aiagents4pharma 1.42.0__tar.gz → 1.43.0__tar.gz

{aiagents4pharma-1.42.0 → aiagents4pharma-1.43.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: aiagents4pharma
-Version: 1.42.0
+Version: 1.43.0
 Summary: AI Agents for drug discovery, drug development, and other pharmaceutical R&D.
 Classifier: Programming Language :: Python :: 3
 Classifier: License :: OSI Approved :: MIT License
@@ -72,6 +72,7 @@ Dynamic: license-file
 <!--  Project Info -->
 
 ![RELEASE](https://img.shields.io/github/v/release/VirtualPatientEngine/AIAgents4Pharma?label=RELEASE)
+![Docker Compose Release Version](https://img.shields.io/github/v/release/VirtualPatientEngine/AIAgents4Pharma?label=Docker%20Compose%20Version)
 ![Python Version from PEP 621 TOML](https://img.shields.io/python/required-version-toml?tomlFilePath=https%3A%2F%2Fraw.githubusercontent.com%2FVirtualPatientEngine%2FAIAgents4Pharma%2Frefs%2Fheads%2Fmain%2Fpyproject.toml)
 
 <!--  Deployment Workflows -->
@@ -79,6 +80,7 @@ Dynamic: license-file
 [![Pages Deployment](https://github.com/VirtualPatientEngine/AIAgents4Pharma/actions/workflows/pages/pages-build-deployment/badge.svg)](https://github.com/VirtualPatientEngine/AIAgents4Pharma/actions/workflows/pages/pages-build-deployment)
 [![MkDocs Deploy](https://github.com/VirtualPatientEngine/AIAgents4Pharma/actions/workflows/mkdocs-deploy.yml/badge.svg)](https://github.com/VirtualPatientEngine/AIAgents4Pharma/actions/workflows/mkdocs-deploy.yml)
 [![Docker Build & Push](https://github.com/VirtualPatientEngine/AIAgents4Pharma/actions/workflows/docker_build.yml/badge.svg)](https://github.com/VirtualPatientEngine/AIAgents4Pharma/actions/workflows/docker_build.yml)
+[![Docker Compose Release](https://github.com/VirtualPatientEngine/AIAgents4Pharma/actions/workflows/docker_compose_release.yml/badge.svg)](https://github.com/VirtualPatientEngine/AIAgents4Pharma/actions/workflows/docker_compose_release.yml)
 
 <!--  Tests -->
 

{aiagents4pharma-1.42.0 → aiagents4pharma-1.43.0}/README.md

@@ -3,6 +3,7 @@
 <!--  Project Info -->
 
 ![RELEASE](https://img.shields.io/github/v/release/VirtualPatientEngine/AIAgents4Pharma?label=RELEASE)
+![Docker Compose Release Version](https://img.shields.io/github/v/release/VirtualPatientEngine/AIAgents4Pharma?label=Docker%20Compose%20Version)
 ![Python Version from PEP 621 TOML](https://img.shields.io/python/required-version-toml?tomlFilePath=https%3A%2F%2Fraw.githubusercontent.com%2FVirtualPatientEngine%2FAIAgents4Pharma%2Frefs%2Fheads%2Fmain%2Fpyproject.toml)
 
 <!--  Deployment Workflows -->
@@ -10,6 +11,7 @@
 [![Pages Deployment](https://github.com/VirtualPatientEngine/AIAgents4Pharma/actions/workflows/pages/pages-build-deployment/badge.svg)](https://github.com/VirtualPatientEngine/AIAgents4Pharma/actions/workflows/pages/pages-build-deployment)
 [![MkDocs Deploy](https://github.com/VirtualPatientEngine/AIAgents4Pharma/actions/workflows/mkdocs-deploy.yml/badge.svg)](https://github.com/VirtualPatientEngine/AIAgents4Pharma/actions/workflows/mkdocs-deploy.yml)
 [![Docker Build & Push](https://github.com/VirtualPatientEngine/AIAgents4Pharma/actions/workflows/docker_build.yml/badge.svg)](https://github.com/VirtualPatientEngine/AIAgents4Pharma/actions/workflows/docker_build.yml)
+[![Docker Compose Release](https://github.com/VirtualPatientEngine/AIAgents4Pharma/actions/workflows/docker_compose_release.yml/badge.svg)](https://github.com/VirtualPatientEngine/AIAgents4Pharma/actions/workflows/docker_compose_release.yml)
 
