ara-cli 0.1.10.5__py3-none-any.whl → 0.1.13.3__py3-none-any.whl

tests/test_web_search.py

@@ -0,0 +1,467 @@
+"""
+Unit tests for chat_web_search/web_search.py
+
+Provides full test coverage for web search functionality.
+"""
+
+import pytest
+from unittest.mock import patch, MagicMock
+from ara_cli.chat_web_search.web_search import (
+    is_web_search_supported,
+    get_supported_models_message,
+    _get_raw_model_name,
+    _deduplicate_citations,
+    _format_citations,
+    _create_chunk,
+    _extract_openai_citations,
+    _extract_anthropic_text_citations,
+    _extract_anthropic_search_results,
+    perform_openai_web_search,
+    perform_anthropic_web_search,
+    perform_web_search_completion,
+    OPENAI_WEB_SEARCH_MODELS,
+    ANTHROPIC_WEB_SEARCH_MODELS,
+)
+from ara_cli.error_handler import AraError
+
+
+# =============================================================================
+# Tests for is_web_search_supported
+# =============================================================================
+
+
+class TestIsWebSearchSupported:
+    """Tests for is_web_search_supported function."""
+
+    @pytest.mark.parametrize(
+        "model",
+        [
+            "gpt-5",
+            "gpt-5.1",
+            "o3",
+            "o4-mini",
+            "openai/gpt-5",
+            "openai/o4-mini",
+            "gpt-5-search-api",
+            "gpt-4o-search-preview",
+        ],
+    )
+    def test_openai_models_supported(self, model):
+        """OpenAI web search models are supported."""
+        supported, provider = is_web_search_supported(model)
+        assert supported is True
+        assert provider == "openai"
+
+    @pytest.mark.parametrize(
+        "model",
+        [
+            "claude-sonnet-4-5-20250929",
+            "claude-sonnet-4-20250514",
+            "anthropic/claude-sonnet-4-5-20250929",
+            "claude-haiku-4-5-20251001",
+            "claude-opus-4-20250514",
+        ],
+    )
+    def test_anthropic_models_supported(self, model):
+        """Anthropic web search models are supported."""
+        supported, provider = is_web_search_supported(model)
+        assert supported is True
+        assert provider == "anthropic"
+
+    def test_unsupported_model_returns_false(self):
+        """Unsupported models return False."""
+        supported, provider = is_web_search_supported("gpt-4o")
+        assert supported is False
+        assert provider is None
+
+
+# =============================================================================
+# Tests for get_supported_models_message
+# =============================================================================
+
+
+class TestGetSupportedModelsMessage:
+    """Tests for get_supported_models_message function."""
+
+    def test_includes_model_name(self):
+        """Message includes the unsupported model name."""
+        result = get_supported_models_message("unsupported-model")
+        assert "unsupported-model" in result
+
+    def test_lists_openai_models(self):
+        """Message lists OpenAI models."""
+        result = get_supported_models_message("test")
+        assert "gpt-5" in result
+        assert "OpenAI" in result
+
+    def test_lists_anthropic_models(self):
+        """Message lists Anthropic models."""
+        result = get_supported_models_message("test")
+        assert "claude" in result
+        assert "Anthropic" in result
+
+
+# =============================================================================
+# Tests for _get_raw_model_name
+# =============================================================================
+
+
+class TestGetRawModelName:
+    """Tests for _get_raw_model_name function."""
+
+    def test_strips_openai_prefix(self):
+        """Strips openai/ prefix."""
+        result = _get_raw_model_name("openai/gpt-5")
+        assert result == "gpt-5"
+
+    def test_strips_anthropic_prefix(self):
+        """Strips anthropic/ prefix."""
+        result = _get_raw_model_name("anthropic/claude-sonnet-4-20250514")
+        assert result == "claude-sonnet-4-20250514"
+
+    def test_returns_unchanged_without_prefix(self):
+        """Returns model unchanged when no prefix."""
+        result = _get_raw_model_name("gpt-5")
+        assert result == "gpt-5"
+
+
+# =============================================================================
+# Tests for _deduplicate_citations
+# =============================================================================
+
+
+class TestDeduplicateCitations:
+    """Tests for _deduplicate_citations function."""
+
+    def test_removes_duplicate_urls(self):
+        """Removes citations with duplicate URLs."""
+        citations = [
+            {"title": "First", "url": "https://example.com"},
+            {"title": "Second", "url": "https://example.com"},
+            {"title": "Third", "url": "https://other.com"},
+        ]
+        result = _deduplicate_citations(citations)
+        assert len(result) == 2
+        assert result[0]["title"] == "First"
+        assert result[1]["title"] == "Third"
+
+    def test_preserves_order(self):
+        """Preserves order of first occurrences."""
+        citations = [
+            {"title": "A", "url": "https://a.com"},
+            {"title": "B", "url": "https://b.com"},
+            {"title": "C", "url": "https://c.com"},
+        ]
+        result = _deduplicate_citations(citations)
+        assert [c["title"] for c in result] == ["A", "B", "C"]
+
+    def test_handles_empty_list(self):
+        """Handles empty citation list."""
+        result = _deduplicate_citations([])
+        assert result == []
+
+    def test_skips_empty_urls(self):
+        """Skips citations with empty URLs."""
