tooluniverse 1.0.5__py3-none-any.whl → 1.0.6__py3-none-any.whl

tooluniverse/__init__.py

@@ -224,6 +224,19 @@ if not LAZY_LOADING_ENABLED:
         ODPHPTopicSearch,
         ODPHPOutlinkFetch,
     )
+    # Literature search tools
+    from .arxiv_tool import ArXivTool
+    from .crossref_tool import CrossrefTool
+    from .dblp_tool import DBLPTool
+    from .pubmed_tool import PubMedTool
+    from .doaj_tool import DOAJTool
+    from .unpaywall_tool import UnpaywallTool
+    from .biorxiv_tool import BioRxivTool
+    from .medrxiv_tool import MedRxivTool
+    from .hal_tool import HALTool
+    from .core_tool import CoreTool
+    from .pmc_tool import PMCTool
+    from .zenodo_tool import ZenodoTool
 else:
     # With lazy loading, create lazy import proxies that import modules only when accessed
     MonarchTool = _LazyImportProxy("restful_tool", "MonarchTool")
@@ -305,6 +318,19 @@ else:
     ODPHPMyHealthfinder = _LazyImportProxy("odphp_tool", "ODHPHPMyHealthfinder")
     ODPHPTopicSearch = _LazyImportProxy("odphp_tool", "ODPHPTopicSearch")
     ODPHPOutlinkFetch = _LazyImportProxy("odphp_tool", "ODPHPOutlinkFetch")
+    # Literature search tools
+    ArXivTool = _LazyImportProxy("arxiv_tool", "ArXivTool")
+    CrossrefTool = _LazyImportProxy("crossref_tool", "CrossrefTool")
+    DBLPTool = _LazyImportProxy("dblp_tool", "DBLPTool")
+    PubMedTool = _LazyImportProxy("pubmed_tool", "PubMedTool")
+    DOAJTool = _LazyImportProxy("doaj_tool", "DOAJTool")
+    UnpaywallTool = _LazyImportProxy("unpaywall_tool", "UnpaywallTool")
+    BioRxivTool = _LazyImportProxy("biorxiv_tool", "BioRxivTool")
+    MedRxivTool = _LazyImportProxy("medrxiv_tool", "MedRxivTool")
+    HALTool = _LazyImportProxy("hal_tool", "HALTool")
+    CoreTool = _LazyImportProxy("core_tool", "CoreTool")
+    PMCTool = _LazyImportProxy("pmc_tool", "PMCTool")
+    ZenodoTool = _LazyImportProxy("zenodo_tool", "ZenodoTool")
 
 __all__ = [
     "__version__",
@@ -376,4 +402,17 @@ __all__ = [
     "ODPHPItemList",
     "ODPHPTopicSearch",
     "ODPHPOutlinkFetch",
+    # Literature search tools
+    "ArXivTool",
+    "CrossrefTool",
+    "DBLPTool",
+    "PubMedTool",
+    "DOAJTool",
+    "UnpaywallTool",
+    "BioRxivTool",
+    "MedRxivTool",
+    "HALTool",
+    "CoreTool",
+    "PMCTool",
+    "ZenodoTool",
 ]

tooluniverse/agentic_tool.py

@@ -3,7 +3,7 @@ from __future__ import annotations
 import os
 import json
 from datetime import datetime
-from typing import Any, Dict, List, Optional
+from typing import Any, Callable, Dict, List, Optional
 
 from .base_tool import BaseTool
 from .tool_registry import register_tool
@@ -30,6 +30,8 @@ API_KEY_ENV_VARS = {
 class AgenticTool(BaseTool):
     """Generic wrapper around LLM prompting supporting JSON-defined configs with prompts and input arguments."""
 