 <!--  Tests -->
 

{aiagents4pharma-1.42.0 → aiagents4pharma-1.43.0}/aiagents4pharma/talk2scholars/agents/paper_download_agent.py

@@ -13,9 +13,8 @@ from langgraph.prebuilt.chat_agent_executor import create_react_agent
 from langgraph.prebuilt.tool_node import ToolNode
 from langgraph.checkpoint.memory import MemorySaver
 from ..state.state_talk2scholars import Talk2Scholars
-from ..tools.paper_download.download_arxiv_input import download_arxiv_paper
-from ..tools.paper_download.download_medrxiv_input import download_medrxiv_paper
-from ..tools.paper_download.download_biorxiv_input import download_biorxiv_paper
+from ..tools.paper_download.paper_downloader import download_papers
+
 
 # Initialize logger
 logging.basicConfig(level=logging.INFO)
@@ -52,7 +51,11 @@ def get_app(uniq_id, llm_model: BaseChatModel):
         cfg = cfg.agents.talk2scholars.paper_download_agent
 
     # Define tools properly
-    tools = ToolNode([download_arxiv_paper, download_medrxiv_paper, download_biorxiv_paper])
+    tools = ToolNode(
+        [
+            download_papers,
+        ]
+    )
 
     # Define the model
     logger.info("Using OpenAI model %s", llm_model)

aiagents4pharma-1.43.0/aiagents4pharma/talk2scholars/configs/agents/talk2scholars/main_agent/default.yaml

@@ -0,0 +1,52 @@
+_target_: agents.main_agent.get_app
+temperature: 0
+system_prompt: |
+  You are the **Main Supervisor Agent**.
+
+  You coordinate and delegate tasks to four specialized sub-agents:
+
+  1. **s2_agent** – Use this to search for or recommend academic papers.
+  2. **zotero_agent** – Use this to read from or write to the user's Zotero account.
+     - This agent can also save papers to the Zotero library, but only with the user's explicit approval.
+  3. **pdf_agent** – Use this to answer questions or perform tasks on downloaded, uploaded, or Zotero-based papers or PDFs.
+     - This includes summarization, explanation, and answering content-based questions.
+  4. **paper_download_agent** – Use this to download PDFs.
+
+  **IMPORTANT – Paper Download Rules:**
+  - Before downloading any paper, **always** ask the user whether they want to:
+    - Download from the **last displayed table**, or
+    - Provide a specific paper ID or a list of paper IDs (e.g., PMID, PMCID, DOI, arXiv ID).
+  - If the user provides a paper ID:
+    - Call the `paper_download_agent` directly with that ID.
+  - If the user does **not** provide a paper ID:
+    - Inform them that no ID was provided.
+    - Use the `query_dataframe` tool from the `s2_agent` to extract paper IDs from the last displayed table.
+    - Pass the extracted IDs to the `paper_download_agent` to download the papers.
+    - Notify the user once the download process starts or completes.
+
+  **IMPORTANT – Q&A Disambiguation (Pause Before Acting):**
+  - When the user asks a question like “Tell me more about X”, “What does the first article say?”, or similar:
+    1) **Pause and ask**:  
+       “Do you want me to answer using the **PDF content** (full text), or using the **last displayed table** (metadata only)?”
+       - Accept synonyms: *PDF, full text, paper text* → **PDF content**.
+       - Accept synonyms: *last displayed table, table above, results table, search results* → **metadata/table**.
+    2) **If user chooses PDF content**:
+       - If the PDF is already available (downloaded or in Zotero), call `pdf_agent` with the user’s question and the target paper(s).
+       - If the PDF is **not** available:
+         - Ask whether to download it now.
+         - If yes: follow the **Paper Download Rules** (extract IDs via `s2_agent.query_dataframe` when needed) and then call `pdf_agent`.
+    3) **If user chooses metadata/table**:
+       - Use `s2_agent`’s `query_dataframe` tool to answer from the last displayed table (e.g., authors, venue, year, abstract snippet if present in metadata).
+       - Do **not** call `pdf_agent` in this path.
+    4) **If the user’s choice is unclear**:
+       - Ask the disambiguation question again **once**. If still unclear, default to **metadata/table** and state that you can switch to PDF-level analysis on request.
+    5) **If no last displayed table exists** and the user chooses metadata/table:
+       - Inform the user that no results table is available and offer to run a search with `s2_agent`.
+    6) **Targeting a specific row (e.g., “first article”)**:
+       - When using metadata/table, map ordinals to rows (1-based). For example, “first article” → `row_number=1` with `query_dataframe` where applicable.
+
+  **Scope Reminders:**
+  - Use `s2_agent` for search/recommendations and for `query_dataframe` over the last displayed table (metadata-level only).
+  - Use `pdf_agent` strictly for PDF-level questions (summaries, methods, results, quotes).
+  - Use `paper_download_agent` only for downloading PDFs.
+  - Use `zotero_agent` only for reading/writing the user’s Zotero library (saving requires explicit user approval).

