cext: factor out header for charencode.c

This merges a part of util.h with the header which should exist for
charencode.c.

mercurial/cext/charencode.c

@@ -1,209 +1,210 b''
 /*
  charencode.c - miscellaneous character encoding
 
  Copyright 2008 Matt Mackall <[e-mail]> and others
 
  This software may be used and distributed according to the terms of
  the GNU General Public License, incorporated herein by reference.
 */
 
 #include <Python.h>
 
+#include "charencode.h"
 #include "util.h"
 
 static const char lowertable[128] = {
 	'\x00', '\x01', '\x02', '\x03', '\x04', '\x05', '\x06', '\x07',
 	'\x08', '\x09', '\x0a', '\x0b', '\x0c', '\x0d', '\x0e', '\x0f',
 	'\x10', '\x11', '\x12', '\x13', '\x14', '\x15', '\x16', '\x17',
 	'\x18', '\x19', '\x1a', '\x1b', '\x1c', '\x1d', '\x1e', '\x1f',
 	'\x20', '\x21', '\x22', '\x23', '\x24', '\x25', '\x26', '\x27',
 	'\x28', '\x29', '\x2a', '\x2b', '\x2c', '\x2d', '\x2e', '\x2f',
 	'\x30', '\x31', '\x32', '\x33', '\x34', '\x35', '\x36', '\x37',
 	'\x38', '\x39', '\x3a', '\x3b', '\x3c', '\x3d', '\x3e', '\x3f',
 	'\x40',
 	        '\x61', '\x62', '\x63', '\x64', '\x65', '\x66', '\x67', /* A-G */
 	'\x68', '\x69', '\x6a', '\x6b', '\x6c', '\x6d', '\x6e', '\x6f', /* H-O */
 	'\x70', '\x71', '\x72', '\x73', '\x74', '\x75', '\x76', '\x77', /* P-W */
 	'\x78', '\x79', '\x7a',                                         /* X-Z */
 	                        '\x5b', '\x5c', '\x5d', '\x5e', '\x5f',
 	'\x60', '\x61', '\x62', '\x63', '\x64', '\x65', '\x66', '\x67',
 	'\x68', '\x69', '\x6a', '\x6b', '\x6c', '\x6d', '\x6e', '\x6f',
 	'\x70', '\x71', '\x72', '\x73', '\x74', '\x75', '\x76', '\x77',
 	'\x78', '\x79', '\x7a', '\x7b', '\x7c', '\x7d', '\x7e', '\x7f'
 };
 
 static const char uppertable[128] = {
 	'\x00', '\x01', '\x02', '\x03', '\x04', '\x05', '\x06', '\x07',
 	'\x08', '\x09', '\x0a', '\x0b', '\x0c', '\x0d', '\x0e', '\x0f',
 	'\x10', '\x11', '\x12', '\x13', '\x14', '\x15', '\x16', '\x17',
 	'\x18', '\x19', '\x1a', '\x1b', '\x1c', '\x1d', '\x1e', '\x1f',
 	'\x20', '\x21', '\x22', '\x23', '\x24', '\x25', '\x26', '\x27',
 	'\x28', '\x29', '\x2a', '\x2b', '\x2c', '\x2d', '\x2e', '\x2f',
 	'\x30', '\x31', '\x32', '\x33', '\x34', '\x35', '\x36', '\x37',
 	'\x38', '\x39', '\x3a', '\x3b', '\x3c', '\x3d', '\x3e', '\x3f',
 	'\x40', '\x41', '\x42', '\x43', '\x44', '\x45', '\x46', '\x47',
 	'\x48', '\x49', '\x4a', '\x4b', '\x4c', '\x4d', '\x4e', '\x4f',
 	'\x50', '\x51', '\x52', '\x53', '\x54', '\x55', '\x56', '\x57',
 	'\x58', '\x59', '\x5a', '\x5b', '\x5c', '\x5d', '\x5e', '\x5f',
 	'\x60',
 		'\x41', '\x42', '\x43', '\x44', '\x45', '\x46', '\x47', /* a-g */
 	'\x48', '\x49', '\x4a', '\x4b', '\x4c', '\x4d', '\x4e', '\x4f', /* h-o */
 	'\x50', '\x51', '\x52', '\x53', '\x54', '\x55', '\x56', '\x57', /* p-w */
 	'\x58', '\x59', '\x5a', 					/* x-z */
 				'\x7b', '\x7c', '\x7d', '\x7e', '\x7f'
 };
 
 /*
  * Turn a hex-encoded string into binary.
  */
 PyObject *unhexlify(const char *str, int len)
 {
 	PyObject *ret;
 	char *d;
 	int i;
 
 	ret = PyBytes_FromStringAndSize(NULL, len / 2);
 
 	if (!ret)
 		return NULL;
 
 	d = PyBytes_AsString(ret);
 
 	for (i = 0; i < len;) {
 		int hi = hexdigit(str, i++);
 		int lo = hexdigit(str, i++);
 		*d++ = (hi << 4) | lo;
 	}
 
 	return ret;
 }
 
 static inline PyObject *_asciitransform(PyObject *str_obj,
 					const char table[128],
 					PyObject *fallback_fn)
 {
 	char *str, *newstr;
 	Py_ssize_t i, len;
 	PyObject *newobj = NULL;
 	PyObject *ret = NULL;
 
 	str = PyBytes_AS_STRING(str_obj);
 	len = PyBytes_GET_SIZE(str_obj);
 
 	newobj = PyBytes_FromStringAndSize(NULL, len);
 	if (!newobj)
 		goto quit;
 
 	newstr = PyBytes_AS_STRING(newobj);
 
 	for (i = 0; i < len; i++) {
 		char c = str[i];
 		if (c & 0x80) {
 			if (fallback_fn != NULL) {
 				ret = PyObject_CallFunctionObjArgs(fallback_fn,
 					str_obj, NULL);
 			} else {
 				PyObject *err = PyUnicodeDecodeError_Create(
 					"ascii", str, len, i, (i + 1),
 					"unexpected code byte");
 				PyErr_SetObject(PyExc_UnicodeDecodeError, err);
 				Py_XDECREF(err);
 			}
 			goto quit;
 		}
 		newstr[i] = table[(unsigned char)c];
 	}
 
 	ret = newobj;
 	Py_INCREF(ret);
 quit:
 	Py_XDECREF(newobj);
 	return ret;
 }
 
 PyObject *asciilower(PyObject *self, PyObject *args)
 {
 	PyObject *str_obj;
 	if (!PyArg_ParseTuple(args, "O!:asciilower", &PyBytes_Type, &str_obj))
 		return NULL;
 	return _asciitransform(str_obj, lowertable, NULL);
 }
 
 PyObject *asciiupper(PyObject *self, PyObject *args)
 {
 	PyObject *str_obj;
 	if (!PyArg_ParseTuple(args, "O!:asciiupper", &PyBytes_Type, &str_obj))
 		return NULL;
 	return _asciitransform(str_obj, uppertable, NULL);
 }
 
 PyObject *make_file_foldmap(PyObject *self, PyObject *args)
 {
 	PyObject *dmap, *spec_obj, *normcase_fallback;
 	PyObject *file_foldmap = NULL;
 	enum normcase_spec spec;
 	PyObject *k, *v;
 	dirstateTupleObject *tuple;
 	Py_ssize_t pos = 0;
 	const char *table;
 
 	if (!PyArg_ParseTuple(args, "O!O!O!:make_file_foldmap",
 			      &PyDict_Type, &dmap,
 			      &PyInt_Type, &spec_obj,
 			      &PyFunction_Type, &normcase_fallback))
 		goto quit;
 
 	spec = (int)PyInt_AS_LONG(spec_obj);
 	switch (spec) {
 	case NORMCASE_LOWER:
 		table = lowertable;
 		break;
 	case NORMCASE_UPPER:
 		table = uppertable;
 		break;
 	case NORMCASE_OTHER:
 		table = NULL;
 		break;
 	default:
 		PyErr_SetString(PyExc_TypeError, "invalid normcasespec");
 		goto quit;
 	}
 
 	/* Add some more entries to deal with additions outside this
 	   function. */
 	file_foldmap = _dict_new_presized((PyDict_Size(dmap) / 10) * 11);
 	if (file_foldmap == NULL)
 		goto quit;
 
 	while (PyDict_Next(dmap, &pos, &k, &v)) {
 		if (!dirstate_tuple_check(v)) {
 			PyErr_SetString(PyExc_TypeError,
 					"expected a dirstate tuple");
 			goto quit;
 		}
 
 		tuple = (dirstateTupleObject *)v;
 		if (tuple->state != 'r') {
 			PyObject *normed;
 			if (table != NULL) {
 				normed = _asciitransform(k, table,
 					normcase_fallback);
 			} else {
 				normed = PyObject_CallFunctionObjArgs(
 					normcase_fallback, k, NULL);
 			}
 
 			if (normed == NULL)
 				goto quit;
 			if (PyDict_SetItem(file_foldmap, normed, k) == -1) {
 				Py_DECREF(normed);
 				goto quit;
 			}
 			Py_DECREF(normed);
 		}
 	}
 	return file_foldmap;
 quit:
 	Py_XDECREF(file_foldmap);
 	return NULL;
 }

mercurial/cext/charencode.h

@@ -1,88 +1,57 b''
 /*
- util.h - utility functions for interfacing with the various python APIs.
+ charencode.h - miscellaneous character encoding
 