+    STREAM_FLAG_KEY = "_tooluniverse_stream"
+
     @staticmethod
     def has_any_api_keys() -> bool:
         """
@@ -250,9 +252,18 @@ class AgenticTool(BaseTool):
             raise ValueError("max_new_tokens must be positive or None")
 
     # ------------------------------------------------------------------ public API --------------
-    def run(self, arguments: Dict[str, Any]) -> Dict[str, Any]:
+    def run(
+        self,
+        arguments: Dict[str, Any],
+        stream_callback: Optional[Callable[[str], None]] = None,
+    ) -> Dict[str, Any]:
         start_time = datetime.now()
 
+        # Work on a copy so we can remove control flags without mutating caller data
+        arguments = dict(arguments or {})
+        stream_flag = bool(arguments.pop("_tooluniverse_stream", False))
+        streaming_requested = stream_flag or stream_callback is not None
+
         # Check if tool is available before attempting to run
         if not self._is_available:
             error_msg = f"Tool '{self.name}' is not available due to initialization error: {self._initialization_error}"
@@ -300,16 +311,52 @@ class AgenticTool(BaseTool):
             custom_format = arguments.get("response_format", None)
 
             # Delegate to client; client handles provider-specific logic
-            response = self._llm_client.infer(
-                messages=messages,
-                temperature=self._temperature,
-                max_tokens=None,  # client resolves per-model defaults/env
-                return_json=self._return_json,
-                custom_format=custom_format,
-                max_retries=self._max_retries,
-                retry_delay=self._retry_delay,
+            response = None
+
+            streaming_permitted = (
+                streaming_requested and not self._return_json and custom_format is None
             )
 
+            if streaming_permitted and hasattr(self._llm_client, "infer_stream"):
+                try:
+                    chunks_collected: List[str] = []
+                    stream_iter = self._llm_client.infer_stream(
+                        messages=messages,
+                        temperature=self._temperature,
+                        max_tokens=None,
+                        return_json=self._return_json,
+                        custom_format=custom_format,
+                        max_retries=self._max_retries,
+                        retry_delay=self._retry_delay,
+                    )
+                    for chunk in stream_iter:
+                        if not chunk:
+                            continue
+                        chunks_collected.append(chunk)
+                        self._emit_stream_chunk(chunk, stream_callback)
+                    if chunks_collected:
+                        response = "".join(chunks_collected)
+                except Exception as stream_error:  # noqa: BLE001
+                    self.logger.warning(
+                        f"Streaming failed for tool '{self.name}': {stream_error}. Falling back to buffered response."
+                    )
+                    response = None
+
+            if response is None:
+                response = self._llm_client.infer(
+                    messages=messages,
+                    temperature=self._temperature,
+                    max_tokens=None,  # client resolves per-model defaults/env
+                    return_json=self._return_json,
+                    custom_format=custom_format,
+                    max_retries=self._max_retries,
+                    retry_delay=self._retry_delay,
+                )
+
+                if streaming_requested and response:
+                    for chunk in self._iter_chunks(response):
+                        self._emit_stream_chunk(chunk, stream_callback)
+
             end_time = datetime.now()
             execution_time = (end_time - start_time).total_seconds()
 
@@ -338,7 +385,8 @@ class AgenticTool(BaseTool):
                 }
             else:
                 return response
-        except Exception as e:
+
+        except Exception as e:  # noqa: BLE001
             end_time = datetime.now()
             execution_time = (end_time - start_time).total_seconds()
             self.logger.error(f"Error executing {self.name}: {str(e)}")
@@ -359,13 +407,35 @@ class AgenticTool(BaseTool):
                         "model_info": {
                             "api_type": self._api_type,
                             "model_id": self._model_id,
+                            "temperature": self._temperature,
+                            "max_new_tokens": self._max_new_tokens,
                         },
                         "execution_time_seconds": execution_time,
                         "timestamp": start_time.isoformat(),
                     },
                 }
             else:
-                return "error: " + str(e) + " error_type: " + type(e).__name__
+                return f"error: {str(e)} error_type: {type(e).__name__}"
+
+    @staticmethod
+    def _iter_chunks(text: str, size: int = 800):
+        if not text:
+            return
+        for idx in range(0, len(text), size):
+            yield text[idx : idx + size]
+
+    def _emit_stream_chunk(
+        self, chunk: Optional[str], stream_callback: Optional[Callable[[str], None]]
+    ) -> None:
+        if not stream_callback or not chunk:
+            return
+        try:
+            stream_callback(chunk)
+        except Exception as callback_error:  # noqa: BLE001
+            # Streaming callbacks should not break tool execution; log and continue
+            self.logger.debug(
+                f"Stream callback for tool '{self.name}' raised an exception: {callback_error}"
+            )
 
