sapt::Bitext< TKN > Class Template Reference

#include <ug_bitext.h>

Inheritance diagram for sapt::Bitext< TKN >:

Collaboration diagram for sapt::Bitext< TKN >:

List of all members.

Classes

class agenda

Public Types

typedef TKN Token

typedef TSA< Token >::tree_iterator iter

typedef std::vector
< PhrasePair< Token > > vec_ppair

typedef lru_cache::LRU_Cache
< uint64_t, vec_ppair > pplist_cache_t

typedef TSA< Token > tsa

Public Member Functions

bool find_trg_phr_bounds (PhraseExtractionRecord &rec) const

given the source phrase sid[start:stop]

bool find_trg_phr_bounds (size_t const sid, size_t const start, size_t const stop, size_t &s1, size_t &s2, size_t &e1, size_t &e2, int &po_fwd, int &po_bwd, std::vector< unsigned char > *core_alignment, bitvector *full_alignment, bool const flip) const

SPTR< pstats > prep2 (iter const &phrase, int max_sample=-1) const

SPTR< pstats > prep2 (ttasksptr const &ttask, iter const &phrase, bool const track_sids, int max_sample=-1) const

virtual void open (std::string const base, std::string const L1, std::string const L2)=0

SPTR< pstats > lookup (iter const &phrase, int max_sample=-1) const

void prep (iter const &phrase) const

SPTR< pstats > lookup (ttasksptr const &ttask, iter const &phrase, int max_sample=-1) const

void prep (ttasksptr const &ttask, iter const &phrase, bool const track_sids) const

void setDefaultSampleSize (size_t const max_samples)

size_t getDefaultSampleSize () const

std::string toString (uint64_t pid, int isL2) const

virtual size_t revision () const

SPTR< SentenceBias > loadSentenceBias (std::string const &fname) const

SPTR< DocumentBias > SetupDocumentBias (std::string const &bserver, std::string const &text, std::ostream *log) const

SPTR< DocumentBias > SetupDocumentBias (std::map< std::string, float > context_weights, std::ostream *log) const

void mark_match (Token const *start, Token const *end, iter const &m, bitvector &check) const

void write_yawat_alignment (id_type const sid, iter const *m1, iter const *m2, std::ostream &out) const

std::string sid2docname (id_type const sid) const

std::string docid2name (id_type const sid) const

int docname2docid (std::string const &name) const

std::vector< id_type > const * sid2did () const

int sid2did (uint32_t sid) const

Public Attributes

SPTR< Ttrack< char > > Tx

SPTR< Ttrack< Token > > T1

SPTR< Ttrack< Token > > T2

SPTR< TokenIndex > V1

SPTR< TokenIndex > V2

SPTR< TSA< Token > > I1

SPTR< TSA< Token > > I2

Protected Member Functions

Bitext (size_t const max_sample=1000, size_t const xnum_workers=16)

Bitext (Ttrack< Token > *const t1, Ttrack< Token > *const t2, Ttrack< char > *const tx, TokenIndex *const v1, TokenIndex *const v2, TSA< Token > *const i1, TSA< Token > *const i2, size_t const max_sample=1000, size_t const xnum_workers=16)

Protected Attributes

boost::shared_mutex m_lock

SPTR< agenda > ag

size_t m_num_workers

size_t m_default_sample_size

size_t m_pstats_cache_threshold

SPTR< pstats::cache_t > m_cache1

SPTR< pstats::cache_t > m_cache2

std::vector< std::string > m_docname

std::map< std::string, id_type > m_docname2docid

SPTR< std::vector< id_type > > m_sid2docid

pplist_cache_t m_pplist_cache1

pplist_cache_t m_pplist_cache2

Friends

class BitextSampler

class Moses::Mmsapt

Detailed Description

template<typename TKN>
class sapt::Bitext< TKN >

Definition at line 105 of file ug_bitext.h.

