view mercurial/cext/util.h @ 35616:706aa203b396

fileset: add a lightweight file filtering language

This patch was inspired by one that Jun Wu authored for the fb-experimental
repo, to avoid using matcher for efficiency[1].  We want a way to specify what
files will be converted to LFS at commit time.  And per discussion, we also want
to specify what files to skip, text diff, or merge in another config option.
The current `lfs.threshold` config option could not satisfy complex needs.  I'm
putting it in a core package because Augie floated the idea of also using it for
narrow and sparse.

Yuya suggested farming out to fileset.parse(), which added support for more
symbols.  The only fileset element not supported here is 'negate'.  (List isn't
supported by filesets either.)  I also changed the 'always' token to the 'all()'
predicate for consistency, and introduced 'none()' to improve readability in a
future tracked file based config.  The extension operator was changed from '.'
to '**', to match how recursive path globs are specified.  Finally, I changed
the path matcher from '/' to 'path:' at Yuya's suggestion, for consistency with
matcher.  Unfortunately, ':' is currently reserved in filesets, so this has to
be quoted to be processed as a string instead of a symbol[2].  We should
probably revisit that, because it's seriously ugly.  But it's only used by an
experimental extension, and I think using a file based config for LFS may drive
some more tweaks, so I'm settling for this for now.

I reserved all of the glob characters in fileset except '.' and '_' for the
extension test because those are likely valid extension characters.

Sample filter settings:

  all()                        # everything
  size(">20MB")                # larger than 20MB
  !**.txt                      # except for .txt files
  **.zip | **.tar.gz | **.7z   # some types of compressed files
  "path:bin"                   # files under "bin" in the project root

[1] https://www.mercurial-scm.org/pipermail/mercurial-devel/2017-December/109387.html
[2] https://www.mercurial-scm.org/pipermail/mercurial-devel/2018-January/109729.html

/*
 util.h - utility functions for interfacing with the various python APIs.

 This software may be used and distributed according to the terms of
 the GNU General Public License, incorporated herein by reference.
*/

#ifndef _HG_UTIL_H_
#define _HG_UTIL_H_

#include "compat.h"

#if PY_MAJOR_VERSION >= 3
#define IS_PY3K
#endif

/* clang-format off */
typedef struct {
	PyObject_HEAD
	char state;
	int mode;
	int size;
	int mtime;
} dirstateTupleObject;
/* clang-format on */

extern PyTypeObject dirstateTupleType;
#define dirstate_tuple_check(op) (Py_TYPE(op) == &dirstateTupleType)

#define MIN(a, b) (((a) < (b)) ? (a) : (b))
/* VC9 doesn't include bool and lacks stdbool.h based on my searching */
#if defined(_MSC_VER) || __STDC_VERSION__ < 199901L
#define true 1
#define false 0
typedef unsigned char bool;
#else
#include <stdbool.h>
#endif

static inline PyObject *_dict_new_presized(Py_ssize_t expected_size)
{
	/* _PyDict_NewPresized expects a minused parameter, but it actually
	   creates a dictionary that's the nearest power of two bigger than the
	   parameter. For example, with the initial minused = 1000, the
	   dictionary created has size 1024. Of course in a lot of cases that
	   can be greater than the maximum load factor Python's dict object
	   expects (= 2/3), so as soon as we cross the threshold we'll resize
	   anyway. So create a dictionary that's at least 3/2 the size. */
	return _PyDict_NewPresized(((1 + expected_size) / 2) * 3);
}

#endif /* _HG_UTIL_H_ */