LLM/DocsGPT
1
0
Fork 0
mirror of https://github.com/arc53/DocsGPT.git synced 2025-11-29 08:33:20 +00:00
Code Issues Packages Projects Releases Wiki Activity

feat: template-based prompt rendering with dynamic namespace injection (#2091)

Browse Source 
* feat: template-based prompt rendering with dynamic namespace injection

* refactor: improve template engine initialization with clearer formatting

* refactor: streamline ReActAgent methods and improve content extraction logic

feat: enhance error handling in NamespaceManager and TemplateEngine

fix: update NewAgent component to ensure consistent form data submission

test: modify tests for ReActAgent and prompt renderer to reflect method changes and improve coverage

* feat: tools namespace + three-tier token budget

* refactor: remove unused variable assignment in message building tests

* Enhance prompt customization and tool pre-fetching functionality

* ruff lint fix

* refactor: cleaner error handling and reduce code clutter

---------

Co-authored-by: Alex <a@tushynski.me>
This commit is contained in:
Siddhant Rai
2025-10-31 18:17:44 +05:30
committed by GitHub
parent a7d61b9d59
commit 21e5c261ef
33 changed files with 2917 additions and 646 deletions
Download Patch File Download Diff File Expand all files Collapse all files

76
application/retriever/classic_rag.py
View File

@@ -4,7 +4,7 @@ import os
from application.core.settings import settings
from application.llm.llm_creator import LLMCreator
from application.retriever.base import BaseRetriever
from application.utils import num_tokens_from_string
from application.vectorstore.vector_creator import VectorCreator
@@ -15,14 +15,13 @@ class ClassicRAG(BaseRetriever):
chat_history=None,
prompt="",
chunks=2,
token_limit=150,
doc_token_limit=50000,
gpt_model="docsgpt",
user_api_key=None,
llm_name=settings.LLM_PROVIDER,
api_key=settings.API_KEY,
decoded_token=None,
):
"""Initialize ClassicRAG retriever with vectorstore sources and LLM configuration"""
self.original_question = source.get("question", "")
self.chat_history = chat_history if chat_history is not None else []
self.prompt = prompt
@@ -42,16 +41,7 @@ class ClassicRAG(BaseRetriever):
f"sources={'active_docs' in source and source['active_docs'] is not None}"
)
self.gpt_model = gpt_model
self.token_limit = (
token_limit
if token_limit
< settings.LLM_TOKEN_LIMITS.get(
self.gpt_model, settings.DEFAULT_MAX_HISTORY
)
else settings.LLM_TOKEN_LIMITS.get(
self.gpt_model, settings.DEFAULT_MAX_HISTORY
)
)
self.doc_token_limit = doc_token_limit
self.user_api_key = user_api_key
self.llm_name = llm_name
self.api_key = api_key
@@ -118,21 +108,17 @@ class ClassicRAG(BaseRetriever):
return self.original_question
def _get_data(self):
"""Retrieve relevant documents from configured vectorstores"""
if self.chunks == 0 or not self.vectorstores:
logging.info(
f"ClassicRAG._get_data: Skipping retrieval - chunks={self.chunks}, "
f"vectorstores_count={len(self.vectorstores) if self.vectorstores else 0}"
)
return []
all_docs = []
chunks_per_source = max(1, self.chunks // len(self.vectorstores))
logging.info(
f"ClassicRAG._get_data: Starting retrieval with chunks={self.chunks}, "
f"vectorstores={self.vectorstores}, chunks_per_source={chunks_per_source}, "
f"query='{self.question[:50]}...'"
)
token_budget = max(int(self.doc_token_limit * 0.9), 100)
cumulative_tokens = 0
for vectorstore_id in self.vectorstores:
if vectorstore_id:
@@ -140,15 +126,21 @@ class ClassicRAG(BaseRetriever):
docsearch = VectorCreator.create_vectorstore(
settings.VECTOR_STORE, vectorstore_id, settings.EMBEDDINGS_KEY
)
docs_temp = docsearch.search(self.question, k=chunks_per_source)
docs_temp = docsearch.search(
self.question, k=max(chunks_per_source * 2, 20)
)
for doc in docs_temp:
if cumulative_tokens >= token_budget:
break
if hasattr(doc, "page_content") and hasattr(doc, "metadata"):
page_content = doc.page_content
metadata = doc.metadata
else:
page_content = doc.get("text", doc.get("page_content", ""))
metadata = doc.get("metadata", {})
title = metadata.get(
"title", metadata.get("post_title", page_content)
)
@@ -168,23 +160,35 @@ class ClassicRAG(BaseRetriever):
if not filename:
filename = title
source_path = metadata.get("source") or vectorstore_id
all_docs.append(
{
"title": title,
"text": page_content,
"source": source_path,
"filename": filename,
}
)
doc_text_with_header = f"{filename}\n{page_content}"
doc_tokens = num_tokens_from_string(doc_text_with_header)
if cumulative_tokens + doc_tokens < token_budget:
all_docs.append(
{
"title": title,
"text": page_content,
"source": source_path,
"filename": filename,
}
)
cumulative_tokens += doc_tokens
if cumulative_tokens >= token_budget:
break
except Exception as e:
logging.error(
f"Error searching vectorstore {vectorstore_id}: {e}",
exc_info=True,
)
continue
logging.info(
f"ClassicRAG._get_data: Retrieval complete - retrieved {len(all_docs)} documents "
f"(requested chunks={self.chunks}, chunks_per_source={chunks_per_source})"
f"(requested chunks={self.chunks}, chunks_per_source={chunks_per_source}, "
f"cumulative_tokens={cumulative_tokens}/{token_budget})"
)
return all_docs
@@ -194,15 +198,3 @@ class ClassicRAG(BaseRetriever):
self.original_question = query
self.question = self._rephrase_query()
return self._get_data()
def get_params(self):
"""Return current retriever configuration parameters"""
return {
"question": self.original_question,
"rephrased_question": self.question,
"sources": self.vectorstores,
"chunks": self.chunks,
"token_limit": self.token_limit,
"gpt_model": self.gpt_model,
"user_api_key": self.user_api_key,
}