  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
+#ifndef _HG_CHARENCODE_H_
+#define _HG_CHARENCODE_H_
 
-typedef struct {
-	PyObject_HEAD
-	char state;
-	int mode;
-	int size;
-	int mtime;
-} dirstateTupleObject;
-
-extern PyTypeObject dirstateTupleType;
-#define dirstate_tuple_check(op) (Py_TYPE(op) == &dirstateTupleType)
+#include <Python.h>
+#include "compat.h"
 
 /* This should be kept in sync with normcasespecs in encoding.py. */
 enum normcase_spec {
 	NORMCASE_LOWER = -1,
 	NORMCASE_UPPER = 1,
 	NORMCASE_OTHER = 0
 };
 
-#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);
-}
+PyObject *unhexlify(const char *str, int len);
+PyObject *asciilower(PyObject *self, PyObject *args);
+PyObject *asciiupper(PyObject *self, PyObject *args);
+PyObject *make_file_foldmap(PyObject *self, PyObject *args);
 
 static const int8_t hextable[256] = {
 	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
 	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
 	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
 	 0,  1,  2,  3,  4,  5,  6,  7,  8,  9, -1, -1, -1, -1, -1, -1, /* 0-9 */
 	-1, 10, 11, 12, 13, 14, 15, -1, -1, -1, -1, -1, -1, -1, -1, -1, /* A-F */
 	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
 	-1, 10, 11, 12, 13, 14, 15, -1, -1, -1, -1, -1, -1, -1, -1, -1, /* a-f */
 	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
 	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
 	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
 	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
 	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
 	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
 	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
 	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
 	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1
 };
 
 static inline int hexdigit(const char *p, Py_ssize_t off)
 {
 	int8_t val = hextable[(unsigned char)p[off]];
 
 	if (val >= 0) {
 		return val;
 	}
 
 	PyErr_SetString(PyExc_ValueError, "input contains non-hex character");
 	return 0;
 }
 
-#endif /* _HG_UTIL_H_ */
+#endif /* _HG_CHARENCODE_H_ */

mercurial/cext/manifest.c

@@ -1,939 +1,937 b''
 /*
  * manifest.c - manifest type that does on-demand parsing.
  *
  * Copyright 2015, Google Inc.
  *
  * This software may be used and distributed according to the terms of
  * the GNU General Public License, incorporated herein by reference.
  */
 #include <Python.h>
 
 #include <assert.h>
 #include <string.h>
 #include <stdlib.h>
 
+#include "charencode.h"
 #include "util.h"
 
 #define DEFAULT_LINES 100000
 
 typedef struct {
 	char *start;
 	Py_ssize_t len; /* length of line including terminal newline */
 	char hash_suffix;
 	bool from_malloc;
 	bool deleted;
 } line;
 
 typedef struct {
 	PyObject_HEAD
 	PyObject *pydata;
 	line *lines;
 	int numlines; /* number of line entries */
 	int livelines; /* number of non-deleted lines */
 	int maxlines; /* allocated number of lines */
 	bool dirty;
 } lazymanifest;
 
 #define MANIFEST_OOM -1
 #define MANIFEST_NOT_SORTED -2
 #define MANIFEST_MALFORMED -3
 
-/* defined in charencode.c */
-PyObject *unhexlify(const char *str, int len);
-
 /* get the length of the path for a line */
 static size_t pathlen(line *l) {
 	return strlen(l->start);
 }
 
 /* get the node value of a single line */
 static PyObject *nodeof(line *l) {
 	char *s = l->start;
 	ssize_t llen = pathlen(l);
 	PyObject *hash = unhexlify(s + llen + 1, 40);
 	if (!hash) {
 		return NULL;
 	}
 	if (l->hash_suffix != '\0') {
 		char newhash[21];
 		memcpy(newhash, PyBytes_AsString(hash), 20);
 		Py_DECREF(hash);
 		newhash[20] = l->hash_suffix;
 		hash = PyBytes_FromStringAndSize(newhash, 21);
 	}
 	return hash;
 }
 
 /* get the node hash and flags of a line as a tuple */
 static PyObject *hashflags(line *l)
 {
 	char *s = l->start;
 	size_t plen = pathlen(l);
 	PyObject *hash = nodeof(l);
 
 	/* 40 for hash, 1 for null byte, 1 for newline */
 	size_t hplen = plen + 42;
 	Py_ssize_t flen = l->len - hplen;
 	PyObject *flags;
 	PyObject *tup;
 
 	if (!hash)
 		return NULL;
 	flags = PyBytes_FromStringAndSize(s + hplen - 1, flen);
 	if (!flags) {
 		Py_DECREF(hash);
 		return NULL;
 	}
 	tup = PyTuple_Pack(2, hash, flags);
 	Py_DECREF(flags);
 	Py_DECREF(hash);
 	return tup;
 }
 
 /* if we're about to run out of space in the line index, add more */
 static bool realloc_if_full(lazymanifest *self)
 {
 	if (self->numlines == self->maxlines) {
 		self->maxlines *= 2;
 		self->lines = realloc(self->lines, self->maxlines * sizeof(line));
 	}
 	return !!self->lines;
 }
 
 /*
  * Find the line boundaries in the manifest that 'data' points to and store
  * information about each line in 'self'.
  */
 static int find_lines(lazymanifest *self, char *data, Py_ssize_t len)
 {
 	char *prev = NULL;
 	while (len > 0) {
 		line *l;
 		char *next = memchr(data, '\n', len);
 		if (!next) {
 			return MANIFEST_MALFORMED;
 		}
 		next++; /* advance past newline */
 		if (!realloc_if_full(self)) {
 			return MANIFEST_OOM; /* no memory */
 		}
 		if (prev && strcmp(prev, data) > -1) {
 			/* This data isn't sorted, so we have to abort. */
 			return MANIFEST_NOT_SORTED;
 		}
 		l = self->lines + ((self->numlines)++);
 		l->start = data;
 		l->len = next - data;
 		l->hash_suffix = '\0';
 		l->from_malloc = false;
 		l->deleted = false;
 		len = len - l->len;
 		prev = data;
 		data = next;
 	}
 	self->livelines = self->numlines;
 	return 0;
 }
 
 static int lazymanifest_init(lazymanifest *self, PyObject *args)
 {
 	char *data;
 	Py_ssize_t len;
 	int err, ret;
 	PyObject *pydata;
 	if (!PyArg_ParseTuple(args, "S", &pydata)) {
 		return -1;
 	}
 	err = PyBytes_AsStringAndSize(pydata, &data, &len);
 
 	self->dirty = false;
 	if (err == -1)
 		return -1;
 	self->pydata = pydata;
 	Py_INCREF(self->pydata);
 	Py_BEGIN_ALLOW_THREADS
 	self->lines = malloc(DEFAULT_LINES * sizeof(line));
 	self->maxlines = DEFAULT_LINES;
 	self->numlines = 0;
 	if (!self->lines)
 		ret = MANIFEST_OOM;
 	else
 		ret = find_lines(self, data, len);
 	Py_END_ALLOW_THREADS
 	switch (ret) {
 	case 0:
 		break;
 	case MANIFEST_OOM:
 		PyErr_NoMemory();
 		break;
 	case MANIFEST_NOT_SORTED:
 		PyErr_Format(PyExc_ValueError,
 			     "Manifest lines not in sorted order.");
 		break;
 	case MANIFEST_MALFORMED:
 		PyErr_Format(PyExc_ValueError,
 			     "Manifest did not end in a newline.");
 		break;
 	default:
 		PyErr_Format(PyExc_ValueError,
 			     "Unknown problem parsing manifest.");
 	}
 	return ret == 0 ? 0 : -1;
 }
 
 static void lazymanifest_dealloc(lazymanifest *self)
 {
 	/* free any extra lines we had to allocate */
 	int i;
 	for (i = 0; i < self->numlines; i++) {
 		if (self->lines[i].from_malloc) {
 			free(self->lines[i].start);
 		}
 	}
 	if (self->lines) {
 		free(self->lines);
 		self->lines = NULL;
 	}
 	if (self->pydata) {
 		Py_DECREF(self->pydata);
 		self->pydata = NULL;
 	}
 	PyObject_Del(self);
 }
 
