create minimal working prototype for Groq Deep Search or OpenAI Deep Research

Author: Jason Chuang

README.md

@@ -1,13 +1,23 @@
 # open-deepsearch
 open-deepsearch ( Deep Research but Open-Sourced )
 
+Q&A for more details, research, and report generation.
+
 ## How to install in DEV environment after git clone
 ```bash
 python3 -m venv .venv
 source .venv/bin/activate
+#modify .env file and put in OPENAI_KEY
+cp .env.example .env
 pip install -r requirements.txt
 pip install -e .
 deepsearch
 ```
-⭐ A python port with a little more cli of 
-<https://github.com/dzhng/deep-research>
+⭐ A python port from node.js version
+<https://github.com/dzhng/deep-research>
+
+## As for now (2025-02-21, v0.0.3), it only uses OpenAI to produce output.md
+
+## Future work
+
+Try out FIRECRAWL or TAVILY to craw recent web data

open_deepsearch/deep_research.py

@@ -1,13 +1,13 @@
 import os
-import re
+import asyncio
 from typing import List, Dict, Optional, Any
-from ai.providers import generate_object, custom_model, trim_prompt, FirecrawlApp, SearchResponse
-from prompt import system_prompt
-from output_manager import OutputManager
+
+from .research_progress_results import ResearchProgress, ResearchResult
+from .prompt import system_prompt
+from .output_manager import OutputManager
 from pydantic import BaseModel
-import asyncio
 from concurrent.futures import ThreadPoolExecutor
-from ai.models import ResearchProgress, ResearchResult
+from .ai.providers import generate_object, custom_model, trim_prompt, FirecrawlApp, SearchResponse
 
 from dotenv import load_dotenv
 load_dotenv()
@@ -31,6 +31,8 @@ class SerpResultSchema(BaseModel):
     learnings: List[str]
     followUpQuestions: List[str]
 
+
+
 async def generate_serp_queries(query: str, num_queries: int = 3, learnings: Optional[List[str]] = None) -> List[Dict[str, str]]:
     res = await generate_object({
         'model': custom_model,

open_deepsearch/feedback.py

@@ -1,5 +1,5 @@
-from ai.providers import generate_object, custom_model
-from prompt import system_prompt
+from .ai.providers import generate_object, custom_model
+from .prompt import system_prompt
 from pydantic import BaseModel
 from typing import List
 

open_deepsearch/output_manager.py

@@ -1,5 +1,5 @@
 from typing import Any
-from ai.models import ResearchProgress
+from .research_progress_results import ResearchProgress
 
 class OutputManager:
     def __init__(self):

open_deepsearch/progress_manager.py

@@ -1,4 +1,4 @@
-from deep_research import ResearchProgress
+from .research_progress_results import ResearchProgress
 
 class ProgressManager:
     def __init__(self):

open_deepsearch/ai/models.py renamed to open_deepsearch/research_progress_results.py

@@ -1,4 +1,5 @@
-from typing import List, Optional
+from typing import Optional, List 
+
 
 class ResearchProgress:
     def __init__(self, current_depth: int, total_depth: int, current_breadth: int, total_breadth: int, total_queries: int, completed_queries: int, current_query: Optional[str] = None):

open_deepsearch/run.py

@@ -1,8 +1,8 @@
 import asyncio
 from typing import Any
-from open_deepsearch.deep_research import deep_research, write_final_report
-from feedback import generate_feedback
-from output_manager import OutputManager
+from .deep_research import deep_research, write_final_report
+from .feedback import generate_feedback
+from .output_manager import OutputManager
 
 output = OutputManager()
 
@@ -43,5 +43,8 @@ async def run() -> None:
     print(f"\n\nFinal Report:\n\n{report}")
     print('\nReport has been saved to output.md')
 
+def main():
+    asyncio.run(run())
+
 if __name__ == '__main__':
     asyncio.run(run())

pyproject.toml

@@ -3,9 +3,9 @@ requires = ["setuptools>=70.0.0", "wheel"]
 build-backend = "setuptools.build_meta"
 
 [project]
-name = "open_deepsearch"  
-version = "0.0.2"
-description = "DeepResearch but Open-Sourced, called open_deepsearch"
+name = "open-deepsearch"  
+version = "0.0.3"
+description = "Deep Research but Open-Sourced, called open-deepsearch"
 readme = "README.md"
 authors = [{ name = "Jason Chuang", email = "chuangtcee@gmail.com" }]
 classifiers = [
@@ -24,7 +24,7 @@ dependencies = []
 find = {}
 
 [project.scripts]
-deepsearch = "open_deepsearch:run.main"
+deepsearch = "open_deepsearch.run:main"
 
 [project.urls]
 "Homepage" = "https://github.com/aidatatools/open-deepsearch"

requirements.txt

@@ -1,9 +1,9 @@
+python-dotenv>=1.0.1
 openai==1.63.2
 aiohttp>=3.9.0
 aiofiles>=23.2.1
 tiktoken>=0.5.0
-python-dotenv>=1.0.1
 firecrawl-py>=1.11.1
-typer[all]>=0.9.0
+typer>=0.9.0
 prompt-toolkit>=3.0.0
 pydantic==2.10.6

setup.py

@@ -4,11 +4,11 @@
     long_description = fh.read()
 
 setup(
-    name='open_deepsearch',
-    version='0.0.2',
+    name='open-deepsearch',
+    version='0.0.3',
     author='Jason Chuang',
     author_email='chuangtcee@gmail.com',
-    description='Open DeepResearch',
+    description='Deep Research but Open-Sourced, called open-deepsearch',
     long_description=long_description,
     long_description_content_type='text/markdown',
     url='https://github.com/aidatatools/open-deepsearch',