dirstate: fix leak of entry object in dirstate_item_from_v1_data()

dirstate: fix parse_dirstate() to error out if NULL entry created Since 83f0e93ec34b "dirstate-item: move the C implementation to the same logic", dirstate_item_from_v1_data() gets more likely to return NULL, and the fuzzer crashes because of that.

rust: Remove EntryState::Unknown This enum variant represented the `state == '?'` case, which was used to represent the absence of a dirstate entry/item (and therefore of that entry’s state). Now that previous refactors have removed this use in the Python/Rust FFI APIs, the remaining uses can be removed by replacing `EntryState` by `Option<EntryState>` where appropriate, using `None` to represent the absence of an entry. Differential Revision: https://phab.mercurial-scm.org/D11465

dirstate: Remove the `state == ' '` special case Previously this was used to create a `DirstateItem` representing dirstate tree nodes that semantically don’t have an associated `DirtateItem`. This isn’t used anymore now that `dirstatemap.debug_iter` yields plain tuples. Differential Revision: https://phab.mercurial-scm.org/D11464

debugsate: Change debug_iter() to yield tuples instead of DirstateItem This removes the need for `DirstateItem` to support the `state == ' '` special case which represents dirstate tree nodes without an item. Differential Revision: https://phab.mercurial-scm.org/D11463

debugstate: Always call dirstatemap.debug_iter() … passing it a new `all` argument for the `--all` CLI option, instead of conditionally calling `.debug_iter()` or `.items()` This prepares for the next commit. Differential Revision: https://phab.mercurial-scm.org/D11462

rust: Make the fields of DirstateEntry private This is a first step toward making its internal structure equivalent to Python’s DirstateItem. Differential Revision: https://phab.mercurial-scm.org/D11461

rust: Remove support for passing a dict to the Rust pathutil.dirs() That is only used by the Python dirstatemap, which not used when Rust is enabled. This allows removing the private `extract_dirstate` function which creates `DirstateEntry` values. This in turn will make easier upcoming changes to `DirstateEntry`. Differential Revision: https://phab.mercurial-scm.org/D11460

rust: Remove the `rustext.parsers` module It only exported Rust implementations of the parse_dirstate and pack_dirtate functions, which are only used (anymore) when Rust is not enabled. fakedirstatewritetime.py was detecting the presence of `rustext.parsers` but what it really wants to know is whether the Rust implementation of `dirstatemap` is used. This changes it to detect `rustext.dirstate` instead. Differential Revision: https://phab.mercurial-scm.org/D11459

dirstate: remove now-unused .directories() method It calls _rustmap.directories() which was already removed Differential Revision: https://phab.mercurial-scm.org/D11458

rust: Move DirstateEntry to its own module … and RawEntry to the dirstate::parsers module, the only one that uses it. Differential Revision: https://phab.mercurial-scm.org/D11457