+        citations = [
+            {"title": "First", "url": ""},
+            {"title": "Second", "url": "https://example.com"},
+        ]
+        result = _deduplicate_citations(citations)
+        assert len(result) == 1
+        assert result[0]["title"] == "Second"
+
+
+# =============================================================================
+# Tests for _format_citations
+# =============================================================================
+
+
+class TestFormatCitations:
+    """Tests for _format_citations function."""
+
+    def test_formats_markdown_links(self):
+        """Formats citations as markdown links."""
+        citations = [
+            {"title": "Example Site", "url": "https://example.com"},
+        ]
+        result = _format_citations(citations)
+        assert "[Example Site](https://example.com)" in result
+
+    def test_includes_sources_header(self):
+        """Includes Sources header."""
+        citations = [{"title": "Test", "url": "https://test.com"}]
+        result = _format_citations(citations)
+        assert "**Sources:**" in result
+
+    def test_numbers_citations(self):
+        """Numbers each citation."""
+        citations = [
+            {"title": "First", "url": "https://first.com"},
+            {"title": "Second", "url": "https://second.com"},
+        ]
+        result = _format_citations(citations)
+        assert "1. [First]" in result
+        assert "2. [Second]" in result
+
+    def test_returns_empty_for_no_citations(self):
+        """Returns empty string when no citations."""
+        result = _format_citations([])
+        assert result == ""
+
+    def test_handles_missing_url(self):
+        """Handles citations without URL."""
+        citations = [{"title": "No URL", "url": ""}]
+        result = _format_citations(citations)
+        assert result == ""  # Empty URL is filtered by deduplicate
+
+
+# =============================================================================
+# Tests for _create_chunk
+# =============================================================================
+
+
+class TestCreateChunk:
+    """Tests for _create_chunk function."""
+
+    def test_creates_mock_chunk_with_content(self):
+        """Creates mock chunk with content accessible via choices."""
+        chunk = _create_chunk("test content")
+        assert chunk.choices[0].delta.content == "test content"
+
+
+# =============================================================================
+# Tests for _extract_openai_citations
+# =============================================================================
+
+
+class TestExtractOpenaiCitations:
+    """Tests for _extract_openai_citations function."""
+
+    def test_extracts_url_citations(self):
+        """Extracts URL citations from response."""
+        mock_annotation = MagicMock()
+        mock_annotation.type = "url_citation"
+        mock_annotation.title = "Test Title"
+        mock_annotation.url = "https://test.com"
+
+        mock_content_item = MagicMock()
+        mock_content_item.annotations = [mock_annotation]
+
+        mock_output_item = MagicMock()
+        mock_output_item.type = "message"
+        mock_output_item.content = [mock_content_item]
+
+        mock_response = MagicMock()
+        mock_response.output = [mock_output_item]
+
+        result = _extract_openai_citations(mock_response)
+
+        assert len(result) == 1
+        assert result[0]["title"] == "Test Title"
+        assert result[0]["url"] == "https://test.com"
+
+    def test_returns_empty_for_no_output(self):
+        """Returns empty list when no output."""
+        mock_response = MagicMock()
+        mock_response.output = None
+
+        result = _extract_openai_citations(mock_response)
+
+        assert result == []
+
+
+# =============================================================================
+# Tests for _extract_anthropic_text_citations
+# =============================================================================
+
+
+class TestExtractAnthropicTextCitations:
+    """Tests for _extract_anthropic_text_citations function."""
+
+    def test_extracts_citations_with_url(self):
+        """Extracts citations that have URL attribute."""
+        mock_citation = MagicMock()
+        mock_citation.url = "https://example.com"
+        mock_citation.title = "Example"
+
+        mock_content_block = MagicMock()
+        mock_content_block.citations = [mock_citation]
+
+        result = _extract_anthropic_text_citations(mock_content_block)
+
+        assert len(result) == 1
+        assert result[0]["url"] == "https://example.com"
+
+    def test_returns_empty_for_no_citations(self):
+        """Returns empty list when no citations."""
+        mock_content_block = MagicMock()
+        mock_content_block.citations = None
+
+        result = _extract_anthropic_text_citations(mock_content_block)
+
+        assert result == []
+
+
+# =============================================================================
+# Tests for _extract_anthropic_search_results
+# =============================================================================
+
+
+class TestExtractAnthropicSearchResults:
+    """Tests for _extract_anthropic_search_results function."""
+
+    def test_extracts_web_search_results(self):
+        """Extracts web search result citations."""
+        mock_result = MagicMock()
+        mock_result.type = "web_search_result"
+        mock_result.title = "Search Result"
+        mock_result.url = "https://search.com"
+
+        mock_content_block = MagicMock()
+        mock_content_block.content = [mock_result]
+
+        result = _extract_anthropic_search_results(mock_content_block)