     # ------------------------------------------------------------------ helpers -----------------
     def _validate_arguments(self, arguments: Dict[str, Any]):

tooluniverse/arxiv_tool.py

@@ -0,0 +1,113 @@
+import requests
+import xml.etree.ElementTree as ET
+from .base_tool import BaseTool
+from .tool_registry import register_tool
+
+
+@register_tool("ArXivTool")
+class ArXivTool(BaseTool):
+    """
+    Search arXiv for papers by keyword using the public arXiv API.
+    """
+
+    def __init__(
+        self,
+        tool_config,
+        base_url="http://export.arxiv.org/api/query",
+    ):
+        super().__init__(tool_config)
+        self.base_url = base_url
+
+    def run(self, arguments):
+        query = arguments.get("query")
+        limit = int(arguments.get("limit", 10))
+        # sort_by: relevance | lastUpdatedDate | submittedDate
+        sort_by = arguments.get("sort_by", "relevance")
+        # sort_order: ascending | descending
+        sort_order = arguments.get("sort_order", "descending")
+
+        if not query:
+            return {"error": "`query` parameter is required."}
+
+        return self._search(query, limit, sort_by, sort_order)
+
+    def _search(self, query, limit, sort_by, sort_order):
+        params = {
+            "search_query": f"all:{query}",
+            "start": 0,
+            "max_results": max(1, min(limit, 200)),
+            "sortBy": sort_by,
+            "sortOrder": sort_order,
+        }
+
+        try:
+            response = requests.get(self.base_url, params=params, timeout=20)
+        except requests.RequestException as e:
+            return {
+                "error": "Network error calling arXiv API",
+                "reason": str(e),
+            }
+
+        if response.status_code != 200:
+            return {
+                "error": f"arXiv API error {response.status_code}",
+                "reason": response.reason,
+            }
+
+        # Parse Atom XML
+        try:
+            root = ET.fromstring(response.text)
+        except ET.ParseError as e:
+            return {
+                "error": "Failed to parse arXiv response",
+                "reason": str(e),
+            }
+
+        ns = {"atom": "http://www.w3.org/2005/Atom"}
+        entries = []
+        for entry in root.findall("atom:entry", ns):
+            title_text = entry.findtext(
+                "atom:title",
+                default="",
+                namespaces=ns,
+            )
+            title = (title_text or "").strip()
+            summary_text = entry.findtext(
+                "atom:summary",
+                default="",
+                namespaces=ns,
+            )
+            summary = (summary_text or "").strip()
+            link_el = entry.find("atom:link[@type='text/html']", ns)
+            if link_el is not None:
+                link = link_el.get("href")
+            else:
+                link = entry.findtext("atom:id", default="", namespaces=ns)
+            published = entry.findtext(
+                "atom:published", default="", namespaces=ns
+            )
+            updated = entry.findtext("atom:updated", default="", namespaces=ns)
+            authors = [
+                a.findtext("atom:name", default="", namespaces=ns)
+                for a in entry.findall("atom:author", ns)
+            ]
+            primary_category = ""
+            cat_el = entry.find(
+                "{http://arxiv.org/schemas/atom}primary_category"
+            )
+            if cat_el is not None:
+                primary_category = cat_el.get("term", "")
+
+            entries.append(
+                {
+                    "title": title,
+                    "abstract": summary,
+                    "authors": authors,
+                    "published": published,
+                    "updated": updated,
+                    "category": primary_category,
+                    "url": link,
+                }
+            )
+
+        return entries