 /* iteration support */
 
 typedef struct {
 	PyObject_HEAD lazymanifest *m;
 	Py_ssize_t pos;
 } lmIter;
 
 static void lmiter_dealloc(PyObject *o)
 {
 	lmIter *self = (lmIter *)o;
 	Py_DECREF(self->m);
 	PyObject_Del(self);
 }
 
 static line *lmiter_nextline(lmIter *self)
 {
 	do {
 		self->pos++;
 		if (self->pos >= self->m->numlines) {
 			return NULL;
 		}
 		/* skip over deleted manifest entries */
 	} while (self->m->lines[self->pos].deleted);
 	return self->m->lines + self->pos;
 }
 
 static PyObject *lmiter_iterentriesnext(PyObject *o)
 {
 	size_t pl;
 	line *l;
 	Py_ssize_t consumed;
 	PyObject *ret = NULL, *path = NULL, *hash = NULL, *flags = NULL;
 	l = lmiter_nextline((lmIter *)o);
 	if (!l) {
 		goto done;
 	}
 	pl = pathlen(l);
 	path = PyBytes_FromStringAndSize(l->start, pl);
 	hash = nodeof(l);
 	consumed = pl + 41;
 	flags = PyBytes_FromStringAndSize(l->start + consumed,
 					   l->len - consumed - 1);
 	if (!path || !hash || !flags) {
 		goto done;
 	}
 	ret = PyTuple_Pack(3, path, hash, flags);
 done:
 	Py_XDECREF(path);
 	Py_XDECREF(hash);
 	Py_XDECREF(flags);
 	return ret;
 }
 
 #ifdef IS_PY3K
 #define LAZYMANIFESTENTRIESITERATOR_TPFLAGS Py_TPFLAGS_DEFAULT
 #else
 #define LAZYMANIFESTENTRIESITERATOR_TPFLAGS Py_TPFLAGS_DEFAULT \
 	| Py_TPFLAGS_HAVE_ITER
 #endif
 
 static PyTypeObject lazymanifestEntriesIterator = {
 	PyVarObject_HEAD_INIT(NULL, 0)
 	"parsers.lazymanifest.entriesiterator", /*tp_name */
 	sizeof(lmIter),                  /*tp_basicsize */
 	0,                               /*tp_itemsize */
 	lmiter_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 */
 	LAZYMANIFESTENTRIESITERATOR_TPFLAGS, /* tp_flags */
 	"Iterator for 3-tuples in a lazymanifest.",  /* tp_doc */
 	0,                               /* tp_traverse */
 	0,                               /* tp_clear */
 	0,                               /* tp_richcompare */
 	0,                               /* tp_weaklistoffset */
 	PyObject_SelfIter,               /* tp_iter: __iter__() method */
 	lmiter_iterentriesnext,          /* tp_iternext: next() method */
 };
 
 static PyObject *lmiter_iterkeysnext(PyObject *o)
 {
 	size_t pl;
 	line *l = lmiter_nextline((lmIter *)o);
 	if (!l) {
 		return NULL;
 	}
 	pl = pathlen(l);
 	return PyBytes_FromStringAndSize(l->start, pl);
 }
 
 #ifdef IS_PY3K
 #define LAZYMANIFESTKEYSITERATOR_TPFLAGS Py_TPFLAGS_DEFAULT
 #else
 #define LAZYMANIFESTKEYSITERATOR_TPFLAGS Py_TPFLAGS_DEFAULT \
 	| Py_TPFLAGS_HAVE_ITER
 #endif
 
 static PyTypeObject lazymanifestKeysIterator = {
 	PyVarObject_HEAD_INIT(NULL, 0)
 	"parsers.lazymanifest.keysiterator", /*tp_name */
 	sizeof(lmIter),                  /*tp_basicsize */
 	0,                               /*tp_itemsize */
 	lmiter_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 */
 	LAZYMANIFESTKEYSITERATOR_TPFLAGS, /* tp_flags */
 	"Keys iterator for a lazymanifest.",  /* tp_doc */
 	0,                               /* tp_traverse */
 	0,                               /* tp_clear */
 	0,                               /* tp_richcompare */
 	0,                               /* tp_weaklistoffset */
 	PyObject_SelfIter,               /* tp_iter: __iter__() method */
 	lmiter_iterkeysnext,             /* tp_iternext: next() method */
 };
 
 static lazymanifest *lazymanifest_copy(lazymanifest *self);
 
 static PyObject *lazymanifest_getentriesiter(lazymanifest *self)
 {
 	lmIter *i = NULL;
 	lazymanifest *t = lazymanifest_copy(self);
 	if (!t) {
 		PyErr_NoMemory();
 		return NULL;
 	}
 	i = PyObject_New(lmIter, &lazymanifestEntriesIterator);
 	if (i) {
 		i->m = t;
 		i->pos = -1;
 	} else {
 		Py_DECREF(t);
 		PyErr_NoMemory();
 	}
 	return (PyObject *)i;
 }
 
 static PyObject *lazymanifest_getkeysiter(lazymanifest *self)
 {
 	lmIter *i = NULL;
 	lazymanifest *t = lazymanifest_copy(self);
 	if (!t) {
 		PyErr_NoMemory();
 		return NULL;
 	}
 	i = PyObject_New(lmIter, &lazymanifestKeysIterator);
 	if (i) {
 		i->m = t;
 		i->pos = -1;
 	} else {
 		Py_DECREF(t);
 		PyErr_NoMemory();
 	}
 	return (PyObject *)i;
 }
 
 /* __getitem__ and __setitem__ support */
 
 static Py_ssize_t lazymanifest_size(lazymanifest *self)
 {
 	return self->livelines;
 }
 
 static int linecmp(const void *left, const void *right)
 {
 	return strcmp(((const line *)left)->start,
 		      ((const line *)right)->start);
 }
 
 static PyObject *lazymanifest_getitem(lazymanifest *self, PyObject *key)
 {
 	line needle;
 	line *hit;
 	if (!PyBytes_Check(key)) {
 		PyErr_Format(PyExc_TypeError,
 			     "getitem: manifest keys must be a string.");
 		return NULL;
 	}
 	needle.start = PyBytes_AsString(key);
 	hit = bsearch(&needle, self->lines, self->numlines, sizeof(line),
 		      &linecmp);
 	if (!hit || hit->deleted) {
 		PyErr_Format(PyExc_KeyError, "No such manifest entry.");
 		return NULL;
 	}
 	return hashflags(hit);
 }
 
 static int lazymanifest_delitem(lazymanifest *self, PyObject *key)
 {
 	line needle;
 	line *hit;
 	if (!PyBytes_Check(key)) {
 		PyErr_Format(PyExc_TypeError,
 			     "delitem: manifest keys must be a string.");
 		return -1;
 	}
 	needle.start = PyBytes_AsString(key);
 	hit = bsearch(&needle, self->lines, self->numlines, sizeof(line),
 		      &linecmp);
 	if (!hit || hit->deleted) {
 		PyErr_Format(PyExc_KeyError,
 			     "Tried to delete nonexistent manifest entry.");
 		return -1;
 	}
 	self->dirty = true;
 	hit->deleted = true;
 	self->livelines--;
 	return 0;
 }
 
 /* Do a binary search for the insertion point for new, creating the
  * new entry if needed. */
 static int internalsetitem(lazymanifest *self, line *new) {
 	int start = 0, end = self->numlines;
 	while (start < end) {
 		int pos = start + (end - start) / 2;
 		int c = linecmp(new, self->lines + pos);
 		if (c < 0)
 			end = pos;
 		else if (c > 0)
 			start = pos + 1;
 		else {
 			if (self->lines[pos].deleted)
 				self->livelines++;
 			if (self->lines[pos].from_malloc)
 				free(self->lines[pos].start);
 			start = pos;
 			goto finish;
 		}
 	}
 	/* being here means we need to do an insert */
 	if (!realloc_if_full(self)) {
 		PyErr_NoMemory();
 		return -1;
 	}
 	memmove(self->lines + start + 1, self->lines + start,
 		(self->numlines - start) * sizeof(line));
 	self->numlines++;
 	self->livelines++;
 finish:
 	self->lines[start] = *new;
 	self->dirty = true;
 	return 0;
 }
 
 static int lazymanifest_setitem(
 	lazymanifest *self, PyObject *key, PyObject *value)
 {
 	char *path;
 	Py_ssize_t plen;
 	PyObject *pyhash;
 	Py_ssize_t hlen;
 	char *hash;
 	PyObject *pyflags;
 	char *flags;
 	Py_ssize_t flen;
 	size_t dlen;
 	char *dest;
 	int i;
 	line new;
 	if (!PyBytes_Check(key)) {
 		PyErr_Format(PyExc_TypeError,
 			     "setitem: manifest keys must be a string.");
 		return -1;
 	}
 	if (!value) {
 		return lazymanifest_delitem(self, key);
 	}
 	if (!PyTuple_Check(value) || PyTuple_Size(value) != 2) {
 		PyErr_Format(PyExc_TypeError,
 			     "Manifest values must be a tuple of (node, flags).");
 		return -1;
 	}
 	if (PyBytes_AsStringAndSize(key, &path, &plen) == -1) {
 		return -1;
 	}
 