patchbomb: fix traceback on outgoing when not specifying revisions A user reported that `hg email -n -o` raised with the following traceback: ``` ** unknown exception encountered, please report by visiting ** https://mercurial-scm.org/wiki/BugTracker ** Python 3.8.12 (default, Sep 13 2021, 22:12:54) [GCC 7.5.0] ** Mercurial Distributed SCM (version 5.9.1) ** Extensions loaded: patchbomb Traceback (most recent call last): File "/usr/pkg/bin/hg", line 61, in <module> dispatch.run() File "/usr/pkg/lib/python3.8/site-packages/mercurial/dispatch.py", line 144, in run status = dispatch(req) File "/usr/pkg/lib/python3.8/site-packages/mercurial/dispatch.py", line 250, in dispatch status = _rundispatch(req) File "/usr/pkg/lib/python3.8/site-packages/mercurial/dispatch.py", line 294, in _rundispatch ret = _runcatch(req) or 0 File "/usr/pkg/lib/python3.8/site-packages/mercurial/dispatch.py", line 470, in _runcatch return _callcatch(ui, _runcatchfunc) File "/usr/pkg/lib/python3.8/site-packages/mercurial/dispatch.py", line 480, in _callcatch return scmutil.callcatch(ui, func) File "/usr/pkg/lib/python3.8/site-packages/mercurial/scmutil.py", line 153, in callcatch return func() File "/usr/pkg/lib/python3.8/site-packages/mercurial/dispatch.py", line 460, in _runcatchfunc return _dispatch(req) File "/usr/pkg/lib/python3.8/site-packages/mercurial/dispatch.py", line 1273, in _dispatch return runcommand( File "/usr/pkg/lib/python3.8/site-packages/mercurial/dispatch.py", line 918, in runcommand ret = _runcommand(ui, options, cmd, d) File "/usr/pkg/lib/python3.8/site-packages/mercurial/dispatch.py", line 1285, in _runcommand return cmdfunc() File "/usr/pkg/lib/python3.8/site-packages/mercurial/dispatch.py", line 1271, in <lambda> d = lambda: util.checksignature(func)(ui, *args, **strcmdopt) File "/usr/pkg/lib/python3.8/site-packages/mercurial/util.py", line 1886, in check return func(*args, **kwargs) File "/usr/pkg/lib/python3.8/site-packages/hgext/patchbomb.py", line 817, in email revs = _getoutgoing(repo, dest, revs) File "/usr/pkg/lib/python3.8/site-packages/hgext/patchbomb.py", line 537, in _getoutgoing safe_paths = [urlutil.hidepassword(p.rawloc) for p in paths] File "/usr/pkg/lib/python3.8/site-packages/hgext/patchbomb.py", line 537, in <listcomp> safe_paths = [urlutil.hidepassword(p.rawloc) for p in paths] File "/usr/pkg/lib/python3.8/site-packages/mercurial/utils/urlutil.py", line 501, in get_push_paths msg %= dest TypeError: %b requires a bytes-like object, or an object that implements __bytes__, not 'NoneType' ``` This is due to patchbomb passing `[None]` in the `dests` parameter of the multi-paths outgoing function, we fix this in this change and add a non-regression test. Differential Revision: https://phab.mercurial-scm.org/D11456

extensions: prevent a crash on py3 with a `minimumhgversion` str value The expectation is that this field is bytes, but unported extensions are a thing and it shouldn't explode on a bad value. We already do this transformation in the version reporting mechanism. Differential Revision: https://phab.mercurial-scm.org/D11476

extensions: prevent a crash on py3 when testing a bad extension minimum A `None` placeholder is populated for each missing component by `util.versiontuple()`, which could safely be used with `>` on py2, but not py3. I guess there's another hole here where if the string is entirely bogus (i.e no numbers), it will be treated as 0.0, and always load. But that's always been the case. Differential Revision: https://phab.mercurial-scm.org/D11475

