While writing a full-blown compiler for a programming language is a difficult and frustrating task, writing a smaller and more specific parser can be surprisingly easy if you know a small trick.
On the other hand, parsing problems pops up at several places in modern-day programming. So, learning this useful trick can be rewarding.
You need to know the basics of Python, more specifically, you should know the concepts of recursion and flow of control.
Objectives
After reading and understanding this post, you will be able to create simple calculators, interactive interpreters, parsers, very limited and small programming languages, etc. In general, you should be able to take input, tokenize it, perform whatever actions you want to on the tokens, and output the result of the process.At the end of this post, you will have created a simple, Lisp-like prefix calculator. Following is a demonstration of how it's going to look:> ( + 3 2 )= 5> ( / 2 0 )DivisionByZero> ( - -3 2 )= -5> -2= -2> ( + ( * 3 2 ) 5 )= 11
Step 1: Writing the Grammar
The first step to writing a parser is to write a clear grammar for its syntax. The grammar determines what is and what is not right. Once you have written the grammar, translating it to Python code is a trivial chore. The grammar will serve as our first pseudocode.For our tiny calculator, we know that the input can come in two forms: a Number (-2, .5, +8, 8.5, 9.) or a more complicated Expression begins with a (, followed by an operator, etc.).For writing a grammar, we need to identify different elements of the syntax. So far, we have Expression, Number, and Operator. The next important thing to do is to structure the elements (known as terms) into a hierarchical form. This is shown below:Expression:Number( Operator Expression Expression )Number:a floating-point number ([-+][0-9][*\.][0-9]*)Operators:
+
-
*
/
You will notice that Operator and Expression have no parent; they are independent terms.A grammar is read from the bottom up and different choices appear on distinct lines. Our grammar says that:
an Operator is one of +, -, *, /.
a Number is a floating-point number which matches the RegEx [-+][0-9]*[\.][0-9]*
an Expression is either a Number or a ( followed by an Operator, followed by two other Expressions, and finally ends in a ). Note that the definition of an Expression is recursive.
Step 2: Translating the Grammar into Pseudocode
Pseudocode is fake code resembling English which is supposed to be an intermediate code that can easily be converted into real code. Although writing pseudocode is optional, it is really helpful.The trick here is to put each term from our grammar into a separate function. Whenever we need to apply the grammar of a certain term, we only have to call the function. Following is the pseudocode implementing the grammar above:https://gist.github.com/HackerEarthBlog/f0a5a4304326936142da39b0d853f944This is our rough pseudo-code that should be good enough for our purpose. In the next step, we will write the real code.
3. Writing the Code
It is said very profoundly about Python that reading and writing Python feels like doing pseudocode. The same applies here, but there is one small caveat— Python doesn't provide any function for “unreading” or putting a character back in the input buffer.For this, I have created a small class which extends the file object to include this feature. To keep things simple, I have avoided inheritance and my class is not compatible with the file object provided by Python. Treat it like a black-box if you don't want to understand it.https://gist.github.com/HackerEarthBlog/6465f93e1ca155ded5e8b0c8294f16baHere is the buffer.py file which handles buffered input:https://gist.github.com/HackerEarthBlog/5330e5f11f96a22608b45affa61fa858
Explanation
expression():
expression() is our top-level function and maps the Expression grammar term. We first ignore all the whitespace. After that, it takes a single non-whitespace character as input and checks it against several possibilities.If the input string starts with +, -, ., or a digit, it is a number. We put the character back and input the entire number.If the input string starts with (, a complete expression is to follow. We input the operator, two more expressions which will serve as the operands, and finally the closing parenthesis. We then calculate the result and return it.
number():
The number function maps the Number grammar term and is very simple—just a wrapper around getword. We input a whole word and if it converts to a float, we return it, otherwise the function returns an error message.
operator():
The operator function inputs a single character and tests it for equality against several known operators. Like the above two functions, it also maps a grammar term, i.e., Operator. In case the given operator is not valid, an error message is returned.
calc():
The calc function is actually not necessary but makes the code substantially better. In an ideal program, each function should do only one logical task. calc removes some burden from expression.
