|
Public Methods |
| TaggedTextParser () |
| ~TaggedTextParser () |
void | setTags (const std::vector< std::string > &include, const std::vector< std::string > &exclude, const std::vector< std::string > &index, const std::vector< std::string > &metadata, const std::map< std::string, std::string > &conflations) |
ParsedDocument * | parse (UnparsedDocument *document) |
void | handle (UnparsedDocument *document) |
void | setHandler (ObjectHandler< ParsedDocument > &h) |
Protected Types |
typedef HashTable< std::string,
std::string, StringHash,
StringComparator > | StrHashTable |
Protected Methods |
virtual void | handleToken (char *token, int type, long pos) |
virtual void | initialize (UnparsedDocument *unparsed, ParsedDocument *parsed) |
virtual void | cleanup (UnparsedDocument *unparsed, ParsedDocument *parsed) |
void | addTag (const char *s, const char *c, int pos) |
void | endTag (const char *s, const char *c, int pos) |
void | addMetadataTag (const char *s, const char *c, int pos) |
void | endMetadataTag (const char *s, const char *c, int pos) |
void | writeToken (char *token) |
tag_properties * | _findTag (const char *name) |
tag_properties * | _buildTag (const std::string &name, const std::map< std::string, std::string > &conflations) |
void | handleTag (char *token, long pos) |
Protected Attributes |
TagList * | tl |
TagList * | _metaList |
Buffer | _termBuffer |
HashTable< const char *, tag_properties * > | _tagTable |
const tag_properties * | _startExcludeRegion |
const tag_properties * | _startIncludeRegion |
bool | _exclude |
bool | _include |
bool | _defaultInclude |