tooluniverse/biorxiv_tool.py

@@ -0,0 +1,97 @@
+import requests
+from .base_tool import BaseTool
+from .tool_registry import register_tool
+
+
+@register_tool("BioRxivTool")
+class BioRxivTool(BaseTool):
+    """
+    Search bioRxiv preprints using the public bioRxiv API.
+
+    Arguments:
+        query (str): Search term
+        max_results (int): Max results to return (default 10, max 200)
+    """
+
+    def __init__(
+        self,
+        tool_config,
+        base_url="https://api.medrxiv.org/details",
+    ):
+        super().__init__(tool_config)
+        self.base_url = base_url
+
+    def run(self, arguments=None):
+        arguments = arguments or {}
+        query = arguments.get("query")
+        max_results = int(arguments.get("max_results", 10))
+        if not query:
+            return {"error": "`query` parameter is required."}
+        return self._search(query, max_results)
+
+    def _search(self, query, max_results):
+        # Use date range search for recent preprints
+        # Format: /biorxiv/{start_date}/{end_date}/{cursor}/json
+        from datetime import datetime, timedelta
+
+        # Search last 30 days
+        end_date = datetime.now()
+        start_date = end_date - timedelta(days=30)
+
+        url = (f"{self.base_url}/biorxiv/"
+               f"{start_date.strftime('%Y-%m-%d')}/"
+               f"{end_date.strftime('%Y-%m-%d')}/0/json")
+
+        try:
+            resp = requests.get(url, timeout=20)
+            resp.raise_for_status()
+            data = resp.json()
+        except requests.RequestException as e:
+            return {
+                "error": "Network/API error calling bioRxiv",
+                "reason": str(e),
+            }
+        except ValueError:
+            return {"error": "Failed to decode bioRxiv response as JSON"}
+
+        results = []
+        # The API returns a dictionary with a 'collection' key
+        collection = data.get("collection", [])
+        if not isinstance(collection, list):
+            return {"error": "Unexpected API response format"}
+
+        for item in collection:
+            title = item.get("title")
+            authors = item.get("authors", "")
+            if isinstance(authors, str):
+                authors = [a.strip() for a in authors.split(";") if a.strip()]
+            elif isinstance(authors, list):
+                authors = [str(a).strip() for a in authors if str(a).strip()]
+            else:
+                authors = []
+
+            year = None
+            date = item.get("date")
+            if date and len(date) >= 4 and date[:4].isdigit():
+                year = int(date[:4])
+
+            doi = item.get("doi")
+            url = f"https://www.biorxiv.org/content/{doi}" if doi else None
+
+            # Filter by query if provided
+            if query and query.lower() not in (title or "").lower():
+                continue
+
+            results.append(
+                {
+                    "title": title,
+                    "authors": authors,
+                    "year": year,
+                    "doi": doi,
+                    "url": url,
+                    "abstract": item.get("abstract", ""),
+                    "source": "bioRxiv",
+                }
+            )
+
+        return results[:max_results]

tooluniverse/core_tool.py

@@ -0,0 +1,153 @@
+#!/usr/bin/env python3
+"""
+CORE API Tool for searching open access academic papers.
+
+CORE is the world's largest collection of open access research papers.
+This tool provides access to over 200 million open access papers from
+repositories and journals worldwide.
+"""
+
+import requests
+from typing import Dict, List, Any, Optional
+from .base_tool import BaseTool
+from .tool_registry import register_tool
+
+
+@register_tool("CoreTool")
+class CoreTool(BaseTool):
+    """Tool for searching CORE open access academic papers."""
+
+    def __init__(self, tool_config=None):
+        super().__init__(tool_config)
+        self.base_url = "https://api.core.ac.uk/v3"
+        self.session = requests.Session()
+        self.session.headers.update({
+            'User-Agent': 'ToolUniverse/1.0',
+            'Accept': 'application/json'
+        })
+
+    def _search(self, query: str, limit: int = 10,
+                year_from: Optional[int] = None,
+                year_to: Optional[int] = None,
+                language: Optional[str] = None) -> List[Dict[str, Any]]:
+        """
+        Search for papers using CORE API.
+
+        Args:
+            query: Search query
+            limit: Maximum number of results
+            year_from: Start year filter
+            year_to: End year filter
+            language: Language filter (e.g., 'en', 'es', 'fr')
+
+        Returns:
+            List of paper dictionaries
+        """
+        try:
+            # Build search parameters
+            params = {
+                'q': query,
+                'limit': min(limit, 100),  # CORE API max limit is 100
+                'page': 1
+            }
+
+            # Add year filters if provided
+            if year_from or year_to:
+                year_filter = []
+                if year_from:
+                    year_filter.append(f"year:>={year_from}")
+                if year_to:
+                    year_filter.append(f"year:<={year_to}")
+                params['q'] += f" {' '.join(year_filter)}"
+
+            # Add language filter if provided
+            if language:
+                params['q'] += f" language:{language}"
+
+            # Make API request
+            response = self.session.get(
+                f"{self.base_url}/search/works",
+                params=params,
+                timeout=30
+            )
+            response.raise_for_status()
+
+            data = response.json()
+            results = []
+
+            # Parse results
+            for item in data.get('results', []):
+                paper = {
+                    'title': item.get('title', 'No title'),
+                    'abstract': item.get('abstract', 'No abstract available'),
+                    'authors': self._extract_authors(item.get('authors', [])),
+                    'year': self._extract_year(item.get('publishedDate')),
+                    'doi': item.get('doi'),
+                    'url': (item.get('downloadUrl') or
+                            item.get('links', [{}])[0].get('url')),
+                    'venue': item.get('publisher'),
+                    'language': item.get('language', {}).get('code', 'Unknown'),
+                    'open_access': True,  # CORE only contains open access papers
+                    'source': 'CORE',
+                    'citations': item.get('citationCount', 0),
+                    'downloads': item.get('downloadCount', 0)
+                }
+                results.append(paper)
+
+            return results
+
+        except requests.exceptions.RequestException as e:
+            return [{'error': f'CORE API request failed: {str(e)}'}]
+        except Exception as e:
+            return [{'error': f'CORE API error: {str(e)}'}]
+
+    def _extract_authors(self, authors: List[Dict]) -> List[str]:
+        """Extract author names from CORE API response."""
+        if not authors:
+            return []
+
+        author_names = []
+        for author in authors:
+            name = author.get('name', '')
+            if name:
+                author_names.append(name)
+
+        return author_names
+
+    def _extract_year(self, published_date: str) -> str:
+        """Extract year from published date."""
+        if not published_date:
+            return 'Unknown'
+
+        try:
+            # CORE API returns dates in ISO format
+            return published_date[:4]
+        except Exception:
+            return 'Unknown'
+
+    def run(self, tool_arguments) -> List[Dict[str, Any]]:
+        """
+        Execute the CORE search.
+
+        Args:
+            tool_arguments: Dictionary containing search parameters
+
+        Returns:
+            List of paper dictionaries
+        """
+        query = tool_arguments.get('query', '')
+        if not query:
+            return [{'error': 'Query parameter is required'}]
+
+        limit = tool_arguments.get('limit', 10)
+        year_from = tool_arguments.get('year_from')
+        year_to = tool_arguments.get('year_to')
+        language = tool_arguments.get('language')
+
+        return self._search(
+            query=query,
+            limit=limit,
+            year_from=year_from,
+            year_to=year_to,
+            language=language
+        )