Member Typedef Documentation

template<typename TKN>

typedef TSA<Token>::tree_iterator sapt::Bitext< TKN >::iter

Definition at line 110 of file ug_bitext.h.

template<typename TKN>

typedef lru_cache::LRU_Cache<uint64_t, vec_ppair> sapt::Bitext< TKN >::pplist_cache_t

Definition at line 112 of file ug_bitext.h.

template<typename TKN>

typedef TKN sapt::Bitext< TKN >::Token

Definition at line 109 of file ug_bitext.h.

template<typename TKN>

typedef TSA<Token> sapt::Bitext< TKN >::tsa

Definition at line 113 of file ug_bitext.h.

template<typename TKN>

typedef std::vector<PhrasePair<Token> > sapt::Bitext< TKN >::vec_ppair

Definition at line 111 of file ug_bitext.h.

Constructor & Destructor Documentation

template<typename Token >

sapt::Bitext< Token >::Bitext	(	size_t const	max_sample = `1000`,
		size_t const	xnum_workers = `16`
	)			`[inline, protected]`

Definition at line 876 of file ug_bitext.h.

template<typename Token >

sapt::Bitext< Token >::Bitext	(	Ttrack< Token > *const	t1,
		Ttrack< Token > *const	t2,
		Ttrack< char > *const	tx,
		TokenIndex *const	v1,
		TokenIndex *const	v2,
		TSA< Token > *const	i1,
		TSA< Token > *const	i2,
		size_t const	max_sample = `1000`,
		size_t const	xnum_workers = `16`
	)			`[inline, protected]`

Definition at line 886 of file ug_bitext.h.

Member Function Documentation

template<typename Token >

std::string sapt::Bitext< Token >::docid2name ( id_type const sid ) const [inline]

Definition at line 782 of file ug_bitext.h.

Referenced by print_evidence_list().

Here is the caller graph for this function:

template<typename Token >

int sapt::Bitext< Token >::docname2docid ( std::string const & name ) const [inline]

Definition at line 771 of file ug_bitext.h.

template<typename Token >

bool sapt::Bitext< Token >::find_trg_phr_bounds	(	size_t const	sid,
		size_t const	start,
		size_t const	stop,
		size_t &	s1,
		size_t &	s2,
		size_t &	e1,
		size_t &	e2,
		int &	po_fwd,
		int &	po_bwd,
		std::vector< unsigned char > *	core_alignment,
		bitvector *	full_alignment,
		bool const	flip
	)			const `[inline]`

Definition at line 936 of file ug_bitext.h.

template<typename Token >

bool sapt::Bitext< Token >::find_trg_phr_bounds ( PhraseExtractionRecord & rec ) const [inline]

given the source phrase sid[start:stop]

Definition at line 924 of file ug_bitext.h.

template<typename Token >

size_t sapt::Bitext< Token >::getDefaultSampleSize ( ) const [inline]

Definition at line 856 of file ug_bitext.h.

template<typename Token >

SPTR< SentenceBias > sapt::Bitext< Token >::loadSentenceBias ( std::string const & fname ) const [inline]

Definition at line 823 of file ug_bitext.h.

template<typename Token >

SPTR< pstats > sapt::Bitext< Token >::lookup	(	ttasksptr const &	ttask,
		iter const &	phrase,
		int	max_sample = `-1`
	)			const `[inline]`

Definition at line 9 of file ug_bitext_moses.h.

References UTIL_THROW_IF2.

template<typename TKN>

SPTR<pstats> sapt::Bitext< TKN >::lookup	(	iter const &	phrase,
		int	max_sample = `-1`
	)			const

Referenced by main().

Here is the caller graph for this function:

template<typename Token >

void sapt::Bitext< Token >::mark_match	(	Token const *	start,
		Token const *	end,
		iter const &	m,
		bitvector &	check
	)			const `[inline]`

Definition at line 1193 of file ug_bitext.h.

template<typename TKN>

virtual void sapt::Bitext< TKN >::open	(	std::string const	base,
		std::string const	L1,
		std::string const	L2
	)			`[pure virtual]`

Implemented in sapt::imBitext< TKN >, and sapt::mmBitext< TKN >.

template<typename Token >

void sapt::Bitext< Token >::prep	(	ttasksptr const &	ttask,
		iter const &	phrase,
		bool const	track_sids
	)			const `[inline]`

Definition at line 33 of file ug_bitext_moses.h.

