functions are available for manipulating the MessageList. Certain fatal errors will require an immediate abort from the engine. By default fatal error messages will be written to the stderr, but can be written to a text file specified in DtSearchInit. All API aborts are implemented through a call to DtSearchExit. DtSearchExit() ensures cleanup of a number of system resources before the final call to exit. Developers can add an additional user exit to DtSearchExit to specify additional emergency clean up before process exit. Most API functions return one of a set of standard integer return codes. The return code DtSrOK means complete success, other return codes indicate various levels of negative results or failure. DtSrOK Normal, affirmative, successful response. DtSrNOTAVAIL Generic negative response. For example, no hits on search, no such record, etc. DtSrFAIL Miscellaneous unsuccessful engine returns. DtSrREINIT Engine reinitialized, request canceled. Often returned when invalid database name detected. Caller should clean up and call DtSearchReinit(). DtSrERROR Fatal caller programming error. DtSrABORT Fatal engine failure, caller must abort. Each DtSearch database is associated with an integer representing among other things the natural language of its documents. These constants are used throughout the API to identify the supported languages. DtSrLaENG 0 English, ASCII char set (default) DtSrLaENG2 1 English, ISO Latin-1 char set DtSrLaESP 2 Spanish, ISO Latin-1 char set DtSrLaFRA 3 French, ISO Latin-1 char set DtSrLaITA 4 Italian, ISO Latin-1 char set DtSrLaDEU 5 German, ISO Latin-1 char set DtSrLaJPN 6 Japanese, EUC, auto kanji compounds DtSrLaJPN2 7 Japanese, EUC, listed kanji compounds DtSrLaLAST 7 Last supported DtSrLa constant DtSrVERSION DtSearch version number string. DtSrMAX_KTNAME Maximum string length of a keytype name. DtSrMAX_DB_KEYSIZE Maximum size of the unique document key. DtSrMAXWIDTH_HWORD Largest possible word or stem size. DtSrMAX_STEMCOUNT Maximum number of boolean search terms. DtSrObjdate is a typdef for an unsigned integer used as a date/time stamp for documents. DtSearch queries may be qualified by document date ranges. The data type packs certain standard struct tm fields into bitmap fields to minimize space. DtSrObjdate are based on the western Gregorian calendar and are not guaranteed to map to other time locales. DtSearch objdates have a range from 1900 to 5995 inclusive and a resolution of 1 minute. From hi order bits to low: 12 bits = tm_year (0 - 4095, years since 1900 (1900 - 5995)) 4 bits = tm_mon (0 - 11, month name index) 5 bits = tm_mday (1 - 31, day of month) 5 bits = tm_hour (0 - 23, hours since midnight) 6 bits = tm_min (0 - 59, minutes since top of hour) typedef struct { char is_selected; char ktchar; char name [ DtSrMAX_KTNAME+1]; } DtSrKeytype; A DtSearch keytype references a logical subset of the database. The primary identifier for a keytype is the keytype character ktchar. The ktchar identifies the subset of the database that has that character as the first character of its document keys. The DtSrKeytype structure associates the ktchar with a short name string for use in user GUI labels identifying the keytype, and provides a boolean selection toggle for the keytype. An array of DtSrKeytype structures is maintained by the API for each database after API initialization. The API function DgSearchGetKeytypes() is used to access the array. The is_selected boolean in each array node indicates whether the user has selected that keytype to be returned in the current search. The application must ensure that the boolean reflects the current state of the user's desires prior to any search. Typically this is done by having the keytypes array track user interface toggle buttons for the database. typedef struct _DtSrResult { struct _DtSrResult *link; long flags; long objflags; long objuflags; long objsize; DtSrObjdates objdate; short objtype; short objcost; int dbn; DB_ADDR dba; short language; char reckey [ DtSrMAX_DB_KEYSIZE]; int proximity; char *abstractp; } DtSrResult; The API function DtSearchQuery returns a results list upon successful completion of a search. A results list is a linked list of DtSrResult structures, where each node represents a database document that satisfied the query. link Pointer to the next results list node. flags (reserved) objflags The constant DtSrFlNOTAVAIL means that the object is not retrievable from the search engine. objuflags User flags from database record. These are not used by DtSearch and are available for application definition. objsize In uncompressed bytes. objdate Zero is the null date; document is 'undated'. objtype Document type from database header record. Objtype is typically used by application code to identify and launch browsers. Values above x1000 (4096) are set aside for application definition. The following constants identify defined values: DtSrObjUNKNOWN Document type unknown or not applicable DtSrObjTEXT Generic, unformatted flat text DtSrObjBINARY Generic binary object DtSrObjSGML Generic SGML formatted document DtSrObjHTML HTML formatted document DtSrObjPOSTSCR Postscript document DtSrObjINTERLF Interleaf document DtSrObjDTINFO DtInfo document objcost (reserved) dbn Database number; index into dbnames array from DtSearchInit and DtSearchReinit. dba Atomic document identifier within a database. language Language number of the database DtSrLa... constant). reckey Document's unique database key. The first character of reckey is the keytype character. proximity Sort field for ranking results lists. Derived from frequency of occurrence statistics for the query words in the document. Often displayed to users as the subjective 'distance' between the document and the query, in other words a measure of the likelihood that the document will satisfy the user's needs. abstractp Document's abstract string from the database. typedef struct { long offset; /* word location in cleartext */ long length; /* length of word */ } DtSrHitword; Given a text string and the array of search terms returned from DtSearchQuery, DtSearchHighlight will generate a table of offsets and lengths where the search terms are located in the text. The table is typically used to highlight the search terms in the text is a manner appropriate to the application's user interface. The DtSrHitword structure is one element in the table. For each search term to be highlighted, offset specifies the beginning byte for the term, and length specifies the extent of the term in bytes. &cdeman.DtSrAPI;, &cdeman.DtSearchInit;, &cdeman.DtSearchReinit;, &cdeman.DtSearchExit;, &cdeman.DtSearchGetKeytypes;, &cdeman.DtSearchSetMaxResults;, &cdeman.DtSearchGetMaxResults;, &cdeman.DtSearchQuery;, &cdeman.DtSearchRetrieve;, &cdeman.DtSearchHighlight;, &cdeman.DtSearchValidDateString;, &cdeman.DtSearchMergeResults;, &cdeman.DtSearchSortResults;, &cdeman.DtSearchFreeResults;, &cdeman.DtSearchHasMessages;, &cdeman.DtSearchAddMessages;, &cdeman.DtSearchGetMessages;, &cdeman.DtSearchFreeMessages;, &cdeman.DtSearch;