| from __future__ import annotations |
|
|
| from dataclasses import dataclass |
| from datetime import datetime, timezone |
| import json |
|
|
| import pytest |
|
|
| import aura.agents.iris.tools as iris |
| from aura.core.config import AppConfig, FeatureFlags, ModelSettings, PathsSettings |
| import aura.memory.mneme.tools as mneme_tools |
|
|
|
|
| @dataclass |
| class FakeResponse: |
| content: str |
|
|
|
|
| class FakeRouter: |
| def generate(self, prompt: str) -> FakeResponse: |
| return FakeResponse(content="Synthesized answer") |
|
|
|
|
| @pytest.fixture() |
| def iris_config(tmp_path): |
| config = AppConfig( |
| name="AURA", |
| offline_mode=True, |
| log_level="INFO", |
| primary_model=ModelSettings(provider="ollama", name="llama3", host="http://127.0.0.1:11434"), |
| fallback_models=[], |
| paths=PathsSettings( |
| allowed_roots=[tmp_path], |
| data_dir=tmp_path, |
| log_dir=tmp_path / "logs", |
| memory_dir=tmp_path / "memory", |
| ipc_socket=tmp_path / "aura.sock", |
| ), |
| features=FeatureFlags(hotkey=True, tray=True, ipc=True, api=True), |
| source_path=tmp_path / "config.yaml", |
| ) |
| mneme_tools.set_config(config) |
| iris.set_config(config) |
| iris.set_router(FakeRouter()) |
| return config |
|
|
|
|
| def test_web_search_caches_and_summarizes(monkeypatch, iris_config, tmp_path): |
| calls = {"count": 0} |
|
|
| def fake_search(query, num_results, date_filter): |
| calls["count"] += 1 |
| return [iris.SearchResult(title="Doc", url=str(tmp_path / "page.html"), snippet="Snippet", source_domain="example.com", relevance_score=1.0)] |
|
|
| page = tmp_path / "page.html" |
| page.write_text("<html><head><title>Doc</title></head><body><h1>Alpha</h1><p>Beta.</p></body></html>", encoding="utf-8") |
| monkeypatch.setattr(iris, "_search_backend", fake_search) |
| first = iris.web_search("alpha", num_results=1) |
| second = iris.web_search("alpha", num_results=1) |
| summary = iris.summarize_content(str(page)) |
|
|
| assert calls["count"] == 1 |
| assert first[0].title == "Doc" |
| assert second[0].title == "Doc" |
| assert summary.summary_text |
|
|
|
|
| def test_deep_research_and_fact_check(monkeypatch, iris_config, tmp_path): |
| page = tmp_path / "source.html" |
| page.write_text("<html><body><p>Python is a language.</p></body></html>", encoding="utf-8") |
|
|
| monkeypatch.setattr(iris, "web_search", lambda query, num_results=10, date_filter=None, safe_search=True: [iris.SearchResult(title="S", url=str(page), snippet="Python is a language", source_domain="example.com", relevance_score=1.0)]) |
| monkeypatch.setattr(iris, "fetch_url", lambda url, extract_main_content=True: iris.PageContent(url=url, title="S", main_text="Python is a language.", word_count=4, fetched_at=datetime.now(timezone.utc), links=[])) |
|
|
| deep = iris.deep_research("python") |
| fact = iris.fact_check("Python is a language") |
| papers = iris.search_academic("neural nets", source="arxiv", max_results=1) |
|
|
| assert deep.synthesized_answer |
| assert fact.verdict in {"supported", "contradicted", "unverified"} |
| assert isinstance(papers, list) |
|
|
|
|
| def test_compare_and_citations(monkeypatch, iris_config, tmp_path): |
| page = tmp_path / "compare.html" |
| page.write_text("<html><body><p>Reference 2024. Doe, J.</p></body></html>", encoding="utf-8") |
| monkeypatch.setattr(iris, "fetch_url", lambda url, extract_main_content=True: iris.PageContent(url=url, title="T", main_text="Reference 2024. Doe, J.", word_count=4, fetched_at=datetime.now(timezone.utc), links=[])) |
| compare = iris.compare_sources([str(page)], "question") |
| citations = iris.extract_citations("Reference 2024. Doe, J.") |
| assert compare.sources == [str(page)] |
| assert citations |
|
|
|
|
| class FakeHTTPResponse: |
| def __init__(self, payload: bytes, status: int = 200): |
| self._payload = payload |
| self.status = status |
|
|
| def read(self): |
| return self._payload |
|
|
| def __enter__(self): |
| return self |
|
|
| def __exit__(self, exc_type, exc, tb): |
| return False |
|
|
|
|
| def test_fetch_and_academic_search(monkeypatch, iris_config, tmp_path): |
| page = tmp_path / "raw.html" |
| page.write_text("<html><head><title>Raw</title></head><body>Body text</body></html>", encoding="utf-8") |
| raw = iris.fetch_url(page.as_uri(), extract_main_content=False) |
| assert raw.title == "Raw" |
| assert "Body text" in raw.main_text |
|
|
| arxiv_xml = b"""<?xml version='1.0' encoding='utf-8'?> |
| <feed xmlns='http://www.w3.org/2005/Atom'> |
| <entry> |
| <title>Paper One</title> |
| <summary>Abstract one</summary> |
| <id>http://arxiv.org/abs/1234.5678</id> |
| <author><name>Alice</name></author> |
| </entry> |
| </feed>""" |
| sem_json = json.dumps({"data": [{"title": "Paper Two", "abstract": "Abstract two", "authors": [{"name": "Bob"}], "url": "https://example.com", "year": 2024, "citationCount": 5}]}).encode("utf-8") |
|
|
| def fake_urlopen(url, timeout=20): |
| if "arxiv" in url: |
| return FakeHTTPResponse(arxiv_xml) |
| return FakeHTTPResponse(sem_json) |
|
|
| monkeypatch.setattr(iris.urllib.request, "urlopen", fake_urlopen) |
| papers = iris.search_academic("neural nets", source="both", max_results=2) |
| assert len(papers) == 2 |
| assert {paper.source for paper in papers} == {"arxiv", "semantic_scholar"} |
|
|