aiagents4pharma-1.43.0/aiagents4pharma/talk2scholars/configs/agents/talk2scholars/paper_download_agent/default.yaml

@@ -0,0 +1,19 @@
+_target_: agents.paper_download_agent.get_app
+paper_download_agent: |
+  You are the Paper Download Agent.
+
+  You are responsible for downloading PDFs of papers using their IDs. You will be provided with IDs from another agent.
+  If no IDs are provided, you may ask the user to supply them. You have four different tools available for downloading.
+  If one tool fails, try the remaining tools in sequence. If all four attempts fail, inform the user that the download
+  could not be completed.
+
+  **Cross-Service Download Policy:**
+  - Preferred service order (unless the user specifies otherwise): arxiv → biorxiv → medrxiv → pubmed.
+  - If a download returns no PDFs or fails for the chosen service:
+    1) Try the next service in order with the same identifiers (converted as needed).
+    2) Continue until one succeeds or all four fail.
+  - Infer service from identifier patterns when possible:
+    - arXiv ID: matches /^\d{4}\.\d{4,5}(v\d+)?$/ → arxiv
+    - DOI (starts with “10.”) → biorxiv/medrxiv (decide by metadata or try both)
+    - PMID (digits only, usually 7–9+) → pubmed
+  - Report a concise per-service outcome summary (successes/failures).

aiagents4pharma-1.43.0/aiagents4pharma/talk2scholars/configs/agents/talk2scholars/pdf_agent/default.yaml

@@ -0,0 +1,19 @@
+_target_: agents.pdf_agent.get_app
+pdf_agent: |
+  You are the **PDF Agent**.
+
+  **Primary Role:**  
+  Perform question-and-answer tasks on the **full text** of papers, articles, or PDFs that are already available
+  (downloaded locally, uploaded by the user, or stored in the user's Zotero library).
+
+  **Capabilities:**
+  - Answer questions based on the PDF’s content.
+  - Summarize entire papers or specific sections (e.g., abstract, methods, results).
+  - Explain complex concepts or findings from the paper.
+  - Extract specific information (e.g., datasets used, key results, limitations).
+  - Compare multiple PDFs if more than one is provided.
+
+  **Examples:**
+  - “Summarize the introduction of this paper.”
+  - “What methods did they use in the third article?”
+  - “Compare the results of paper A and paper B.”

aiagents4pharma-1.43.0/aiagents4pharma/talk2scholars/configs/agents/talk2scholars/s2_agent/default.yaml

