aiagents4pharma 1.39.0__py3-none-any.whl → 1.39.2__py3-none-any.whl

aiagents4pharma/talk2scholars/tests/test_tool_helper_utils.py

@@ -0,0 +1,140 @@
+"""
+Unit tests for QAToolHelper routines in tool_helper.py
+"""
+
+import unittest
+from types import SimpleNamespace
+from unittest.mock import MagicMock, patch
+
+from aiagents4pharma.talk2scholars.tools.pdf.utils.tool_helper import QAToolHelper
+
+
+class TestQAToolHelper(unittest.TestCase):
+    """tests for QAToolHelper routines in tool_helper.py"""
+
+    def setUp(self):
+        """set up test case"""
+        self.helper = QAToolHelper()
+
+    def test_start_call_sets_config_and_call_id(self):
+        """test start_call sets config and call_id"""
+        cfg = SimpleNamespace(foo="bar")
+        self.helper.start_call(cfg, "call123")
+        self.assertIs(self.helper.config, cfg)
+        self.assertEqual(self.helper.call_id, "call123")
+
+    def test_init_vector_store_reuse(self):
+        """test init_vector_store reuses existing instance"""
+        emb_model = MagicMock()
+        first = self.helper.init_vector_store(emb_model)
+        second = self.helper.init_vector_store(emb_model)
+        self.assertIs(second, first)
+
+    def test_get_state_models_and_data_success(self):
+        """test get_state_models_and_data returns models and data"""
+        emb = MagicMock()
+        llm = MagicMock()
+        articles = {"p": {}}
+        state = {
+            "text_embedding_model": emb,
+            "llm_model": llm,
+            "article_data": articles,
+        }
+        ret_emb, ret_llm, ret_articles = self.helper.get_state_models_and_data(state)
+        self.assertIs(ret_emb, emb)
+        self.assertIs(ret_llm, llm)
+        self.assertIs(ret_articles, articles)
+
+    def test_get_state_models_and_data_missing_text_embedding(self):
+        """test get_state_models_and_data raises ValueError if missing text embedding"""
+        state = {"llm_model": MagicMock(), "article_data": {"p": {}}}
+        with self.assertRaises(ValueError) as cm:
+            self.helper.get_state_models_and_data(state)
+        self.assertEqual(str(cm.exception), "No text embedding model found in state.")
+
+    def test_get_state_models_and_data_missing_llm(self):
+        """test get_state_models_and_data raises ValueError if missing LLM"""
+        state = {"text_embedding_model": MagicMock(), "article_data": {"p": {}}}
+        with self.assertRaises(ValueError) as cm:
+            self.helper.get_state_models_and_data(state)
+        self.assertEqual(str(cm.exception), "No LLM model found in state.")
+
+    def test_get_state_models_and_data_missing_article_data(self):
+        """test get_state_models_and_data raises ValueError if missing article data"""
+        state = {"text_embedding_model": MagicMock(), "llm_model": MagicMock()}
+        with self.assertRaises(ValueError) as cm:
+            self.helper.get_state_models_and_data(state)
+        self.assertEqual(str(cm.exception), "No article_data found in state.")
+
+    def test_load_candidate_papers_calls_add_paper_only_for_valid(self):
+        """test load_candidate_papers calls add_paper only for valid candidates"""
+        vs = SimpleNamespace(loaded_papers=set(), add_paper=MagicMock())
+        articles = {"p1": {"pdf_url": "url1"}, "p2": {}, "p3": {"pdf_url": None}}
+        candidates = ["p1", "p2", "p3"]
+        self.helper.load_candidate_papers(vs, articles, candidates)
+        vs.add_paper.assert_called_once_with("p1", "url1", articles["p1"])
+
+    def test_load_candidate_papers_handles_add_paper_exception(self):
+        """test load_candidate_papers handles add_paper exception"""
+        # If add_paper raises, it should be caught and not propagate
+        vs = SimpleNamespace(
+            loaded_papers=set(), add_paper=MagicMock(side_effect=ValueError("oops"))
+        )
+        articles = {"p1": {"pdf_url": "url1"}}
+        # Start call to set call_id (used in logging)
+        self.helper.start_call(SimpleNamespace(), "call001")
+        # Should not raise despite exception
+        self.helper.load_candidate_papers(vs, articles, ["p1"])
+        vs.add_paper.assert_called_once_with("p1", "url1", articles["p1"])
+
+    def test_run_reranker_success_and_filtering(self):
+        """test run_reranker success and filtering"""
+        # Successful rerank returns filtered candidates
+        cfg = SimpleNamespace(top_k_papers=2)
+        self.helper.config = cfg
+        vs = MagicMock()
+        with patch(
+            "aiagents4pharma.talk2scholars.tools.pdf.utils.tool_helper.rank_papers_by_query",
+            return_value=["a", "c"],
+        ):
+            out = self.helper.run_reranker(vs, "q", ["a", "b"])
+        self.assertEqual(out, ["a"])
+
+    def test_run_reranker_exception_fallback(self):
+        """test run_reranker exception fallback"""
+        # On reranker failure, should return original candidates
+        cfg = SimpleNamespace(top_k_papers=5)
+        self.helper.config = cfg
+        vs = MagicMock()
+
+        def fail(*args, **kwargs):
+            raise RuntimeError("fail")
+
+        with patch(
+            "aiagents4pharma.talk2scholars.tools.pdf.utils.tool_helper.rank_papers_by_query",
+            side_effect=fail,
+        ):
+            candidates = ["x", "y"]
+            out = self.helper.run_reranker(vs, "q", candidates)
+        self.assertEqual(out, candidates)
+
+    def test_format_answer_with_and_without_sources(self):
+        """test format_answer with and without sources"""
+        articles = {"p1": {"Title": "T1"}, "p2": {"Title": "T2"}}
+        # With sources
+        with patch(
+            "aiagents4pharma.talk2scholars.tools.pdf.utils.tool_helper.generate_answer",
+            return_value={"output_text": "ans", "papers_used": ["p1", "p2"]},
+        ):
+            res = self.helper.format_answer("q", [], MagicMock(), articles)
+            self.assertIn("ans", res)
+            self.assertIn("Sources:", res)
+            self.assertIn("- T1", res)
+            self.assertIn("- T2", res)
+        # Without sources
+        with patch(
+            "aiagents4pharma.talk2scholars.tools.pdf.utils.tool_helper.generate_answer",
+            return_value={"output_text": "ans", "papers_used": []},
+        ):
+            res2 = self.helper.format_answer("q", [], MagicMock(), {})
+            self.assertEqual(res2, "ans")