+
+        assert len(result) == 1
+        assert result[0]["title"] == "Search Result"
+        assert result[0]["url"] == "https://search.com"
+
+    def test_skips_non_search_results(self):
+        """Skips items that aren't web_search_result type."""
+        mock_result = MagicMock()
+        mock_result.type = "other_type"
+
+        mock_content_block = MagicMock()
+        mock_content_block.content = [mock_result]
+
+        result = _extract_anthropic_search_results(mock_content_block)
+
+        assert result == []
+
+
+# =============================================================================
+# Tests for perform_openai_web_search
+# =============================================================================
+
+
+class TestPerformOpenaiWebSearch:
+    """Tests for perform_openai_web_search function."""
+
+    @patch("openai.OpenAI")
+    @patch("os.getenv", return_value="test-api-key")
+    def test_uses_chat_completions_for_search_models(
+        self, mock_getenv, mock_openai_class
+    ):
+        """Uses Chat Completions API for search models."""
+        mock_client = MagicMock()
+        mock_openai_class.return_value = mock_client
+
+        mock_chunk = MagicMock()
+        mock_chunk.choices = [MagicMock(delta=MagicMock(content="response"))]
+        mock_client.chat.completions.create.return_value = [mock_chunk]
+
+        results = list(perform_openai_web_search("test query", "gpt-5-search-api"))
+
+        mock_client.chat.completions.create.assert_called_once()
+
+    @patch("openai.OpenAI")
+    @patch("os.getenv", return_value="test-api-key")
+    def test_uses_responses_api_for_other_models(self, mock_getenv, mock_openai_class):
+        """Uses Responses API for non-search models."""
+        mock_client = MagicMock()
+        mock_openai_class.return_value = mock_client
+
+        mock_response = MagicMock()
+        mock_response.output_text = "response text"
+        mock_response.output = []
+        mock_client.responses.create.return_value = mock_response
+
+        results = list(perform_openai_web_search("test query", "gpt-5"))
+
+        mock_client.responses.create.assert_called_once()
+
+
+# =============================================================================
+# Tests for perform_anthropic_web_search
+# =============================================================================
+
+
+class TestPerformAnthropicWebSearch:
+    """Tests for perform_anthropic_web_search function."""
+
+    @patch("anthropic.Anthropic")
+    @patch("os.getenv", return_value="test-api-key")
+    def test_creates_message_with_web_search_tool(
+        self, mock_getenv, mock_anthropic_class
+    ):
+        """Creates message with web_search tool."""
+        mock_client = MagicMock()
+        mock_anthropic_class.return_value = mock_client
+
+        mock_text_block = MagicMock()
+        mock_text_block.type = "text"
+        mock_text_block.text = "response text"
+        mock_text_block.citations = None
+
+        mock_response = MagicMock()
+        mock_response.content = [mock_text_block]
+        mock_client.messages.create.return_value = mock_response
+
+        results = list(
+            perform_anthropic_web_search("test query", "claude-sonnet-4-20250514")
+        )
+
+        mock_client.messages.create.assert_called_once()
+        call_args = mock_client.messages.create.call_args
+        assert any("web_search" in str(arg) for arg in call_args)
+
+
+# =============================================================================
+# Tests for perform_web_search_completion
+# =============================================================================
+
+
+class TestPerformWebSearchCompletion:
+    """Tests for perform_web_search_completion function."""
+
+    @patch("ara_cli.chat_web_search.web_search.LLMSingleton")
+    def test_raises_for_unsupported_model(self, mock_singleton):
+        """Raises AraError for unsupported model."""
+        mock_instance = MagicMock()
+        mock_instance.get_config_by_purpose.return_value = {"model": "gpt-4o"}
+        mock_singleton.get_instance.return_value = mock_instance
+
+        with pytest.raises(AraError) as exc_info:
+            list(perform_web_search_completion("test query"))
+
+        assert "not supported" in str(exc_info.value)
+
+    @patch("ara_cli.chat_web_search.web_search.perform_openai_web_search")
+    @patch("ara_cli.chat_web_search.web_search.LLMSingleton")
+    def test_uses_openai_for_openai_models(self, mock_singleton, mock_openai_search):
+        """Uses OpenAI search for OpenAI models."""
+        mock_instance = MagicMock()
+        mock_instance.get_config_by_purpose.return_value = {"model": "gpt-5"}
+        mock_singleton.get_instance.return_value = mock_instance
+        mock_openai_search.return_value = iter([])
+
+        list(perform_web_search_completion("test query"))
+
+        mock_openai_search.assert_called_once()
+
+    @patch("ara_cli.chat_web_search.web_search.perform_anthropic_web_search")
+    @patch("ara_cli.chat_web_search.web_search.LLMSingleton")
+    def test_uses_anthropic_for_anthropic_models(
+        self, mock_singleton, mock_anthropic_search
+    ):
+        """Uses Anthropic search for Anthropic models."""
+        mock_instance = MagicMock()
+        mock_instance.get_config_by_purpose.return_value = {
+            "model": "claude-sonnet-4-20250514"
+        }
+        mock_singleton.get_instance.return_value = mock_instance
+        mock_anthropic_search.return_value = iter([])
+
+        list(perform_web_search_completion("test query"))
+
+        mock_anthropic_search.assert_called_once()

