OrgMod.cpp File Reference

#include <ncbi_pch.hpp>
#include <util/static_map.hpp>
#include <util/util_misc.hpp>
#include <util/line_reader.hpp>
#include <util/compile_time.hpp>
#include <serial/enumvalues.hpp>
#include <objects/general/general_macros.hpp>
#include <objects/seqfeat/OrgMod.hpp>
#include "institution_codes.inc"

Include dependency graph for OrgMod.cpp:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Go to the SVN repository for this file.

Functions
	DEFINE_STATIC_FAST_MUTEX (s_InstitutionCollectionCodeMutex)
static void	s_ProcessInstitutionCollectionCodeLine (const CTempString &line)
static void	s_InitializeInstitutionCollectionCodeMaps (void)
bool	FindInstCodeAndSpecID (COrgMod::TInstitutionCodeMap &code_map, string &val)
bool	s_IsAllDigits (string str)
bool	s_FixStrainForPrefix (const string &prefix, string &strain)
string	s_FixOneStrain (const string &strain)
void	s_HarmonizeString (string &s)
static bool	s_init_LookForMultipleIsolates (void)

Variables
static COrgMod::TInstitutionCodeMap	s_BiomaterialInstitutionCodeMap
static COrgMod::TInstitutionCodeMap	s_SpecimenVoucherInstitutionCodeMap
static COrgMod::TInstitutionCodeMap	s_CultureCollectionInstitutionCodeMap
static COrgMod::TInstitutionCodeMap	s_CompleteInstitutionCodeMap
static COrgMod::TInstitutionCodeMap	s_CompleteInstitutionFullNameMap
static COrgMod::TInstitutionCodeMap	s_InstitutionCodeTypeMap
static COrgMod::TInstitutionCodeMap	s_InstitutionCodeSynonymsMap
static bool	s_InstitutionCollectionCodeMapInitialized = false
const string	kMissingInst = "Voucher is missing institution code"
const string	kMissingId = "Voucher is missing specific identifier"
const char *	sm_BadStrainValues []
const char *	sm_KnownHostWords []
static constexpr auto	s_hostFixupMap
static const COrgMod::TSubtype	sUnexpectedViralOrgModQualifiers []
static const size_t	sNumUnexpectedViralOrgModQualifiers = sizeof(sUnexpectedViralOrgModQualifiers) / sizeof(COrgMod::TSubtype)
static const string	sValidTypeMaterialPrefixes []
static const int	sNumValidTypeMaterialPrefixes = sizeof(sValidTypeMaterialPrefixes) / sizeof(string)
static const string	sValidCultureTypeMaterialPrefixes []
static const int	sNumValidCultureTypeMaterialPrefixes = sizeof(sValidCultureTypeMaterialPrefixes) / sizeof(string)

Function Documentation

DEFINE_STATIC_FAST_MUTEX()

DEFINE_STATIC_FAST_MUTEX

(

s_InstitutionCollectionCodeMutex

)

FindInstCodeAndSpecID()

bool FindInstCodeAndSpecID	(	COrgMod::TInstitutionCodeMap &	code_map,
		string &	val
	)

Definition at line 562 of file OrgMod.cpp.

References map_checker< Container >::end(), map_checker< Container >::find(), isalpha(), NStr::IsBlank(), isdigit(), len, NStr::TruncateSpacesInPlace(), and val.

Referenced by COrgMod::AddStructureToVoucher().

s_FixOneStrain()

string s_FixOneStrain

(

const string &

strain

)

Definition at line 785 of file OrgMod.cpp.

References kEmptyStr, and s_FixStrainForPrefix().

Referenced by COrgMod::FixStrain().

s_FixStrainForPrefix()

bool s_FixStrainForPrefix	(	const string &	prefix,
		string &	strain
	)

Definition at line 765 of file OrgMod.cpp.

References NStr::eNocase, NStr::IsBlank(), s_IsAllDigits(), NStr::StartsWith(), tmp, and NStr::TruncateSpacesInPlace().

Referenced by s_FixOneStrain().

s_HarmonizeString()

void s_HarmonizeString

(

string &

s

)

Definition at line 1084 of file OrgMod.cpp.

References NStr::ReplaceInPlace().

Referenced by COrgMod::FuzzyStrainMatch().

s_init_LookForMultipleIsolates()

static bool s_init_LookForMultipleIsolates ( void  )

static

Definition at line 1234 of file OrgMod.cpp.

