NCBI C++ ToolKit
Public Types | Static Public Member Functions | Static Private Member Functions | Static Private Attributes | List of all members
CCountries Class Reference

Search Toolkit Book for CCountries

#include <objects/seqfeat/SubSource.hpp>

Public Types

enum  EStateCleanup {
  e_NoResult = 0 , e_Valid = 1 , e_Corrected = 2 , e_Ambiguous = 3 ,
  e_Missing = 4 , e_NotUSA = 5
}
 
typedef CStaticPairArrayMap< const char *, const char *, PCase_CStr > TCStringPairsMap
 
typedef map< string, string, PNocase > TUsaExceptionMap
 

Static Public Member Functions

static bool IsValid (const string &country)
 
static bool IsValid (const string &country, bool &is_miscapitalized)
 
static bool WasValid (const string &country)
 
static bool WasValid (const string &country, bool &is_miscapitalized)
 
static string CapitalizeFirstLetterOfEveryWord (const string &phrase)
 
static string WholeCountryFix (string country)
 
static bool IsSubstringOfStringInList (const string &phrase, const string &country1, size_t pos1)
 
static bool ContainsMultipleCountryNames (const string &phrase)
 
static string GetCorrectedCountryCapitalization (const string &country)
 
static string NewFixCountry (const string &input, bool us_territories=false)
 
static bool ChangeExtraColonsToCommas (string &country)
 
static string CountryFixupItem (const string &input, bool capitalize_after_colon)
 
static void ReadUSAExceptionMap (TUsaExceptionMap &exceptions, const string &filepath)
 
static void LoadUSAExceptionMap (const TUsaExceptionMap &exceptions)
 
static void LoadUSAExceptionMap (const string &exception_file)
 
static string USAStateCleanup (const string &country)
 
static string USAStateCleanup (const string &country, EStateCleanup &type)
 

Static Private Member Functions

static void x_RemoveDelimitersFromEnds (string &val, bool except_paren=false)
 
static vector< string > x_Tokenize (const string &val)
 
static void x_FindCountryName (const TCStringPairsMap &fix_map, const vector< string > &countries, string &valid_country, string &orig_valid_country, bool &too_many_countries, bool &bad_cap)
 

Static Private Attributes

static const string sm_Countries []
 
static const string sm_Former_Countries []
 

Detailed Description

Definition at line 282 of file SubSource.hpp.

Member Typedef Documentation

◆ TCStringPairsMap

typedef CStaticPairArrayMap<const char*, const char*, PCase_CStr> CCountries::TCStringPairsMap

Definition at line 307 of file SubSource.hpp.

◆ TUsaExceptionMap

typedef map<string, string, PNocase> CCountries::TUsaExceptionMap

Definition at line 309 of file SubSource.hpp.

Member Enumeration Documentation

◆ EStateCleanup

enum CCountries::EStateCleanup
Enumerator
e_NoResult 
e_Valid 
e_Corrected 
e_Ambiguous 
e_Missing 
e_NotUSA 

Definition at line 286 of file SubSource.hpp.

Member Function Documentation

◆ CapitalizeFirstLetterOfEveryWord()

string CCountries::CapitalizeFirstLetterOfEveryWord ( const string &  phrase)
static

Definition at line 3851 of file SubSource.cpp.

References isalpha(), NStr::Join(), NStr::Split(), and toupper().

Referenced by CountryFixupItem().

◆ ChangeExtraColonsToCommas()

bool CCountries::ChangeExtraColonsToCommas ( string &  country)
static

Definition at line 4794 of file SubSource.cpp.

References i.

Referenced by NewFixCountry().

◆ ContainsMultipleCountryNames()

bool CCountries::ContainsMultipleCountryNames ( const string &  phrase)
static

Definition at line 3901 of file SubSource.cpp.

References NStr::FindNoCase(), isalpha(), IsSubstringOfStringInList(), ITERATE, NPOS, and s_CountriesSet.

Referenced by NewFixCountry().

