RetroSearch Browse

.GetImp().IsSetLanguage() &&

.GetImp().GetLanguage() ==

"Eng"

)

167 journal

.SetImp().ResetLanguage();

177 if

(medlines.size() != 1) {

181 CPub

& pub = *medlines.front();

203

medlines.push_back(pmid);

206

medlines.push_back(cit_art);

235 static const size_t

ISSN_SIZE = 9;

236 static const size_t

ISSN_DASH_POS = 4;

237 static const size_t

ISSN_X_POS = 8;

243 for

(

size_t i

= 0;

< ISSN_SIZE; ++

) {

245 if

(

isdigit

(ch) || (ch ==

'-'

== ISSN_DASH_POS) || (ch ==

'X'

== ISSN_X_POS)) {

321 for

(

int

retry=0; retry<10; ++retry) {

324 auto

pRes =

Ref

(

new

esearch::CESearchResult());

327 if

(istr.GetStatusCode() == 200) {

328 if

(pRes->IsSetData()) {

329 if

(pRes->GetData().IsInfo() &&

330

pRes->GetData().GetInfo().IsSetContent() &&

331

pRes->GetData().GetInfo().GetContent().IsSetIdList()) {

333 const auto

& idList = pRes->GetData().GetInfo().GetContent().GetIdList();

334 if

(idList.IsSetId()) {

335

ids = idList.GetId();

341 if

(pRes->GetData().IsERROR()) {

343

pRes->GetData().GetERROR());

354 int

sleepSeconds =

int

(sqrt(retry));

361 "failed to execute query: "

+ term);

366 const string

id

) {

373 bool

success=

false

;

374 for

(

int

retry=0; retry<10; ++retry) {

378 if

(istr.GetStatusCode() == 200) {

387 int

sleepSeconds =

int

(sqrt(retry));

395 "failed to execute esummary request: "

+ request.

GetQueryString

());

398 static const string

indexingElement {

"<CurrentIndexingStatus>Y</CurrentIndexingStatus>"

};

400 if

(firstPos ==

NPOS

) {

405 return

firstPos == lastPos;

445 for

(

const string

id

: ids) {

471

last_name = namestd.

GetLast

();

490 const CImprint

* imprint =

nullptr

;

491 const CTitle

* title =

nullptr

;

513

imprint = &book.

GetImp

();

518 static const string

UNKNOWN_JOURNAL(

"journal unknown"

);

519 string

title_str(UNKNOWN_JOURNAL);

521 if

(title && title->

IsSet

() && !title->

Get

().empty()) {

524 const string

str

= title->

GetTitle

(first_title);

531 static const string

NO_PAGE(

"no page number"

);

532 static const string

NO_VOL(

"no volume number"

);

538 bool

in_press =

false

;

558

ostr << last_name <<

" "

<< first_name <<

"|"

<< title_str <<

"|("

<< year <<

")|"

<< vol <<

"|"

<< page;

559 const string

tail = ostr.str();

563 "Too many author name differences: "

<< muid <<

"|"

<< tail);

570 static const int

YEAR_MAX_DIFF = 2;

572 if

(year && cur_year - year > YEAR_MAX_DIFF) {

574 "encountered in-press article more than "

<< YEAR_MAX_DIFF <<

" years old: "

<< tail);

580

muid <<

"|"

<< tail);

585 ">>"

<< muid <<

"<<|"

<< tail);

595 ">>"

<< muid <<

"<<|"

<< tail);

626 auto

StrNotEmpty = [](

const string

str

) ->

bool

{

return

str

.empty(); };

627 size_t

new_num_of_authors = count_if(new_names.

().begin(), new_names.

().end(), StrNotEmpty),

628

num_of_authors = count_if(old_names.

().begin(), old_names.

().end(), StrNotEmpty);

631 for

(

const string

& name : old_names.

()) {

640 size_t

min_num_of_authors =

min

(num_of_authors, new_num_of_authors);

646 static const size_t

MAX_AUTHORS = 10;

647 if

(min_num_of_authors > MAX_AUTHORS) {

656 size_t

num_of_names = 0;

658 for

(

const auto

& name :

names

)

667

extracted.push_back(cur_consortium);

678 if

(!consortia.empty()) {

682 for

(

const auto

& pAuthor : authors) {

684

pAuthor->IsSetName() &&

685

pAuthor->GetName().IsConsortium()) {

686

consortia.push_back(pAuthor->GetName().GetConsortium());

690 if

(!consortia.empty()) {

707 if

(x.size() != y.size()) {

713 for

(

auto

xit = begin(x); xit != end(x); ++xit, ++yit) {

720 if

(xval_starts_with_the && !yval_starts_with_the) {

726 else if

(yval_starts_with_the && !xval_starts_with_the) {

755

list<string> old_consortia;

760

list<string> new_consortia;

762 auto

old_consortia_string =

(old_consortia,

";"

);

764 "Publication as returned by MedArch lacks consortium authors of the original publication : \"" 765

<< old_consortia_string <<

"\"."

