#ifndef _IX_INCLUDED_ #define _IX_INCLUDED_ #import #import #import #import "btree/BTreeFile.h" #define ixMaxKeyLength 255 #define ixMinKeyLength 2 #define ixMaxRegExps 512 // enumeration of error codes reported typedef enum { ixErrorNoError, // no error: the normal state of affairs ixErrorNoMemory, // sufficient memory is not available ixErrorInternal, // an internal inconsistency was found ixErrorSystem = 100 // base value added to system variable errno } ixErrorType; typedef struct { unsigned short regexCount; struct regex *regexTable[ixMaxRegExps]; } ixRegexTable; typedef struct { float threshold; float percentage; unsigned char absolute; unsigned char peculiar; unsigned char discontract; unsigned char foldToLower; unsigned char foldPlurals; unsigned long minCount; unsigned long maxWords; unsigned long windowShift; unsigned long windowSize; WFTable *wordDomain; } ixHistRecord; typedef struct { void *userRecord; unsigned char writeOffsets; unsigned char writeUnknown; unsigned char dispunctuate; void (*controlWord)(char *, void *); vm_offset_t (*readOffset)(void *); void (*writeOffset)(vm_offset_t, void *); int (*charReader)(void *); int (*charWriter)(int, void *); int (*charReturn)(int, void *); } ixRTFRecord; typedef struct FileCell { char *file; // name of file referenced char *body; // pointer to out-of-line data char *desc; // a short human readable description char *type; // the type of the object, if known char *regex; // a regular expression for object char *message; // a message to be sent for object char *data; // a copy of the image of the FileCell struct FileCell *next; // for creating a linked list unsigned long id; // unique key and id number unsigned long pointer; // a pointer to another FileCell unsigned long size; // the size of the FileCell unsigned long date; // the date the FileCell was written unsigned long mode; // access modes at the time of writing unsigned long offset; // offset of referenced object unsigned long length; // length of referenced object unsigned long line; // line no. of referenced object unsigned long bodysize; // amount of data stored at 'body' struct stat statbuf; // for saving the results of a 'stat' char status; // the status of the FileCell } FileCell; // A Reference is a weighted pointer to a FileCell typedef struct Reference { float weight; FileCell *f; } Reference; // A RefList is a list of FileCells, as returned by ixGetRefList typedef struct RefList { unsigned long n; Reference *r; } RefList; // A Registration associates a FileCell and indexing date with a file name typedef struct Registration { unsigned long id; unsigned long date; } Registration; // An Index is a descriptor providing access to an index. typedef struct Index { char *name; // the name of the index file char *folder; // the name of the folder int flag; // a flag word for the index BTreeCursor *registryCursor; BTreeCursor *keywordsCursor; BTreeCursor *fileCellCursor; BTreeFile *btreeFile; // a handle to the index file } Index; // flag bits used in the flag word of the index descriptor #define ixFlagNewIndex (1 << 0) // new index under construction #define ixFlagCreating (1 << 1) // during file creation process // macro for closing an array of Index descriptors returned by ixOpenArray #define ixCloseArray(ix) while (*(ix)) ixClose(*(ix)++) typedef enum { ixSortByDate, ixSortByWeight, ixSortByDescription, ixSortByType, ixSortByName } ixSortMethod; typedef enum { ixSortReverse, ixSortForward } ixSortDirection; typedef enum { ixSearchByPrefix, ixSearchByFullWord, ixSearchAnywhere } ixSearchMethod; typedef enum { ixMatchFileName, ixMatchContent, ixMatchFileNameOrContent } ixMatchAgainst; typedef enum { ixLiteralString, ixRegularExpression } ixQuerySemantic; extern ixErrorType ixErrorNo; extern int ixErrors; extern int ixError(ixErrorType errno, char *fmt, ...); // flags used to specify indexing characteristics extern char *ixFileExtension; // file extension for index files extern char *ixStopWord; // name of stop word list file extern char *ixStopWordDefault; // appended to /usr/lib/indexing/files extern int ixIndexFileNames; // if true, put file names in index // indexing characteristics examined at query time extern int ixFoldToLower; // if true, fold keys to lower case extern int ixFoldPlurals; // if true, plural forms are folded // flags used to specify querying characteristics extern int ixNoCaseQuery; // if true, query is case insensitive extern int ixFindPlurals; // if true, check plural and singular // regular expression tables used by indexing and searching extern ixRegexTable _ixBadNameTable; extern ixRegexTable _ixGoodNameTable; // routines for working with regular expression tables inline static int _ixCheckRegexTable(char *itemString, ixRegexTable *regexTable) { int i; for (i = 0; i < regexTable->regexCount; ++i) if (re_match_exactly(itemString, regexTable->regexTable[i])) return 1; return 0; } inline static int _ixIsGoodName(char *fileName) { return _ixGoodNameTable.regexCount ? _ixCheckRegexTable(fileName, &_ixGoodNameTable) : ! _ixCheckRegexTable(fileName, &_ixBadNameTable); } extern int _ixItemToRegexTable(char *itemString, ixRegexTable *regexTable, int shellSyntax); extern int _ixFileToRegexTable(char *fileName, ixRegexTable *regexTable, int shellSyntax); extern int _ixPathToRegexTable(char *parentName, char *fileName, ixRegexTable *regexTable, int shellSyntax); // routines for opening, closing, flushing and removing indices extern int _ixUnlink(char *name); extern int ixUnlink(char *name); extern Index *ixOpen(char *name, char *mode); extern Index **ixOpenArray(char **path, char *mode); extern Index *ixOpenParent(char *file); extern Index *ixCleanIndex(Index *ix, int verbose); extern void ixFlush(Index *ix); extern void ixClose(Index *ix); // routines for managing FileCells extern void _ixLoadFileCell(unsigned long id, FileCell *fileCell); extern int _ixMergeFileCell(FileCell *fileCell, FileCell *tempFileCell); extern FileCell *_ixReadFileCell(Index *ix, unsigned long id); extern FileCell *_ixReadFirstFileCell(Index *ix); extern FileCell *_ixReadNextFileCell(Index *ix); extern int _ixWriteFileCell(Index *ix, FileCell *fileCell); extern FileCell *ixCopyFileCell(FileCell *fileCell, char *file); extern FileCell *ixFreeFileCell(FileCell *fileCell); extern FileCell *ixInitFileCell(char *file, char *type, char *desc, char *message, char *regex, struct stat *b); extern FileCell *ixLoadFileCell(Index *ix, FileCell *fileCell); extern FileCell *ixMergeFileCell(FileCell *fileCell, FileCell *tempFileCell); extern FileCell *ixNewFileCell(); extern FileCell *ixReadFileCell(Index *ix, unsigned long id); extern FileCell *ixReadFirstFileCell(Index *ix); extern FileCell *ixReadNextFileCell(Index *ix); extern int ixRemoveFileCell(Index *ix, FileCell *fileCell, int remove); extern int ixRemoveFileCellById(Index *ix, unsigned long id, int remove); extern int ixRemoveFileCellByName(Index *ix, char *name, int remove); extern int ixWriteFileCell(Index *ix, FileCell *fileCell); extern int ixWriteStatus(Index *ix, FileCell *fileCell, char status); // routines for managing FileCell registration extern Registration *ixGetRegistration(Index *ix, char *name); extern unsigned long ixGetRegistrationId(Index *ix, char *name); extern FileCell *ixGetRegistrationFileCell(Index *ix, char *name); extern unsigned long ixGetRegistrationDate(Index *ix, char *name); extern int ixPutRegistration(Index *ix, FileCell *f); extern int ixClearRegistrationByName(Index *ix, char *name); extern int ixClearRegistration(Index *ix, FileCell *f); // routines for printing the contents of FileCells and indices extern void