FineEnvs/repo2rlenv-swe-next
repo2rlenv-swe-next: Harbor dataset on Hugging Face with 100 tasks. Mine real repository history for supported code changes and associated tests, preserve regression behavior, and export the historical implementation change as the reference. Contains 100 Harbor tasks generated with the owned swe…
Tasks
- Header injection / HeaderWriteError when metadata values contain non- \n line boundaries
- Signer and Serializer should support a list of secret keys for key rotation
- Bug: Malformed Description-Content-Type values cause IndexError or are silently accepted
- Bug: strip comments=True leaves the last comment when a query contains only comments
- more itertools.pairwise missing from public API
- IntervalError message for weekday properties is not descriptive enough
- get real name() returns wrong component for fully-qualified names with 3+ parts
- iter index raises ValueError with negative start / stop for general iterables
- ALTER TABLE ... ROW FORMAT=Dynamic parses the table name incorrectly
- parse tag() raises ValueError instead of InvalidTag for tags with wrong number of components
- Bug: Wheel filename with \n in distribution name is incorrectly accepted as valid
- IndexedSet.update() adds iterables as members instead of their elements when called with multiple arguments
- JSONLIterator raises TypeError when using rel seek with a binary file stream
- SpecifierSet is missing is subset , is superset , and is disjoint methods
- pluralize() returns wrong plural for words ending in "x" (e.g. "boxs" instead of "boxes")
- every().hour.at('MM:SS') format not parsed correctly
- cache.invalidate() raises KeyError when the key is not in the cache
- value chain silently swallows TypeError raised during iteration
- random ordered range not accessible from the more itertools module
- args2sh and args2cmd ignore the sep argument
- Bug: Passing platforms= falls back to host platform tags instead of yielding nothing
- Bits with zero length serializes as '0' instead of ''
- reraise does not support callable as into argument
- Description-Content-Type with lowercase charset=utf-8 is incorrectly rejected
- packaging.tags is missing a pure python tags() function
- schedule.next run() does not support filtering by tag
- BETWEEN keyword misidentified as a Name when followed by a leading-dot float literal
- USTimeZone ignores datetime.fold during US DST fall-back transitions
- interleave evenly crashes with IndexError when given an empty list of iterables
- has path function missing from funcy.colls
- tail() silently returns empty result for negative n on sized iterables
- iequals incorrectly returns True when comparing iterables of different lengths with equality-wildcard objects
- complement set <= and = comparisons raise AttributeError
- Pylock.select() does not support prefer sdist predicate argument
- Bug: reversed() on an empty numeric range raises IndexError
- TimestampSigner.unsign does not raise SignatureExpired when the token was signed in the future
- Missing interpreter abi() function in packaging.tags
- log durations uses time.time instead of a patchable timer reference
- Bug: GUIDerator error message says 20 < size <= 36 but 20 is actually a valid size
- Title: relative time / decimal relative time raises TypeError when called with an aware datetime and no comparison time
- BarrelList slicing raises ValueError for negative stop/start and gives wrong results for reverse steps
- AtomicSaver and iter find files crash when given a pathlib.Path
- iter splitlines incorrectly splits text containing 28 or 29 , and doesn't recognize Unicode line separators
- BarrelList has three index-boundary bugs: out-of-bounds getitem , wrong insert position, and pop() crash on drained tail
- concurrent tee is missing from more itertools
- Bug: backoff raises ZeroDivisionError when factor=1.0 and count is not specified
- MATERIALIZED is not recognized as a keyword, breaks keyword case formatting
- Off-by-one in Bits length validation allows values that exceed the declared bit width
- Bug: Version string containing {field} is silently rewritten in InvalidMetadata error messages
- Bits. getitem always returns False for negative indices
- constrained batches silently accepts max count=0 and negative values
- Add subfactorial() function to more itertools
- JSONLIterator rejects rel seek=1.0 despite it being a documented valid value
- Title: JSONWebSignatureSerializer and TimedJSONWebSignatureSerializer should emit a DeprecationWarning on instantiation
- combination with replacement index gives wrong results and doesn't raise errors when None is involved
- In boltons/urlutils.py , the parse qsl function accepts an encoding keyword argument but never forwards it to the internal unquote calls…
- schedule.get jobs() is missing — no way to retrieve scheduled jobs by tag
- Bug: ZeroDivisionError when computing histogram for constant or near-constant data
- date signed is a raw integer instead of datetime on bad signature
- once per / once per args raises TypeError when arguments are unhashable (list, dict, set)
- str() on a scheduled functools.partial job raises AttributeError
- numeric range membership and len() are inconsistent with iteration for float steps
- Bug: Marker value serialization always uses double quotes, breaking round-trips for values containing "
- peekable is not subscriptable as a generic type
- split before / split after / split when return for an empty iterable when maxsplit=0
- Bug: Wheel filename derived from URL is not percent-decoded, causing invalid filenames for local versions
- autocurry does not preserve function metadata (e.g., docstrings) when used as a decorator
- Bug: Requirement leaks InvalidSpecifier instead of raising InvalidRequirement for malformed specifier lists
- Bug: Marker and Requirement parsers silently accept a trailing LF ( \n )
- sliced() silently returns wrong results for negative n
- InvalidRequirement leaks raw from dependency group parsing, dropping sibling errors
- running min and running max are not stable when values are equal
- BarrelList raises IndexError when deleting a slice that spans multiple barrels
- Bug: ~= operator accepts non-ASCII Unicode characters in pre/post-release labels
- Function object has no get window() method and get parameters() breaks with OVER clause
- Bug: parse tag and parse wheel filename silently accept invalid interpreter tags
- ecoutils.get profile(scrub=True) still performs hostname resolution and reads the working directory
- wraps(g)(f) raises TypeError when f has keyword-only parameters that g accepts positionally
- serialize does not support generator methods ( send , throw , close )
- InvalidMetadata and CyclicDependencyGroup cannot be pickled/unpickled
- numeric range equality doesn't match built-in range semantics for single-element ranges
- Bug: "X" in extras and "X" in dependency groups markers are not normalized at parse time
- Wrong next run date when using .at() with a timezone
- OrderedMultiDict.addlist() silently drops values from one-shot iterables
- singularize() corrupts words ending in double s ( glass , boss , address , etc.)
- Title: TimestampSigner.timestamp to datetime() returns a naive datetime instead of a timezone-aware UTC datetime
- Function grouping skipped for lowercase as in CREATE TABLE ... AS SELECT
- Metadata.from email drops from raw validation errors when unparsed headers are present
- Bug: partition and chunks give wrong results when step n and input is an iterator
- Bug: LowerBound(None, True) and UpperBound(None, True) don't normalize inclusive to False
- repr() crashes on partially-constructed Job when job func is None
- memoize silently drops key func when called directly (non-decorator form)
- Long-running jobs skip their next scheduled period
- Ordered identifiers inside parentheses are not grouped correctly
- chunked() raises a confusing ValueError for negative n
- When a signed token contains a timestamp value so large that it would correspond to a year far beyond what Python's datetime supports…
- rpartial does not accept keyword arguments
- Bug: Pylock.select() silently succeeds when environments is an empty list
- namedtuple and namedlist raise IndexError instead of ValueError for empty names
- Pylock.select(tags= ) ignores the empty tag list and falls back to host tags