◆ CountryFixupItem()

string CCountries::CountryFixupItem ( const string &  input,
bool  capitalize_after_colon 
)
static

Definition at line 4812 of file SubSource.cpp.

References CapitalizeFirstLetterOfEveryWord(), NStr::eTrunc_Begin, NStr::Find(), input(), isspace(), NewFixCountry(), NPOS, and NStr::TruncateSpacesInPlace().

Referenced by BOOST_AUTO_TEST_CASE(), and CCountryFixup::x_ApplyToBioSource().

◆ GetCorrectedCountryCapitalization()

string CCountries::GetCorrectedCountryCapitalization ( const string &  country)
static

Definition at line 3921 of file SubSource.cpp.

References NStr::EqualNocase(), ITERATE, output, and s_CountriesSet.

Referenced by NewFixCountry().

◆ IsSubstringOfStringInList()

bool CCountries::IsSubstringOfStringInList ( const string &  phrase,
const string &  country1,
size_t  pos1 
)
static

Definition at line 3881 of file SubSource.cpp.

References NStr::FindNoCase(), ITERATE, NPOS, r(), and s_CountriesSet.

Referenced by ContainsMultipleCountryNames().

◆ IsValid() [1/2]

bool CCountries::IsValid ( const string &  country)
static

Definition at line 3304 of file SubSource.cpp.

References CStaticArraySearchBase< KeyValueGetter, KeyCompare >::end(), CStaticArraySearchBase< KeyValueGetter, KeyCompare >::find(), NPOS, s_CountriesSet, s_Former_CountriesSet, and s_Null_CountriesSet.

Referenced by CAutoColValidatorcountry::DoValidate(), NewFixCountry(), UpdateRawBioSource(), CValidError_imp::ValidateSubSource(), CCountryFixup::x_ApplyToBioSource(), and x_FindCountryName().

◆ IsValid() [2/2]

bool CCountries::IsValid ( const string &  country,
bool &  is_miscapitalized 
)
static

Definition at line 3329 of file SubSource.cpp.

References CStaticArraySearchBase< KeyValueGetter, KeyCompare >::end(), NStr::EqualNocase(), CStaticArraySearchBase< KeyValueGetter, KeyCompare >::find(), ITERATE, NPOS, s_CountriesSet, s_Former_CountriesSet, and s_Null_CountriesSet.

◆ LoadUSAExceptionMap() [1/2]

void CCountries::LoadUSAExceptionMap ( const string &  exception_file)
static

Definition at line 4617 of file SubSource.cpp.

References LoadUSAExceptionMap(), and ReadUSAExceptionMap().

◆ LoadUSAExceptionMap() [2/2]

void CCountries::LoadUSAExceptionMap ( const TUsaExceptionMap &  exceptions)
static

Definition at line 4592 of file SubSource.cpp.

References map_checker< Container >::clear(), exception_map, exceptions_initialized, NStr::SplitInTwo(), and NStr::TruncateSpacesInPlace().

Referenced by BOOST_AUTO_TEST_CASE(), and LoadUSAExceptionMap().

◆ NewFixCountry()

string CCountries::NewFixCountry ( const string &  input,
bool  us_territories = false 
)
static

Definition at line 4664 of file SubSource.cpp.

References ChangeExtraColonsToCommas(), ContainsMultipleCountryNames(), NStr::EndsWith(), NStr::eNocase, NStr::Equal(), NStr::EqualCase(), NStr::EqualNocase(), NStr::Find(), GetCorrectedCountryCapitalization(), input(), IsValid(), kEmptyStr, NPOS, s_SuppressCountryFix(), NStr::SplitInTwo(), NStr::StartsWith(), test, NStr::TruncateSpacesInPlace(), USAStateCleanup(), WholeCountryFix(), x_FindCountryName(), x_RemoveDelimitersFromEnds(), and x_Tokenize().

Referenced by CSubSource::AutoFix(), BOOST_AUTO_TEST_CASE(), CountryFixupItem(), and s_DoUSAStateCleanup().