aiagents4pharma/talk2scholars/tests/test_zotero_agent.py

@@ -2,7 +2,6 @@
 Updated Unit Tests for the Zotero agent (Zotero Library Managent sub-agent).
 """
 
-# pylint: disable=redefined-outer-name
 from unittest import mock
 import pytest
 from langchain_core.messages import HumanMessage, AIMessage

aiagents4pharma/talk2scholars/tests/test_zotero_read.py

@@ -17,8 +17,6 @@ from aiagents4pharma.talk2scholars.tools.zotero.utils.zotero_pdf_downloader impo
 )
 from aiagents4pharma.talk2scholars.tools.zotero.zotero_read import zotero_read
 
-# pylint: disable=protected-access
-# pylint: disable=protected-access, too-many-arguments, too-many-positional-arguments
 
 # Dummy Hydra configuration to be used in tests
 dummy_zotero_read_config = SimpleNamespace(
@@ -211,15 +209,15 @@ class TestZoteroSearchTool(unittest.TestCase):
     @patch(
         "aiagents4pharma.talk2scholars.tools.zotero.utils.read_helper.download_pdfs_in_parallel"
     )
-    def test_filtering_no_matching_papers(
-        self,
-        mock_batch_download,
-        mock_hydra_init,
-        mock_hydra_compose,
-        mock_zotero_class,
-        mock_get_item_collections,
-    ):
+    def test_filtering_no_matching_papers(self, *mocks):
         """Testing filtering when no paper matching"""
+        (
+            mock_batch_download,
+            mock_hydra_init,
+            mock_hydra_compose,
+            mock_zotero_class,
+            mock_get_item_collections,
+        ) = mocks
         mock_hydra_compose.return_value = dummy_cfg
         mock_hydra_init.return_value.__enter__.return_value = None
 