 	pyhash = PyTuple_GetItem(value, 0);
 	if (!PyBytes_Check(pyhash)) {
 		PyErr_Format(PyExc_TypeError,
 			     "node must be a 20-byte string");
 		return -1;
 	}
 	hlen = PyBytes_Size(pyhash);
 	/* Some parts of the codebase try and set 21 or 22
 	 * byte "hash" values in order to perturb things for
 	 * status. We have to preserve at least the 21st
 	 * byte. Sigh. If there's a 22nd byte, we drop it on
 	 * the floor, which works fine.
 	 */
 	if (hlen != 20 && hlen != 21 && hlen != 22) {
 		PyErr_Format(PyExc_TypeError,
 			     "node must be a 20-byte string");
 		return -1;
 	}
 	hash = PyBytes_AsString(pyhash);
 
 	pyflags = PyTuple_GetItem(value, 1);
 	if (!PyBytes_Check(pyflags) || PyBytes_Size(pyflags) > 1) {
 		PyErr_Format(PyExc_TypeError,
 			     "flags must a 0 or 1 byte string");
 		return -1;
 	}
 	if (PyBytes_AsStringAndSize(pyflags, &flags, &flen) == -1) {
 		return -1;
 	}
 	/* one null byte and one newline */
 	dlen = plen + 41 + flen + 1;
 	dest = malloc(dlen);
 	if (!dest) {
 		PyErr_NoMemory();
 		return -1;
 	}
 	memcpy(dest, path, plen + 1);
 	for (i = 0; i < 20; i++) {
 		/* Cast to unsigned, so it will not get sign-extended when promoted
 		 * to int (as is done when passing to a variadic function)
 		 */
 		sprintf(dest + plen + 1 + (i * 2), "%02x", (unsigned char)hash[i]);
 	}
 	memcpy(dest + plen + 41, flags, flen);
 	dest[plen + 41 + flen] = '\n';
 	new.start = dest;
 	new.len = dlen;
 	new.hash_suffix = '\0';
 	if (hlen > 20) {
 		new.hash_suffix = hash[20];
 	}
 	new.from_malloc = true;     /* is `start` a pointer we allocated? */
 	new.deleted = false;        /* is this entry deleted? */
 	if (internalsetitem(self, &new)) {
 		return -1;
 	}
 	return 0;
 }
 
 static PyMappingMethods lazymanifest_mapping_methods = {
 	(lenfunc)lazymanifest_size,             /* mp_length */
 	(binaryfunc)lazymanifest_getitem,       /* mp_subscript */
 	(objobjargproc)lazymanifest_setitem,    /* mp_ass_subscript */
 };
 
 /* sequence methods (important or __contains__ builds an iterator) */
 
 static int lazymanifest_contains(lazymanifest *self, PyObject *key)
 {
 	line needle;
 	line *hit;
 	if (!PyBytes_Check(key)) {
 		/* Our keys are always strings, so if the contains
 		 * check is for a non-string, just return false. */
 		return 0;
 	}
 	needle.start = PyBytes_AsString(key);
 	hit = bsearch(&needle, self->lines, self->numlines, sizeof(line),
 		      &linecmp);
 	if (!hit || hit->deleted) {
 		return 0;
 	}
 	return 1;
 }
 
 static PySequenceMethods lazymanifest_seq_meths = {
 	(lenfunc)lazymanifest_size, /* sq_length */
 	0, /* sq_concat */
 	0, /* sq_repeat */
 	0, /* sq_item */
 	0, /* sq_slice */
 	0, /* sq_ass_item */
 	0, /* sq_ass_slice */
 	(objobjproc)lazymanifest_contains, /* sq_contains */
 	0, /* sq_inplace_concat */
 	0, /* sq_inplace_repeat */
 };
 
 
 /* Other methods (copy, diff, etc) */
 static PyTypeObject lazymanifestType;
 
 /* If the manifest has changes, build the new manifest text and reindex it. */
 static int compact(lazymanifest *self) {
 	int i;
 	ssize_t need = 0;
 	char *data;
 	line *src, *dst;
 	PyObject *pydata;
 	if (!self->dirty)
 		return 0;
 	for (i = 0; i < self->numlines; i++) {
 		if (!self->lines[i].deleted) {
 			need += self->lines[i].len;
 		}
 	}
 	pydata = PyBytes_FromStringAndSize(NULL, need);
 	if (!pydata)
 		return -1;
 	data = PyBytes_AsString(pydata);
 	if (!data) {
 		return -1;
 	}
 	src = self->lines;
 	dst = self->lines;
 	for (i = 0; i < self->numlines; i++, src++) {
 		char *tofree = NULL;
 		if (src->from_malloc) {
 			tofree = src->start;
 		}
 		if (!src->deleted) {
 			memcpy(data, src->start, src->len);
 			*dst = *src;
 			dst->start = data;
 			dst->from_malloc = false;
 			data += dst->len;
 			dst++;
 		}
 		free(tofree);
 	}
 	Py_DECREF(self->pydata);
 	self->pydata = pydata;
 	self->numlines = self->livelines;
 	self->dirty = false;
 	return 0;
 }
 
 static PyObject *lazymanifest_text(lazymanifest *self)
 {
 	if (compact(self) != 0) {
 		PyErr_NoMemory();
 		return NULL;
 	}
 	Py_INCREF(self->pydata);
 	return self->pydata;
 }
 
 static lazymanifest *lazymanifest_copy(lazymanifest *self)
 {
 	lazymanifest *copy = NULL;
 	if (compact(self) != 0) {
 		goto nomem;
 	}
 	copy = PyObject_New(lazymanifest, &lazymanifestType);
 	if (!copy) {
 		goto nomem;
 	}
 	copy->numlines = self->numlines;
 	copy->livelines = self->livelines;
 	copy->dirty = false;
 	copy->lines = malloc(self->maxlines *sizeof(line));
 	if (!copy->lines) {
 		goto nomem;
 	}
 	memcpy(copy->lines, self->lines, self->numlines * sizeof(line));
 	copy->maxlines = self->maxlines;
 	copy->pydata = self->pydata;
 	Py_INCREF(copy->pydata);
 	return copy;
 nomem:
 	PyErr_NoMemory();
 	Py_XDECREF(copy);
 	return NULL;
 }
 
 static lazymanifest *lazymanifest_filtercopy(
 	lazymanifest *self, PyObject *matchfn)
 {
 	lazymanifest *copy = NULL;
 	int i;
 	if (!PyCallable_Check(matchfn)) {
 		PyErr_SetString(PyExc_TypeError, "matchfn must be callable");
 		return NULL;
 	}
 	/* compact ourselves first to avoid double-frees later when we
 	 * compact tmp so that it doesn't have random pointers to our
 	 * underlying from_malloc-data (self->pydata is safe) */
 	if (compact(self) != 0) {
 		goto nomem;
 	}
 	copy = PyObject_New(lazymanifest, &lazymanifestType);
 	if (!copy) {
 		goto nomem;
 	}
 	copy->dirty = true;
 	copy->lines = malloc(self->maxlines * sizeof(line));
 	if (!copy->lines) {
 		goto nomem;
 	}
 	copy->maxlines = self->maxlines;
 	copy->numlines = 0;
 	copy->pydata = self->pydata;
 	Py_INCREF(self->pydata);
 	for (i = 0; i < self->numlines; i++) {
 		PyObject *arglist = NULL, *result = NULL;
 		arglist = Py_BuildValue("(s)", self->lines[i].start);
 		if (!arglist) {
 			return NULL;
 		}
 		result = PyObject_CallObject(matchfn, arglist);
 		Py_DECREF(arglist);
 		/* if the callback raised an exception, just let it
 		 * through and give up */
 		if (!result) {
 			free(copy->lines);
 			Py_DECREF(self->pydata);
 			return NULL;
 		}
 		if (PyObject_IsTrue(result)) {
 			assert(!(self->lines[i].from_malloc));
 			copy->lines[copy->numlines++] = self->lines[i];
 		}
 		Py_DECREF(result);
 	}
 	copy->livelines = copy->numlines;
 	return copy;
 nomem:
 	PyErr_NoMemory();
 	Py_XDECREF(copy);
 	return NULL;
 }
 
