test: expand prompt_builder unit tests for log_context, multi-turn hi… (#204)

* test: expand prompt_builder unit tests for log_context, multi-turn history, and edge cases

* test: add tests for handling whitespace-only log context in prompt builder

---------

Co-authored-by: Bervianto Leo Pratama <[email protected]>

Author: Yugansh5013

chatbot-core/tests/unit/prompts/test_prompt_builder.py

@@ -2,6 +2,7 @@
 
 from langchain.memory import ConversationBufferMemory
 from api.prompts.prompt_builder import build_prompt, SYSTEM_INSTRUCTION
+from api.prompts.prompts import LOG_ANALYSIS_INSTRUCTION
 
 
 def test_build_prompt_with_full_history_and_context():
@@ -76,6 +77,182 @@ def test_build_prompt_with_none_memory():
     assert context in context_section
     assert user_query in question_section
 
+def test_build_prompt_with_log_context_uses_log_analysis_instruction():
+    """Test that providing log_context switches to LOG_ANALYSIS_INSTRUCTION
+    and includes the 'User-Provided Log Data' section in the prompt."""
+    memory = ConversationBufferMemory(return_messages=True)
+    context = "Jenkins pipeline documentation."
+    user_query = "Why did my build fail?"
+    log_context = "ERROR: Build step 'Execute shell' marked build as failure"
+
+    prompt = build_prompt(user_query, context, memory, log_context=log_context)
+
+    # Should use LOG_ANALYSIS_INSTRUCTION, NOT SYSTEM_INSTRUCTION
+    assert LOG_ANALYSIS_INSTRUCTION.strip() in prompt
+    assert SYSTEM_INSTRUCTION.strip() not in prompt
+
+    # Log section must be present with the actual log data
+    assert "User-Provided Log Data:" in prompt
+    assert log_context in prompt
+
+    # Standard structural sections still present and in order
+    chat_idx, context_idx, question_idx, answer_idx = get_prompt_indexes(prompt)
+    assert chat_idx < context_idx < question_idx < answer_idx
+
+    # User query still appears correctly
+    _, _, question_section = get_prompt_sections(prompt)
+    assert user_query.strip() in question_section
+
+
+def test_build_prompt_with_log_context_and_history():
+    """Test the full combination: log_context + conversation history + context."""
+    memory = ConversationBufferMemory(return_messages=True)
+    memory.chat_memory.add_user_message("My build failed.") # pylint: disable=no-member
+    memory.chat_memory.add_ai_message("Can you share the logs?") # pylint: disable=no-member
+    context = "Check Jenkins console output for errors."
+    user_query = "Here are my logs, can you analyze?"
+    log_context = "java.lang.OutOfMemoryError: Java heap space"
+
+    prompt = build_prompt(user_query, context, memory, log_context=log_context)
+
+    # Uses log analysis instruction
+    assert LOG_ANALYSIS_INSTRUCTION.strip() in prompt
+    assert SYSTEM_INSTRUCTION.strip() not in prompt
+
+    # History is preserved
+    history_section, context_section, question_section = get_prompt_sections(prompt)
+    assert "User: My build failed." in history_section
+    assert "Jenkins Assistant: Can you share the logs?" in history_section
+
+    # Context and log data are both present
+    assert context in context_section
+    assert "User-Provided Log Data:" in prompt
+    assert log_context in prompt
+
+    # Question appears correctly
+    assert user_query.strip() in question_section
+
+
+def test_build_prompt_with_empty_string_log_context_uses_system_instruction():
+    """Test that an empty string log_context (falsy) does NOT trigger the
+    log analysis branch — should fall back to SYSTEM_INSTRUCTION."""
+    memory = ConversationBufferMemory(return_messages=True)
+    context = "Some context."
+    user_query = "How do I configure agents?"
+
+    prompt = build_prompt(user_query, context, memory, log_context="")
+
+    # Should use standard instruction since "" is falsy
+    assert SYSTEM_INSTRUCTION.strip() in prompt
+    assert LOG_ANALYSIS_INSTRUCTION.strip() not in prompt
+    assert "User-Provided Log Data:" not in prompt
+
+
+def test_build_prompt_with_none_log_context_uses_system_instruction():
+    """Test that log_context=None (the default) does NOT trigger the
+    log analysis branch."""
+    memory = ConversationBufferMemory(return_messages=True)
+    context = "Some context."
+    user_query = "How do I set up a pipeline?"
+
+    prompt = build_prompt(user_query, context, memory, log_context=None)
+
+    assert SYSTEM_INSTRUCTION.strip() in prompt
+    assert LOG_ANALYSIS_INSTRUCTION.strip() not in prompt
+    assert "User-Provided Log Data:" not in prompt
+
+def test_build_prompt_with_multiple_conversation_turns():
+    """Test that multiple rounds of user/assistant messages are all