);

767 auto

& std_list = new_cit.

SetAuthors

().SetNames().SetStd();

770

back_inserter(std_list),

771

[](

const string

& consortium) {

773

pAuthor->SetName().SetConsortium(consortium);

777 auto

old_consortia_string =

(old_consortia,

";"

);

778 auto

new_consortia_string =

(new_consortia,

";"

);

780 "Consortium author names differ. Original is \"" 781

<< old_consortia_string

782

"\". MedArch's is \"" 783

<< new_consortia_string <<

"\"."

);

791 static const size_t

MAX_EXTRACTED = 10;

792 size_t

extracted = 0;

794 for

(

const auto

& name :

names

) {

795 if

(name->IsSetName() && name->GetName().IsName() && name->GetName().GetName().IsSetLast()) {

796

res.push_back(name->GetName().GetName().GetLast());

799 if

(extracted == MAX_EXTRACTED) {

831 if

(!old_consortiums.empty()) {

833 string

old_cons_list =

(old_consortiums,

";"

);

834 if

(new_consortiums.empty()) {

837 "Publication as returned by MedArch lacks consortium authors of the original publication : \""

<< old_cons_list <<

"\"."

);

839

for_each(old_consortiums.begin(), old_consortiums.end(),

840

[&new_cit](

const string

& consortium) {

842

CRef<CAuthor> auth(new CAuthor);

843

auth->SetName().SetConsortium(consortium);

845

new_cit.SetAuthors().SetNames().SetStd().push_front(auth);

850 string

new_cons_list =

(new_consortiums,

";"

);

853 "Consortium author names differ. Original is \""

<< old_cons_list <<

"\". MedArch's is \""

<< new_cons_list <<

"\"."

);

857 if

(num_names == 0) {

862

list<CTempString> new_author_names;

872 if

(find_if(new_author_names.begin(), new_author_names.end(),

875

return NStr::EqualNocase(last_name, cur_name);

876

}) != new_author_names.end()) {

883 size_t

min_num_names =

min

(num_names, new_author_names.size());

888 bool

replace_authors = new_num_names == 0;

889 if

(!replace_authors && new_num_names < num_names) {

917 if

(!replace_authors)

919 static const int

MIN_FIRST_AUTHORS_THRESHOLD_1995 = 10;

920 static const int

MAX_FIRST_AUTHORS_THRESHOLD_1995 = 12;

922 static const int

MIN_SECOND_AUTHORS_THRESHOLD_1999 = 25;

923 static const int

MAX_SECOND_AUTHORS_THRESHOLD_1999 = 27;

925

replace_authors = (new_num_names < MAX_FIRST_AUTHORS_THRESHOLD_1995 && num_names > MIN_FIRST_AUTHORS_THRESHOLD_1995) ||

926

(new_num_names < MAX_SECOND_AUTHORS_THRESHOLD_1999 && num_names > MIN_SECOND_AUTHORS_THRESHOLD_1999);

930 if

(replace_authors) {

946 for

(

const auto

& cur_id : old_ids.

Get

()) {

948 if

(!cur_id->IsDoi() && !cur_id->IsOther()) {

955 auto

& new_ids = cit_new.

GetIds

().

Get

();

956

found = find_if(new_ids.begin(), new_ids.end(),

959

if (cur_id->Which() != new_id->Which()) {

967 bool

res = cur_id->GetOther().IsSetDb() == new_id->

GetOther

().

IsSetDb

();

968 if

(res && cur_id->GetOther().IsSetDb()) {

969

res = cur_id->GetOther().GetDb() == new_id->GetOther().GetDb();

976

cit_new.

SetIds

().Set().push_front(cur_id);

995 if

.GetImp().IsSetVolume() || !

.GetImp().IsSetPages() || !

.GetImp().IsSetDate()) {

1018

imprint = &cit_art.

SetFrom

().SetJournal().SetImp();

1024

imprint = &cit_art.

SetFrom

().SetBook().SetImp();

1030

imprint = &cit_art.

SetFrom

().SetProc().SetBook().SetImp();

1054 if

(pub_equiv.

IsSet

()) {

1055 for

(

auto

& pub : pub_equiv.

Set

())

1057 if

(pub->IsMuid()) {

1058

muids.push_back(pub);

1060 else if

(pub->IsPmid()) {

1061

pmids.push_back(pub);

1063 else if

(pub->IsArticle()) {

1065

others.push_back(pub);

1068

cit_arts.push_back(pub);

1071 else if

(pub->IsMedline()) {

1072

medlines.push_back(pub);

1075

others.push_back(pub);

1080 auto

& pub_list = pub_equiv.