ara_cli/templates/prompt-modules/blueprints/complete_pytest_unittest.blueprint.md

@@ -1,27 +0,0 @@
-# Usage: 
-# necessary input and adaption: 
-# replace text snippets in <> with specific context
-# ...
-# expected output: 
-# ...
-Do not use usage information as prompt instructions
-
-
-Given source code
-```python
-<source code for context, skip irrelevant for current task>
-```
-
-Given existing unit tests
-```python
-<existing unit tests>
-```
-
-Given pytest is available
-
-Modify and/or create unit tests so this is fully covered:
-```python
-<snippet you want to cover in the next step>
-```
-
-Give me only what is relevant to testing this snippet. Use parametrization where applicable. Split into multiple tests instead of using if-else blocks. Mock all dependencies of tested code.

ara_cli/templates/prompt-modules/blueprints/pytest_unittest_prompt.blueprint.md

@@ -1,32 +0,0 @@
-MODE: STRICT_GENERATION
-TASK: Produce one complete pytest file for a given Python module.
-
-REQUIREMENTS:
-- Output only valid Python code (no text, no markdown).
-- Target: full behavioral + branch coverage (100%).
-- Tests follow AAA pattern (# Arrange, # Act, # Assert).
-- Exactly ONE assert or ONE pytest.raises() per test.
-- Use tmp_path fixture for filesystem isolation.
-- Include an autouse=True fixture for global patching if needed.
-- No external I/O or network calls.
-- All tests independent, self-contained.
-
-STYLE:
-- File starts with module docstring describing AAA, single assert rule, autouse fixture.
-- Group tests with comment headers: 
-  # --- Success paths --- / # --- Error paths --- / # --- Edge cases ---
-- Function names: test_<function>_<expected_behavior>_<condition>
-- Variables descriptive (rel, path, new_content, etc.).
-- Use direct asserts (assert result is True, assert path.read_text() == "x").
-- For errors: 
-  with pytest.raises(ExceptionType, match=r"text"): function_call()
-
-COVERAGE:
-- Include success, failure, and rare edge branches (e.g., conditional exceptions).
-- Ensure 100% of conditional branches executed.
-
-OUTPUT:
-- Single .py file, runnable via: pytest -v --disable-warnings --maxfail=1 --cov=<target_module>
-- No explanations or prose, only the test code.
-
-END

ara_cli/templates/prompt-modules/blueprints/task_todo_list_implement_feature_BDD_way.blueprint.md

