def emit_url(self, document, url):
if isinstance(document, CrawlerMetadata):
doc = self.create_document(foreign_id=document.foreign_id,
content_hash=document.content_hash)
doc.meta.update(document.meta)
document = doc
if document.source_url is None:
document.source_url = url
db.session.commit()
ingest_url.delay(document.id, url)
self.increment_count()
def emit_url(self, meta, url):
ingest_url.delay(self.collection.id, meta.clone().data, url)
def emit_url(self, meta, url):
ingest_url.delay(self.source.id, meta.clone().data, url)
def emit_url(self, source, meta, url):
db.session.commit()
ingest_url.delay(source.id, meta.clone().data, url)
def emit_url(self, meta, url):
ingest_url.delay(self.collection.id, meta.to_attr_dict(), url)
self.increment_count()
def emit_url(self, meta, url):
ingest_url.delay(self.collection.id, meta.clone().data, url)
def emit_url(self, meta, url):
ingest_url.delay(self.collection.id, meta.to_attr_dict(), url)
self.increment_count()
def emit_url(self, source, meta, url):
db.session.commit()
ingest_url.delay(source.id, meta.clone().data, url)