References env, CNcbiApplicationAPI::GetConfig(), CNcbiApplicationAPI::GetEnvironment(), IRegistry::GetString(), CNcbiApplication::Instance(), and NStr::ToLower().

Referenced by COrgMod::NCBI_ValidateForMultipleIsolates().

s_InitializeInstitutionCollectionCodeMaps()

static void s_InitializeInstitutionCollectionCodeMaps ( void  )

static

Definition at line 279 of file OrgMod.cpp.

References file, g_FindDataFile(), g_IsDataFileOld(), i, LOG_POST, lr(), NCBI_CATCH, ILineReader::New(), s_InstitutionCollectionCodeMapInitialized, and s_ProcessInstitutionCollectionCodeLine().

Referenced by COrgMod::AddStructureToVoucher(), COrgMod::FixStructuredVoucher(), COrgMod::GetInstitutionFullName(), COrgMod::GetInstitutionShortName(), and COrgMod::IsInstitutionCodeValid().

s_IsAllDigits()

bool s_IsAllDigits

(

string

str

)

Definition at line 759 of file OrgMod.cpp.

References NPOS, and str().

Referenced by s_FixStrainForPrefix().

s_ProcessInstitutionCollectionCodeLine()

static void s_ProcessInstitutionCollectionCodeLine ( const CTempString &  line )

static

Definition at line 230 of file OrgMod.cpp.

References i, NStr::IsBlank(), NON_CONST_ITERATE, s_BiomaterialInstitutionCodeMap, s_CompleteInstitutionCodeMap, s_CompleteInstitutionFullNameMap, s_CultureCollectionInstitutionCodeMap, s_InstitutionCodeSynonymsMap, s_InstitutionCodeTypeMap, s_SpecimenVoucherInstitutionCodeMap, NStr::Split(), NStr::StartsWith(), and NStr::TruncateSpacesInPlace().

Referenced by s_InitializeInstitutionCollectionCodeMaps().

Variable Documentation

kMissingId

const string kMissingId = "Voucher is missing specific identifier"

Definition at line 458 of file OrgMod.cpp.

Referenced by COrgMod::IsStructuredVoucherValid().

kMissingInst

const string kMissingInst = "Voucher is missing institution code"

Definition at line 457 of file OrgMod.cpp.

Referenced by COrgMod::IsStructuredVoucherValid().

s_BiomaterialInstitutionCodeMap

COrgMod::TInstitutionCodeMap s_BiomaterialInstitutionCodeMap

static

Definition at line 215 of file OrgMod.cpp.

Referenced by COrgMod::AddStructureToVoucher(), and s_ProcessInstitutionCollectionCodeLine().

s_CompleteInstitutionCodeMap

COrgMod::TInstitutionCodeMap s_CompleteInstitutionCodeMap

static

Definition at line 220 of file OrgMod.cpp.

Referenced by COrgMod::GetInstitutionFullName(), and s_ProcessInstitutionCollectionCodeLine().

s_CompleteInstitutionFullNameMap

COrgMod::TInstitutionCodeMap s_CompleteInstitutionFullNameMap

static

Definition at line 221 of file OrgMod.cpp.

Referenced by COrgMod::GetInstitutionShortName(), and s_ProcessInstitutionCollectionCodeLine().

s_CultureCollectionInstitutionCodeMap

COrgMod::TInstitutionCodeMap s_CultureCollectionInstitutionCodeMap

static

Definition at line 217 of file OrgMod.cpp.

Referenced by COrgMod::AddStructureToVoucher(), and s_ProcessInstitutionCollectionCodeLine().

s_hostFixupMap

constexpr auto s_hostFixupMap

staticconstexpr

Initial value:

=
ct::const_map<ct::tagStrNocase, const char*>::construct({
    { "-", "missing" },
    { "no", "missing" },
    { "none", "missing" },
    { "NA", "not available" },
    { "N/A", "not available" },
    { "n/a", "not available" },
    { "free-living", "natural / free-living" },
    { "natural", "natural / free-living" },
    { "not available", "not available" },
    { "not collected", "not collected" },
    { "not applicable", "not applicable" },
    { "NR", "not applicable" },
    { "not known", "unknown" },
    { "other", "missing" },
    { "misc", "missing" },
    { "not determined", "unknown" },
    { "unknown", "unknown" },
    { "not available: to be reported later", "not available" },
    { "obscured", "obscured" },
    { "human", "Homo sapiens" },
    { "homo sapiens", "Homo sapiens" }
})

Definition at line 981 of file OrgMod.cpp.