UngetableInput
Although Python 3 supports buffered input through stdin.buffer, Python 2 has no such facility. Plus, Python 3's stdin.buffer would still require us to create some wrapper of our own.The UngetableInput class wraps Python's basic input to go through a buffer. We take input into the buffer and put a character back into the buffer when ungetc is called. Unless the buffer is empty, all input comes from the buffer.
Homework
This code works and leaves a lot of cleaning as homework for the reader. :) Following is a list of things you can do to improve and extend the rudimentary calculator:Improve buffer.py to handle input whitespace more accurately. Hint: You might want to use a string as the buffer.Implement a function to get a single character while skipping all whitespace and replace the whitespace skipping loop with it.Add the ability to create variables. Following the Lisp syntax, it should look something like the following:( define var_name 839457.892 )
What's Coming Next?
One of the most important parts of our program is the input buffer we created. Unfortunately, it's not general purpose and can break when used in something more complicated than our tiny calculator program. In the next article, we will examine a bigger module which does this chore better.
Subscribe to The HackerEarth Blog
Get expert tips, hacks, and how-tos from the world of tech recruiting to stay on top of your hiring!
Thank you for subscribing!
We're so pumped you're here! Welcome to the most amazing bunch that we are, the HackerEarth community. Happy reading!
I Used AI to Build a "Simple Image Carousel" at VibeCodeArena. It Found 15+ Issues and Taught Me How to Fix Them.
My Learning Journey
I wanted to understand what separates working code from good code. So I used VibeCodeArena.ai to pick a problem statement where different LLMs produce code for the same prompt. Upon landing on the main page of VibeCodeArena, I could see different challenges. Since I was interested in an Image carousal application, I picked the challenge with the prompt "Make a simple image carousel that lets users click 'next' and 'previous' buttons to cycle through images."
Within seconds, I had code from multiple LLMs, including DeepSeek, Mistral, GPT, and Llama. Each code sample also had an objective evaluation score. I was pleasantly surprised to see so many solutions for the same problem. I picked gpt-oss-20b model from OpenAI. For this experiment, I wanted to focus on learning how to code better so either one of the LLMs could have worked. But VibeCodeArena can also be used to evaluate different LLMs to help make a decision about which model to use for what problem statement.
The model had produced a clean HTML, CSS, and JavaScript. The code looked professional. I could see the preview of the code by clicking on the render icon. It worked perfectly in my browser. The carousel was smooth, and the images loaded beautifully.
But was it actually good code?
I had no idea. That's when I decided to look at the evaluation metrics
What I Thought Was "Good Code"
A working image carousel with:
Clean, semantic HTML
Smooth CSS transitions
Keyboard navigation support
ARIA labels for accessibility
Error handling for failed images
It looked like something a senior developer would write. But I had questions:
Was it secure? Was it optimized? Would it scale? Were there better ways to structure it?
Without objective evaluation, I had no answers. So, I proceeded to look at the detailed evaluation metrics for this code
What VibeCodeArena's Evaluation Showed
The platform's objective evaluation revealed issues I never would have spotted:
Security Vulnerabilities (The Scary Ones)
No Content Security Policy (CSP): My carousel was wide open to XSS attacks. Anyone could inject malicious scripts through the image URLs or manipulate the DOM. VibeCodeArena flagged this immediately and recommended implementing CSP headers.
Missing Input Validation: The platform pointed out that while the code handles image errors, it doesn't validate or sanitize the image sources. A malicious actor could potentially exploit this.
Hardcoded Configuration: Image URLs and settings were hardcoded directly in the code. The platform recommended using environment variables instead - a best practice I completely overlooked.
SQL Injection Vulnerability Patterns: Even though this carousel doesn't use a database, the platform flagged coding patterns that could lead to SQL injection in similar contexts. This kind of forward-thinking analysis helps prevent copy-paste security disasters.
Performance Problems (The Silent Killers)
DOM Structure Depth (15 levels): VibeCodeArena measured my DOM at 15 levels deep. I had no idea. This creates unnecessary rendering overhead that would get worse as the carousel scales.
