presage
0.9.2~beta
|
#include <forwardTokenizer.h>
Public Member Functions | |
ForwardTokenizer (std::istream &stream, const std::string blankspaces, const std::string separators) | |
virtual | ~ForwardTokenizer () |
virtual int | countTokens () |
virtual bool | hasMoreTokens () const |
virtual std::string | nextToken () |
virtual double | progress () const |
Public Member Functions inherited from Tokenizer | |
Tokenizer (std::istream &stream, const std::string blankspaces, const std::string separators) | |
virtual | ~Tokenizer () |
void | blankspaceChars (const std::string) |
std::string | blankspaceChars () const |
void | separatorChars (const std::string) |
std::string | separatorChars () const |
void | lowercaseMode (const bool) |
bool | lowercaseMode () const |
std::string | streamToString () const |
Additional Inherited Members | |
Protected Member Functions inherited from Tokenizer | |
bool | isBlankspace (const int character) const |
bool | isSeparator (const int character) const |
Protected Attributes inherited from Tokenizer | |
std::istream & | stream |
std::ios::iostate | sstate |
std::streamoff | offbeg |
std::streamoff | offend |
std::streamoff | offset |
Tokenizes a stream from the beginning to the end.
|_|_|_|_|_|_|_| | | | | | offend | offset offbeg
Definition at line 38 of file forwardTokenizer.h.
ForwardTokenizer::ForwardTokenizer | ( | std::istream & | stream, |
const std::string | blankspaces, | ||
const std::string | separators | ||
) |
Definition at line 27 of file forwardTokenizer.cpp.
References Tokenizer::offbeg, and Tokenizer::offset.
|
virtual |
Definition at line 36 of file forwardTokenizer.cpp.
|
virtual |
Returns the number of tokens left.
Implements Tokenizer.
Definition at line 39 of file forwardTokenizer.cpp.
References hasMoreTokens(), nextToken(), Tokenizer::offbeg, Tokenizer::offset, and Tokenizer::stream.
|
virtual |
Tests if there are more tokens.
Implements Tokenizer.
Definition at line 61 of file forwardTokenizer.cpp.
References Tokenizer::offend, and Tokenizer::offset.
Referenced by countTokens(), ContextTracker::learn(), and main().
|
virtual |
Returns the next token.
Implements Tokenizer.
Definition at line 72 of file forwardTokenizer.cpp.
References endl(), Tokenizer::isBlankspace(), Tokenizer::isSeparator(), Tokenizer::lowercaseMode(), Tokenizer::offend, Tokenizer::offset, and Tokenizer::stream.
Referenced by countTokens(), ContextTracker::learn(), and main().
|
virtual |
Returns progress percentage.
Implements Tokenizer.
Definition at line 151 of file forwardTokenizer.cpp.
References Tokenizer::offend, and Tokenizer::offset.
Referenced by main().