References sapt::Bitext< TKN >::m_default_sample_size, and sapt::Bitext< TKN >::prep2().

Here is the call graph for this function:

template<typename Token >

void sapt::Bitext< Token >::prep ( iter const & phrase ) const [inline]

Definition at line 1076 of file ug_bitext.h.

template<typename Token >

SPTR< pstats > sapt::Bitext< Token >::prep2	(	ttasksptr const &	ttask,
		iter const &	phrase,
		bool const	track_sids,
		int	max_sample = `-1`
	)			const `[inline]`

Definition at line 47 of file ug_bitext_moses.h.

References sapt::TSA_tree_iterator< TKN >::approxOccurrenceCount(), sapt::TSA_tree_iterator< TKN >::getPid(), I1, NULL, sapt::TSA_tree_iterator< TKN >::root, and UTIL_THROW_IF2.

Here is the call graph for this function:

template<typename Token >

SPTR< pstats > sapt::Bitext< Token >::prep2	(	iter const &	phrase,
		int	max_sample = `-1`
	)			const `[inline]`

Definition at line 1091 of file ug_bitext.h.

Referenced by sapt::Bitext< TKN >::prep().

Here is the caller graph for this function:

template<typename TKN>

virtual size_t sapt::Bitext< TKN >::revision ( ) const [inline, virtual]

Reimplemented in sapt::imBitext< TKN >.

Definition at line 200 of file ug_bitext.h.

template<typename Token >

void sapt::Bitext< Token >::setDefaultSampleSize ( size_t const max_samples ) [inline]

Definition at line 863 of file ug_bitext.h.

template<typename Token >

SPTR< DocumentBias > sapt::Bitext< Token >::SetupDocumentBias	(	std::map< std::string, float >	context_weights,
		std::ostream *	log
	)			const `[inline]`

Definition at line 1063 of file ug_bitext.h.

template<typename Token >

SPTR< DocumentBias > sapt::Bitext< Token >::SetupDocumentBias	(	std::string const &	bserver,
		std::string const &	text,
		std::ostream *	log
	)			const `[inline]`

Definition at line 1049 of file ug_bitext.h.

template<typename Token >

int sapt::Bitext< Token >::sid2did ( uint32_t sid ) const [inline]

Definition at line 812 of file ug_bitext.h.

template<typename Token >

std::vector< id_type > const * sapt::Bitext< Token >::sid2did ( ) const [inline]

Definition at line 804 of file ug_bitext.h.

Referenced by main().

Here is the caller graph for this function:

template<typename Token >

std::string sapt::Bitext< Token >::sid2docname ( id_type const sid ) const [inline]

Definition at line 793 of file ug_bitext.h.

Referenced by main().

Here is the caller graph for this function:

template<typename Token >

std::string sapt::Bitext< Token >::toString	(	uint64_t	pid,
		int	isL2
	)			const `[inline]`

Definition at line 838 of file ug_bitext.h.

template<typename Token >

void sapt::Bitext< Token >::write_yawat_alignment	(	id_type const	sid,
		iter const *	m1,
		iter const *	m2,
		std::ostream &	out
	)			const `[inline]`

Definition at line 1223 of file ug_bitext.h.

Friends And Related Function Documentation

template<typename TKN>

friend class BitextSampler [friend]

Definition at line 108 of file ug_bitext.h.

template<typename TKN>

friend class Moses::Mmsapt [friend]

Definition at line 114 of file ug_bitext.h.