Expensive DOM Queries: The JavaScript was repeatedly querying the DOM without caching results. Under load, this would create performance bottlenecks I'd never notice in local testing.
Missing Performance Optimizations: The platform provided a checklist of optimizations I didn't even know existed:
Each of these seems minor, but together they compound into a poor user experience.
Code Quality Issues (The Technical Debt)
High Nesting Depth (4 levels): My JavaScript had logic nested 4 levels deep. VibeCodeArena flagged this as a maintainability concern and suggested flattening the logic.
Overly Specific CSS Selectors (depth: 9): My CSS had selectors 9 levels deep, making it brittle and hard to refactor. I thought I was being thorough; I was actually creating maintenance nightmares.
Code Duplication (7.9%): The platform detected nearly 8% code duplication across files. That's technical debt accumulating from day one.
Moderate Maintainability Index (67.5): While not terrible, the platform showed there's significant room for improvement in code maintainability.
Missing Best Practices (The Professional Touches)
The platform also flagged missing elements that separate hobby projects from professional code:
No 'use strict' directive in JavaScript
Missing package.json for dependency management
No test files
Missing README documentation
No .gitignore or version control setup
Could use functional array methods for cleaner code
Missing CSS animations for enhanced UX
The "Aha" Moment
Here's what hit me: I had no framework for evaluating code quality beyond "does it work?"
The carousel functioned. It was accessible. It had error handling. But I couldn't tell you if it was secure, optimized, or maintainable.
VibeCodeArena gave me that framework. It didn't just point out problems, it taught me what production-ready code looks like.
My New Workflow: The Learning Loop
This is when I discovered the real power of the platform. Here's my process now:
Step 1: Generate Code Using VibeCodeArena
I start with a prompt and let the AI generate the initial solution. This gives me a working baseline.
Step 2: Analyze Across Several Metrics
I can get comprehensive analysis across:
Security vulnerabilities
Performance/Efficiency issues
Performance optimization opportunities
Code Quality improvements
This is where I learn. Each issue includes explanation of why it matters and how to fix it.
Step 3: Click "Challenge" and Improve
Here's the game-changer: I click the "Challenge" button and start fixing the issues based on the suggestions. This turns passive reading into active learning.
Do I implement CSP headers correctly? Does flattening the nested logic actually improve readability? What happens when I add dns-prefetch hints?
I can even use AI to help improve my code. For this action, I can use from a list of several available models that don't need to be the same one that generated the code. This helps me to explore which models are good at what kind of tasks.
For my experiment, I decided to work on two suggestions provided by VibeCodeArena by preloading critical CSS/JS resources with <link rel="preload"> for faster rendering in index.html and by adding explicit width and height attributes to images to prevent layout shift in index.html. The code editor gave me change summary before I submitted by code for evaluation.
Step 4: Submit for Evaluation
After making improvements, I submit my code for evaluation. Now I see:
What actually improved (and by how much)
What new issues I might have introduced
Where I still have room to grow
Step 5: Hey, I Can Beat AI
My changes helped improve the performance metric of this simple code from 82% to 83% - Yay! But this was just one small change. I now believe that by acting upon multiple suggestions, I can easily improve the quality of the code that I write versus just relying on prompts.
Each improvement can move me up the leaderboard. I'm not just learning in isolation—I'm seeing how my solutions compare to other developers and AI models.
So, this is the loop: Generate → Analyze → Challenge → Improve → Measure → Repeat.
Every iteration makes me better at both evaluating AI code and writing better prompts.
What This Means for Learning to Code with AI
This experience taught me three critical lessons:
1. Working ≠ Good Code
AI models are incredible at generating code that functions. But "it works" tells you nothing about security, performance, or maintainability.
The gap between "functional" and "production-ready" is where real learning happens. VibeCodeArena makes that gap visible and teachable.
2. Improvement Requires Measurement
I used to iterate on code blindly: "This seems better... I think?"