+    captured in the Chat History section in the correct order."""
+    memory = ConversationBufferMemory(return_messages=True)
+    memory.chat_memory.add_user_message("What is a Jenkinsfile?") # pylint: disable=no-member
+    memory.chat_memory.add_ai_message("A Jenkinsfile defines your pipeline.") # pylint: disable=no-member
+    memory.chat_memory.add_user_message("Can I use it with GitHub?") # pylint: disable=no-member
+    memory.chat_memory.add_ai_message("Yes, you can integrate it with GitHub.") # pylint: disable=no-member
+    context = "Jenkinsfile supports declarative and scripted syntax."
+    user_query = "Show me an example."
+
+    prompt = build_prompt(user_query, context, memory)
+
+    history_section, _, _ = get_prompt_sections(prompt)
+
+    # All four messages must appear in history
+    assert "User: What is a Jenkinsfile?" in history_section
+    assert "Jenkins Assistant: A Jenkinsfile defines your pipeline." in history_section
+    assert "User: Can I use it with GitHub?" in history_section
+    assert "Jenkins Assistant: Yes, you can integrate it with GitHub." in history_section
+
+    # Order: first user message appears before second user message
+    first_user_pos = history_section.index("User: What is a Jenkinsfile?")
+    second_user_pos = history_section.index("User: Can I use it with GitHub?")
+    assert first_user_pos < second_user_pos
+
+
+def test_build_prompt_with_none_content_message():
+    """Test that a message with None content is handled gracefully
+    (the `msg.content or ''` guard in the source)."""
+    memory = ConversationBufferMemory(return_messages=True)
+    memory.chat_memory.add_user_message("Hello") # pylint: disable=no-member
+    # Simulate a None content message by directly manipulating memory
+    memory.chat_memory.messages[-1].content = None  # pylint: disable=no-member
+    context = "Jenkins docs."
+    user_query = "Test query"
+
+    prompt = build_prompt(user_query, context, memory)
+
+    history_section, _, _ = get_prompt_sections(prompt)
+    # Should show "User: " with empty content, not crash
+    assert "User: " in history_section
+    assert "Hello" not in history_section
+
+
+def test_build_prompt_with_special_characters_in_query():
+    """Test that special characters (unicode, newlines) in the user query
+    are preserved correctly in the output prompt."""
+    memory = ConversationBufferMemory(return_messages=True)
+    context = "Jenkins configuration docs."
+    user_query = "  How do I use the 'pipeline' with \"quotes\" & <angle> brackets?\n  "
+
+    prompt = build_prompt(user_query, context, memory)
+
+    _, _, question_section = get_prompt_sections(prompt)
+    assert user_query.strip() in question_section
+
+
+def test_build_prompt_log_data_section_appears_between_context_and_question():
+    """Test that the 'User-Provided Log Data' section is placed
+    after the context section and before the user question."""
+    memory = ConversationBufferMemory(return_messages=True)
+    context = "Pipeline troubleshooting guide."
+    user_query = "Analyze this error."
+    log_context = "FATAL: command execution failed"
+
+    prompt = build_prompt(user_query, context, memory, log_context=log_context)
+
+    context_idx = prompt.index("Context (Documentation & Knowledge Base):")
+    log_data_idx = prompt.index("User-Provided Log Data:")
+    question_idx = prompt.index("User Question:")
+
+    # Log data must appear AFTER context and BEFORE question
+    assert context_idx < log_data_idx < question_idx
+
+def test_build_prompt_with_whitespace_only_log_context_triggers_log_branch():
+    """Test that a whitespace-only log_context is truthy and triggers the
+    log analysis branch. This is the current intended behavior — the
+    log_context check uses `if log_context:` which treats non-empty
+    whitespace strings as truthy. If this behavior should change,
+    the source should be updated to use `if log_context and log_context.strip():`."""
+    memory = ConversationBufferMemory(return_messages=True)
+    context = "Some context."
+    user_query = "Why did my build fail?"
+
+    prompt = build_prompt(user_query, context, memory, log_context="   ")
+
+    # Whitespace-only string is truthy, so log analysis branch is triggered
+    assert LOG_ANALYSIS_INSTRUCTION.strip() in prompt
+    assert SYSTEM_INSTRUCTION.strip() not in prompt
+    assert "User-Provided Log Data:" in prompt
+
 def get_prompt_indexes(prompt: str) -> tuple[int, int, int, int]:
     """Helper to extract section positions in the prompt."""
     chat_idx = prompt.index("Chat History:")