def normalize_document_data(self, issue, name):
self.json_data['f']['120'] = 'XML_' + return_singleval(self.json_data['f'], '120')
self.json_data['f']['42'] = '1'
if self.json_data['f']['32'] == 'ahead':
self.json_data['f']['121'] = name
if 'epub' in self.json_data['f'].keys():
self.json_data['f'] = self.json_normalizer.normalize_dates(self.json_data['f'], 'epub', '223', 'epub')
if 'epub' in self.json_data['f'].keys():
del self.json_data['f']['epub']
section = Section(return_singleval(self.json_data['f'], '49'))
self.section = issue.toc.return_section(section)
if self.section == None:
self.section = section
self.json_data['f']['49'] = self.section.code
self.normalize_metadata_authors()
self.normalize_illustrative_materials()
self.normalize_affiliations()
self.normalize_keywords()
self.json_data['f'] = self.json_normalizer.convert_value(self.json_data['f'], '71', 'doctopic')
self.json_data['f'] = self.json_normalizer.normalize_dates(self.json_data['f'], '111', '112', '111')
self.json_data['f'] = self.json_normalizer.normalize_dates(self.json_data['f'], '113', '114', '113')
self.json_data['h'] = self.json_normalizer.format_for_indexing(self.json_data['f'])
self.json_data['l'] = self.json_normalizer.format_for_indexing(self.json_data['h'])
def normalize(self, issn_id):
self.json_data['f']['35'] = issn_id
self.json_data['f']['120'] = 'XML_' + return_singleval(self.json_data['f'], '120')
self.json_data['f']['42'] = '1'
self.section = Section(return_singleval(self.json_data['f'], '49'))
#self.json_data['f']['49'] = self.section.code
self.normalize_metadata_authors()
self.normalize_illustrative_materials()
self.normalize_affiliations()
self.normalize_keywords()
self.json_data['f'] = self.json_normalizer.convert_value(self.json_data['f'], '71', 'doctopic')
self.json_data['f'] = self.json_normalizer.normalize_dates(self.json_data['f'], '64', '65', '64')
self.json_data['f'] = self.json_normalizer.normalize_dates(self.json_data['f'], '112', '111', '112')
self.json_data['f'] = self.json_normalizer.normalize_dates(self.json_data['f'], '114', '113', '114')
self.publication_dateiso = return_singleval(self.json_data['f'], '65')
self.json_data['h'] = self.json_normalizer.format_for_indexing(self.json_data['f'])
self.json_data['l'] = self.json_normalizer.format_for_indexing(self.json_data['h'])
def journal_title(self):
if 'f' in self.json_data.keys():
json_data = self.json_data['f']
else:
json_data = self.json_data
r = return_singleval(json_data, '100')
if r == '':
r = return_singleval(json_data, '130')
return r
def return_article(self, name):
titles = return_multval(self.json_data['f'], '12')
authors = return_multval(self.json_data['f'], '10')
first_page = ''
last_page = ''
page = return_singleval(self.json_data['f'], '14')
if type(page) == type({}):
if 'f' in page:
first_page = page['f']
if 'l' in page:
last_page = page['l']
if first_page == '' or last_page == '':
self.article_report.write('\n'+ ' ! ERROR: Missing first and last pages', True, True, False)
data4id = name
article = Article(data4id, first_page, last_page)
article.titles = self.format_titles(titles)
article.authors = self.format_author_names(authors)
article.section = self.section
article.json_data = self.json_data
return article
def return_issue(self, journal):
suppl = ''
order = ''
vol = ''
num = ''
date = ''
if 'f' in self.json_data.keys():
data = self.json_data['f']
else:
data = self.json_data
suppl = return_singleval(data, '131')
suppl = return_singleval(data, '132')
vol = return_singleval(data, '31')
num = return_singleval(data, '32')
date = return_singleval(data, '65')
order = return_singleval(data, '36')
compl = return_singleval(data, '41')
if 'suppl' in num.lower():
if ' ' in num:
if '(' in num:
suppl = num[num.find('(')+1:]
suppl = suppl[0:suppl.find(')')]
else:
suppl = num[num.rfind(' ')+1:]
num = num[0:num.find(' ')]
issue = JournalIssue(journal, vol, num, date, suppl, compl, order)
i_record = {}
keep_list = [30, 31, 32, 132, 35, 42, 65, 100, 480, ]
for key, item in data.items():
if key.isdigit():
if int(key) in keep_list:
i_record[key] = item
i_record['706'] = 'i'
i_record['700'] = '0'
i_record['701'] = '1'
i_record['48'] = []
i_record['48'].append({'l': 'en', 'h': 'Table of Contents'})
i_record['48'].append({'l': 'pt', 'h': 'Sumário'})
i_record['48'].append({'l': 'es', 'h': 'Sumario'})
i_record['36'] = issue.order
i_record['35'] = issue.journal.issn_id
i_record['2'] = 'br1.1'
i_record['930'] = issue.journal.acron.upper()
if issue.number != num:
if '31' in i_record.keys():
del i_record['31']
i_record['32'] = issue.number
issue.json_data = i_record
return issue
def normalize_illustrative_materials(self):
types = {'900': 'TAB', '901': 'GRA'}
illustrative_materials = []
for tag, type in types.items():
count = return_singleval(self.json_data['f'], tag)
if len(count)>0:
if int(count)>0:
illustrative_materials.append(type)
del self.json_data['f'][tag]
if len(illustrative_materials) > 0:
self.json_data['f']['38'] = illustrative_materials
else:
self.json_data['f']['38'] = 'ND'
def normalize_issue_data(self, issn_id):
self.json_data['f']['35'] = issn_id
test_vol = self.json_data['f']['31'].replace('0','')
test_num = self.json_data['f']['32'].replace('0','')
if test_vol + test_num == '':
if '31' in self.json_data['f'].keys():
del self.json_data['f']['31']
self.json_data['f']['32'] = 'ahead'
#self.section = Section(return_singleval(self.json_data['f'], '49'))
#self.json_data['f']['49'] = self.section.code
self.json_data['f'] = self.json_normalizer.normalize_dates(self.json_data['f'], '64', '65', '64')
self.publication_dateiso = return_singleval(self.json_data['f'], '65')
def normalize_citation_title_language(self, citation):
lang = return_singleval(citation, '40')
if len(lang) == 0:
lang = 'en'
if '18' in citation.keys():
monog_title = citation['18']
citation['18'] = { 'l': lang, '_': monog_title}
if '12' in citation.keys():
if type(citation['12']) == type([]):
for title in citation['12']:
if not 'l' in title.keys():
title['l'] = lang
elif type(citation['12']) == type({}):
if not 'l' in citation['12']:
citation['12']['l'] = lang
else:
citation['12'] = { 'l': lang, '_':citation['12']}
return citation
def journal_issn_id(self):
if 'f' in self.json_data.keys():
json_data = self.json_data['f']
else:
json_data = self.json_data
return return_singleval(json_data, '400')
def journal_acron(self):
if 'f' in self.json_data.keys():
json_data = self.json_data['f']
else:
json_data = self.json_data
return return_singleval(json_data, '68')
def validate_pages(self, errors, warnings):
pages = return_singleval(self.json_data['f'], '14')
if not 'f' in pages:
errors.append('Missing pages')
return (errors, warnings)