Now I know exactly what improved. When I flatten nested logic, I see the maintainability index go up. When I add CSP headers, I see security scores improve. When I optimize selectors, I see performance gains.
Measurement transforms vague improvement into concrete progress.
3. Competition Accelerates Learning
The leaderboard changed everything for me. I'm not just trying to write "good enough" code—I'm trying to climb past other developers and even beat the AI models.
This competitive element keeps me pushing to learn one more optimization, fix one more issue, implement one more best practice.
How the Platform Helps Me Become A Better Programmer
VibeCodeArena isn't just an evaluation tool—it's a structured learning environment. Here's what makes it effective:
Immediate Feedback: I see issues the moment I submit code, not weeks later in code review.
Contextual Education: Each issue comes with explanation and guidance. I learn why something matters, not just that it's wrong.
Iterative Improvement: The "Challenge" button transforms evaluation into action. I learn by doing, not just reading.
Measurable Progress: I can track my improvement over time—both in code quality scores and leaderboard position.
Comparative Learning: Seeing how my solutions stack up against others shows me what's possible and motivates me to reach higher.
What I've Learned So Far
Through this iterative process, I've gained practical knowledge I never would have developed just reading documentation:
How to implement Content Security Policy correctly
Why DOM depth matters for rendering performance
What CSS containment does and when to use it
How to structure code for better maintainability
Which performance optimizations actually make a difference
Each "Challenge" cycle teaches me something new. And because I'm measuring the impact, I know what actually works.
The Bottom Line
AI coding tools are incredible for generating starting points. But they don't produce high quality code and can't teach you what good code looks like or how to improve it.
✓ Objective analysis that shows you what's actually wrong ✓ Educational feedback that explains why it matters ✓ A "Challenge" system that turns learning into action ✓ Measurable improvement tracking so you know what works ✓ Competitive motivation through leaderboards
My "simple image carousel" taught me an important lesson: The real skill isn't generating code with AI. It's knowing how to evaluate it, improve it, and learn from the process.
The future of AI-assisted development isn't just about prompting better. It's about developing the judgment to make AI-generated code production-ready. That requires structured learning, objective feedback, and iterative improvement. And that's exactly what VibeCodeArena delivers.
Here is a link to the code for the image carousal I used for my learning journey
Revolutionizing Mobile Talent Hiring: The HackerEarth Advantage
The demand for mobile applications is exploding, but finding and verifying developers with proven, real-world skills is more difficult than ever. Traditional assessment methods often fall short, failing to replicate the complexities of modern mobile development.
Introducing a New Era in Mobile Assessment
At HackerEarth, we're closing this critical gap with two groundbreaking features, seamlessly integrated into our Full Stack IDE:
Now, assess mobile developers in their true native environment. Our enhanced Full Stack questions now offer full support for both Java and Kotlin, the core languages powering the Android ecosystem. This allows you to evaluate candidates on authentic, real-world app development skills, moving beyond theoretical knowledge to practical application.
Say goodbye to setup drama and tool-switching. Candidates can now build, test, and debug Android and React Native applications directly within the browser-based IDE. This seamless, in-browser experience provides a true-to-life evaluation, saving valuable time for both candidates and your hiring team.
Assess the Skills That Truly Matter
With native Android support, your assessments can now delve into a candidate's ability to write clean, efficient, and functional code in the languages professional developers use daily. Kotlin's rapid adoption makes proficiency in it a key indicator of a forward-thinking candidate ready for modern mobile development.
This chart illustrates the importance of assessing proficiency in both modern (Kotlin) and established (Java) codebases.
Streamlining Your Assessment Workflow
The integrated mobile emulator fundamentally transforms the assessment process. By eliminating the friction of fragmented toolchains and complex local setups, we enable a faster, more effective evaluation and a superior candidate experience.
Visualize the stark difference: Our streamlined workflow removes technical hurdles, allowing candidates to focus purely on demonstrating their coding and problem-solving abilities.