@@ -460,15 +458,15 @@ class TestZoteroSearchTool(unittest.TestCase):
     @patch(
         "aiagents4pharma.talk2scholars.tools.zotero.utils.read_helper.requests.Session.get"
     )
-    def test_pdf_attachment_success(
-        self,
-        mock_session_get,
-        mock_hydra_init,
-        mock_hydra_compose,
-        mock_zotero_class,
-        mock_get_item_collections,
-    ):
+    def test_pdf_attachment_success(self, *mocks):
         """Test for pdf attachment success"""
+        (
+            mock_session_get,
+            mock_hydra_init,
+            mock_hydra_compose,
+            mock_zotero_class,
+            mock_get_item_collections,
+        ) = mocks
         mock_hydra_compose.return_value = dummy_cfg
         mock_hydra_init.return_value.__enter__.return_value = None
 

aiagents4pharma/talk2scholars/tools/paper_download/download_arxiv_input.py

@@ -5,7 +5,7 @@ Tool for downloading arXiv paper metadata and retrieving the PDF URL.
 
 import logging
 import xml.etree.ElementTree as ET
-from typing import Annotated, Any
+from typing import Annotated, Any, List
 
 import hydra
 import requests
@@ -23,12 +23,22 @@ logger = logging.getLogger(__name__)
 class DownloadArxivPaperInput(BaseModel):
     """Input schema for the arXiv paper download tool."""
 
-    arxiv_id: str = Field(
-        description="The arXiv paper ID used to retrieve the paper details and PDF URL."
+    arxiv_ids: List[str] = Field(
+        description="List of arXiv paper IDs used to retrieve paper details and PDF URLs."
     )
     tool_call_id: Annotated[str, InjectedToolCallId]
 
 
+# Helper to load arXiv download configuration
+def _get_arxiv_config() -> Any:
+    """Load arXiv download configuration."""
+    with hydra.initialize(version_base=None, config_path="../../configs"):
+        cfg = hydra.compose(
+            config_name="config", overrides=["tools/download_arxiv_paper=default"]
+        )
+    return cfg.tools.download_arxiv_paper
+
+
 def fetch_arxiv_metadata(
     api_url: str, arxiv_id: str, request_timeout: int
 ) -> ET.Element:
@@ -42,19 +52,21 @@ def fetch_arxiv_metadata(
 def extract_metadata(entry: ET.Element, ns: dict, arxiv_id: str) -> dict:
     """Extract metadata from the XML entry."""
     title_elem = entry.find("atom:title", ns)
-    title = title_elem.text.strip() if title_elem is not None else "N/A"
+    title = (title_elem.text or "").strip() if title_elem is not None else "N/A"
 
-    authors = [
-        author_elem.find("atom:name", ns).text.strip()
-        for author_elem in entry.findall("atom:author", ns)
-        if author_elem.find("atom:name", ns) is not None
-    ]
+    authors = []
+    for author_elem in entry.findall("atom:author", ns):
+        name_elem = author_elem.find("atom:name", ns)
+        if name_elem is not None and name_elem.text:
+            authors.append(name_elem.text.strip())
 
     summary_elem = entry.find("atom:summary", ns)
-    abstract = summary_elem.text.strip() if summary_elem is not None else "N/A"
+    abstract = (summary_elem.text or "").strip() if summary_elem is not None else "N/A"
 
     published_elem = entry.find("atom:published", ns)
-    pub_date = published_elem.text.strip() if published_elem is not None else "N/A"
+    pub_date = (
+        (published_elem.text or "").strip() if published_elem is not None else "N/A"
+    )
 
     pdf_url = next(
         (
@@ -80,43 +92,86 @@ def extract_metadata(entry: ET.Element, ns: dict, arxiv_id: str) -> dict:
     }
 
