ConversationHistory

Overview

The ConversationHistory class manages conversation context for the RAG engine by maintaining a rolling window of recent query-response pairs. It automatically prunes old entries when the history exceeds the configured limit.

Class Definition

from src.DSAAssistant.components.memory_buffer import ConversationHistory

history = ConversationHistory(max_history=5)

Constructor Parameters

max_history

int

default:"5"

Maximum number of query-response pairs to retain in history. When exceeded, the oldest entries are automatically removed (FIFO).

Initialization:

Creates an empty history list
Sets the maximum history size
No entries stored initially

Methods

add_query

Add a new query and response pair to the conversation history.

history.add_query(
    query="What is the Two Sum problem?",
    response="The Two Sum problem involves finding two numbers that add up to a target."
)

query

str

required

The user’s query or question

response

str

required

The system’s response to the query

Behavior:

Appends the query-response pair to history as a dictionary: {"query": query, "response": response}
If history length exceeds max_history, removes the oldest entry (index 0)
Maintains FIFO order (first in, first out)
No return value

Example:

history = ConversationHistory(max_history=3)

# Add first query
history.add_query(
    "Explain Two Sum",
    "Two Sum finds pairs that sum to target using hash map."
)

# Add second query
history.add_query(
    "What's the time complexity?",
    "O(n) time and O(n) space."
)

# Add third query
history.add_query(
    "Show me the code",
    "def twoSum(nums, target): ..."
)

print(len(history.history))  # Output: 3

# Add fourth query - oldest entry is removed
history.add_query(
    "Any edge cases?",
    "Handle empty arrays and no solution cases."
)

print(len(history.history))  # Output: 3 (still at max)

get_context

Generate a formatted context string from the conversation history.

context = history.get_context()
print(context)

context

str

Formatted string containing all query-response pairs in the history. Each pair is formatted as:

User: {query}
System: {response}

Returns empty string if history is empty.

Behavior:

Iterates through all entries in history order (oldest to newest)
Formats each entry with “User:” and “System:” prefixes
Joins entries with newlines
Strips trailing whitespace

Example:

history = ConversationHistory(max_history=3)

history.add_query(
    "What is Binary Search?",
    "Binary search is a O(log n) algorithm for sorted arrays."
)

history.add_query(
    "Show example",
    "def binary_search(arr, target): ..."
)

context = history.get_context()
print(context)

# Output:
# User: What is Binary Search?
# System: Binary search is a O(log n) algorithm for sorted arrays.
# User: Show example
# System: def binary_search(arr, target): ...

clear

Clear all entries from the conversation history.

history.clear()

Behavior:

Resets history to an empty list
Preserves max_history setting
No return value

Example:

history = ConversationHistory(max_history=3)

history.add_query("Query 1", "Response 1")
history.add_query("Query 2", "Response 2")

print(len(history.history))  # Output: 2

history.clear()

print(len(history.history))  # Output: 0
print(history.get_context())  # Output: "" (empty string)

Attributes

max_history

int

Maximum number of query-response pairs to retain

history

List[Dict[str, str]]

List of dictionaries, where each dictionary contains:

"query": User’s question (str)
"response": System’s answer (str)

Ordered from oldest (index 0) to newest (index -1)

Usage Examples

from src.DSAAssistant.components.memory_buffer import ConversationHistory

# Create history with limit of 3
history = ConversationHistory(max_history=3)

# Add conversation turns
history.add_query(
    "What is dynamic programming?",
    "Dynamic programming is an optimization technique using memoization."
)

history.add_query(
    "Give an example",
    "Fibonacci sequence: fib(n) = fib(n-1) + fib(n-2)"
)

# Get formatted context
context = history.get_context()
print(context)

# Clear when starting new topic
history.clear()

History Management Strategy

FIFO Behavior

The history uses a First-In-First-Out (FIFO) strategy:

history = ConversationHistory(max_history=2)

history.add_query("Q1", "R1")  # history: [Q1]
history.add_query("Q2", "R2")  # history: [Q1, Q2]
history.add_query("Q3", "R3")  # history: [Q2, Q3] (Q1 removed)
history.add_query("Q4", "R4")  # history: [Q3, Q4] (Q2 removed)

Choosing max_history

Small (1-3)

Best for:

Short-term context
Quick follow-ups
Limited memory

Tradeoff:

Less context
Faster

Medium (3-5)

Best for:

Multi-turn conversations
Balanced context
Default use case

Tradeoff:

Good balance
Recommended

Large (5-10)

Best for:

Complex discussions
Long context needed
Detailed analysis

Tradeoff:

More tokens
Slower

Integration with RAGEngine

The RAGEngine automatically manages conversation history:

# From rag_engine.py initialization:
self.conversation_history = ConversationHistory(max_history)

# In generate_enhanced_prompt():
history_context = self.conversation_history.get_context()
enhanced_prompt = (
    f"Conversation History:\n{history_context}\n\n"
    f"Query: {query}\n\n"
    f"Context: {context}\n\n"
    f"Instruction: {base_prompt}"
)

# In answer_question():
self.conversation_history.add_query(query, response)

Performance Characteristics

add_query

O(1) amortized

Constant time for appending. O(n) worst case when removing oldest entry (list.pop(0)).

get_context

O(n * m)

Linear in history length (n) and average string length (m). Typically very fast for small max_history.

clear

O(1)

Constant time list reassignment.

Memory Usage

# Estimate: ~1KB per entry (typical query + response)
max_history = 5
memory_estimate = max_history * 1  # ~5KB

# For large responses:
max_history = 10
avg_response_size = 2  # 2KB per response
memory_estimate = max_history * avg_response_size  # ~20KB

Testing Example

def test_conversation_history():
    # Test initialization
    history = ConversationHistory(max_history=3)
    assert len(history.history) == 0
    assert history.get_context() == ""
    
    # Test adding queries
    history.add_query("Q1", "R1")
    assert len(history.history) == 1
    
    history.add_query("Q2", "R2")
    history.add_query("Q3", "R3")
    assert len(history.history) == 3
    
    # Test max_history enforcement
    history.add_query("Q4", "R4")
    assert len(history.history) == 3
    assert history.history[0]["query"] == "Q2"  # Q1 removed
    
    # Test context formatting
    context = history.get_context()
    assert "User: Q2" in context
    assert "System: R2" in context
    assert "User: Q4" in context
    
    # Test clear
    history.clear()
    assert len(history.history) == 0
    assert history.get_context() == ""
    
    print("All tests passed!")

test_conversation_history()

Core Components

Web API

ConversationHistory

Overview

Class Definition

Constructor Parameters

Methods

add_query

get_context

clear

Attributes

Usage Examples

History Management Strategy

FIFO Behavior

Choosing max_history

Small (1-3)

Medium (3-5)

Large (5-10)

Integration with RAGEngine

Performance Characteristics

Memory Usage

Testing Example

See Also

Build docs developers (and LLMs) love

Core Components

Web API

​Overview

​Class Definition

​Constructor Parameters

​Methods

​add_query

​get_context

​clear

​Attributes

​Usage Examples

​History Management Strategy

​FIFO Behavior

​Choosing max_history

Small (1-3)

Medium (3-5)

Large (5-10)

​Integration with RAGEngine

​Performance Characteristics

​Memory Usage

​Testing Example

​See Also

Build docs developers (and LLMs) love

Overview

Class Definition

Constructor Parameters

Methods

add_query

get_context

clear

Attributes

Usage Examples

History Management Strategy

FIFO Behavior

Choosing max_history

Integration with RAGEngine

Performance Characteristics

Memory Usage

Testing Example

See Also