typing: drop annotations on `mercurial/logcmdutil.walkopts` attributes Pytype 2021-09-09 must have learned some things about `@attr`, because while these were presumably required to help it before (c1d0f83d62c4::9a26fea2b518 and c11099cc1de4::224c786f4fce), these instances now get flagged as an assignment mismatch: File "/mnt/c/Users/Matt/hg/mercurial/logcmdutil.py", line 717, in walkopts: Type annotation for pats does not match type of assignment [annotation-type-mismatch] Annotation: List[bytes] Assignment: mercurial.thirdparty.attr._make._CountingAttr File "/mnt/c/Users/Matt/hg/mercurial/logcmdutil.py", line 718, in walkopts: Type annotation for opts does not match type of assignment [annotation-type-mismatch] Annotation: Dict[bytes, Any] Assignment: mercurial.thirdparty.attr._make._CountingAttr File "/mnt/c/Users/Matt/hg/mercurial/logcmdutil.py", line 722, in walkopts: Type annotation for revspec does not match type of assignment [annotation-type-mismatch] Annotation: List[bytes] Assignment: mercurial.thirdparty.attr._make._CountingAttr File "/mnt/c/Users/Matt/hg/mercurial/logcmdutil.py", line 725, in walkopts: Type annotation for bookmarks does not match type of assignment [annotation-type-mismatch] Annotation: List[bytes] Assignment: mercurial.thirdparty.attr._make._CountingAttr File "/mnt/c/Users/Matt/hg/mercurial/logcmdutil.py", line 726, in walkopts: Type annotation for branches does not match type of assignment [annotation-type-mismatch] Annotation: List[bytes] Assignment: mercurial.thirdparty.attr._make._CountingAttr File "/mnt/c/Users/Matt/hg/mercurial/logcmdutil.py", line 727, in walkopts: Type annotation for date does not match type of assignment [annotation-type-mismatch] Annotation: Optional[bytes] Assignment: mercurial.thirdparty.attr._make._CountingAttr File "/mnt/c/Users/Matt/hg/mercurial/logcmdutil.py", line 728, in walkopts: Type annotation for keywords does not match type of assignment [annotation-type-mismatch] Annotation: List[bytes] Assignment: mercurial.thirdparty.attr._make._CountingAttr File "/mnt/c/Users/Matt/hg/mercurial/logcmdutil.py", line 729, in walkopts: Type annotation for no_merges does not match type of assignment [annotation-type-mismatch] Annotation: bool Assignment: mercurial.thirdparty.attr._make._CountingAttr File "/mnt/c/Users/Matt/hg/mercurial/logcmdutil.py", line 730, in walkopts: Type annotation for only_merges does not match type of assignment [annotation-type-mismatch] Annotation: bool Assignment: mercurial.thirdparty.attr._make._CountingAttr File "/mnt/c/Users/Matt/hg/mercurial/logcmdutil.py", line 731, in walkopts: Type annotation for prune_ancestors does not match type of assignment [annotation-type-mismatch] Annotation: List[bytes] Assignment: mercurial.thirdparty.attr._make._CountingAttr File "/mnt/c/Users/Matt/hg/mercurial/logcmdutil.py", line 732, in walkopts: Type annotation for users does not match type of assignment [annotation-type-mismatch] Annotation: List[bytes] Assignment: mercurial.thirdparty.attr._make._CountingAttr File "/mnt/c/Users/Matt/hg/mercurial/logcmdutil.py", line 735, in walkopts: Type annotation for include_pats does not match type of assignment [annotation-type-mismatch] Annotation: List[bytes] Assignment: mercurial.thirdparty.attr._make._CountingAttr File "/mnt/c/Users/Matt/hg/mercurial/logcmdutil.py", line 736, in walkopts: Type annotation for exclude_pats does not match type of assignment [annotation-type-mismatch] Annotation: List[bytes] Assignment: mercurial.thirdparty.attr._make._CountingAttr File "/mnt/c/Users/Matt/hg/mercurial/logcmdutil.py", line 739, in walkopts: Type annotation for follow does not match type of assignment [annotation-type-mismatch] Annotation: int Assignment: mercurial.thirdparty.attr._make._CountingAttr File "/mnt/c/Users/Matt/hg/mercurial/logcmdutil.py", line 743, in walkopts: Type annotation for force_changelog_traversal does not match type of assignment [annotation-type-mismatch] Annotation: bool Assignment: mercurial.thirdparty.attr._make._CountingAttr File "/mnt/c/Users/Matt/hg/mercurial/logcmdutil.py", line 747, in walkopts: Type annotation for filter_revisions_by_pats does not match type of assignment [annotation-type-mismatch] Annotation: bool Assignment: mercurial.thirdparty.attr._make._CountingAttr File "/mnt/c/Users/Matt/hg/mercurial/logcmdutil.py", line 750, in walkopts: Type annotation for sort_revisions does not match type of assignment [annotation-type-mismatch] Annotation: Optional[bytes] Assignment: mercurial.thirdparty.attr._make._CountingAttr File "/mnt/c/Users/Matt/hg/mercurial/logcmdutil.py", line 753, in walkopts: Type annotation for limit does not match type of assignment [annotation-type-mismatch] Annotation: Optional[int] Assignment: mercurial.thirdparty.attr._make._CountingAttr Differential Revision: https://phab.mercurial-scm.org/D11474

util: eliminate the possibility of returning None from `versiontuple()` This fixes the following error flagged by pytype: File "/mnt/c/Users/Matt/hg/mercurial/extensions.py", line 228, in load: unsupported operand types for > [unsupported-operands] Primitive types 'Optional[tuple]' and 'curver: Optional[tuple]' are not comparable. Differential Revision: https://phab.mercurial-scm.org/D11473

archival: force a `CompressionError` to bytes before passing to `error.Abort` I'm not sure what changed before pytype 09-09-2021 (from 04-15-2021), but this started getting flagged. This fixes: File "/mnt/c/Users/Matt/hg/mercurial/archival.py", line 199, in taropen: Function bytestr.__init__ was called with the wrong arguments [wrong-arg-types] Expected: (self, ints: Iterable[int]) Actually passed: (self, ints: tarfile.CompressionError) Attributes of protocol Iterable[int] are not implemented on tarfile.CompressionError: __iter__ Differential Revision: https://phab.mercurial-scm.org/D11472

