rust: exposing in parsers module

To build with the Rust code, set the HGWITHRUSTEXT
environment variable.

At this point, it's possible to instantiate and use
a rustlazyancestors object from a Python interpreter.

The changes in setup.py are obviously a quick hack,
just good enough to test/bench without much
refactoring. We'd be happy to improve on that with
help from the community.

Rust bindings crate gets compiled as a static library,
which in turn gets linked within 'parsers.so'

With respect to the plans at
https://www.mercurial-scm.org/wiki/OxidationPlan
this would probably qualify as "roll our own FFI".
Also, it doesn't quite meet the target of getting
rid of C code, since it brings actually more, yet:

- the new C code does nothing else than parsing
  arguments and calling Rust functions.
  In particular, there's no complex allocation involved.
- subsequent changes could rewrite more of revlog.c, this
  time resulting in an overall decrease of C code and
  unsafety.

mercurial/cext/revlog.c

@@ -2290,6 +2290,152 b' bail:'
 	return NULL;
 }
 
+#ifdef WITH_RUST
+
+/* rustlazyancestors: iteration over ancestors implemented in Rust
+ *
+ * This class holds a reference to an index and to the Rust iterator.
+ */
+typedef struct rustlazyancestorsObjectStruct rustlazyancestorsObject;
+
+struct rustlazyancestorsObjectStruct {
+	PyObject_HEAD
+	/* Type-specific fields go here. */
+	indexObject *index;    /* Ref kept to avoid GC'ing the index */
+	void *iter;        /* Rust iterator */
+};
+
+/* FFI exposed from Rust code */
+rustlazyancestorsObject *rustlazyancestors_init(
+	indexObject *index,
+	/* to pass index_get_parents() */
+	int (*)(indexObject *, Py_ssize_t, int*, int),
+	/* intrevs vector */
+	int initrevslen, long *initrevs,
+	long stoprev,
+	int inclusive);
+void rustlazyancestors_drop(rustlazyancestorsObject *self);
+int rustlazyancestors_next(rustlazyancestorsObject *self);
+
+/* CPython instance methods */
+static int rustla_init(rustlazyancestorsObject *self,
+                       PyObject *args) {
+	PyObject *initrevsarg = NULL;
+	PyObject *inclusivearg = NULL;
+	long stoprev = 0;
+	long *initrevs = NULL;
+	int inclusive = 0;
+	Py_ssize_t i;
+
+	indexObject *index;
+	if (!PyArg_ParseTuple(args, "O!O!lO!",
+			      &indexType, &index,
+                              &PyList_Type, &initrevsarg,
+                              &stoprev,
+                              &PyBool_Type, &inclusivearg))
+	return -1;
+
+	Py_INCREF(index);
+	self->index = index;
+
+	if (inclusivearg == Py_True)
+		inclusive = 1;
+
+	Py_ssize_t linit = PyList_GET_SIZE(initrevsarg);
+
+	initrevs = (long*)calloc(linit, sizeof(long));
+
+	if (initrevs == NULL) {
+		PyErr_NoMemory();
+		goto bail;
+	}
+
+	for (i=0; i<linit; i++) {
+		initrevs[i] = PyInt_AsLong(PyList_GET_ITEM(initrevsarg, i));
+	}
+	if (PyErr_Occurred())
+		goto bail;
+
+	self->iter = rustlazyancestors_init(index,
+		                            index_get_parents,
+		                            linit, initrevs,
+		                            stoprev, inclusive);
+	if (self->iter == NULL) {
+		/* if this is because of GraphError::ParentOutOfRange
+		 * index_get_parents() has already set the proper ValueError */
+		goto bail;
+	}
+
+	free(initrevs);
+	return 0;
+
+bail:
+	free(initrevs);
+	return -1;
+};
+
+static void rustla_dealloc(rustlazyancestorsObject *self)
+{
+	Py_XDECREF(self->index);
+	if (self->iter != NULL) { /* can happen if rustla_init failed */
+		rustlazyancestors_drop(self->iter);
+	}
+	PyObject_Del(self);
+}
+
+static PyObject *rustla_next(rustlazyancestorsObject *self) {
+	int res = rustlazyancestors_next(self->iter);
+	if (res == -1) {
+		/* Setting an explicit exception seems unnecessary
+		 * as examples from Python source code (Objects/rangeobjets.c and
+		 * Modules/_io/stringio.c) seem to demonstrate.
+		 */
+		return NULL;
+	}
+	return PyInt_FromLong(res);
+}
+
+static PyTypeObject rustlazyancestorsType = {
+	PyVarObject_HEAD_INIT(NULL, 0) /* header */
+	"parsers.rustlazyancestors",           /* tp_name */
+	sizeof(rustlazyancestorsObject),       /* tp_basicsize */
+	0,                         /* tp_itemsize */
+	(destructor)rustla_dealloc, /* tp_dealloc */
+	0,                         /* tp_print */
+	0,                         /* tp_getattr */
+	0,                         /* tp_setattr */
+	0,                         /* tp_compare */
+	0,                         /* tp_repr */
+	0,                         /* tp_as_number */
+	0,                         /* tp_as_sequence */
+	0,                         /* tp_as_mapping */
+	0,                         /* tp_hash */
+	0,                         /* tp_call */
+	0,                         /* tp_str */
+	0,                         /* tp_getattro */
+	0,                         /* tp_setattro */
+	0,                         /* tp_as_buffer */
+	Py_TPFLAGS_DEFAULT,        /* tp_flags */
+	"Iterator over ancestors, implemented in Rust", /* tp_doc */
+	0,                         /* tp_traverse */
+	0,                         /* tp_clear */
+	0,                         /* tp_richcompare */
+	0,                         /* tp_weaklistoffset */
+	0,                         /* tp_iter */
+	(iternextfunc)rustla_next, /* tp_iternext */
+	0,                         /* tp_methods */
+	0,                         /* tp_members */
+	0,                         /* tp_getset */
+	0,                         /* tp_base */
+	0,                         /* tp_dict */
+	0,                         /* tp_descr_get */
+	0,                         /* tp_descr_set */
+	0,                         /* tp_dictoffset */
+	(initproc)rustla_init,     /* tp_init */
+	0,                         /* tp_alloc */
+};
+#endif /* WITH_RUST */
+
 void revlog_module_init(PyObject *mod)
 {
 	indexType.tp_new = PyType_GenericNew;
@@ -2310,4 +2456,14 b' void revlog_module_init(PyObject *mod)'
 	}
 	if (nullentry)
 		PyObject_GC_UnTrack(nullentry);
+
+#ifdef WITH_RUST
+	rustlazyancestorsType.tp_new = PyType_GenericNew;
+	if (PyType_Ready(&rustlazyancestorsType) < 0)
+		return;
+	Py_INCREF(&rustlazyancestorsType);
+	PyModule_AddObject(mod, "rustlazyancestors",
+		(PyObject *)&rustlazyancestorsType);
+#endif
+
 }

