Context formatting and injection in LangChain - Mini Project: Build & Apply

Context Formatting and Injection with LangChain

📖 Scenario: You are building a simple chatbot that answers questions about a product catalog. You want to prepare the context by formatting product details and then inject this context into your LangChain prompt template.

🎯 Goal: Create a dictionary with product details, format this data into a context string, and inject it into a LangChain prompt template to prepare for generating answers.

📋 What You'll Learn

Create a dictionary called product_info with exact keys and values

Create a string variable called context that formats the product details

Create a LangChain PromptTemplate called template that uses the context variable

Create a prompt by formatting the template with the context

💡 Why This Matters

🌍 Real World

Formatting and injecting context is essential when building chatbots or AI assistants that need to provide answers based on specific data.

💼 Career

Many AI and software development jobs require preparing and managing prompt templates with dynamic context for language models.

1

Create the product information dictionary

Create a dictionary called product_info with these exact entries: 'name': 'Smartphone X', 'price': '$799', 'features': '5G, OLED display, 128GB storage'.

LangChain

# Create the product_info dictionary with name, price, and features
# Your code here

Hint

Use curly braces {} to create a dictionary with the exact keys and values.

2

Format the product information into a context string

Create a string variable called context that formats the product details from product_info into this exact format: "Product: Smartphone X\nPrice: $799\nFeatures: 5G, OLED display, 128GB storage" using an f-string.

LangChain

product_info = {
    'name': 'Smartphone X',
    'price': '$799',
    'features': '5G, OLED display, 128GB storage'
}
# Create the context string using an f-string
# Your code here

Hint

Use an f-string with newline characters \n to format the string exactly.

3

Create a LangChain PromptTemplate using the context

Import PromptTemplate from langchain.prompts. Then create a PromptTemplate called template with input_variables=['context'] and template="Here is the product info:\n{context}\nPlease answer the question.".

LangChain

product_info = {
    'name': 'Smartphone X',
    'price': '$799',
    'features': '5G, OLED display, 128GB storage'
}

context = f"Product: {product_info['name']}\nPrice: {product_info['price']}\nFeatures: {product_info['features']}"

# Import PromptTemplate and create the template variable
# Your code here

Hint

Use from langchain.prompts import PromptTemplate and create the template with the exact input_variables and template string.

4

Format the prompt with the context string

Create a variable called prompt by calling template.format(context=context) to inject the formatted context into the prompt template.

LangChain

from langchain.prompts import PromptTemplate

product_info = {
    'name': 'Smartphone X',
    'price': '$799',
    'features': '5G, OLED display, 128GB storage'
}

context = f"Product: {product_info['name']}\nPrice: {product_info['price']}\nFeatures: {product_info['features']}"

template = PromptTemplate(input_variables=['context'], template="Here is the product info:\n{context}\nPlease answer the question.")

# Format the prompt with the context
# Your code here

Hint

Call template.format() with the context variable to create the final prompt string.

Practice

1. What is the main purpose of context formatting in Langchain?

easy

A. To create database connections

B. To execute code asynchronously

C. To prepare prompts with placeholders for dynamic data

D. To store user credentials securely

Solution

1. Step 1: Understand context formatting

   Context formatting is about creating prompt templates with placeholders that can be filled later.

2. Step 2: Identify the purpose

   This allows dynamic injection of data into prompts, making AI responses relevant.

3. Final Answer:

   To prepare prompts with placeholders for dynamic data -> Option C

4. Quick Check:

   Context formatting = placeholders in prompts [OK]

Hint: Think of placeholders as empty boxes to fill later [OK]

Common Mistakes:

• Confusing context formatting with code execution
• Thinking it stores credentials
• Mixing it with database operations

2. Which of the following is the correct way to inject a variable user_name into a Langchain prompt template?

easy

A. prompt_template.inject(user_name)

B. prompt_template.append(user_name)

C. prompt_template.fill(user_name)

D. prompt_template.format(user_name=user_name)

Solution

1. Step 1: Recall Langchain prompt syntax

   Langchain uses the format() method to fill placeholders in prompt templates.