parser: force a `ValueError` to bytes before passing to `error.ParseError` I'm not sure what changed before pytype 09-09-2021 (from 04-15-2021), but this started getting flagged. I think there's a pytype bug here, because I don't see how `.lower()` can be getting called on a `ValueError` after it is forced to a byte string. That's suppressed for now to make progress. This fixes: File "/mnt/c/Users/Matt/hg/mercurial/parser.py", line 219, in unescapestr: Function bytestr.__init__ was called with the wrong arguments [wrong-arg-types] Expected: (self, ints: Iterable[int]) Actually passed: (self, ints: ValueError) Attributes of protocol Iterable[int] are not implemented on ValueError: __iter__ File "/mnt/c/Users/Matt/hg/mercurial/parser.py", line 219, in unescapestr: No attribute 'lower' on ValueError [attribute-error] In Union[ValueError, mercurial.pycompat.bytestr] Differential Revision: https://phab.mercurial-scm.org/D11471

util: avoid a name-error warning in the `mmapread` exception handler Newly caught by pytype 2021-09-09. File "/mnt/c/Users/Matt/hg/mercurial/util.py", line 458, in mmapread: Name 'fd' is not defined [name-error] Differential Revision: https://phab.mercurial-scm.org/D11470

typing: suppress an name-error warning in `mercurial/windows.py` Newly caught by pytype 2021-09-09. Differential Revision: https://phab.mercurial-scm.org/D11469

typing: suppress an import-error warning in `mercurial/utils/resourceutil.py` Newly caught by pytype 2021-09-09. Differential Revision: https://phab.mercurial-scm.org/D11468

encoding: force a few Errors to bytes before passing to `error.Abort` I'm not sure what changed before pytype 09-09-2021 (from 04-15-2021), but these started getting flagged. PyCharm also flagged these. This fixes the following: File "/mnt/c/Users/Matt/hg/mercurial/encoding.py", line 243, in fromlocal: Function Abort.__init__ was called with the wrong arguments [wrong-arg-types] Expected: (self, message: Union[bytearray, bytes, memoryview], ...) Actually passed: (self, message: LookupError, ...) File "/mnt/c/Users/Matt/hg/mercurial/encoding.py", line 309, in lower: Function Abort.__init__ was called with the wrong arguments [wrong-arg-types] Expected: (self, message: Union[bytearray, bytes, memoryview], ...) Actually passed: (self, message: LookupError, ...) File "/mnt/c/Users/Matt/hg/mercurial/encoding.py", line 336, in upperfallback: Function Abort.__init__ was called with the wrong arguments [wrong-arg-types] Expected: (self, message: Union[bytearray, bytes, memoryview], ...) Actually passed: (self, message: LookupError, ...) Called from (traceback): line 391, in current file line 348, in get line 318, in upper Differential Revision: https://phab.mercurial-scm.org/D11467

typing: add a fake `__init__()` to bytestr to distract pytype I'm not sure what changed before pytype 09-09-2021 (from 04-15-2021), but these started getting flagged. This wrapping an exception in a `bytestr` pattern has been flagged before, and I've fixed it then with `stringutil.forcebytestr()`. But that doesn't work here, because it would create a circular import. I suspect the issue is `bytes.__new__()` wants `Iterable[int]`, so it just assumes the subclass will also take that. The referenced pytype bug isn't an exact match, but seems related and the suggested workaround helps. The specific warnings fixed are: File "/mnt/c/Users/Matt/hg/mercurial/encoding.py", line 212, in tolocal: Function bytestr.__init__ was called with the wrong arguments [wrong-arg-types] Expected: (self, ints: Iterable[int]) Actually passed: (self, ints: LookupError) Attributes of protocol Iterable[int] are not implemented on LookupError: __iter__ Called from (traceback): line 353, in current file File "/mnt/c/Users/Matt/hg/mercurial/encoding.py", line 240, in fromlocal: Function bytestr.__init__ was called with the wrong arguments [wrong-arg-types] Expected: (self, ints: Iterable[int]) Actually passed: (self, ints: UnicodeDecodeError) Attributes of protocol Iterable[int] are not implemented on UnicodeDecodeError: __iter__ Differential Revision: https://phab.mercurial-scm.org/D11466

merge: with stable

pep-517: remove the `build-backend` key to allow for users to build extensions PEP 517 changed how projects should define their build dependencies. The presence of a `pyproject.toml` file changes the behavior of `pip` to conform to PEP 517. Since we haven't updated the `setup.py` file yet (I'm not even sure we're not an edge case which would make it harder/impossible, I have not tried yet), this is a workaround to issue6589: it allows users to pass `--no-use-pep517` to pip to revert to the legacy installer. The build backend is simply assumed to be the `build_meta:__legacy__` one. Differential Revision: https://phab.mercurial-scm.org/D11393