 static PyObject *lazymanifest_diff(lazymanifest *self, PyObject *args)
 {
 	lazymanifest *other;
 	PyObject *pyclean = NULL;
 	bool listclean;
 	PyObject *emptyTup = NULL, *ret = NULL;
 	PyObject *es;
 	int sneedle = 0, oneedle = 0;
 	if (!PyArg_ParseTuple(args, "O!|O", &lazymanifestType, &other, &pyclean)) {
 		return NULL;
 	}
 	listclean = (!pyclean) ? false : PyObject_IsTrue(pyclean);
 	es = PyBytes_FromString("");
 	if (!es) {
 		goto nomem;
 	}
 	emptyTup = PyTuple_Pack(2, Py_None, es);
 	Py_DECREF(es);
 	if (!emptyTup) {
 		goto nomem;
 	}
 	ret = PyDict_New();
 	if (!ret) {
 		goto nomem;
 	}
 	while (sneedle != self->numlines || oneedle != other->numlines) {
 		line *left = self->lines + sneedle;
 		line *right = other->lines + oneedle;
 		int result;
 		PyObject *key;
 		PyObject *outer;
 		/* If we're looking at a deleted entry and it's not
 		 * the end of the manifest, just skip it. */
 		if (left->deleted && sneedle < self->numlines) {
 			sneedle++;
 			continue;
 		}
 		if (right->deleted && oneedle < other->numlines) {
 			oneedle++;
 			continue;
 		}
 		/* if we're at the end of either manifest, then we
 		 * know the remaining items are adds so we can skip
 		 * the strcmp. */
 		if (sneedle == self->numlines) {
 			result = 1;
 		} else if (oneedle == other->numlines) {
 			result = -1;
 		} else {
 			result = linecmp(left, right);
 		}
 		key = result <= 0 ?
 			PyBytes_FromString(left->start) :
 			PyBytes_FromString(right->start);
 		if (!key)
 			goto nomem;
 		if (result < 0) {
 			PyObject *l = hashflags(left);
 			if (!l) {
 				goto nomem;
 			}
 			outer = PyTuple_Pack(2, l, emptyTup);
 			Py_DECREF(l);
 			if (!outer) {
 				goto nomem;
 			}
 			PyDict_SetItem(ret, key, outer);
 			Py_DECREF(outer);
 			sneedle++;
 		} else if (result > 0) {
 			PyObject *r = hashflags(right);
 			if (!r) {
 				goto nomem;
 			}
 			outer = PyTuple_Pack(2, emptyTup, r);
 			Py_DECREF(r);
 			if (!outer) {
 				goto nomem;
 			}
 			PyDict_SetItem(ret, key, outer);
 			Py_DECREF(outer);
 			oneedle++;
 		} else {
 			/* file exists in both manifests */
 			if (left->len != right->len
 			    || memcmp(left->start, right->start, left->len)
 			    || left->hash_suffix != right->hash_suffix) {
 				PyObject *l = hashflags(left);
 				PyObject *r;
 				if (!l) {
 					goto nomem;
 				}
 				r = hashflags(right);
 				if (!r) {
 					Py_DECREF(l);
 					goto nomem;
 				}
 				outer = PyTuple_Pack(2, l, r);
 				Py_DECREF(l);
 				Py_DECREF(r);
 				if (!outer) {
 					goto nomem;
 				}
 				PyDict_SetItem(ret, key, outer);
 				Py_DECREF(outer);
 			} else if (listclean) {
 				PyDict_SetItem(ret, key, Py_None);
 			}
 			sneedle++;
 			oneedle++;
 		}
 		Py_DECREF(key);
 	}
 	Py_DECREF(emptyTup);
 	return ret;
 nomem:
 	PyErr_NoMemory();
 	Py_XDECREF(ret);
 	Py_XDECREF(emptyTup);
 	return NULL;
 }
 
 static PyMethodDef lazymanifest_methods[] = {
 	{"iterkeys", (PyCFunction)lazymanifest_getkeysiter, METH_NOARGS,
 	 "Iterate over file names in this lazymanifest."},
 	{"iterentries", (PyCFunction)lazymanifest_getentriesiter, METH_NOARGS,
 	 "Iterate over (path, nodeid, flags) tuples in this lazymanifest."},
 	{"copy", (PyCFunction)lazymanifest_copy, METH_NOARGS,
 	 "Make a copy of this lazymanifest."},
 	{"filtercopy", (PyCFunction)lazymanifest_filtercopy, METH_O,
 	 "Make a copy of this manifest filtered by matchfn."},
 	{"diff", (PyCFunction)lazymanifest_diff, METH_VARARGS,
 	 "Compare this lazymanifest to another one."},
 	{"text", (PyCFunction)lazymanifest_text, METH_NOARGS,
 	 "Encode this manifest to text."},
 	{NULL},
 };
 
 #ifdef IS_PY3K
 #define LAZYMANIFEST_TPFLAGS Py_TPFLAGS_DEFAULT
 #else
 #define LAZYMANIFEST_TPFLAGS Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_SEQUENCE_IN
 #endif
 
 static PyTypeObject lazymanifestType = {
 	PyVarObject_HEAD_INIT(NULL, 0)
 	"parsers.lazymanifest",                           /* tp_name */
 	sizeof(lazymanifest),                             /* tp_basicsize */
 	0,                                                /* tp_itemsize */
 	(destructor)lazymanifest_dealloc,                 /* tp_dealloc */
 	0,                                                /* tp_print */
 	0,                                                /* tp_getattr */
 	0,                                                /* tp_setattr */
 	0,                                                /* tp_compare */
 	0,                                                /* tp_repr */
 	0,                                                /* tp_as_number */
 	&lazymanifest_seq_meths,                          /* tp_as_sequence */
 	&lazymanifest_mapping_methods,                    /* tp_as_mapping */
 	0,                                                /* tp_hash */
 	0,                                                /* tp_call */
 	0,                                                /* tp_str */
 	0,                                                /* tp_getattro */
 	0,                                                /* tp_setattro */
 	0,                                                /* tp_as_buffer */
 	LAZYMANIFEST_TPFLAGS,                             /* tp_flags */
 	"TODO(augie)",                                    /* tp_doc */
 	0,                                                /* tp_traverse */
 	0,                                                /* tp_clear */
 	0,                                                /* tp_richcompare */
 	0,                                             /* tp_weaklistoffset */
 	(getiterfunc)lazymanifest_getkeysiter,                /* tp_iter */
 	0,                                                /* tp_iternext */
 	lazymanifest_methods,                             /* 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)lazymanifest_init,                      /* tp_init */
 	0,                                                /* tp_alloc */
 };
 
 void manifest_module_init(PyObject * mod)
 {
 	lazymanifestType.tp_new = PyType_GenericNew;
 	if (PyType_Ready(&lazymanifestType) < 0)
 		return;
 	Py_INCREF(&lazymanifestType);
 
 	PyModule_AddObject(mod, "lazymanifest",
 			   (PyObject *)&lazymanifestType);
 }

mercurial/cext/parsers.c

@@ -1,806 +1,801 b''
 /*
  parsers.c - efficient content parsing
 
  Copyright 2008 Matt Mackall <[e-mail]> and others
 
  This software may be used and distributed according to the terms of
  the GNU General Public License, incorporated herein by reference.
 */
 
 #include <Python.h>
 #include <ctype.h>
 #include <stddef.h>
 #include <string.h>
 
+#include "charencode.h"
 #include "util.h"
 #include "bitmanipulation.h"
 
 #ifdef IS_PY3K
 /* The mapping of Python types is meant to be temporary to get Python
  * 3 to compile. We should remove this once Python 3 support is fully
  * supported and proper types are used in the extensions themselves. */
 #define PyInt_Type PyLong_Type
 #define PyInt_Check PyLong_Check
 #define PyInt_FromLong PyLong_FromLong
 #define PyInt_FromSsize_t PyLong_FromSsize_t
 #define PyInt_AS_LONG PyLong_AS_LONG
 #define PyInt_AsLong PyLong_AsLong
 #endif
 
 static const char *const versionerrortext = "Python minor version mismatch";
 
-/* defined in charencode.c */
-PyObject *unhexlify(const char *str, int len);
-PyObject *asciilower(PyObject *self, PyObject *args);
-PyObject *asciiupper(PyObject *self, PyObject *args);
-PyObject *make_file_foldmap(PyObject *self, PyObject *args);
-
 static PyObject *dict_new_presized(PyObject *self, PyObject *args)
 {
 	Py_ssize_t expected_size;
 
 	if (!PyArg_ParseTuple(args, "n:make_presized_dict", &expected_size))
 		return NULL;
 
 	return _dict_new_presized(expected_size);
 }
 
 /*
  * This code assumes that a manifest is stitched together with newline
  * ('\n') characters.
  */
 static PyObject *parse_manifest(PyObject *self, PyObject *args)
 {
 	PyObject *mfdict, *fdict;
 	char *str, *start, *end;
 	int len;
 
 	if (!PyArg_ParseTuple(args, "O!O!s#:parse_manifest",
 			      &PyDict_Type, &mfdict,
 			      &PyDict_Type, &fdict,
 			      &str, &len))
 		goto quit;
 
 	start = str;
 	end = str + len;
 	while (start < end) {
 		PyObject *file = NULL, *node = NULL;
 		PyObject *flags = NULL;
 		char *zero = NULL, *newline = NULL;
 		ptrdiff_t nlen;
 
 		zero = memchr(start, '\0', end - start);
 		if (!zero) {
 			PyErr_SetString(PyExc_ValueError,
 					"manifest entry has no separator");
 			goto quit;
 		}
 
 		newline = memchr(zero + 1, '\n', end - (zero + 1));
 		if (!newline) {
 			PyErr_SetString(PyExc_ValueError,
 					"manifest contains trailing garbage");
 			goto quit;
 		}
 
 		file = PyBytes_FromStringAndSize(start, zero - start);
 
 		if (!file)
 			goto bail;
 
 		nlen = newline - zero - 1;
 
 		node = unhexlify(zero + 1, nlen > 40 ? 40 : (int)nlen);
 		if (!node)
 			goto bail;
 
 		if (nlen > 40) {
 			flags = PyBytes_FromStringAndSize(zero + 41,
 							   nlen - 40);
 			if (!flags)
 				goto bail;
 
 			if (PyDict_SetItem(fdict, file, flags) == -1)
 				goto bail;
 		}
 
 		if (PyDict_SetItem(mfdict, file, node) == -1)
 			goto bail;
 
 		start = newline + 1;
 
