⚡ Make deep-copying of blocks and fields cheap - #613
Closed
MiWeiss wants to merge 1 commit into
Closed
Conversation
`BlockMiddleware.transform_block` deep-copies every block unless `allow_inplace_modification` is set, which is the default for the unparse stack. With the generic `copy.deepcopy` this dominated `write_string`/`write_file` (~85% of a profile on a real-world .bib). Implement `__deepcopy__` for `Block` and `Field` via a shared, attribute-aware helper: immutable primitives are shared, plain lists and dicts and nested blocks/fields are copied directly, anything else is delegated to `copy.deepcopy`. Semantics are unchanged (independent copies, memo honoured, cycles and shared references preserved, subclasses and custom `__deepcopy__` overrides supported). Per-entry deepcopy drops from ~34us to ~11us; `write_string` on a 41 MB / 91 MB .bib from 5.7s / 9.4s to 3.6s / 6.0s. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Problem:
BlockMiddleware.transform_blockdeep-copies every block unlessallow_inplace_modificationis set (the default for the unparse stack). With the genericcopy.deepcopy(~140 calls per entry) this is ~85% of awrite_stringprofile on a real-world .bib.Fix:
BlockandFieldget a__deepcopy__backed by one attribute-aware helper: immutable primitives are shared, plain lists/dicts and nested blocks/fields are copied directly, everything else goes throughcopy.deepcopy. Semantics unchanged: independent copies,memohonoured (shared references and cycles preserved), subclasses and custom__deepcopy__overrides supported. No middleware change.Benchmark (
write_string(lib), min of 3, single core;copy.deepcopy(entry)per entry):Tests: new deepcopy tests for every block type, nested values, memo sharing, cycles, subclasses, and a write round-trip; 2751 passed, 12 skipped.
🤖 Generated with Claude Code