tooluniverse/crossref_tool.py

@@ -0,0 +1,73 @@
+import requests
+from .base_tool import BaseTool
+from .tool_registry import register_tool
+
+
+@register_tool("CrossrefTool")
+class CrossrefTool(BaseTool):
+    """
+    Search Crossref Works API for articles by keyword.
+    """
+
+    def __init__(
+        self,
+        tool_config,
+        base_url="https://api.crossref.org/works",
+    ):
+        super().__init__(tool_config)
+        self.base_url = base_url
+
+    def run(self, arguments):
+        query = arguments.get("query")
+        rows = int(arguments.get("limit", 10))
+        # e.g., 'type:journal-article,from-pub-date:2020-01-01'
+        filter_str = arguments.get("filter")
+        if not query:
+            return {"error": "`query` parameter is required."}
+        return self._search(query, rows, filter_str)
+
+    def _search(self, query, rows, filter_str):
+        params = {"query": query, "rows": max(1, min(rows, 100))}
+        if filter_str:
+            params["filter"] = filter_str
+
+        try:
+            response = requests.get(self.base_url, params=params, timeout=20)
+        except requests.RequestException as e:
+            return {
+                "error": "Network error calling Crossref API",
+                "reason": str(e),
+            }
+
+        if response.status_code != 200:
+            return {
+                "error": f"Crossref API error {response.status_code}",
+                "reason": response.reason,
+            }
+
+        data = response.json().get("message", {}).get("items", [])
+        results = []
+        for item in data:
+            title_list = item.get("title") or []
+            title = title_list[0] if title_list else None
+            abstract = item.get("abstract")
+            year = None
+            issued = item.get("issued", {}).get("date-parts") or []
+            if issued and issued[0]:
+                year = issued[0][0]
+            url = item.get("URL")
+            doi = item.get("DOI")
+            container_title = item.get("container-title") or []
+            journal = container_title[0] if container_title else None
+            results.append(
+                {
+                    "title": title,
+                    "abstract": abstract,
+                    "journal": journal,
+                    "year": year,
+                    "doi": doi,
+                    "url": url,
+                }
+            )
+
+        return results