◆ ReadUSAExceptionMap()

void CCountries::ReadUSAExceptionMap ( CCountries::TUsaExceptionMap &  exceptions,
const string &  filepath 
)
static

Definition at line 4577 of file SubSource.cpp.

References row.

Referenced by LoadUSAExceptionMap().

◆ USAStateCleanup() [1/2]

string CCountries::USAStateCleanup ( const string &  country)
static

Definition at line 4656 of file SubSource.cpp.

References e_NoResult.

Referenced by NewFixCountry(), and s_USAStateTest().

◆ USAStateCleanup() [2/2]

string CCountries::USAStateCleanup ( const string &  country,
CCountries::EStateCleanup &  type 
)
static

Definition at line 4627 of file SubSource.cpp.

References e_Corrected, e_NotUSA, e_Valid, NStr::Equal(), exception_map, exceptions_initialized, s_DoUSAStateCleanup(), and NStr::StartsWith().

◆ WasValid() [1/2]

bool CCountries::WasValid ( const string &  country)
static

Definition at line 3377 of file SubSource.cpp.

References CStaticArraySearchBase< KeyValueGetter, KeyCompare >::end(), CStaticArraySearchBase< KeyValueGetter, KeyCompare >::find(), NPOS, and s_Former_CountriesSet.

Referenced by UpdateRawBioSource(), and CValidError_imp::ValidateSubSource().

◆ WasValid() [2/2]

bool CCountries::WasValid ( const string &  country,
bool &  is_miscapitalized 
)
static

Definition at line 3391 of file SubSource.cpp.

References CStaticArraySearchBase< KeyValueGetter, KeyCompare >::end(), NStr::EqualNocase(), CStaticArraySearchBase< KeyValueGetter, KeyCompare >::find(), ITERATE, NPOS, and s_Former_CountriesSet.

◆ WholeCountryFix()

string CCountries::WholeCountryFix ( string  country)
static

Definition at line 3861 of file SubSource.cpp.

References NStr::EqualNocase(), i, s_USAStates, and NStr::ToLower().

Referenced by NewFixCountry().

◆ x_FindCountryName()

void CCountries::x_FindCountryName ( const TCStringPairsMap &  fix_map,
const vector< string > &  countries,
string &  valid_country,
string &  orig_valid_country,
bool &  too_many_countries,
bool &  bad_cap 
)
staticprivate

Definition at line 4046 of file SubSource.cpp.

References check, CStaticArraySearchBase< KeyValueGetter, KeyCompare >::end(), CStaticArraySearchBase< KeyValueGetter, KeyCompare >::find(), IsValid(), NStr::TruncateSpacesInPlace(), and x_RemoveDelimitersFromEnds().

Referenced by NewFixCountry().

◆ x_RemoveDelimitersFromEnds()

void CCountries::x_RemoveDelimitersFromEnds ( string &  val,
bool  except_paren = false 
)
staticprivate

Definition at line 3933 of file SubSource.cpp.

References NStr::EndsWith(), isalpha(), ispunct(), isspace(), len, NStr::StartsWith(), NStr::TruncateSpacesInPlace(), and val.

Referenced by NewFixCountry(), and x_FindCountryName().

◆ x_Tokenize()

vector< string > CCountries::x_Tokenize ( const string &  val)
staticprivate

Definition at line 3983 of file SubSource.cpp.

References NStr::Find(), first(), NPOS, NStr::Split(), and val.

Referenced by NewFixCountry().

Member Data Documentation

◆ sm_Countries

const string CCountries::sm_Countries[]
staticprivate

Definition at line 317 of file SubSource.hpp.

◆ sm_Former_Countries

const string CCountries::sm_Former_Countries[]
staticprivate

Definition at line 318 of file SubSource.hpp.

The documentation for this class was generated from the following files:
Generated by doxygen 1.9.0
Modified on Tue Apr 30 06:37:54 2024 by modify_doxy.py rev. 669887