TLE Lens
See an estimated time complexity above every C++ function, right in VS Code.
⏱ O(n²) · estimated
int countPairs(vector<int>& a) { ... }
Click the lens to see why (which loops, which recursion, which library calls).
Add a demo GIF here: 
What it detects
| Pattern |
Result |
| No loops |
O(1) |
| Single loop, nested loops |
O(n), O(n²), O(n³) … |
i *= 2, n /= 10, binary-search style lo = mid + 1 |
O(log n) |
i * i <= n |
O(√n) |
Loops with constant bounds (i < 5) |
O(1) |
std::sort, stable_sort |
O(n log n) |
std::find, count, accumulate, reverse… |
O(n) |
lower_bound, upper_bound, binary_search |
O(log n) |
| Recursion: linear, halving, merge/quick sort shape, branching |
O(n), O(log n), O(n log n), O(2ⁿ) |
Limitations (it is a heuristic, not a proof)
- Loop bounds are assumed to depend on input size
n, unless they are literal constants.
- Calls to your other functions are not followed (no whole-program analysis).
- Member calls like
map.find() or set.insert() are not costed.
- Amortized patterns (two pointers, monotonic stack) may be over-estimated.
- Recursion with memoization is reported as
O(?).
- Quicksort-shaped recursion assumes balanced splits (average case).
- C++ only for now.
Develop
npm install # also copies the .wasm parser files into ./parsers
npm test # compiles and checks the analyzer against test-samples/
Then press F5 in VS Code to launch an Extension Development Host and open any .cpp file.
Add a test: drop a .cpp file into test-samples/ starting with // expect: O(...).
How it works
web-tree-sitter (WebAssembly) parses the file into a syntax tree. It tolerates half-typed code.src/analyzer.ts walks each function: loops multiply the cost of their body, sequential code takes the max, and self-calls are classified (how many per path, and whether the input is split or shrinks).src/extension.ts renders results as CodeLens and shows the explanation on click.
Publish
npm install -g @vscode/vsce
vsce login <your-publisher-id>
vsce package # creates big-o-lens-0.1.0.vsix (test it: Extensions → ... → Install from VSIX)
vsce publish
Create the publisher ID at https://marketplace.visualstudio.com/manage (needs a Microsoft account / Azure DevOps token).
Replace publisher, repository and the LICENSE name first.
Roadmap
- Python and Java support
- Follow calls between your own functions
- Space complexity estimate
- Optional "explain this" using an LLM