ixPrintFileCell(FileCell *f, float weight); extern void ixPrintFileCellById(Index *ix, unsigned long id, int pointerFlag); extern void ixPrintIndexQuery(Index *ix, char *queryString, ixSearchMethod searchBy, ixMatchAgainst matchAgainst, ixQuerySemantic querySemantic, ixSortMethod sortBy, ixSortDirection sortDirection); extern void ixPrintSearchQuery(char *applyFolder, char *queryString, ixSearchMethod searchBy, ixMatchAgainst matchAgainst, ixQuerySemantic querySemantic, ixSortMethod sortBy, ixSortDirection sortDirection); extern void ixPrintRefList(Index *ix, char *s, ixSearchMethod searchBy, ixSortMethod sortBy, ixSortDirection sortDirection); // routines for managing the string matching tables typedef NXHashTable *ixMatchTable; extern ixMatchTable _ixStopWordTable; extern ixMatchTable _ixBadTypeTable; extern ixMatchTable _ixGoodTypeTable; extern ixMatchTable _ixTextTypeTable; inline static int _ixItemToMatchTable(char *itemString, ixMatchTable *matchTable) { if (! *matchTable && ! (*matchTable = NXCreateHashTable(NXStrPrototype, 64, 0))) return 0; return ! NXHashInsert(*matchTable, strsave(itemString)); } inline static int _ixCheckMatchTable(char *itemString, ixMatchTable *matchTable) { return *matchTable ? NXHashMember(*matchTable, itemString) : 0; } inline static int _ixIsGoodType(char *fileType) { return _ixGoodTypeTable ? NXHashMember(_ixGoodTypeTable, fileType) : (!_ixBadTypeTable || !NXHashMember(_ixBadTypeTable, fileType)); } extern int _ixFileToMatchTable(char *fileName, ixMatchTable *matchTable); extern int _ixPathToMatchTable(char *parentName, char *fileName, ixMatchTable *matchTable); extern void ixLoadStopWordTable(void); #define _ixLoadStopWordTable(fileName) \ _ixFileToMatchTable(fileName, &_ixStopWordTable) #define _ixStopWordEnter(stopWord) \ _ixItemToMatchTable(stopWord, &_ixStopWordTable) inline static const char *_ixStopWordCheck(const char *stopWord) { return (const char *) (_ixStopWordTable ? NXHashGet(_ixStopWordTable, stopWord) : 0); } inline static void _ixClearStopWordTable(void) { if (_ixStopWordTable) NXResetHashTable(_ixStopWordTable); } // routines for testing and creating keyword/FileCell associations extern int _ixAssociate(Index *ix, char *k, unsigned long id, float w); extern unsigned long ixAssociate(Index *ix, char *k, unsigned long id, float w); extern int ixAssociated(Index *ix, char *s, unsigned long id); extern unsigned long ixAddFileCell(Index *ix, NXStream *keys, FileCell *f); // routines for getting, clearing and sorting RefLists extern int ixOrRefList(RefList *sourceList, RefList *targetList); extern int ixAndRefList(RefList *sourceList, RefList *targetList); extern int ixAndNotRefList(RefList *sourceList, RefList *targetList); extern void ixSortRefList(RefList *refList, ixSortMethod sortBy, ixSortDirection sortDirection); extern RefList *ixClearRefList(RefList *refList); extern RefList *ixGetRefList(Index *ix, char *queryString, RefList *refList, ixSearchMethod searchBy, int (*interrupt)()); extern RefList ixIndexQuery(Index *ix, char *queryString, ixSearchMethod searchBy, ixMatchAgainst matchAgainst, ixQuerySemantic querySemantic, int (*interrupt)()); extern RefList ixSearchQuery(char *applyFolder, char *queryString, ixSearchMethod searchBy, ixMatchAgainst matchAgainst, ixQuerySemantic querySemantic, int (*interrupt)()); // routines for managing file names and paths extern char *ixParentFolder(Index *ix); extern char *_ixFromParentFolder(Index *ix, char *file); extern char *ixFullPath(Index *ix, char *file); // routines for filtering rich text and generating histograms extern int ixHistogram(NXStream *inputStream, NXStream *outputStream, ixHistRecord *histRecord); extern void _ixReadRichText(ixRTFRecord *rtfRecord); extern void ixReadRichText(NXStream *inputStream, NXStream *outputStream, unsigned long outputLimit); #endif