⚡️ Speed up function get_first_top_level_function_or_method_ast by 39% in PR #769 (clean-async-branch)#781
Closed
codeflash-ai[bot] wants to merge 1 commit into
Conversation
The optimized code achieves a 38% speedup through several key micro-optimizations in AST traversal: **Primary optimizations:** 1. **Reduced tuple allocation overhead**: Moving `skip_types = (ast.FunctionDef, ast.AsyncFunctionDef, ast.ClassDef)` to a local variable eliminates repeated tuple construction on each function call (128 calls show 0.5% overhead vs previous inline tuple creation). 2. **Improved iterator efficiency**: Converting `ast.iter_child_nodes(node)` to `list(ast.iter_child_nodes(node))` upfront provides better cache locality and eliminates generator overhead during iteration, though this comes with a memory trade-off. 3. **Optimized control flow**: Restructuring the isinstance checks to handle the common case (finding matching object_type) first, then using early `continue` statements to skip unnecessary processing, reduces the total number of isinstance calls from ~14,000 to ~11,000. 4. **Eliminated walrus operator complexity**: Simplifying the class_node assignment in `get_first_top_level_function_or_method_ast` removes the complex conditional expression, making the code path more predictable. **Performance characteristics:** - The optimizations are most effective for **large-scale test cases** with many classes/functions (500+ nodes), where the reduced overhead per iteration compounds significantly - **Basic test cases** see modest improvements since the overhead reduction is less impactful on smaller AST trees - The memory trade-off of list conversion is worthwhile because AST child node lists are typically small and the improved iteration speed outweighs the memory cost The line profiler shows the optimized version spends more time in the initial list conversion (49.9% vs 46% in the original iterator), but this is offset by faster subsequent processing of the child nodes.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
⚡️ This pull request contains optimizations for PR #769
If you approve this dependent PR, these changes will be merged into the original PR branch
clean-async-branch.📄 39% (0.39x) speedup for
get_first_top_level_function_or_method_astincodeflash/code_utils/static_analysis.py⏱️ Runtime :
1.43 milliseconds→1.03 milliseconds(best of61runs)📝 Explanation and details
The optimized code achieves a 38% speedup through several key micro-optimizations in AST traversal:
Primary optimizations:
Reduced tuple allocation overhead: Moving
skip_types = (ast.FunctionDef, ast.AsyncFunctionDef, ast.ClassDef)to a local variable eliminates repeated tuple construction on each function call (128 calls show 0.5% overhead vs previous inline tuple creation).Improved iterator efficiency: Converting
ast.iter_child_nodes(node)tolist(ast.iter_child_nodes(node))upfront provides better cache locality and eliminates generator overhead during iteration, though this comes with a memory trade-off.Optimized control flow: Restructuring the isinstance checks to handle the common case (finding matching object_type) first, then using early
continuestatements to skip unnecessary processing, reduces the total number of isinstance calls from ~14,000 to ~11,000.Eliminated walrus operator complexity: Simplifying the class_node assignment in
get_first_top_level_function_or_method_astremoves the complex conditional expression, making the code path more predictable.Performance characteristics:
The line profiler shows the optimized version spends more time in the initial list conversion (49.9% vs 46% in the original iterator), but this is offset by faster subsequent processing of the child nodes.
✅ Correctness verification report:
🌀 Generated Regression Tests and Runtime
To edit these changes
git checkout codeflash/optimize-pr769-2025-09-27T03.15.10and push.