def process_instrument(row=None,
db=None,
refresh=False,
tree=None,
latest=False,
strategy={'links': process_instrument_links}):
current_app.logger.info('Processing: %s' % row.get('title'))
if not tree:
tree = etree.fromstring(row.get('document'), parser=large_parser)
if not row.get('latest', True):
tree.attrib['old-version'] = 'true'
definitions = Definitions()
extra_formatting(tree, row.get('version'))
start = time.time()
tree = strategy['links'](tree, db)
current_app.logger.debug('Populated Links in %.2f seconds' %
(time.time() - start))
title = unicode(row.get('title').decode('utf-8'))
start = time.time()
tree, definitions = populate_definitions(tree,
definitions=definitions,
title=title,
expire=True,
document_id=row.get('id'))
current_app.logger.debug('Populated Definitions in %.2f seconds' %
(time.time() - start))
definitions = add_parent_definitions(row,
definitions=definitions,
db=db,
strategy=strategy)
# now mark them
start = time.time()
tree, definitions = process_definitions(tree, definitions)
current_app.logger.debug('Found Definitions in %.2f seconds' %
(time.time() - start))
with (db or get_db()).cursor() as cur:
query = """UPDATE documents d SET processed_document = %(doc)s
WHERE d.id = %(id)s """
cur.execute(
query, {
'id': row.get('id'),
'doc': etree.tostring(tree, encoding='UTF-8', method="html"),
})
defs = definitions.render(document_id=row.get('id'))
if len(defs):
current_app.logger.info('New Definitions: %d' % len(defs))
args_str = ','.join(
cur.mogrify("(%s,%s,%s,%s,%s,%s,%s)", (
row.get('id'), x['id'], x['full_word'], x['keys'],
x['html'], x['expiry_tags'], x['priority'])) for x in defs)
cur.execute("DELETE FROM definitions where document_id = %(id)s",
{'id': row.get('id')})
cur.execute(
"INSERT INTO definitions (document_id, id, full_word, words, html, expiry_tags, priority) VALUES "
+ args_str)
if refresh:
cur.execute("select update_views()")
(db or get_db()).commit()
return tree, definitions