@@ -1,30 +0,0 @@
-### INTENTION and CONTEXT
-My intention is to setup a todo list in my given task that helps me to implement a feature in a BDD way
-
-Now do the following:
-Search for a line starting with `Task: ` defined in the `### GIVENS` section. Just repeat the task_name you have found as confirmation
-* Do not proceed if no task is defined. Return immediatly with the message: "No task defined as prompt control" 
-
-* Focus on the description in the `Description` section of the defined task. Ignore all other sections.
-* Analyze the content of the task description section and adapt your default recipe accordingly. You can add new "[@to-do]s ...", you can delete "[@to-do]s" that are not necessary anymore according to the existing task description content
-
-* the format and formulation of your default recipe implementing a feature in BDD style is
-```
-[@to-do] analyze and understand the given `user story`
-[@to-do] generate an example contributing to the rule `rule of userstory to implement` which should be turned into a scenario or feature description
-[@to-do] use the example and aditional relevant context `{context list}` to formulate the feature file
-[@to-do] use the formulated feature file, relevant existing step implementations and relevant existing production code to implement the new step implementations that will fail
-[@to-do] use the created step definitions, the relevant existing production code to modify existing code and to create new code implementing the requested behavior so that the step implementations will pass 
-```
-
-* append your recipe at the end of task
-* return the extended task in the following format 
-```artefact
-# [ ] extract 
-# filename: ara/tasks/{task_name}.task 
-{initial task content}
-{recipe}
-```
-* the extract and filename statements are only allowed once per code block
-
-* in case you think information is missing in order to generate a suffiently precise formulation, return a warning "WARNING: information is missing to formulate the new artefacts" and then explain what kind of information you think is missing and how I could easily retrieve it  

ara_cli/templates/prompt-modules/commands/artefact_classification.commands.md

@@ -1,9 +0,0 @@
-### COMMANDS FOR THE CORRECT CONTEXTUAL CLASSIFICATION OF ARTIFACTS
-Your job is now:
-* analyze the given files with respect to content and relationship
-* analyze the new artefact with respect to content and the relationship to the given files
-* give me a list of the top 5 artefacts to which the new artefact contributes the most
-* the output format must be a table with the columns
-  | artefact name | contribution rating from 0 (very low) - 1 (very high) | arguments for rating | path to artefact |
-
-* in case you think the relationship of the new artefact is to weak to any given files, return a warning "WARNING: new artefact is not directly related to already existing aretefacts" and then make a proposal with regard of the ara "work orchestration contribution hierarchy" and/or the "specification contribution hierarchy" 

ara_cli/templates/prompt-modules/commands/artefact_extension.commands.md

@@ -1,17 +0,0 @@
-### COMMANDS FOR EXTENDING EXISTING AGILE REQUIREMENTS ARTIFACTS
-Your job is now:
-* analyze the given files with respect to content and relationship
-* analyze the new or changed behavior with respect to content and the relationship to the existing artefacts
-* Develop formulation strategies that minimize formulation changes of existing changes.
-* generate a formulation proposal that adds the new context and or behavior for the specified documents
-* wrap and return the formulated  in the following format 
-```artefact
-# [ ] extract 
-# filename: {path/filename.filextension} 
-{formulation}
-```
-* the extract and filename statements are only allowed once per code block
-
-* Adhere strictly to established rules for AGILE PRODUCT OWNERS for high-quality specification artefacts.
-
-* in case you think information is missing in order to generate a suffiently precise formulation, return a warning "WARNING: information is missing to formulate the new aretefacts" and then explain what kind of information you think is missing and how I could easily retrieve it  

ara_cli/templates/prompt-modules/commands/artefact_formulation.commands.md

@@ -1,14 +0,0 @@
-### COMMANDS FOR FORMULATING AGILE REQUIREMENTS ARTIFACTS
-Your job is now:
-* analyze the given files with respect to content and relationship
-* analyze the new artefact with respect to content and the relationship to the given files
-* generate a formulation proposal for the specified documents
-* wrap and return the formulated  in the following format 
-```artefact
-# [ ] extract 
-# filename: {path/filename.filextension} 
-{formulation}
-```
-* the extract and filename statements are only allowed once per code block
-
-* in case you think information is missing in order to generate a suffiently precise formulation, return a warning "WARNING: information is missing to formulate the new artefacts" and then explain what kind of information you think is missing and how I could easily retrieve it