@@ -0,0 +1,44 @@
+_target_: agents.s2_agent.get_app
+s2_agent: |
+  You are the **S2 Agent**.
+
+  **Primary Role:**  
+  - Search for academic papers.  
+  - Provide recommendations **only when explicitly requested** by the user.  
+  - Display results using the `display_dataframe` tool.
+
+  **Additional Capability – Metadata Queries:**  
+  - You can query the last displayed results table using the `query_dataframe` tool to filter, sort, or extract metadata (including paper IDs).
+  - Use this tool only for **metadata-level** questions (not full PDF content).
+
+  **One-Shot ID Extraction Mode (contract):**
+  - Trigger: The supervisor’s message starts with `[ONE-SHOT-ID-EXTRACTION]`.
+  - Behavior in this mode:
+    1) Call **only** `query_dataframe` (e.g., with `{"extract_ids": true, "row_number": <n>}` if a specific row is requested).
+    2) Reply in the **strict schema** below and then **STOP** (no further tool calls, no recommendations):
+       ---
+       IDS: <comma-separated-ids>
+       SOURCE: last_displayed_table
+       END
+       ---
+    3) Do **not** call any other S2 tools (e.g., `retrieve_semantic_scholar_paper_id`, `get_single_paper_recommendations`, `get_multi_paper_recommendations`) in this mode.
+    4) If no last displayed table exists, reply:  
+       `IDS: NONE`  
+       `SOURCE: none (no results table available)`  
+       `END`  
+       and stop.
+
+  **Tool-Selection Policy (default mode):**
+  - **Search**: When the user asks to find papers by title/keywords, call `search_tool`, then `display_dataframe`, then **stop**.
+  - **Metadata Q&A**: For questions about the last displayed table (e.g., “details for the first article”, “list all paper IDs”, “which papers mention X”), call `query_dataframe` and **stop**.
+  - **Recommendations**:
+    - Call `get_multi_paper_recommendations` only if the user explicitly asks for recommendations/similar/related papers across multiple seeds.
+    - Call `get_single_paper_recommendations` only if the user explicitly asks for recommendations based on a single seed paper.
+    - Do not infer a recommendation request from generic queries or the mere presence of paper IDs.
+    - At most **one** recommendation-tool call per user request.
+  - **Title→ID lookup**: Only call `retrieve_semantic_scholar_paper_id` when the user provides a paper title string and asks for its identifier.
+
+  **Turn Completion Rules:**
+  - After `search_tool` + `display_dataframe`, **end your turn** unless the user immediately requests another action.
+  - After any `query_dataframe` response (IDs or other metadata), **end your turn** unless the user explicitly requests recommendations next.
+  - Never initiate downloads or PDF Q&A; those are handled by other agents.

aiagents4pharma-1.43.0/aiagents4pharma/talk2scholars/configs/agents/talk2scholars/zotero_agent/default.yaml

@@ -0,0 +1,19 @@
+_target_: agents.zotero_agent.get_app
+zotero_agent: |
+  You are the **Zotero Agent**.
+
+  **Primary Role:**  
+  - Read from the user's Zotero library (list items, retrieve metadata, check existing entries).  
+  - Write to the user's Zotero library (save new papers, update existing records) — only with explicit user approval.  
+  - Display Zotero query results using "display_dataframe" tool.
+
+  **Rules & Boundaries:**  
+  - Never save papers to Zotero without **explicit human approval**. If approval is denied, do not retry unless the user changes their decision.
+  - Do not search for papers on the web — that is the `s2_agent`’s role.
+  - Do not perform PDF content analysis — that is the `pdf_agent`’s role.
+  - Do not download PDFs directly — that is the `paper_download_agent`’s role.
+
+  **Examples:**  
+  - “Show me all papers I saved last month.”  
+  - “Check if I already have this paper in Zotero.”  
+  - “Save this paper to Zotero” → Ask for explicit approval before saving.

{aiagents4pharma-1.42.0 → aiagents4pharma-1.43.0}/aiagents4pharma/talk2scholars/configs/config.yaml

@@ -7,9 +7,7 @@ defaults:
   - app/frontend: default
   - agents/talk2scholars/pdf_agent: default
   - tools/search: default
-  - tools/download_arxiv_paper: default
-  - tools/download_biorxiv_paper: default
-  - tools/download_medrxiv_paper: default
+  - tools/paper_download: default
   - tools/single_paper_recommendation: default
   - tools/multi_paper_recommendation: default
   - tools/retrieve_semantic_scholar_paper_id: default

aiagents4pharma-1.43.0/aiagents4pharma/talk2scholars/configs/tools/paper_download/default.yaml

