def _get_text_for_grounding(stmt, agent_text):
"""Get text context for Adeft disambiguation
If the INDRA database is available, attempts to get the fulltext from
which the statement was extracted. If the fulltext is not available, the
abstract is returned. If the indra database is not available, uses the
pubmed client to get the abstract. If no abstract can be found, falls back
on returning the evidence text for the statement.
Parameters
----------
stmt : py:class:`indra.statements.Statement`
Statement with agent we seek to disambiguate.
agent_text : str
Agent text that needs to be disambiguated
Returns
-------
text : str
Text for Adeft disambiguation
"""
text = None
# First we will try to get content from the DB
try:
from indra_db.util.content_scripts \
import get_text_content_from_text_refs
from indra.literature.adeft_tools import universal_extract_text
refs = stmt.evidence[0].text_refs
# Prioritize the pmid attribute if given
if stmt.evidence[0].pmid:
refs['PMID'] = stmt.evidence[0].pmid
logger.debug('Obtaining text for disambiguation with refs: %s' %
refs)
content = get_text_content_from_text_refs(refs)
if not content:
raise ValueError('Text obtained from DB is empty')
text = universal_extract_text(content, contains=agent_text)
if text:
return text
except Exception as e:
logger.info('Could not get text for disambiguation from DB: %s' % e)
# If that doesn't work, we try PubMed next
if text is None:
from indra.literature import pubmed_client
pmid = stmt.evidence[0].pmid
if pmid:
logger.debug('Obtaining abstract for disambiguation for PMID%s' %
pmid)
text = pubmed_client.get_abstract(pmid)
if text:
return text
# Finally, falling back on the evidence sentence
if text is None:
logger.info('Falling back on sentence-based disambiguation')
text = stmt.evidence[0].text
return text
return None
def _get_text_for_grounding(stmt, agent_text):
"""Get text context for Adeft disambiguation
If the INDRA database is available, attempts to get the fulltext from
which the statement was extracted. If the fulltext is not available, the
abstract is returned. If the indra database is not available, uses the
pubmed client to get the abstract. If no abstract can be found, falls back
on returning the evidence text for the statement.
Parameters
----------
stmt : py:class:`indra.statements.Statement`
Statement with agent we seek to disambiguate.
agent_text : str
Agent text that needs to be disambiguated
Returns
-------
text : str
Text for Adeft disambiguation
"""
text = None
# First we will try to get content from the DB
try:
from indra_db.util.content_scripts \
import get_text_content_from_text_refs
from indra.literature.adeft_tools import universal_extract_text
refs = stmt.evidence[0].text_refs
# Prioritize the pmid attribute if given
if stmt.evidence[0].pmid:
refs['PMID'] = stmt.evidence[0].pmid
logger.info('Obtaining text for disambiguation with refs: %s' %
refs)
content = get_text_content_from_text_refs(refs)
text = universal_extract_text(content, contains=agent_text)
if text:
return text
except Exception as e:
logger.info('Could not get text for disambiguation from DB.')
# If that doesn't work, we try PubMed next
if text is None:
from indra.literature import pubmed_client
pmid = stmt.evidence[0].pmid
if pmid:
logger.info('Obtaining abstract for disambiguation for PMID%s' %
pmid)
text = pubmed_client.get_abstract(pmid)
if text:
return text
# Finally, falling back on the evidence sentence
if text is None:
logger.info('Falling back on sentence-based disambiguation')
text = stmt.evidence[0].text
return text
return None
def _get_text_for_grounding(self, stmt, agent_text):
"""Get text context for Adeft disambiguation
If the INDRA database is available, attempts to get the fulltext from
which the statement was extracted. If the fulltext is not available,
the abstract is returned. If the indra database is not available, uses
the pubmed client to get the abstract. If no abstract can be found,
falls back on returning the evidence text for the statement.
Parameters
----------
stmt : py:class:`indra.statements.Statement`
Statement with agent we seek to disambiguate.
agent_text : str
Agent text that needs to be disambiguated
Returns
-------
text : str
Text for Adeft disambiguation
"""
text = None
# First we will try to get content from a local text content DB if
# available since this is the fastest option
if self.has_local_text_db:
try:
from indra_db_lite import get_plaintexts_for_text_ref_ids, \
get_text_ref_ids_for_pmids
refs = stmt.evidence[0].text_refs
trid = refs.get('TRID')
pmid = refs.get('PMID')
if trid:
text_content = get_plaintexts_for_text_ref_ids([trid])
_, content = next(text_content.trid_content_pairs())
if content:
return content
elif pmid:
mappings = get_text_ref_ids_for_pmids([int(pmid)])
if int(pmid) in mappings:
trid = mappings[int(pmid)]
text_content = get_plaintexts_for_text_ref_ids([trid])
_, content = next(text_content.trid_content_pairs())
if content:
return content
except Exception as e:
logger.info('Could not get text from local DB: %s' % e)
# If the above is not available or fails, we try the INDRA DB
# if available.
if self.__tc is not None:
try:
from indra.literature.adeft_tools import universal_extract_text
refs = stmt.evidence[0].text_refs
# Prioritize the pmid attribute if given
if stmt.evidence[0].pmid:
refs['PMID'] = stmt.evidence[0].pmid
logger.debug(
'Obtaining text for disambiguation with refs: %s' % refs)
content = self.__tc.get_text_content_from_text_refs(refs)
if not content:
raise ValueError('Text obtained from DB is empty')
text = universal_extract_text(content, contains=agent_text)
if text:
return text
except Exception as e:
logger.info(
'Could not get text for disambiguation from DB: %s' % e)
# If that doesn't work, we try PubMed next trying to fetch an abstract
if text is None:
from indra.literature import pubmed_client
pmid = stmt.evidence[0].pmid
if pmid:
logger.debug(
'Obtaining abstract for disambiguation for PMID%s' % pmid)
text = pubmed_client.get_abstract(pmid)
if text:
return text
# Finally, falling back on the evidence sentence
if text is None:
logger.info('Falling back on sentence-based disambiguation')
text = stmt.evidence[0].text
return text
return None
if __name__ == '__main__':
parser = argparse.ArgumentParser(description='Get texts for statements'
' with agent from a list of shortforms')
parser.add_argument('vars', nargs='*')
args = parser.parse_args()
shortforms = args.vars
all_stmts = set()
cased_shortforms = [
escape_filename(shortform) for shortform in sorted(shortforms)
]
for shortform in shortforms:
cased_shortform = escape_filename(shortform)
path = os.path.join(DATA_PATH, 'statements',
f'{cased_shortform}_statements.json')
with open(path, 'r') as f:
stmts = json.load(f)
all_stmts.update(stmts)
ref_dict, text_dict = get_text_content_from_stmt_ids(all_stmts)
text_dict = {
text_ref: universal_extract_text(article, contains=shortforms)
for text_ref, article in text_dict.items()
}
agg_name = ':'.join(cased_shortforms)
dir_path = os.path.join(DATA_PATH, 'texts', agg_name)
if not os.path.exists(dir_path):
os.makedirs(dir_path)
with open(os.path.join(dir_path, f'{agg_name}_texts.json'), 'w') as f:
json.dump(text_dict, f)
with open(os.path.join(dir_path, f'{agg_name}_text_map.json'), 'w') as f:
json.dump(ref_dict, f)
def get_text_content(pmid):
content = get_text_content_from_text_refs(text_refs={'PMID': pmid}, db=db)
if content:
text = universal_extract_text(content)
return text
return None
