view release on metacpan or  search on metacpan

libswish3.h  view on Meta::CPAN

struct swish_TagStack
{
    swish_Tag         *head;
    swish_Tag         *temp;
    unsigned int       count;
    char              *name;       // debugging aid -- name of the stack
};

struct swish_Analyzer
{
    unsigned int           maxwordlen;         // max word length
    unsigned int           minwordlen;         // min word length
    boolean                tokenize;           // should we parse into TokenList
    int                  (*tokenizer) (swish_TokenIterator*, xmlChar*, swish_MetaName*, xmlChar*);
    xmlChar*             (*stemmer)   (xmlChar*);
    boolean                lc;                 // should tokens be lowercased
    void                  *stash;              // for script bindings
    void                  *regex;              // optional regex
    int                    ref_cnt;            // for script bindings
};

struct swish_Parser
{
    int                    ref_cnt;             // for script bindings
    void                 (*handler)(swish_ParserData*); // handler reference
    void                  *stash;               // for script bindings
    int                    verbosity;           
};

struct swish_ParserData
{
    swish_3               *s3;                 // main object
    xmlBufferPtr           meta_buf;           // tmp MetaName buffer
    xmlBufferPtr           prop_buf;           // tmp Property buffer
    xmlChar               *tag;                // current tag name
    swish_DocInfo         *docinfo;            // document-specific properties
    unsigned int           ignore_content;     // toggle flag. should buffer be indexed.
    boolean                is_html;            // shortcut flag for html parser
    boolean                bump_word;          // boolean for moving word position/adding space
    unsigned int           offset;             // current offset position
    swish_TagStack        *metastack;          // stacks for tracking the tag => metaname
    swish_TagStack        *propstack;          // stacks for tracking the tag => property
    swish_TagStack        *domstack;           // stacks for tracking xml/html dom tree
    xmlParserCtxtPtr       ctxt;               // so we can free at end
    swish_TokenIterator   *token_iterator;     // token container
    swish_NamedBuffer     *properties;         // buffer all properties
    swish_NamedBuffer     *metanames;          // buffer all metanames
};

/*
=cut
*/

/*
=head2 Global Functions
*/
void            swish_setup();
const char *    swish_lib_version();
const char *    swish_libxml2_version();
void            swish_setenv(char * name, char * value, int override);
/*
=cut
*/

/*
=head2 Top-Level Functions
*/
swish_3 *       swish_3_init( void (*handler) (swish_ParserData *), void *stash );
void            swish_3_free( swish_3 *s3 );
int             swish_parse_file( swish_3 * s3, xmlChar *filename );
unsigned int    swish_parse_fh( swish_3 * s3, FILE * fh );
int             swish_parse_buffer( swish_3 * s3, xmlChar * buf );
unsigned int    swish_parse_directory( swish_3 *s3, xmlChar *dir, boolean follow_symlinks );
/*
=cut
*/

/*
=head2 I/O Functions
*/
xmlChar *   swish_io_slurp_fh( FILE * fh, unsigned long flen, boolean binmode );
xmlChar *   swish_io_slurp_file_len( xmlChar *filename, off_t flen, boolean binmode );
xmlChar *   swish_io_slurp_gzfile_len( xmlChar *filename, off_t *flen, boolean binmode );
xmlChar *   swish_io_slurp_file( xmlChar *filename, off_t flen, boolean is_gzipped, boolean binmode );
long int    swish_io_count_operable_file_lines( xmlChar *filename );
boolean     swish_io_is_skippable_line( xmlChar *str );
/*
=cut
*/

/*
=head2 Filesystem Functions
*/
boolean     swish_fs_file_exists( xmlChar *filename );
boolean     swish_fs_is_dir( xmlChar *path );
boolean     swish_fs_is_file( xmlChar *path );
boolean     swish_fs_is_link( xmlChar *path );
off_t       swish_fs_get_file_size( xmlChar *path );
time_t      swish_fs_get_file_mtime( xmlChar *path );
xmlChar *   swish_fs_get_file_ext( xmlChar *url );
xmlChar *   swish_fs_get_path( xmlChar *url );
boolean     swish_fs_looks_like_gz( xmlChar *file );
/*
=cut
*/


/*
=head2 Hash Functions
*/
int         swish_hash_add( xmlHashTablePtr hash, xmlChar *key, void * value );
int         swish_hash_replace( xmlHashTablePtr hash, xmlChar *key, void *value );
int         swish_hash_delete( xmlHashTablePtr hash, xmlChar *key );
boolean     swish_hash_exists( xmlHashTablePtr hash, xmlChar *key );
int         swish_hash_exists_or_add( xmlHashTablePtr hash, xmlChar *key, xmlChar *value );
void        swish_hash_merge( xmlHashTablePtr hash1, xmlHashTablePtr hash2 );
void *      swish_hash_fetch( xmlHashTablePtr hash, xmlChar *key );
void        swish_hash_dump( xmlHashTablePtr hash, const char *label );
xmlHashTablePtr swish_hash_init(int size);
void        swish_hash_free( xmlHashTablePtr hash );