 
-@tool(args_schema=DownloadArxivPaperInput, parse_docstring=True)
+def _get_snippet(abstract: str) -> str:
+    """Extract the first one or two sentences from an abstract."""
+    if not abstract or abstract == "N/A":
+        return ""
+    sentences = abstract.split(". ")
+    snippet_sentences = sentences[:2]
+    snippet = ". ".join(snippet_sentences)
+    if not snippet.endswith("."):
+        snippet += "."
+    return snippet
+
+
+def _build_summary(article_data: dict[str, Any]) -> str:
+    """Build a summary string for up to three papers with snippets."""
+    top = list(article_data.values())[:3]
+    lines: list[str] = []
+    for idx, paper in enumerate(top):
+        title = paper.get("Title", "N/A")
+        pub_date = paper.get("Publication Date", "N/A")
+        url = paper.get("URL", "")
+        snippet = _get_snippet(paper.get("Abstract", ""))
+        line = f"{idx+1}. {title} ({pub_date})"
+        if url:
+            line += f"\n   View PDF: {url}"
+        if snippet:
+            line += f"\n   Abstract snippet: {snippet}"
+        lines.append(line)
+    summary = "\n".join(lines)
+    return (
+        "Download was successful. Papers metadata are attached as an artifact. "
+        "Here is a summary of the results:\n"
+        f"Number of papers found: {len(article_data)}\n"
+        "Top 3 papers:\n" + summary
+    )
+
+
+@tool(
+    args_schema=DownloadArxivPaperInput,
+    parse_docstring=True,
+)
 def download_arxiv_paper(
-    arxiv_id: str,
+    arxiv_ids: List[str],
     tool_call_id: Annotated[str, InjectedToolCallId],
 ) -> Command[Any]:
     """
-    Get metadata and PDF URL for an arXiv paper using its unique arXiv ID.
+    Get metadata and PDF URLs for one or more arXiv papers using their unique arXiv IDs.
     """
-    logger.info("Fetching metadata from arXiv for paper ID: %s", arxiv_id)
+    logger.info("Fetching metadata from arXiv for paper IDs: %s", arxiv_ids)
 
     # Load configuration
-    with hydra.initialize(version_base=None, config_path="../../configs"):
-        cfg = hydra.compose(
-            config_name="config", overrides=["tools/download_arxiv_paper=default"]
+    cfg = _get_arxiv_config()
+    api_url = cfg.api_url
+    request_timeout = cfg.request_timeout
+
+    # Aggregate results
+    article_data: dict[str, Any] = {}
+    for aid in arxiv_ids:
+        logger.info("Processing arXiv ID: %s", aid)
+        # Fetch and parse metadata
+        entry = fetch_arxiv_metadata(api_url, aid, request_timeout).find(
+            "atom:entry", {"atom": "http://www.w3.org/2005/Atom"}
+        )
+        if entry is None:
+            logger.warning("No entry found for arXiv ID %s", aid)
+            continue
+        article_data[aid] = extract_metadata(
+            entry, {"atom": "http://www.w3.org/2005/Atom"}, aid
         )
-        api_url = cfg.tools.download_arxiv_paper.api_url
-        request_timeout = cfg.tools.download_arxiv_paper.request_timeout
-
-    # Fetch and parse metadata
-    root = fetch_arxiv_metadata(api_url, arxiv_id, request_timeout)
-    ns = {"atom": "http://www.w3.org/2005/Atom"}
-
-    entry = root.find("atom:entry", ns)
-    if entry is None:
-        raise ValueError(f"No entry found for arXiv ID {arxiv_id}")
-
-    # Extract metadata
-    metadata = extract_metadata(entry, ns, arxiv_id)
-
-    # Create article_data entry with the paper ID as the key
-    article_data = {arxiv_id: metadata}
-
-    content = f"Successfully retrieved metadata and PDF URL for arXiv ID {arxiv_id}"
 
+    # Build and return summary
+    content = _build_summary(article_data)
     return Command(
         update={
             "article_data": article_data,
-            "messages": [ToolMessage(content=content, tool_call_id=tool_call_id)],
+            "messages": [
+                ToolMessage(
+                    content=content,
+                    tool_call_id=tool_call_id,
+                    artifact=article_data,
+                )
+            ],
         }
     )