Quantifiable Impact on Hiring Success
A seamless and authentic assessment environment isn't just a convenience, it's a powerful catalyst for efficiency and better hiring outcomes. By removing technical barriers, candidates can focus entirely on demonstrating their skills, leading to faster submissions and higher-quality signals for your recruiters and hiring managers.
A Better Experience for Everyone
Our new features are meticulously designed to benefit the entire hiring ecosystem:
For Recruiters & Hiring Managers:
Accurately assess real-world development skills.
Gain deeper insights into candidate proficiency.
Hire with greater confidence and speed.
Reduce candidate drop-off from technical friction.
For Candidates:
Enjoy a seamless, efficient assessment experience.
No need to switch between different tools or manage complex setups.
Focus purely on showcasing skills, not environment configurations.
Work in a powerful, professional-grade IDE.
Unlock a New Era of Mobile Talent Assessment
Stop guessing and start hiring the best mobile developers with confidence. Explore how HackerEarth can transform your tech recruiting.
Vibe coding is a new method of using natural language prompts and AI tools to generate code. I have seen firsthand that this change makes software more accessible to everyone. In the past, being able to produce functional code was a strong advantage for developers. Today, when code is produced quickly through AI, the true value lies in designing, refining, and optimizing systems. Our role now goes beyond writing code; we must also ensure that our systems remain efficient and reliable.
From Machine Language to Natural Language
I recall the early days when every line of code was written manually. We progressed from machine language to high-level programming, and now we are beginning to interact with our tools using natural language. This development does not only increase speed but also changes how we approach problem solving. Product managers can now create working demos in hours instead of weeks, and founders have a clearer way of pitching their ideas with functional prototypes. It is important for us to rethink our role as developers and focus on architecture and system design rather than simply on typing c
The Promise and the Pitfalls
I have experienced both sides of vibe coding. In cases where the goal was to build a quick prototype or a simple internal tool, AI-generated code provided impressive results. Teams have been able to test new ideas and validate concepts much faster. However, when it comes to more complex systems that require careful planning and attention to detail, the output from AI can be problematic. I have seen situations where AI produces large volumes of code that become difficult to manage without significant human intervention.
AI-powered coding tools like GitHub Copilot and AWS’s Q Developer have demonstrated significant productivity gains. For instance, at the National Australia Bank, it’s reported that half of the production code is generated by Q Developer, allowing developers to focus on higher-level problem-solving . Similarly, platforms like Lovable or Hostinger Horizons enable non-coders to build viable tech businesses using natural language prompts, contributing to a shift where AI-generated code reduces the need for large engineering teams. However, there are challenges. AI-generated code can sometimes be verbose or lack the architectural discipline required for complex systems. While AI can rapidly produce prototypes or simple utilities, building large-scale systems still necessitates experienced engineers to refine and optimize the code.
The Economic Impact
The democratization of code generation is altering the economic landscape of software development. As AI tools become more prevalent, the value of average coding skills may diminish, potentially affecting salaries for entry-level positions. Conversely, developers who excel in system design, architecture, and optimization are likely to see increased demand and compensation. Seizing the Opportunity
Vibe coding is most beneficial in areas such as rapid prototyping and building simple applications or internal tools. It frees up valuable time that we can then invest in higher-level tasks such as system architecture, security, and user experience. When used in the right context, AI becomes a helpful partner that accelerates the development process without replacing the need for skilled engineers.
This is revolutionizing our craft, much like the shift from machine language to assembly to high-level languages did in the past. AI can churn out code at lightning speed, but remember, “Any fool can write code that a computer can understand. Good programmers write code that humans can understand.” Use AI for rapid prototyping, but it’s your expertise that transforms raw output into robust, scalable software. By honing our skills in design and architecture, we ensure our work remains impactful and enduring. Let’s continue to learn, adapt, and build software that stands the test of time.
Ready to streamline your recruitment process? Get a free demo to explore cutting-edge solutions and resources for your hiring needs.
Top Products
Explore HackerEarth’s top products for Hiring & Innovation
Discover powerful tools designed to streamline hiring, assess talent efficiently, and run seamless hackathons. Explore HackerEarth’s top products that help businesses innovate and grow.