 		Py_XDECREF(flags);
 		Py_XDECREF(node);
 		Py_XDECREF(file);
 		continue;
 	bail:
 		Py_XDECREF(flags);
 		Py_XDECREF(node);
 		Py_XDECREF(file);
 		goto quit;
 	}
 
 	Py_INCREF(Py_None);
 	return Py_None;
 quit:
 	return NULL;
 }
 
 static inline dirstateTupleObject *make_dirstate_tuple(char state, int mode,
 						       int size, int mtime)
 {
 	dirstateTupleObject *t = PyObject_New(dirstateTupleObject,
 					      &dirstateTupleType);
 	if (!t)
 		return NULL;
 	t->state = state;
 	t->mode = mode;
 	t->size = size;
 	t->mtime = mtime;
 	return t;
 }
 
 static PyObject *dirstate_tuple_new(PyTypeObject *subtype, PyObject *args,
 				    PyObject *kwds)
 {
 	/* We do all the initialization here and not a tp_init function because
 	 * dirstate_tuple is immutable. */
 	dirstateTupleObject *t;
 	char state;
 	int size, mode, mtime;
 	if (!PyArg_ParseTuple(args, "ciii", &state, &mode, &size, &mtime))
 		return NULL;
 
 	t = (dirstateTupleObject *)subtype->tp_alloc(subtype, 1);
 	if (!t)
 		return NULL;
 	t->state = state;
 	t->mode = mode;
 	t->size = size;
 	t->mtime = mtime;
 
 	return (PyObject *)t;
 }
 
 static void dirstate_tuple_dealloc(PyObject *o)
 {
 	PyObject_Del(o);
 }
 
 static Py_ssize_t dirstate_tuple_length(PyObject *o)
 {
 	return 4;
 }
 
 static PyObject *dirstate_tuple_item(PyObject *o, Py_ssize_t i)
 {
 	dirstateTupleObject *t = (dirstateTupleObject *)o;
 	switch (i) {
 	case 0:
 		return PyBytes_FromStringAndSize(&t->state, 1);
 	case 1:
 		return PyInt_FromLong(t->mode);
 	case 2:
 		return PyInt_FromLong(t->size);
 	case 3:
 		return PyInt_FromLong(t->mtime);
 	default:
 		PyErr_SetString(PyExc_IndexError, "index out of range");
 		return NULL;
 	}
 }
 
 static PySequenceMethods dirstate_tuple_sq = {
 	dirstate_tuple_length,     /* sq_length */
 	0,                         /* sq_concat */
 	0,                         /* sq_repeat */
 	dirstate_tuple_item,       /* sq_item */
 	0,                         /* sq_ass_item */
 	0,                         /* sq_contains */
 	0,                         /* sq_inplace_concat */
 	0                          /* sq_inplace_repeat */
 };
 
 PyTypeObject dirstateTupleType = {
 	PyVarObject_HEAD_INIT(NULL, 0)
 	"dirstate_tuple",          /* tp_name */
 	sizeof(dirstateTupleObject),/* tp_basicsize */
 	0,                         /* tp_itemsize */
 	(destructor)dirstate_tuple_dealloc, /* tp_dealloc */
 	0,                         /* tp_print */
 	0,                         /* tp_getattr */
 	0,                         /* tp_setattr */
 	0,                         /* tp_compare */
 	0,                         /* tp_repr */
 	0,                         /* tp_as_number */
 	&dirstate_tuple_sq,        /* 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 */
 	"dirstate tuple",          /* tp_doc */
 	0,                         /* tp_traverse */
 	0,                         /* tp_clear */
 	0,                         /* tp_richcompare */
 	0,                         /* tp_weaklistoffset */
 	0,                         /* tp_iter */
 	0,                         /* 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 */
 	0,                         /* tp_init */
 	0,                         /* tp_alloc */
 	dirstate_tuple_new,        /* tp_new */
 };
 
 static PyObject *parse_dirstate(PyObject *self, PyObject *args)
 {
 	PyObject *dmap, *cmap, *parents = NULL, *ret = NULL;
 	PyObject *fname = NULL, *cname = NULL, *entry = NULL;
 	char state, *cur, *str, *cpos;
 	int mode, size, mtime;
 	unsigned int flen, len, pos = 40;
 	int readlen;
 
 	if (!PyArg_ParseTuple(args, "O!O!s#:parse_dirstate",
 			      &PyDict_Type, &dmap,
 			      &PyDict_Type, &cmap,
 			      &str, &readlen))
 		goto quit;
 
 	len = readlen;
 
 	/* read parents */
 	if (len < 40) {
 		PyErr_SetString(
 			PyExc_ValueError, "too little data for parents");
 		goto quit;
 	}
 
 	parents = Py_BuildValue("s#s#", str, 20, str + 20, 20);
 	if (!parents)
 		goto quit;
 
 	/* read filenames */
 	while (pos >= 40 && pos < len) {
 		if (pos + 17 > len) {
 			PyErr_SetString(PyExc_ValueError,
 					"overflow in dirstate");
 			goto quit;
 		}
 		cur = str + pos;
 		/* unpack header */
 		state = *cur;
 		mode = getbe32(cur + 1);
 		size = getbe32(cur + 5);
 		mtime = getbe32(cur + 9);
 		flen = getbe32(cur + 13);
 		pos += 17;
 		cur += 17;
 		if (flen > len - pos) {
 			PyErr_SetString(PyExc_ValueError, "overflow in dirstate");
 			goto quit;
 		}
 
 		entry = (PyObject *)make_dirstate_tuple(state, mode, size,
 							mtime);
 		cpos = memchr(cur, 0, flen);
 		if (cpos) {
 			fname = PyBytes_FromStringAndSize(cur, cpos - cur);
 			cname = PyBytes_FromStringAndSize(cpos + 1,
 							   flen - (cpos - cur) - 1);
 			if (!fname || !cname ||
 			    PyDict_SetItem(cmap, fname, cname) == -1 ||
 			    PyDict_SetItem(dmap, fname, entry) == -1)
 				goto quit;
 			Py_DECREF(cname);
 		} else {
 			fname = PyBytes_FromStringAndSize(cur, flen);
 			if (!fname ||
 			    PyDict_SetItem(dmap, fname, entry) == -1)
 				goto quit;
 		}
 		Py_DECREF(fname);
 		Py_DECREF(entry);
 		fname = cname = entry = NULL;
 		pos += flen;
 	}
 
 	ret = parents;
 	Py_INCREF(ret);
 quit:
 	Py_XDECREF(fname);
 	Py_XDECREF(cname);
 	Py_XDECREF(entry);
 	Py_XDECREF(parents);
 	return ret;
 }
 
 /*
  * Build a set of non-normal and other parent entries from the dirstate dmap
 */
 static PyObject *nonnormalotherparententries(PyObject *self, PyObject *args) {
 	PyObject *dmap, *fname, *v;
 	PyObject *nonnset = NULL, *otherpset = NULL, *result = NULL;
 	Py_ssize_t pos;
 
 	if (!PyArg_ParseTuple(args, "O!:nonnormalentries",
 			      &PyDict_Type, &dmap))
 		goto bail;
 
 	nonnset = PySet_New(NULL);
 	if (nonnset == NULL)
 		goto bail;
 
 	otherpset = PySet_New(NULL);
 	if (otherpset == NULL)
 		goto bail;
 
 	pos = 0;
 	while (PyDict_Next(dmap, &pos, &fname, &v)) {
 		dirstateTupleObject *t;
 		if (!dirstate_tuple_check(v)) {
 			PyErr_SetString(PyExc_TypeError,
 					"expected a dirstate tuple");
 			goto bail;
 		}
 		t = (dirstateTupleObject *)v;
 
 		if (t->state == 'n' && t->size == -2) {
 			if (PySet_Add(otherpset, fname) == -1) {
 				goto bail;
 			}
 		}
 
 		if (t->state == 'n' && t->mtime != -1)
 			continue;
 		if (PySet_Add(nonnset, fname) == -1)
 			goto bail;
 	}
 
 	result = Py_BuildValue("(OO)", nonnset, otherpset);
 	if (result == NULL)
 		goto bail;
 	Py_DECREF(nonnset);
 	Py_DECREF(otherpset);
 	return result;
 bail:
 	Py_XDECREF(nonnset);
 	Py_XDECREF(otherpset);
 	Py_XDECREF(result);
 	return NULL;
 }
 