Member Data Documentation

template<typename TKN>

SPTR<agenda> sapt::Bitext< TKN >::ag [mutable, protected]

Definition at line 118 of file ug_bitext.h.

template<typename TKN>

SPTR<TSA<Token> > sapt::Bitext< TKN >::I1

Definition at line 139 of file ug_bitext.h.

Referenced by main().

template<typename TKN>

SPTR<TSA<Token> > sapt::Bitext< TKN >::I2

Definition at line 140 of file ug_bitext.h.

Referenced by main(), and show().

template<typename TKN>

SPTR<pstats::cache_t> sapt::Bitext< TKN >::m_cache1 [protected]

Definition at line 124 of file ug_bitext.h.

template<typename TKN>

SPTR<pstats::cache_t> sapt::Bitext< TKN >::m_cache2 [protected]

Definition at line 124 of file ug_bitext.h.

template<typename TKN>

size_t sapt::Bitext< TKN >::m_default_sample_size [protected]

Definition at line 122 of file ug_bitext.h.

Referenced by sapt::Bitext< TKN >::prep().

template<typename TKN>

std::vector<std::string> sapt::Bitext< TKN >::m_docname [protected]

Definition at line 126 of file ug_bitext.h.

template<typename TKN>

std::map<std::string,id_type> sapt::Bitext< TKN >::m_docname2docid [protected]

Definition at line 127 of file ug_bitext.h.

template<typename TKN>

boost::shared_mutex sapt::Bitext< TKN >::m_lock [mutable, protected]

Definition at line 116 of file ug_bitext.h.

template<typename TKN>

size_t sapt::Bitext< TKN >::m_num_workers [protected]

Definition at line 120 of file ug_bitext.h.

template<typename TKN>

pplist_cache_t sapt::Bitext< TKN >::m_pplist_cache1 [mutable, protected]

Definition at line 130 of file ug_bitext.h.

template<typename TKN>

pplist_cache_t sapt::Bitext< TKN >::m_pplist_cache2 [mutable, protected]

Definition at line 130 of file ug_bitext.h.

template<typename TKN>

size_t sapt::Bitext< TKN >::m_pstats_cache_threshold [protected]

Definition at line 123 of file ug_bitext.h.

template<typename TKN>

SPTR<std::vector<id_type> > sapt::Bitext< TKN >::m_sid2docid [protected]

Definition at line 128 of file ug_bitext.h.

Referenced by sapt::Bitext< TKN >::agenda::job::done().

template<typename TKN>

SPTR<Ttrack<Token> > sapt::Bitext< TKN >::T1

Definition at line 135 of file ug_bitext.h.

Referenced by sapt::PScoreLex1< Token >::operator()(), sapt::PScoreLengthRatio< Token >::operator()(), and sapt::Bitext< TKN >::agenda::worker::worker().

template<typename TKN>

SPTR<Ttrack<Token> > sapt::Bitext< TKN >::T2

Definition at line 136 of file ug_bitext.h.

Referenced by sapt::PScoreLex1< Token >::operator()(), and sapt::PScoreLengthRatio< Token >::operator()().

template<typename TKN>

SPTR<Ttrack<char> > sapt::Bitext< TKN >::Tx

Definition at line 134 of file ug_bitext.h.

template<typename TKN>

SPTR<TokenIndex> sapt::Bitext< TKN >::V1

Definition at line 137 of file ug_bitext.h.

Referenced by main(), sapt::PScoreLex1< Token >::operator()(), show(), and show_pair().

template<typename TKN>

SPTR<TokenIndex> sapt::Bitext< TKN >::V2

Definition at line 138 of file ug_bitext.h.

Referenced by main(), sapt::PScoreLex1< Token >::operator()(), and show_pair().