Referenced by COrgMod::FixHost().

s_InstitutionCodeSynonymsMap

COrgMod::TInstitutionCodeMap s_InstitutionCodeSynonymsMap

static

Definition at line 223 of file OrgMod.cpp.

Referenced by COrgMod::IsInstitutionCodeValid(), and s_ProcessInstitutionCollectionCodeLine().

s_InstitutionCodeTypeMap

COrgMod::TInstitutionCodeMap s_InstitutionCodeTypeMap

static

Definition at line 222 of file OrgMod.cpp.

Referenced by COrgMod::FixStructuredVoucher(), COrgMod::IsInstitutionCodeValid(), and s_ProcessInstitutionCollectionCodeLine().

s_InstitutionCollectionCodeMapInitialized

bool s_InstitutionCollectionCodeMapInitialized = false

static

Definition at line 224 of file OrgMod.cpp.

Referenced by s_InitializeInstitutionCollectionCodeMaps().

s_SpecimenVoucherInstitutionCodeMap

COrgMod::TInstitutionCodeMap s_SpecimenVoucherInstitutionCodeMap

static

Definition at line 216 of file OrgMod.cpp.

Referenced by COrgMod::AddStructureToVoucher(), and s_ProcessInstitutionCollectionCodeLine().

sm_BadStrainValues

const char* sm_BadStrainValues[]

Initial value:

= {
    "yes",
    "no",
    "-",
    "bacteria",
    "sp.",
    "sp",
    "strain",
    "environmental",
    "soil",
    "clinical isolate",
    "NA",
    "whole organism",
    "microbial"
}

Definition at line 820 of file OrgMod.cpp.

Referenced by COrgMod::IsStrainValid().

sm_KnownHostWords

const char* sm_KnownHostWords[]

Definition at line 848 of file OrgMod.cpp.

Referenced by COrgMod::FixHostCapitalization().

sNumUnexpectedViralOrgModQualifiers

const size_t sNumUnexpectedViralOrgModQualifiers = sizeof(sUnexpectedViralOrgModQualifiers) / sizeof(COrgMod::TSubtype)

static

Definition at line 1138 of file OrgMod.cpp.

Referenced by COrgMod::IsUnexpectedViralOrgModQualifier().

sNumValidCultureTypeMaterialPrefixes

const int sNumValidCultureTypeMaterialPrefixes = sizeof(sValidCultureTypeMaterialPrefixes) / sizeof(string)

static

Definition at line 1200 of file OrgMod.cpp.

Referenced by COrgMod::IsValidTypeMaterial().

sNumValidTypeMaterialPrefixes

const int sNumValidTypeMaterialPrefixes = sizeof(sValidTypeMaterialPrefixes) / sizeof(string)

static

Definition at line 1179 of file OrgMod.cpp.

Referenced by COrgMod::IsValidTypeMaterial().

sUnexpectedViralOrgModQualifiers

const COrgMod::TSubtype sUnexpectedViralOrgModQualifiers[]

static

Initial value:

= {
    COrgMod::eSubtype_breed,
    COrgMod::eSubtype_cultivar,
    COrgMod::eSubtype_specimen_voucher
}

Definition at line 1132 of file OrgMod.cpp.

Referenced by COrgMod::IsUnexpectedViralOrgModQualifier().

sValidCultureTypeMaterialPrefixes

const string sValidCultureTypeMaterialPrefixes[]

static

Initial value:

= {
    "epitype",
    "hapantotype",
    "holotype",
    "isoepitype",
    "isoepitype",
    "isolectotype",
    "isoneotype",
    "isoparatype",
    "isosyntype",
    "isotype",
    "lectotype",
    "neotype",
    "paratype",
    "reference",
    "syntype",
    "type material"
}

Definition at line 1181 of file OrgMod.cpp.

Referenced by COrgMod::IsValidTypeMaterial().

sValidTypeMaterialPrefixes

const string sValidTypeMaterialPrefixes[]

static

Initial value:

= {
    "type material",
    "type strain",
    "reference material",
    "reference strain",
    "neotype strain",
    "paralectotype",
    "hapantotype",
    "allotype",
    "culture from reference material",
    "culture from type material",
    "ex-type",
    "culture from hapantotype",
    "pathotype strain"
}

Definition at line 1163 of file OrgMod.cpp.

Referenced by COrgMod::IsValidTypeMaterial().