feat: seperating test cases for deepresearch and ai_scrape.

Khurdhula-Harshavardhan · Khurdhula-Harshavardhan · commit ba56ab1b30b5 · 2025-09-12T09:03:57.000-07:00
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
@@ -8,35 +8,8 @@ on:
 
 jobs:
   ruff-format-check:
-    name: Ruff Format Check - ${{ matrix.file }}
+    name: Ruff Format Check
     runs-on: ubuntu-latest
-    strategy:
-      fail-fast: false
-      matrix:
-        file:
-          - __init__.py
-          - _config.py
-          - _types.py
-          - async_request.py
-          - audio.py
-          - classification.py
-          - embedding_v2.py
-          - embedding.py
-          - exceptions.py
-          - helpers.py
-          - image_generation.py
-          - prediction.py
-          - prompt_engine.py
-          - request.py
-          - search.py
-          - sentiment.py
-          - sql.py
-          - store.py
-          - summary.py
-          - translate.py
-          - validate.py
-          - vision.py
-          - web.py
     steps:
       - uses: actions/checkout@v4
       
@@ -48,9 +21,10 @@ jobs:
       - name: Install ruff
         run: pip install ruff
       
-      - name: Check formatting for ${{ matrix.file }}
+      - name: Check all files with ruff
         run: |
-          ruff check jigsawstack/${{ matrix.file }} --config .github/ruff.toml
+          ruff check jigsawstack/ --config .github/ruff.toml
+          ruff format --check jigsawstack/ --config .github/ruff.toml
 
   test:
     name: Test - ${{ matrix.test-file }}
@@ -72,6 +46,8 @@ jobs:
           - test_translate.py
           - test_validate.py
           - test_web.py
+          - test_deep_research.py
+          - test_ai_scrape.py
     steps:
       - uses: actions/checkout@v4
       