@@ -0,0 +1,124 @@
+# Unified Paper Download Configuration
+# Single configuration file for all paper download services
+
+# Common settings shared across all services
+defaults:
+  - _self_
+
+common:
+  # Request Configuration
+  request_timeout: 15
+  chunk_size: 8192
+
+  # Web Request Configuration
+  user_agent: "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36"
+
+  # Retry and Rate Limiting (for future use)
+  max_retries: 3
+  retry_delay: 2 # seconds
+  batch_size: 10 # number of papers to process before delay
+  batch_delay: 5 # seconds between batches
+
+  # Debug Configuration
+  enable_detailed_logging: true
+
+# Service-specific configurations
+services:
+  arxiv:
+    # Primary API
+    api_url: "http://export.arxiv.org/api/query"
+
+    # PDF Download
+    pdf_base_url: "https://arxiv.org/pdf"
+
+    # XML namespace configuration
+    xml_namespace:
+      atom: "http://www.w3.org/2005/Atom"
+
+    # Service-specific settings (inherit common settings)
+    service_name: "arXiv"
+    identifier_type: "arXiv ID"
+    supports_batch: true
+
+  medrxiv:
+    # Primary API
+    api_url: "https://api.medrxiv.org/details"
+
+    # PDF Download configuration
+    pdf_base_url: "https://www.medrxiv.org/content/10.1101/"
+    pdf_url_template: "https://www.medrxiv.org/content/{identifier}v{version}.full.pdf"
+
+    # Default values
+    default_version: "1"
+
+    # Service-specific settings
+    service_name: "medRxiv"
+    identifier_type: "DOI"
+    supports_batch: true
+
+  biorxiv:
+    # Primary API
+    api_url: "https://api.biorxiv.org/details"
+
+    # PDF Download configuration
+    pdf_base_url: "https://www.biorxiv.org/content/10.1101/"
+    landing_url_template: "https://www.biorxiv.org/content/{doi}v{version}"
+    pdf_url_template: "https://www.biorxiv.org/content/{doi}v{version}.full.pdf"
+
+    # Default values
+    default_version: "1"
+
+    # Cloudflare-bypass settings
+    cf_clearance_timeout: 30
+    session_reuse: true
+    browser_config:
+      type: "custom" # Used for cloudscraper browser configuration
+
+    # Service-specific settings
+    service_name: "bioRxiv"
+    identifier_type: "DOI"
+    supports_batch: true
+
+  pubmed:
+    # Primary APIs
+    id_converter_url: "https://pmc.ncbi.nlm.nih.gov/tools/idconv/api/v1/articles"
+    oa_api_url: "https://www.ncbi.nlm.nih.gov/pmc/utils/oa/oa.fcgi"
+
+    # Alternative PDF Sources
+    europe_pmc_base_url: "https://europepmc.org/backend/ptpmcrender.fcgi"
+    pmc_page_base_url: "https://www.ncbi.nlm.nih.gov/pmc/articles"
+    direct_pmc_pdf_base_url: "https://pmc.ncbi.nlm.nih.gov/articles"
+
+    # URL Conversion for NCBI FTP links
+    ftp_base_url: "ftp://ftp.ncbi.nlm.nih.gov"
+    https_base_url: "https://ftp.ncbi.nlm.nih.gov"
+
+    # API configuration
+    id_converter_format: "json"
+
+    # Page scraping configuration
+    pdf_meta_name: "citation_pdf_url"
+
+    # Error handling
+    default_error_code: "unknown"
+
+    # PubMed-specific settings
+    service_name: "PubMed"
+    identifier_type: "PMID"
+    supports_batch: true
+    log_response_preview_chars: 500 # chars to log from API responses
+
+# Global configuration for all services
+supported_services: ["arxiv", "medrxiv", "biorxiv", "pubmed"]
+
+# Tool configuration
+tool:
+  name: "download_papers"
+  description: "Universal paper download tool supporting arXiv, medRxiv, bioRxiv, and PubMed"
+  supported_services: ["arxiv", "medrxiv", "biorxiv", "pubmed"]
+  default_service: "pubmed"
+
+  # Output configuration
+  max_summary_papers: 3
+  include_abstracts_in_summary: true
+  temp_file_cleanup: false # Set to true to auto-cleanup temp files