 /*
  * Efficiently pack a dirstate object into its on-disk format.
  */
 static PyObject *pack_dirstate(PyObject *self, PyObject *args)
 {
 	PyObject *packobj = NULL;
 	PyObject *map, *copymap, *pl, *mtime_unset = NULL;
 	Py_ssize_t nbytes, pos, l;
 	PyObject *k, *v = NULL, *pn;
 	char *p, *s;
 	int now;
 
 	if (!PyArg_ParseTuple(args, "O!O!Oi:pack_dirstate",
 			      &PyDict_Type, &map, &PyDict_Type, &copymap,
 			      &pl, &now))
 		return NULL;
 
 	if (!PySequence_Check(pl) || PySequence_Size(pl) != 2) {
 		PyErr_SetString(PyExc_TypeError, "expected 2-element sequence");
 		return NULL;
 	}
 
 	/* Figure out how much we need to allocate. */
 	for (nbytes = 40, pos = 0; PyDict_Next(map, &pos, &k, &v);) {
 		PyObject *c;
 		if (!PyBytes_Check(k)) {
 			PyErr_SetString(PyExc_TypeError, "expected string key");
 			goto bail;
 		}
 		nbytes += PyBytes_GET_SIZE(k) + 17;
 		c = PyDict_GetItem(copymap, k);
 		if (c) {
 			if (!PyBytes_Check(c)) {
 				PyErr_SetString(PyExc_TypeError,
 						"expected string key");
 				goto bail;
 			}
 			nbytes += PyBytes_GET_SIZE(c) + 1;
 		}
 	}
 
 	packobj = PyBytes_FromStringAndSize(NULL, nbytes);
 	if (packobj == NULL)
 		goto bail;
 
 	p = PyBytes_AS_STRING(packobj);
 
 	pn = PySequence_ITEM(pl, 0);
 	if (PyBytes_AsStringAndSize(pn, &s, &l) == -1 || l != 20) {
 		PyErr_SetString(PyExc_TypeError, "expected a 20-byte hash");
 		goto bail;
 	}
 	memcpy(p, s, l);
 	p += 20;
 	pn = PySequence_ITEM(pl, 1);
 	if (PyBytes_AsStringAndSize(pn, &s, &l) == -1 || l != 20) {
 		PyErr_SetString(PyExc_TypeError, "expected a 20-byte hash");
 		goto bail;
 	}
 	memcpy(p, s, l);
 	p += 20;
 
 	for (pos = 0; PyDict_Next(map, &pos, &k, &v); ) {
 		dirstateTupleObject *tuple;
 		char state;
 		int mode, size, mtime;
 		Py_ssize_t len, l;
 		PyObject *o;
 		char *t;
 
 		if (!dirstate_tuple_check(v)) {
 			PyErr_SetString(PyExc_TypeError,
 					"expected a dirstate tuple");
 			goto bail;
 		}
 		tuple = (dirstateTupleObject *)v;
 
 		state = tuple->state;
 		mode = tuple->mode;
 		size = tuple->size;
 		mtime = tuple->mtime;
 		if (state == 'n' && mtime == now) {
 			/* See pure/parsers.py:pack_dirstate for why we do
 			 * this. */
 			mtime = -1;
 			mtime_unset = (PyObject *)make_dirstate_tuple(
 				state, mode, size, mtime);
 			if (!mtime_unset)
 				goto bail;
 			if (PyDict_SetItem(map, k, mtime_unset) == -1)
 				goto bail;
 			Py_DECREF(mtime_unset);
 			mtime_unset = NULL;
 		}
 		*p++ = state;
 		putbe32((uint32_t)mode, p);
 		putbe32((uint32_t)size, p + 4);
 		putbe32((uint32_t)mtime, p + 8);
 		t = p + 12;
 		p += 16;
 		len = PyBytes_GET_SIZE(k);
 		memcpy(p, PyBytes_AS_STRING(k), len);
 		p += len;
 		o = PyDict_GetItem(copymap, k);
 		if (o) {
 			*p++ = '\0';
 			l = PyBytes_GET_SIZE(o);
 			memcpy(p, PyBytes_AS_STRING(o), l);
 			p += l;
 			len += l + 1;
 		}
 		putbe32((uint32_t)len, t);
 	}
 
 	pos = p - PyBytes_AS_STRING(packobj);
 	if (pos != nbytes) {
 		PyErr_Format(PyExc_SystemError, "bad dirstate size: %ld != %ld",
                              (long)pos, (long)nbytes);
 		goto bail;
 	}
 
 	return packobj;
 bail:
 	Py_XDECREF(mtime_unset);
 	Py_XDECREF(packobj);
 	Py_XDECREF(v);
 	return NULL;
 }
 
 #define BUMPED_FIX 1
 #define USING_SHA_256 2
 #define FM1_HEADER_SIZE (4 + 8 + 2 + 2 + 1 + 1 + 1)
 
 static PyObject *readshas(
 	const char *source, unsigned char num, Py_ssize_t hashwidth)
 {
 	int i;
 	PyObject *list = PyTuple_New(num);
 	if (list == NULL) {
 		return NULL;
 	}
 	for (i = 0; i < num; i++) {
 		PyObject *hash = PyBytes_FromStringAndSize(source, hashwidth);
 		if (hash == NULL) {
 			Py_DECREF(list);
 			return NULL;
 		}
 		PyTuple_SET_ITEM(list, i, hash);
 		source += hashwidth;
 	}
 	return list;
 }
 
 static PyObject *fm1readmarker(const char *databegin, const char *dataend,
 			       uint32_t *msize)
 {
 	const char *data = databegin;
 	const char *meta;
 
 	double mtime;
 	int16_t tz;
 	uint16_t flags;
 	unsigned char nsuccs, nparents, nmetadata;
 	Py_ssize_t hashwidth = 20;
 
 	PyObject *prec = NULL, *parents = NULL, *succs = NULL;
 	PyObject *metadata = NULL, *ret = NULL;
 	int i;
 
 	if (data + FM1_HEADER_SIZE > dataend) {
 		goto overflow;
 	}
 
 	*msize = getbe32(data);
 	data += 4;
 	mtime = getbefloat64(data);
 	data += 8;
 	tz = getbeint16(data);
 	data += 2;
 	flags = getbeuint16(data);
 	data += 2;
 
 	if (flags & USING_SHA_256) {
 		hashwidth = 32;
 	}
 
 	nsuccs = (unsigned char)(*data++);
 	nparents = (unsigned char)(*data++);
 	nmetadata = (unsigned char)(*data++);
 
 	if (databegin + *msize > dataend) {
 		goto overflow;
 	}
 	dataend = databegin + *msize;  /* narrow down to marker size */
 
 	if (data + hashwidth > dataend) {
 		goto overflow;
 	}
 	prec = PyBytes_FromStringAndSize(data, hashwidth);
 	data += hashwidth;
 	if (prec == NULL) {
 		goto bail;
 	}
 
 	if (data + nsuccs * hashwidth > dataend) {
 		goto overflow;
 	}
 	succs = readshas(data, nsuccs, hashwidth);
 	if (succs == NULL) {
 		goto bail;
 	}
 	data += nsuccs * hashwidth;
 
 	if (nparents == 1 || nparents == 2) {
 		if (data + nparents * hashwidth > dataend) {
 			goto overflow;
 		}
 		parents = readshas(data, nparents, hashwidth);
 		if (parents == NULL) {
 			goto bail;
 		}
 		data += nparents * hashwidth;
 	} else {
 		parents = Py_None;
 		Py_INCREF(parents);
 	}
 
 	if (data + 2 * nmetadata > dataend) {
 		goto overflow;
 	}
 	meta = data + (2 * nmetadata);
 	metadata = PyTuple_New(nmetadata);
 	if (metadata == NULL) {
 		goto bail;
 	}
 	for (i = 0; i < nmetadata; i++) {
 		PyObject *tmp, *left = NULL, *right = NULL;
 		Py_ssize_t leftsize = (unsigned char)(*data++);
 		Py_ssize_t rightsize = (unsigned char)(*data++);
 		if (meta + leftsize + rightsize > dataend) {
 			goto overflow;
 		}
 		left = PyBytes_FromStringAndSize(meta, leftsize);
 		meta += leftsize;
 		right = PyBytes_FromStringAndSize(meta, rightsize);
 		meta += rightsize;
 		tmp = PyTuple_New(2);
 		if (!left || !right || !tmp) {
 			Py_XDECREF(left);
 			Py_XDECREF(right);
 			Py_XDECREF(tmp);
 			goto bail;
 		}
 		PyTuple_SET_ITEM(tmp, 0, left);
 		PyTuple_SET_ITEM(tmp, 1, right);
 		PyTuple_SET_ITEM(metadata, i, tmp);
 	}
 	ret = Py_BuildValue("(OOHO(di)O)", prec, succs, flags,
 			    metadata, mtime, (int)tz * 60, parents);
 	goto bail;  /* return successfully */
 