The documentation for this class was generated from the following files:

/disk4/html/www/moses/doxygen/mosesdecoder/moses/TranslationModel/UG/mm/ug_bitext.h
/disk4/html/www/moses/doxygen/mosesdecoder/moses/TranslationModel/UG/mm/ug_bitext_moses.h


Classes
class	agenda
Public Types
typedef TKN	Token
typedef TSA< Token >::tree_iterator	iter
typedef std::vector < PhrasePair< Token > >	vec_ppair
typedef lru_cache::LRU_Cache < uint64_t, vec_ppair >	pplist_cache_t
typedef TSA< Token >	tsa
Public Member Functions
bool	find_trg_phr_bounds (PhraseExtractionRecord &rec) const
	given the source phrase sid[start:stop]
bool	find_trg_phr_bounds (size_t const sid, size_t const start, size_t const stop, size_t &s1, size_t &s2, size_t &e1, size_t &e2, int &po_fwd, int &po_bwd, std::vector< unsigned char > core_alignment, bitvector full_alignment, bool const flip) const
SPTR< pstats >	prep2 (iter const &phrase, int max_sample=-1) const
SPTR< pstats >	prep2 (ttasksptr const &ttask, iter const &phrase, bool const track_sids, int max_sample=-1) const
virtual void	open (std::string const base, std::string const L1, std::string const L2)=0
SPTR< pstats >	lookup (iter const &phrase, int max_sample=-1) const
void	prep (iter const &phrase) const
SPTR< pstats >	lookup (ttasksptr const &ttask, iter const &phrase, int max_sample=-1) const
void	prep (ttasksptr const &ttask, iter const &phrase, bool const track_sids) const
void	setDefaultSampleSize (size_t const max_samples)
size_t	getDefaultSampleSize () const
std::string	toString (uint64_t pid, int isL2) const
virtual size_t	revision () const
SPTR< SentenceBias >	loadSentenceBias (std::string const &fname) const
SPTR< DocumentBias >	SetupDocumentBias (std::string const &bserver, std::string const &text, std::ostream *log) const
SPTR< DocumentBias >	SetupDocumentBias (std::map< std::string, float > context_weights, std::ostream *log) const
void	mark_match (Token const start, Token const end, iter const &m, bitvector &check) const
void	write_yawat_alignment (id_type const sid, iter const m1, iter const m2, std::ostream &out) const
std::string	sid2docname (id_type const sid) const
std::string	docid2name (id_type const sid) const
int	docname2docid (std::string const &name) const
std::vector< id_type > const *	sid2did () const
int	sid2did (uint32_t sid) const
Public Attributes
SPTR< Ttrack< char > >	Tx
SPTR< Ttrack< Token > >	T1
SPTR< Ttrack< Token > >	T2
SPTR< TokenIndex >	V1
SPTR< TokenIndex >	V2
SPTR< TSA< Token > >	I1
SPTR< TSA< Token > >	I2
Protected Member Functions
	Bitext (size_t const max_sample=1000, size_t const xnum_workers=16)
	Bitext (Ttrack< Token > const t1, Ttrack< Token > const t2, Ttrack< char > const tx, TokenIndex const v1, TokenIndex const v2, TSA< Token > const i1, TSA< Token > *const i2, size_t const max_sample=1000, size_t const xnum_workers=16)
Protected Attributes
boost::shared_mutex	m_lock
SPTR< agenda >	ag
size_t	m_num_workers
size_t	m_default_sample_size
size_t	m_pstats_cache_threshold
SPTR< pstats::cache_t >	m_cache1
SPTR< pstats::cache_t >	m_cache2
std::vector< std::string >	m_docname
std::map< std::string, id_type >	m_docname2docid
SPTR< std::vector< id_type > >	m_sid2docid
pplist_cache_t	m_pplist_cache1
pplist_cache_t	m_pplist_cache2
Friends
class	BitextSampler
class	Moses::Mmsapt

sapt::Bitext< TKN > Class Template Reference

Classes

Public Types

Public Member Functions

Public Attributes

Protected Member Functions

Protected Attributes

Friends

Detailed Description

template<typename TKN> class sapt::Bitext< TKN >

Member Typedef Documentation

Constructor & Destructor Documentation

Member Function Documentation

Friends And Related Function Documentation

Member Data Documentation

template<typename TKN>
class sapt::Bitext< TKN >