setup.py

@@ -132,6 +132,8 b' else:'
 
 ispypy = "PyPy" in sys.version
 
+iswithrustextensions = 'HGWITHRUSTEXT' in os.environ
+
 import ctypes
 import stat, subprocess, time
 import re
@@ -460,6 +462,8 b' class hgbuildext(build_ext):'
         return build_ext.build_extensions(self)
 
     def build_extension(self, ext):
+        if isinstance(ext, RustExtension):
+            ext.rustbuild()
         try:
             build_ext.build_extension(self, ext)
         except CCompilerError:
@@ -884,6 +888,54 b' xdiff_headers = ['
     'mercurial/thirdparty/xdiff/xutils.h',
 ]
 
+class RustExtension(Extension):
+    """A C Extension, conditionnally enhanced with Rust code.
+
+    if iswithrustextensions is False, does nothing else than plain Extension
+    """
+
+    rusttargetdir = os.path.join('rust', 'target', 'release')
+
+    def __init__(self, mpath, sources, rustlibname, subcrate, **kw):
+        Extension.__init__(self, mpath, sources, **kw)
+        if not iswithrustextensions:
+            return
+        srcdir = self.rustsrcdir = os.path.join('rust', subcrate)
+        self.libraries.append(rustlibname)
+        self.extra_compile_args.append('-DWITH_RUST')
+
+        # adding Rust source and control files to depends so that the extension
+        # gets rebuilt if they've changed
+        self.depends.append(os.path.join(srcdir, 'Cargo.toml'))
+        cargo_lock = os.path.join(srcdir, 'Cargo.lock')
+        if os.path.exists(cargo_lock):
+            self.depends.append(cargo_lock)
+        for dirpath, subdir, fnames in os.walk(os.path.join(srcdir, 'src')):
+            self.depends.extend(os.path.join(dirpath, fname)
+                                for fname in fnames
+                                if os.path.splitext(fname)[1] == '.rs')
+
+    def rustbuild(self):
+        if not iswithrustextensions:
+            return
+        env = os.environ.copy()
+        if 'HGTEST_RESTOREENV' in env:
+            # Mercurial tests change HOME to a temporary directory,
+            # but, if installed with rustup, the Rust toolchain needs
+            # HOME to be correct (otherwise the 'no default toolchain'
+            # error message is issued and the build fails).
+            # This happens currently with test-hghave.t, which does
+            # invoke this build.
+
+            # Unix only fix (os.path.expanduser not really reliable if
+            # HOME is shadowed like this)
+            import pwd
+            env['HOME'] = pwd.getpwuid(os.getuid()).pw_dir
+
+        subprocess.check_call(['cargo', 'build', '-vv', '--release'],
+                              env=env, cwd=self.rustsrcdir)
+        self.library_dirs.append(self.rusttargetdir)
+
 extmodules = [
     Extension('mercurial.cext.base85', ['mercurial/cext/base85.c'],
               include_dirs=common_include_dirs,
@@ -896,14 +948,19 b' extmodules = ['
                                         'mercurial/cext/mpatch.c'],
               include_dirs=common_include_dirs,
               depends=common_depends),
-    Extension('mercurial.cext.parsers', ['mercurial/cext/charencode.c',
-                                         'mercurial/cext/dirs.c',
-                                         'mercurial/cext/manifest.c',
-                                         'mercurial/cext/parsers.c',
-                                         'mercurial/cext/pathencode.c',
-                                         'mercurial/cext/revlog.c'],
-              include_dirs=common_include_dirs,
-              depends=common_depends + ['mercurial/cext/charencode.h']),
+    RustExtension('mercurial.cext.parsers', ['mercurial/cext/charencode.c',
+                                             'mercurial/cext/dirs.c',
+                                             'mercurial/cext/manifest.c',
+                                             'mercurial/cext/parsers.c',
+                                             'mercurial/cext/pathencode.c',
+                                             'mercurial/cext/revlog.c'],
+                  'hgdirectffi',
+                  'hg-direct-ffi',
+                  include_dirs=common_include_dirs,
+                  depends=common_depends + ['mercurial/cext/charencode.h',
+                                            'mercurial/rust/src/lib.rs',
+                                            'mercurial/rust/src/ancestors.rs',
+                                            'mercurial/rust/src/cpython.rs']),
     Extension('mercurial.cext.osutil', ['mercurial/cext/osutil.c'],
               include_dirs=common_include_dirs,
               extra_compile_args=osutil_cflags,