Set

();

1083 if

((!muids.empty() || !pmids.empty()) && !m_always_lookup) {

1085

pub_list.splice(pub_list.end(), cit_arts);

1086

pub_list.splice(pub_list.end(), muids);

1087

pub_list.splice(pub_list.end(), pmids);

1088

pub_list.splice(pub_list.end(), medlines);

1089

pub_list.splice(pub_list.end(), others);

1093

pub_list.splice(pub_list.end(), others);

1095 if

(!medlines.empty())

1097 if

(medlines.size() > 1) {

1103

pub_list.splice(pub_list.end(), medlines);

1107 if

(!pmids.empty()) {

1109

oldpmid = pmids.front()->GetPmid();

1112 for

(

const auto

& pub : pmids) {

1113 if

(pub->GetPmid() != oldpmid) {

1115 "Two different pmids in Pub-equiv ["

<< oldpmid <<

"] ["

<< pub->GetPmid() <<

"]"

);

1122 if

(!muids.empty()) {

1124

oldmuid = muids.front()->GetMuid();

1127 for

(

const auto

& pub : muids) {

1128 if

(pub->GetMuid() != oldmuid) {

1130 "Two different muids in Pub-equiv ["

<< oldmuid <<

"] ["

<< pub->GetMuid() <<

"]"

);

1136 if

(!cit_arts.empty()) {

1137 if

(cit_arts.size() > 1) {

1143 CCit_art

* cit_art = &cit_arts.front()->SetArticle();

1152

pmid = m_upd->CitMatch(*new_pub);

1165 "OldPMID="

<< oldpmid <<

" doesn't match lookup ("

<< pmid <<

"). Keeping lookup."

);

1168 bool

set_pmid =

true

;

1169 if

(m_replace_cit) {

1175 bool

new_cit_is_valid(

false

);

1180

new_cit_is_valid =

true

;

1185

new_cit_is_valid =

true

;

1188

new_cit_is_valid =

false

;

1191 throw

logic_error(

"Invalid outcome returned by CAuthListValidator::validate(): "

+ std::to_string(outcome));

1198 if

(new_cit_is_valid) {

1199 if

(pmids.empty()) {

1201

pmids.push_back(pmid_pub);

1204

pmids.front()->SetPmid().Set(pmid);

1205

pub_list.splice(pub_list.end(), pmids);

1209

pub_list.push_back(cit_pub);

1216

cit_arts.push_back(cit_pub);

1217

cit_art = new_cit_art;

1223

pub_list.splice(pub_list.end(), cit_arts);

1230 "Failed to get pub from MedArch server for pmid = "

<< pmid <<

". Input one is preserved."

);

1235 if

(pmids.empty()) {

1237

pmids.push_back(pmid_pub);

1240

pmids.front()->SetPmid().Set(pmid);

1241

pub_list.splice(pub_list.end(), pmids);

1245

pub_list.splice(pub_list.end(), cit_arts);

1254

pub_list.splice(pub_list.end(), cit_arts);

1266

pub_list.splice(pub_list.end(), pmids);

1268 if

(m_replace_cit) {

1272

pub_list.push_back(cit_pub);

1278 "Cant find article for pmid ["

<< oldpmid <<

"]"

);

1282

pub_list.splice(pub_list.end(), pmids);

1285

pub_list.splice(pub_list.end(), muids);

1293 switch

(pub.

Which

()) {

1299

pub_equiv->

Set

().front()->Assign(pub);

1316

pmid = m_upd->CitMatch(pub);

1324 if

(m_replace_cit) {

1373

pub = upd->

GetPub

(pmid);

1396

enabled = cfg.

GetBool

(section,

"enabled"

, enabled);

1397

cfg_matched_to_min = cfg.

GetDouble

(section,

"matched_to_min"

, cfg_matched_to_min);

1398

cfg_removed_to_gb = cfg.

GetDouble

(section,

"removed_to_gb"

, cfg_removed_to_gb);

1403

: outcome(

eNotSet

), pub_year(0), reported_limit(

"not initialized"

), m_err_log(err_log)

1415 if

(pub_year < 1900 || pub_year > 3000) {

1416 throw

logic_error(

"Publication from PubMed has invalid year: "

+ std::to_string(

pub_year

));

1432 "Too many authors removed ("

cnt_removed

") compared to total Genbank authors ("

")"

);

1443 "Preserving original "

" GB authors, ignoring "

cnt_pm

" PubMed authors " 1455 "Preserving original "

" GB authors, ignoring "

cnt_pm

" PubMed authors " 1466 "Only "

cnt_matched

" authors matched between "

" Genbank and " 1476 out

"\n--- Debug Dump of CAuthListValidator object ---\n"

;

1483 out

"GB author list type: "

gb_type

"; # of entries: "