diff --git a/tests/test_ai_scrape.py b/tests/test_ai_scrape.py
@@ -0,0 +1,141 @@
+import logging
+import os
+
+import pytest
+from dotenv import load_dotenv
+
+import jigsawstack
+from jigsawstack.exceptions import JigsawStackError
+
+load_dotenv()
+
+logging.basicConfig(level=logging.INFO)
+logger = logging.getLogger(__name__)
+
+jigsaw = jigsawstack.JigsawStack(api_key=os.getenv("JIGSAWSTACK_API_KEY"))
+async_jigsaw = jigsawstack.AsyncJigsawStack(api_key=os.getenv("JIGSAWSTACK_API_KEY"))
+
+URL = "https://jigsawstack.com"
+
+# AI Scrape Test Cases
+AI_SCRAPE_TEST_CASES = [
+    {
+        "name": "scrape_with_element_prompts",
+        "params": {
+            "url": URL,
+            "element_prompts": ["title", "main content", "navigation links"],
+        },
+    },
+    {
+        "name": "scrape_with_selectors",
+        "params": {
+            "url": URL,
+            "selectors": ["h1", "p", "a"],
+        },
+    },
+    {
+        "name": "scrape_with_features",
+        "params": {
+            "url": URL,
+            "element_prompts": ["title"],
+            "features": ["meta", "link"],
+        },
+    },
+    {
+        "name": "scrape_with_root_element",
+        "params": {
+            "url": URL,
+            "element_prompts": ["content"],
+            "root_element_selector": "main",
+        },
+    },
+    {
+        "name": "scrape_with_wait_for_timeout",
+        "params": {
+            "url": URL,
+            "element_prompts": ["content"],
+            "wait_for": {"mode": "timeout", "value": 3000},
+        },
+    },
+    {
+        "name": "scrape_mobile_view",
+        "params": {
+            "url": URL,
+            "element_prompts": ["mobile menu"],
+            "is_mobile": True,
+        },
+    },
+    {
+        "name": "scrape_with_cookies",
+        "params": {
+            "url": URL,
+            "element_prompts": ["user data"],
+            "cookies": [{"name": "session", "value": "test123", "domain": "example.com"}],
+        },
+    },
+    {
+        "name": "scrape_with_advance_config",
+        "params": {
+            "url": URL,
+            "element_prompts": ["content"],
+            "advance_config": {"console": True, "network": True, "cookies": True},
+        },
+    },
+]
+
+
+class TestAIScrapeSync:
+    """Test synchronous AI scrape methods"""
+
+    @pytest.mark.parametrize(
+        "test_case",
+        AI_SCRAPE_TEST_CASES,
+        ids=[tc["name"] for tc in AI_SCRAPE_TEST_CASES],
+    )
+    def test_ai_scrape(self, test_case):
+        """Test synchronous AI scrape with various inputs"""
+        try:
+            result = jigsaw.web.ai_scrape(test_case["params"])
+
+            assert result["success"]
+            assert "data" in result
+            assert isinstance(result["data"], list)
+
+            # Check for optional features
+            if "meta" in test_case["params"].get("features", []):
+                assert "meta" in result
+            if "link" in test_case["params"].get("features", []):
+                assert "link" in result
+                assert isinstance(result["link"], list)
+
+        except JigsawStackError as e:
+            pytest.fail(f"Unexpected JigsawStackError in {test_case['name']}: {e}")
+
+
+class TestAIScrapeAsync:
+    """Test asynchronous AI scrape methods"""
+
+    @pytest.mark.parametrize(
+        "test_case",
+        AI_SCRAPE_TEST_CASES,
+        ids=[tc["name"] for tc in AI_SCRAPE_TEST_CASES],
+    )
+    @pytest.mark.asyncio
+    async def test_ai_scrape_async(self, test_case):
+        """Test asynchronous AI scrape with various inputs"""
+        try:
+            result = await async_jigsaw.web.ai_scrape(test_case["params"])
+
+            assert result["success"]
+            assert "data" in result
+            assert isinstance(result["data"], list)
+
+            # Check for optional features
+            if "meta" in test_case["params"].get("features", []):
+                assert "meta" in result
+            if "link" in test_case["params"].get("features", []):
+                assert "link" in result
+                assert isinstance(result["link"], list)
+
+        except JigsawStackError as e:
+            pytest.fail(f"Unexpected JigsawStackError in {test_case['name']}: {e}")
diff --git a/tests/test_deep_research.py b/tests/test_deep_research.py
@@ -0,0 +1,95 @@
+import logging
+import os
+
+import pytest
+from dotenv import load_dotenv
+
+import jigsawstack
+from jigsawstack.exceptions import JigsawStackError
+
+load_dotenv()
+
+logging.basicConfig(level=logging.INFO)
+logger = logging.getLogger(__name__)
+
+jigsaw = jigsawstack.JigsawStack(api_key=os.getenv("JIGSAWSTACK_API_KEY"))
+async_jigsaw = jigsawstack.AsyncJigsawStack(api_key=os.getenv("JIGSAWSTACK_API_KEY"))
+
+URL = "https://jigsawstack.com"
+
+
+# Deep Research Test Cases
+DEEP_RESEARCH_TEST_CASES = [
+    {
+        "name": "basic_deep_research",
+        "params": {
+            "query": "climate change effects",
+        },
+    },
+    {
+        "name": "technical_deep_research",
+        "params": {
+            "query": "quantum computing applications in cryptography",
+        },
+    },
+    {
+        "name": "deep_research_with_depth",
+        "params": {
+            "query": "renewable energy sources",
+            "depth": 2,
+        },
+    },
+]
+
+
+class TestDeepResearchSync:
+    """Test synchronous deep research methods"""
+
+    @pytest.mark.parametrize(
+        "test_case",
+        DEEP_RESEARCH_TEST_CASES,
+        ids=[tc["name"] for tc in DEEP_RESEARCH_TEST_CASES],
+    )
+    def test_deep_research(self, test_case):
+        """Test synchronous deep research with various inputs"""
+        try:
+            result = jigsaw.web.deep_research(test_case["params"])
+
+            assert result["success"]
+            assert "results" in result
+            assert isinstance(result["results"], str)
+            assert len(result["results"]) > 0
+
+            # Check for sources
+            if "sources" in result:
+                assert isinstance(result["sources"], list)
+
+        except JigsawStackError as e:
+            pytest.fail(f"Unexpected JigsawStackError in {test_case['name']}: {e}")
+
+
+class TestDeepResearchAsync:
+    """Test asynchronous deep research methods"""
+
+    @pytest.mark.parametrize(
+        "test_case",
+        DEEP_RESEARCH_TEST_CASES,
+        ids=[tc["name"] for tc in DEEP_RESEARCH_TEST_CASES],
+    )
+    @pytest.mark.asyncio
+    async def test_deep_research_async(self, test_case):
+        """Test asynchronous deep research with various inputs"""
+        try:
+            result = await async_jigsaw.web.deep_research(test_case["params"])
+
+            assert result["success"]
+            assert "results" in result
+            assert isinstance(result["results"], str)
+            assert len(result["results"]) > 0
+
+            # Check for sources
+            if "sources" in result:
+                assert isinstance(result["sources"], list)
+
+        except JigsawStackError as e:
+            pytest.fail(f"Unexpected JigsawStackError in {test_case['name']}: {e}")
diff --git a/tests/test_web.py b/tests/test_web.py
