def test_pipeline_example():
document_store = LocalDocumentStore()
document_store.put("test.doc", create_document())
pipeline = Pipeline(document_store)
stats = pipeline.run().statistics
assert stats.documents_processed == 1
def test_url_pipeline():
document = Document.from_url("http://www.google.com")
new_document_store = LocalDocumentStore()
stats = Pipeline(document).add_step(TextParser(encoding='ISO-8859-1')).add_step(
DocumentStoreWriter(new_document_store)).run().statistics
assert stats.documents_processed == 1
assert stats.document_exceptions == 0
assert new_document_store.count() == 1
new_doc = new_document_store.get_latest_document("http://www.google.com")
print(new_doc.content_node.get_all_content())
def test_fluent_pipeline():
def my_function(doc):
doc.metadata.cheese = "fishstick"
logging.error("Hello")
return doc
document = create_document()
new_document_store = LocalDocumentStore()
stats = Pipeline(document).add_step(my_function).add_step(my_function).add_step(
DocumentStoreWriter(new_document_store)).run().statistics
assert stats.documents_processed == 1
assert stats.document_exceptions == 0
assert new_document_store.count() == 1
assert new_document_store.get_latest_document("test.doc").metadata.cheese == 'fishstick'
def test_basic_local_document_store():
JSON_STORE = "/tmp/test-json-store.jsonkey"
document_store = LocalDocumentStore(store_path=JSON_STORE, force_initialize=True)
document_store.put("test.doc", create_document())
new_document_store = LocalDocumentStore(store_path=JSON_STORE)
assert (new_document_store.count() == 1)
def test_table_stores_with_extractor():
document_store = LocalDocumentStore()
document_store.put("test.doc", create_document())
pipeline = Pipeline(document_store, stop_on_exception=False)
pipeline.add_store('output', TableDataStore(columns=['cheese']))
def extractor(document, context):
# An example of how we might
# extract into a dict
#
context.get_store('output').add(['test'])
return document
pipeline.add_step(extractor)
context = pipeline.run()
assert context.get_store('output').count() == 1
def test_function_step_with_context():
document_store = LocalDocumentStore()
document_store.put("test.doc", create_document())
new_document_store = LocalDocumentStore()
def my_function(doc, context):
doc.metadata.cheese = context.execution_id
logging.error("Hello")
return doc
assert new_document_store.count() == 0
pipeline = Pipeline(document_store)
pipeline.add_step(my_function)
pipeline.add_step(DocumentStoreWriter(new_document_store))
stats = pipeline.run().statistics
assert stats.documents_processed == 1
assert stats.document_exceptions == 0
assert new_document_store.count() == 1
assert new_document_store.get_latest_document("test.doc").metadata.cheese == pipeline.context.execution_id
def test_function_step_with_exception():
document_store = LocalDocumentStore()
document_store.put("test.doc", create_document())
new_document_store = LocalDocumentStore()
def my_function(doc):
doc.metadata.cheese = "fishstick"
raise Exception("hello world")
assert new_document_store.count() == 0
pipeline = Pipeline(document_store, stop_on_exception=False)
pipeline.add_step(my_function)
pipeline.add_step(DocumentStoreWriter(new_document_store))
stats = pipeline.run().statistics
assert stats.documents_processed == 1
assert stats.document_exceptions == 1
assert new_document_store.count() == 1
assert len(new_document_store.get_latest_document("test.doc").exceptions) == 1
def test_class_step_step_with_context():
document_store = LocalDocumentStore()
document_store.put('test.doc', create_document())
new_document_store = LocalDocumentStore()
class MyProcessingStep:
def get_name(self):
return "test-step"
def process(self, doc, context):
doc.metadata.cheese = context.execution_id
logging.error("Hello")
return doc
pipeline = Pipeline(document_store)
pipeline.add_step(MyProcessingStep())
pipeline.add_step(DocumentStoreWriter(new_document_store))
ctxt = pipeline.run()
assert ctxt.statistics.documents_processed == 1
assert ctxt.statistics.document_exceptions == 0
assert new_document_store.get_latest_document("test.doc").metadata.cheese == pipeline.context.execution_id