opensource/dify
2
0
Fork 0
mirror of https://github.com/langgenius/dify.git synced 2026-04-29 20:48:01 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix knowledge single retrieve when function call response is none

Browse Source
This commit is contained in:
jyong 2024-03-19 14:17:22 +08:00
parent 1c7573a686
commit 4ec14d8d91
2 changed files with 71 additions and 61 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
api/core/model_runtime/model_providers/anthropic/llm/llm.py
View File

@ -342,12 +342,21 @@ class AnthropicLargeLanguageModel(LargeLanguageModel):
Convert prompt messages to dict list and system Convert prompt messages to dict list and system
""" """
system = "" system = ""
first_loop = True
for message in prompt_messages:
if isinstance(message, SystemPromptMessage):
message.content = message.content.strip()
if first_loop:
system = message.content
first_loop = False
else:
system += "\n"
system += message.content
prompt_message_dicts = [] prompt_message_dicts = []
for message in prompt_messages: for message in prompt_messages:
if isinstance(message, SystemPromptMessage): if not isinstance(message, SystemPromptMessage):
system += message.content + ("\n" if not system else "")
else:
prompt_message_dicts.append(self._convert_prompt_message_to_dict(message)) prompt_message_dicts.append(self._convert_prompt_message_to_dict(message))
return system, prompt_message_dicts return system, prompt_message_dicts

117
api/core/workflow/nodes/knowledge_retrieval/knowledge_retrieval_node.py
View File

@ -103,69 +103,69 @@ class KnowledgeRetrievalNode(BaseNode):
elif node_data.retrieval_mode == DatasetRetrieveConfigEntity.RetrieveStrategy.MULTIPLE.value: elif node_data.retrieval_mode == DatasetRetrieveConfigEntity.RetrieveStrategy.MULTIPLE.value:
all_documents = self._multiple_retrieve(available_datasets, node_data, query) all_documents = self._multiple_retrieve(available_datasets, node_data, query)
document_score_list = {}
for item in all_documents:
if 'score' in item.metadata and item.metadata['score']:
document_score_list[item.metadata['doc_id']] = item.metadata['score']
document_context_list = []
index_node_ids = [document.metadata['doc_id'] for document in all_documents]
segments = DocumentSegment.query.filter(
DocumentSegment.dataset_id.in_(dataset_ids),
DocumentSegment.completed_at.isnot(None),
DocumentSegment.status == 'completed',
DocumentSegment.enabled == True,
DocumentSegment.index_node_id.in_(index_node_ids)
).all()
context_list = [] context_list = []
if segments: if all_documents:
index_node_id_to_position = {id: position for position, id in enumerate(index_node_ids)} document_score_list = {}
sorted_segments = sorted(segments, for item in all_documents:
key=lambda segment: index_node_id_to_position.get(segment.index_node_id, if 'score' in item.metadata and item.metadata['score']:
float('inf'))) document_score_list[item.metadata['doc_id']] = item.metadata['score']
for segment in sorted_segments:
if segment.answer:
document_context_list.append(f'question:{segment.content} answer:{segment.answer}')
else:
document_context_list.append(segment.content)
for segment in sorted_segments: document_context_list = []
dataset = Dataset.query.filter_by( index_node_ids = [document.metadata['doc_id'] for document in all_documents]
id=segment.dataset_id segments = DocumentSegment.query.filter(
).first() DocumentSegment.dataset_id.in_(dataset_ids),
document = Document.query.filter(Document.id == segment.document_id, DocumentSegment.completed_at.isnot(None),
Document.enabled == True, DocumentSegment.status == 'completed',
Document.archived == False, DocumentSegment.enabled == True,
).first() DocumentSegment.index_node_id.in_(index_node_ids)
resource_number = 1 ).all()
if dataset and document: if segments:
index_node_id_to_position = {id: position for position, id in enumerate(index_node_ids)}
source = { sorted_segments = sorted(segments,
'metadata': { key=lambda segment: index_node_id_to_position.get(segment.index_node_id,
'_source': 'knowledge', float('inf')))
'position': resource_number, for segment in sorted_segments:
'dataset_id': dataset.id,
'dataset_name': dataset.name,
'document_id': document.id,
'document_name': document.name,
'document_data_source_type': document.data_source_type,
'segment_id': segment.id,
'retriever_from': 'workflow',
'score': document_score_list.get(segment.index_node_id, None),
'segment_hit_count': segment.hit_count,
'segment_word_count': segment.word_count,
'segment_position': segment.position,
'segment_index_node_hash': segment.index_node_hash,
},
'title': document.name
}
if segment.answer: if segment.answer:
source['content'] = f'question:{segment.content} \nanswer:{segment.answer}' document_context_list.append(f'question:{segment.content} answer:{segment.answer}')
else: else:
source['content'] = segment.content document_context_list.append(segment.content)
context_list.append(source)
resource_number += 1
for segment in sorted_segments:
dataset = Dataset.query.filter_by(
id=segment.dataset_id
).first()
document = Document.query.filter(Document.id == segment.document_id,
Document.enabled == True,
Document.archived == False,
).first()
resource_number = 1
if dataset and document:
source = {
'metadata': {
'_source': 'knowledge',
'position': resource_number,
'dataset_id': dataset.id,
'dataset_name': dataset.name,
'document_id': document.id,
'document_name': document.name,
'document_data_source_type': document.data_source_type,
'segment_id': segment.id,
'retriever_from': 'workflow',
'score': document_score_list.get(segment.index_node_id, None),
'segment_hit_count': segment.hit_count,
'segment_word_count': segment.word_count,
'segment_position': segment.position,
'segment_index_node_hash': segment.index_node_hash,
},
'title': document.name
}
if segment.answer:
source['content'] = f'question:{segment.content} \nanswer:{segment.answer}'
else:
source['content'] = segment.content
context_list.append(source)
resource_number += 1
return context_list return context_list
@classmethod @classmethod
@ -257,6 +257,7 @@ class KnowledgeRetrievalNode(BaseNode):
top_k=top_k, score_threshold=score_threshold, top_k=top_k, score_threshold=score_threshold,
reranking_model=reranking_model) reranking_model=reranking_model)
return results return results
return []
def _fetch_model_config(self, node_data: KnowledgeRetrievalNodeData) -> tuple[ def _fetch_model_config(self, node_data: KnowledgeRetrievalNodeData) -> tuple[
ModelInstance, ModelConfigWithCredentialsEntity]: ModelInstance, ModelConfigWithCredentialsEntity]: