def _write_chartsheet_rels_files(self):
# Write the chartsheet .rels files for links to drawing files.
index = 0
for worksheet in self.workbook.worksheets():
if not worksheet.is_chartsheet:
continue
index += 1
external_links = worksheet.external_drawing_links
if not external_links:
continue
# Create the chartsheet .rels xlsx_dir.
rels = Relationships()
for link_data in external_links:
rels._add_worksheet_relationship(*link_data)
# Create .rels file such as /xl/chartsheets/_rels/sheet1.xml.rels.
rels._set_xml_writer(self._filename('xl/chartsheets/_rels/sheet'
+ str(index) + '.xml.rels'))
rels._assemble_xml_file()
def _write_worksheet_rels_files(self):
# Write data such as hyperlinks or drawings.
index = 0
for worksheet in self.workbook.worksheets():
if worksheet.is_chartsheet:
continue
index += 1
external_links = (worksheet.external_hyper_links +
worksheet.external_drawing_links +
worksheet.external_vml_links +
worksheet.external_table_links +
worksheet.external_comment_links)
if not external_links:
continue
# Create the worksheet .rels dirs.
rels = Relationships()
for link_data in external_links:
rels._add_worksheet_relationship(*link_data)
# Create .rels file such as /xl/worksheets/_rels/sheet1.xml.rels.
rels._set_xml_writer(self._filename('xl/worksheets/_rels/sheet'
+ str(index) + '.xml.rels'))
rels._assemble_xml_file()
def _write_workbook_rels_file(self):
# Write the _rels/.rels xml file.
rels = Relationships()
worksheet_index = 1
chartsheet_index = 1
for worksheet in self.workbook.worksheets():
if worksheet.is_chartsheet:
rels._add_document_relationship('/chartsheet',
'chartsheets/sheet'
+ str(chartsheet_index)
+ '.xml')
chartsheet_index += 1
else:
rels._add_document_relationship('/worksheet',
'worksheets/sheet'
+ str(worksheet_index)
+ '.xml')
worksheet_index += 1
rels._add_document_relationship('/theme', 'theme/theme1.xml')
rels._add_document_relationship('/styles', 'styles.xml')
# Add the sharedString rel if there is string data in the workbook.
if self.workbook.str_table.count:
rels._add_document_relationship('/sharedStrings',
'sharedStrings.xml')
# Add vbaProject if present.
if self.workbook.vba_project:
rels._add_ms_package_relationship('/vbaProject', 'vbaProject.bin')
rels._set_xml_writer(self._filename('xl/_rels/workbook.xml.rels'))
rels._assemble_xml_file()
def get_relationships(self):
if self.relationships:
return self.relationships
doc_path = os.path.join(self.file_path, "word/_rels/document.xml.rels")
with open(doc_path, encoding="UTF-8") as f:
doc = f.read()
doc = BeautifulSoup(doc, "xml")
self.relationships = Relationships(doc)
return self.relationships
def _find_relationships(self, list_tagged, global_entities):
relationships = Relationships()
relation_stops_type = ['CONJ', 'WPRO', ',', '(', ')']
relationship_stop_words = ['ex']
for tagged in list_tagged:
for index_sentence, sentence in enumerate(tagged):
last_entity = None
last_entity_index = 0
last_relation = None
for index, item in enumerate(sentence):
# In order to avoid stop words
if( len(item[0]) == 1 or item[0].lower() in relationship_stop_words):
continue
# to get the entity already identified
elif( item[1] == 'NE'):
# In order to build the relationship
if(last_entity is not None and self._contain_main_entity(last_entity[0], item[0])):
# to build a relationship with anything between entities
# just if there is only one token between entities
if(index-last_entity_index == 2 and len(sentence[index-1][0])>1 ):
id1 = self._search_parent_entity(last_entity[2], global_entities).id()
id2 = self._search_parent_entity(item[2], global_entities).id()
relation = (sentence[index-1][0], id1, last_entity[0], id2, item[0])
relationships.add(relation)
# In order to build a relationship from relation already identified
elif(last_relation is not None):
id1 = self._search_parent_entity(last_entity[2], global_entities).id()
id2 = self._search_parent_entity(item[2], global_entities).id()
relation = (last_relation, id1, last_entity[0], id2, item[0])
relationships.add(relation)
last_entity = item
last_entity_index = index
last_relation = None
# In order to get just relationships between entities
if(last_entity is None):
continue
# In order to get relationship composed by verb and noun
elif('N' in item[1]):
last_relation = self._compose_verb_noun(sentence, index, last_entity_index, relation_stops_type)
# In order to get relationship composed by one or more verbs
elif('VB' in item[1]):
last_relation = self._get_composed_verbs(sentence, index, last_entity_index, relation_stops_type)
# In order to break relationships
elif(item[1] in relation_stops_type):
last_relation = None
last_entity = None
last_entity_index = 0
# In order to remove relationships if a conjuction is found
if last_relation is not None and last_relation[0].isupper():
last_relation = None
return relationships
def _write_root_rels_file(self):
# Write the _rels/.rels xml file.
rels = Relationships()
rels._add_document_relationship('/officeDocument', 'xl/workbook.xml')
rels._add_package_relationship('/metadata/core-properties',
'docProps/core.xml')
rels._add_document_relationship('/extended-properties',
'docProps/app.xml')
rels._set_xml_writer(self._filename('_rels/.rels'))
rels._assemble_xml_file()
def _write_vml_drawing_rels_file(self, worksheet, index):
# Write the vmlDdrawing .rels files for worksheets with images in
# headers or footers.
# Create the drawing .rels dir.
rels = Relationships()
for drawing_data in worksheet.vml_drawing_links:
rels._add_document_relationship(*drawing_data)
# Create .rels file such as /xl/drawings/_rels/vmlDrawing1.vml.rels.
rels._set_xml_writer(self._filename('xl/drawings/_rels/vmlDrawing'
+ str(index)
+ '.vml.rels'))
rels._assemble_xml_file()
def _write_drawing_rels_files(self):
# Write the drawing .rels files for worksheets with charts or drawings.
index = 0
for worksheet in self.workbook.worksheets():
if not worksheet.drawing_links:
continue
index += 1
# Create the drawing .rels xlsx_dir.
rels = Relationships()
for drawing_data in worksheet.drawing_links:
rels._add_document_relationship(*drawing_data)
# Create .rels file such as /xl/drawings/_rels/sheet1.xml.rels.
rels._set_xml_writer(self._filename('xl/drawings/_rels/drawing'
+ str(index) + '.xml.rels'))
rels._assemble_xml_file()