dirstate: drop the `dirstatemap.dropfile` method All use have been migrated. Differential Revision: https://phab.mercurial-scm.org/D11443

dirstate: use `reset_state` instead of `dropfile` in test-rebuildstate.t As `dirstatemap.dropfile` is on its way out. Differential Revision: https://phab.mercurial-scm.org/D11442

dirstate: use `reset_state` instead of `dropfile` in largefile As `dirstatemap.dropfile` is on its way out. Differential Revision: https://phab.mercurial-scm.org/D11441

dirstate: use `reset_state` in `rebuild` instead of `dropfile` As `dirstatemap.dropfile` is on its way out. Differential Revision: https://phab.mercurial-scm.org/D11440

dirstate: use `reset_state` to drop file in `update_file_p1` The `dropfile` function is on its way out. Differential Revision: https://phab.mercurial-scm.org/D11439

dirstate: support file tracked nowhere in `reset_state` This let the dirstatemap decide when to drop files. Differential Revision: https://phab.mercurial-scm.org/D11438

dirstate: drop dirstatemap.addfile All users have been migrated. Differential Revision: https://phab.mercurial-scm.org/D11437

dirstate: update the documentation of the dirstatemap API The API changed so should the documentation. Differential Revision: https://phab.mercurial-scm.org/D11436

dirstate: drop the `_addpath` method All user have been migrated. Differential Revision: https://phab.mercurial-scm.org/D11435

dirstate: remove the `normallookup` function All use have been migrated. Differential Revision: https://phab.mercurial-scm.org/D11434

dirstate: stop using `_normallookup` in the adddrop extension That extension is just trying to produce a "corrupted" dirstate. Differential Revision: https://phab.mercurial-scm.org/D11433

dirstate: replace the use of `_normallookup` in `rebuild` Normal lookup is a complicated function that we want to get rid of. Differential Revision: https://phab.mercurial-scm.org/D11432

dirstate: replace the use of _normallookup in `setparents` normal lookup is a complicated function that we want to get rid of. Differential Revision: https://phab.mercurial-scm.org/D11431

dirstate: introduce a set_tracked method on "map" and "item" This move more implementation details withing the DirstateItem itself, which is what we have been doing for a while. Differential Revision: https://phab.mercurial-scm.org/D11430

dirstate: fix restoration of "merged" state after a remove Before this change, "merged" file that get removed and re-added later were recorded as "from_p2" instead. This came from 8fe74328f700, a 2014 changeset that start explicitly doing so for reason I have not been able to fully grasp. The graft test mentioned in the description are still happy after this changeset. So this changeset restore what seems to be the intended behavior. Restoring information as it was before the removal. Differential Revision: https://phab.mercurial-scm.org/D11429

dirstate: inline the last two `_drop` usage The function is small and having the associated code directly inline help use to cleanup the dirstate API. Differential Revision: https://phab.mercurial-scm.org/D11428

dirstate: make a conditionnal easier to read in `setparents` By grabing `nullid` in the local space, we get the conditionnal to fit in one line. Differential Revision: https://phab.mercurial-scm.org/D11427

dirstate: removed the now unused `_add` method Differential Revision: https://phab.mercurial-scm.org/D11426

dirstate: replace `_add` call in `setparent` Same logic as the previous changeset, it get us closer to removed older API on the dirstate. Differential Revision: https://phab.mercurial-scm.org/D11425

dirstate: directly call the right function in `set_tracked` This get use closer to removing some older API on the dirstate. Differential Revision: https://phab.mercurial-scm.org/D11424

dirstate: same logic as what we did for `_drop` This is part of the dirstatemap so let the dirstatemap deal with it. Differential Revision: https://phab.mercurial-scm.org/D11423

dirstate: use `tracked` property in `_addpath` It is semantically better. Differential Revision: https://phab.mercurial-scm.org/D11422

rust: Update the memmap2 crate to version 0.4.0 This version: * Optionally implements the `StableDeref` trait, removing the need for a wrapper type doing that * Adds support for mapping empty files, which can routinely happen for dirstate-v2 data files. This was a cause of some failing tests when running with `run-tests.py --rhg --extra-config-opt rhg.status=1` Differential Revision: https://phab.mercurial-scm.org/D11446