Stop buffering huge ANSI styled lines - #450
Merged
Merged
Conversation
walles
force-pushed
the
fix/ansi-huge-line-limit
branch
from
July 29, 2026 08:23
af48cc9 to
384c26c
Compare
The test feeds single byte characters, so the flushed chunk is exactly maxBytesPerCell bytes per requested cell. Asserting that exact size says more than an upper bound does, and gives a clearer failure message. 🤖 Committed by Claude (Opus 5)
Nothing covered this: the whole suite, TestCellsLimitOnlyTruncates included, still passes if styledStringsFromString() truncates its input to the escape sequence scan window instead of buffering up to it. That returns zero cells for a line whose escapes outweigh its text. 🤖 Committed by Claude (Opus 5)
b.Loop() both resets the timer and runs the loop, so the separate b.ResetTimer() call goes away. 🤖 Committed by Claude (Opus 5)
It exercises StyledRunesFromString(), which lives in ansiTokenizer.go, and the package's other two benchmarks are both at the end of that file's test file. Also say where the 81 comes from. 🤖 Committed by Claude (Opus 5)
Two of the three call sites had to remember to reset the builder afterwards, and the third one must not care since the splitter is done by then. Doing it inside finalizeCurrentPart() removes that obligation from the callers. 🤖 Committed by Claude (Opus 5)
The escape sequence scan window and the part flushing threshold are the same number for the same reason: maxBytesPerCell bytes per cell is an upper bound, so that many bytes per requested cell covers everything that can reach the result. Deriving it twice meant explaining it twice. 🤖 Committed by Claude (Opus 5)
An escape sequence we can't parse gets replayed as plain runes, and that replay loop iterates the whole sequence without noticing that the cells limit has been reached. So a huge malformed sequence reaches the callback in full, 2341 calls for the five megabytes in this test. 🤖 Committed by Claude (Opus 5)
Reaching the cells limit used to mean pointing nextByteIndex past the end of the input, which stops nextChar() and with it run()'s main loop. But the loop replaying an unparseable escape sequence iterates that sequence directly, so nextByteIndex never came up and the replay ran to the end. A named flag says what it means and can be tested where it matters. On a five megabyte malformed sequence that takes us from 39.7 ms and 31.8 MB per operation to 15.9 ms and 5.3 MB. 🤖 Committed by Claude (Opus 5)
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
2 participants
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.
Follow-up to #358.
A leading ANSI sequence bypassed the plain-text shortcut, after which
styledStringSplitterbuffered the entire same-style run before invoking the cells-counting callback. Flush the run after the existing conservative bytes-per-cell window so the callback can stop parsing once the requested cells are filled.On Windows/amd64, the 5 MiB leading-SGR benchmark improved from 39.17 ms and 26.5 MB allocated per operation to 37–41 µs and 16.9 KB.
Testing
MOOR_SKIP_INTERACTIVE_TESTS=1 ./test.shMOOR_SKIP_INTERACTIVE_TESTS=1 GOARCH=386 ./test.shgo build -race ./...go test -race -timeout 60s ./internal/textstyles