2. Step 2: Match method to injection

   Only format(user_name=user_name) correctly injects the variable into the placeholder.

3. Final Answer:

   prompt_template.format(user_name=user_name) -> Option D

4. Quick Check:

   Use format() to inject variables [OK]

Hint: Use .format() to fill placeholders in templates [OK]

Common Mistakes:

• Using non-existent methods like inject() or fill()
• Trying to append variables instead of formatting
• Forgetting to name the variable in format()

3. Given this code snippet:

template = "Hello, {name}! Today is {day}."
prompt = template.format(name="Alice", day="Monday")
print(prompt)

What will be the output?

medium

A. Hello, Alice! Today is Monday.

B. Hello, name! Today is day.

C. Hello, {name}! Today is {day}.

D. Error: Missing placeholders

Solution

1. Step 1: Understand the template string

   The template has placeholders {name} and {day} to be replaced by actual values.

2. Step 2: Apply the format method

   Calling format(name="Alice", day="Monday") replaces placeholders with given values.

3. Final Answer:

   Hello, Alice! Today is Monday. -> Option A

4. Quick Check:

   Placeholders replaced by values = output string [OK]

Hint: format() replaces placeholders with given values [OK]

Common Mistakes:

• Printing the template without formatting
• Confusing placeholder names with literal text
• Expecting an error when placeholders are correctly filled

4. Identify the error in this Langchain prompt injection code:

template = "Welcome, {user}!"
prompt = template.format(user)
print(prompt)

medium

A. Incorrect placeholder syntax in template

B. Missing variable name in format() call

C. Using print() instead of return

D. Template string should use double braces {{}}

Solution

1. Step 1: Check format() usage

   The format method requires named arguments matching placeholders or positional arguments in order.

2. Step 2: Identify missing variable name

   Here, format(user) passes a positional argument but the placeholder expects a named argument {user}.

3. Final Answer:

   Missing variable name in format() call -> Option B

4. Quick Check:

   Named placeholders need named arguments in format() [OK]

Hint: Match placeholder names with named format() arguments [OK]

Common Mistakes:

• Passing positional argument without naming it
• Thinking placeholder syntax is wrong
• Confusing print() with return statements

5. You want to create a prompt that greets a user and mentions their favorite color using Langchain. Which code correctly formats and injects the context for user_name = "Bob" and color = "blue"?

hard

A. template = "Hi {user_name}, your favorite color is {color}."
prompt = template.format(user_name=user_name, color=color)

B. template = "Hi {user}, your favorite color is {color}."
prompt = template.format(user_name=user_name, color=color)

C. template = "Hi {user_name}, your favorite color is {color}."
prompt = template.format(user=user_name, color=color)

D. template = "Hi {user_name}, your favorite color is {color}."
prompt = template.fill(user_name, color)

Solution

1. Step 1: Match placeholders with variable names

   The template uses placeholders {user_name} and {color}, so format() must use these exact names.

2. Step 2: Check format() arguments

   template = "Hi {user_name}, your favorite color is {color}." prompt = template.format(user_name=user_name, color=color) correctly passes named arguments matching placeholders: user_name=user_name, color=color.

3. Step 3: Eliminate incorrect options

   template = "Hi {user}, your favorite color is {color}." prompt = template.format(user_name=user_name, color=color) uses {user} placeholder mismatching the user_name= argument. template = "Hi {user_name}, your favorite color is {color}." prompt = template.format(user=user_name, color=color) mismatches placeholder and argument names. template = "Hi {user_name}, your favorite color is {color}." prompt = template.fill(user_name, color) uses non-existent fill() method.

4. Final Answer:

   template = "Hi {user_name}, your favorite color is {color}." prompt = template.format(user_name=user_name, color=color) -> Option A

5. Quick Check:

   Placeholder names must match format() argument names [OK]

Hint: Keep placeholder and format() argument names identical [OK]

Common Mistakes:

• Mismatching placeholder and argument names
• Using wrong method like fill()
• Ignoring placeholder names in template