def run_modules(self, modules: list):
module: BaseModule
cluster = Cluster()
for module in modules:
command = module.get_command()
cluster = command(cluster)
CoreNLP().close()
print()
def _summarize(self, cluster: Cluster) -> Cluster:
doc_id: str
doc: Document
for doc_id, doc in cluster.items():
combined_doc = self._combine_document(doc)
summ_doc = self._extract_summ_doc(combined_doc)
summary = self._generate_summ(summ_doc)
setattr(doc, 'summary', summary)
return cluster
def _gen_summaries(self, cluster: Cluster) -> Cluster:
doc_id: int
doc: Document
for doc_id, doc in cluster.items():
if not hasattr(doc, 'summary'):
raise Exception(f'Summary not found for document {doc_id}')
file_name = f'{doc_id}.system'
file = open(os.path.join(self.folder, file_name), 'w')
file.write(' '.join([sent.text for sent in doc.summary]))
file.close()
return cluster
def _summarize(self, cluster: Cluster):
doc_id: str
doc: Document
for doc_id, doc in cluster.items():
summary = list()
sentence: Sentence
for sentence in doc.bodies[:self._params['n_lines']]:
summary_sent = Sentence()
summary_sent.text = sentence.text
summary_sent.pos = len(summary)
summary.append(summary_sent)
setattr(doc, 'summary', summary)
return cluster
def _annotate(self, cluster: Cluster):
doc_id: str
doc: Document
for doc_id, doc in cluster.items():
sent: Sentence
for sent in doc.bodies:
if self._context.annotation == 'tokenize':
self._tokenize(sent)
elif self._context.annotation == 'pos-tag':
self._pos_tag(sent)
elif self._context.annotation == 'dependency':
self._parse_dep(sent)
return cluster
def add_module_code(self, cluster: Cluster) -> Cluster:
if self._context.annotation == 'tokenize':
cluster.append_code('tokenize')
elif self._context.annotation == 'pos-tag':
cluster.append_code('pos-tag')
elif self._context.annotation == 'dependency':
cluster.append_code('dependency')
return cluster
def _gen_summaries(self, cluster: Cluster) -> Cluster:
doc_id: int
doc: Document
for doc_id, doc in cluster.items():
file_name = f'{doc_id}'
file = open(os.path.join(self.folder, file_name), 'w')
result = ''
example: Example
for example in doc.gold_summaries:
result += ' '.join([sent.text
for sent in example.sentences]) + '\n'
file.write(result.strip())
file.close()
return cluster
def add_module_code(self, cluster: Cluster) -> Cluster:
cluster.append_code('lead')
return cluster