 overflow:
 	PyErr_SetString(PyExc_ValueError, "overflow in obsstore");
 bail:
 	Py_XDECREF(prec);
 	Py_XDECREF(succs);
 	Py_XDECREF(metadata);
 	Py_XDECREF(parents);
 	return ret;
 }
 
 
 static PyObject *fm1readmarkers(PyObject *self, PyObject *args) {
 	const char *data, *dataend;
 	int datalen;
 	Py_ssize_t offset, stop;
 	PyObject *markers = NULL;
 
 	if (!PyArg_ParseTuple(args, "s#nn", &data, &datalen, &offset, &stop)) {
 		return NULL;
 	}
 	dataend = data + datalen;
 	data += offset;
 	markers = PyList_New(0);
 	if (!markers) {
 		return NULL;
 	}
 	while (offset < stop) {
 		uint32_t msize;
 		int error;
 		PyObject *record = fm1readmarker(data, dataend, &msize);
 		if (!record) {
 			goto bail;
 		}
 		error = PyList_Append(markers, record);
 		Py_DECREF(record);
 		if (error) {
 			goto bail;
 		}
 		data += msize;
 		offset += msize;
 	}
 	return markers;
 bail:
 	Py_DECREF(markers);
 	return NULL;
 }
 
 static char parsers_doc[] = "Efficient content parsing.";
 
 PyObject *encodedir(PyObject *self, PyObject *args);
 PyObject *pathencode(PyObject *self, PyObject *args);
 PyObject *lowerencode(PyObject *self, PyObject *args);
 PyObject *parse_index2(PyObject *self, PyObject *args);
 
 static PyMethodDef methods[] = {
 	{"pack_dirstate", pack_dirstate, METH_VARARGS, "pack a dirstate\n"},
 	{"nonnormalotherparententries", nonnormalotherparententries, METH_VARARGS,
 	"create a set containing non-normal and other parent entries of given "
 	"dirstate\n"},
 	{"parse_manifest", parse_manifest, METH_VARARGS, "parse a manifest\n"},
 	{"parse_dirstate", parse_dirstate, METH_VARARGS, "parse a dirstate\n"},
 	{"parse_index2", parse_index2, METH_VARARGS, "parse a revlog index\n"},
 	{"asciilower", asciilower, METH_VARARGS, "lowercase an ASCII string\n"},
 	{"asciiupper", asciiupper, METH_VARARGS, "uppercase an ASCII string\n"},
 	{"dict_new_presized", dict_new_presized, METH_VARARGS,
 	 "construct a dict with an expected size\n"},
 	{"make_file_foldmap", make_file_foldmap, METH_VARARGS,
 	 "make file foldmap\n"},
 	{"encodedir", encodedir, METH_VARARGS, "encodedir a path\n"},
 	{"pathencode", pathencode, METH_VARARGS, "fncache-encode a path\n"},
 	{"lowerencode", lowerencode, METH_VARARGS, "lower-encode a path\n"},
 	{"fm1readmarkers", fm1readmarkers, METH_VARARGS,
 			"parse v1 obsolete markers\n"},
 	{NULL, NULL}
 };
 
 void dirs_module_init(PyObject *mod);
 void manifest_module_init(PyObject *mod);
 void revlog_module_init(PyObject *mod);
 
 static const int version = 1;
 
 static void module_init(PyObject *mod)
 {
 	PyModule_AddIntConstant(mod, "version", version);
 
 	/* This module constant has two purposes.  First, it lets us unit test
 	 * the ImportError raised without hard-coding any error text.  This
 	 * means we can change the text in the future without breaking tests,
 	 * even across changesets without a recompile.  Second, its presence
 	 * can be used to determine whether the version-checking logic is
 	 * present, which also helps in testing across changesets without a
 	 * recompile.  Note that this means the pure-Python version of parsers
 	 * should not have this module constant. */
 	PyModule_AddStringConstant(mod, "versionerrortext", versionerrortext);
 
 	dirs_module_init(mod);
 	manifest_module_init(mod);
 	revlog_module_init(mod);
 
 	if (PyType_Ready(&dirstateTupleType) < 0)
 		return;
 	Py_INCREF(&dirstateTupleType);
 	PyModule_AddObject(mod, "dirstatetuple",
 			   (PyObject *)&dirstateTupleType);
 }
 
 static int check_python_version(void)
 {
 	PyObject *sys = PyImport_ImportModule("sys"), *ver;
 	long hexversion;
 	if (!sys)
 		return -1;
 	ver = PyObject_GetAttrString(sys, "hexversion");
 	Py_DECREF(sys);
 	if (!ver)
 		return -1;
 	hexversion = PyInt_AsLong(ver);
 	Py_DECREF(ver);
 	/* sys.hexversion is a 32-bit number by default, so the -1 case
 	 * should only occur in unusual circumstances (e.g. if sys.hexversion
 	 * is manually set to an invalid value). */
 	if ((hexversion == -1) || (hexversion >> 16 != PY_VERSION_HEX >> 16)) {
 		PyErr_Format(PyExc_ImportError, "%s: The Mercurial extension "
 			"modules were compiled with Python " PY_VERSION ", but "
 			"Mercurial is currently using Python with sys.hexversion=%ld: "
 			"Python %s\n at: %s", versionerrortext, hexversion,
 			Py_GetVersion(), Py_GetProgramFullPath());
 		return -1;
 	}
 	return 0;
 }
 
 #ifdef IS_PY3K
 static struct PyModuleDef parsers_module = {
 	PyModuleDef_HEAD_INIT,
 	"parsers",
 	parsers_doc,
 	-1,
 	methods
 };
 
 PyMODINIT_FUNC PyInit_parsers(void)
 {
 	PyObject *mod;
 
 	if (check_python_version() == -1)
 		return NULL;
 	mod = PyModule_Create(&parsers_module);
 	module_init(mod);
 	return mod;
 }
 #else
 PyMODINIT_FUNC initparsers(void)
 {
 	PyObject *mod;
 
 	if (check_python_version() == -1)
 		return;
 	mod = Py_InitModule3("parsers", methods, parsers_doc);
 	module_init(mod);
 }
 #endif

mercurial/cext/revlog.c

@@ -1,2089 +1,2090 b''
 /*
  parsers.c - efficient content parsing
 
  Copyright 2008 Matt Mackall <[e-mail]> and others
 
  This software may be used and distributed according to the terms of
  the GNU General Public License, incorporated herein by reference.
 */
 
 #include <Python.h>
 #include <assert.h>
 #include <ctype.h>
 #include <stddef.h>
 #include <string.h>
 
+#include "charencode.h"
 #include "util.h"
 #include "bitmanipulation.h"
 
 #ifdef IS_PY3K
 /* The mapping of Python types is meant to be temporary to get Python
  * 3 to compile. We should remove this once Python 3 support is fully
  * supported and proper types are used in the extensions themselves. */
 #define PyInt_Check PyLong_Check
 #define PyInt_FromLong PyLong_FromLong
 #define PyInt_FromSsize_t PyLong_FromSsize_t
 #define PyInt_AS_LONG PyLong_AS_LONG
 #define PyInt_AsLong PyLong_AsLong
 #endif
 
 /*
  * A base-16 trie for fast node->rev mapping.
  *
  * Positive value is index of the next node in the trie
  * Negative value is a leaf: -(rev + 1)
  * Zero is empty
  */
 typedef struct {
 	int children[16];
 } nodetree;
 
 /*
  * This class has two behaviors.
  *
  * When used in a list-like way (with integer keys), we decode an
  * entry in a RevlogNG index file on demand. Our last entry is a
  * sentinel, always a nullid.  We have limited support for
  * integer-keyed insert and delete, only at elements right before the
  * sentinel.
  *
  * With string keys, we lazily perform a reverse mapping from node to
  * rev, using a base-16 trie.
  */
 typedef struct {
 	PyObject_HEAD
 	/* Type-specific fields go here. */
 	PyObject *data;        /* raw bytes of index */
 	Py_buffer buf;         /* buffer of data */
 	PyObject **cache;      /* cached tuples */
 	const char **offsets;  /* populated on demand */
 	Py_ssize_t raw_length; /* original number of elements */
 	Py_ssize_t length;     /* current number of elements */
 	PyObject *added;       /* populated on demand */
 	PyObject *headrevs;    /* cache, invalidated on changes */
 	PyObject *filteredrevs;/* filtered revs set */
 	nodetree *nt;          /* base-16 trie */
 	unsigned ntlength;          /* # nodes in use */
 	unsigned ntcapacity;        /* # nodes allocated */
 	int ntdepth;           /* maximum depth of tree */
 	int ntsplits;          /* # splits performed */
 	int ntrev;             /* last rev scanned */
 	int ntlookups;         /* # lookups */
 	int ntmisses;          /* # lookups that miss the cache */
 	int inlined;
 } indexObject;
 
 static Py_ssize_t index_length(const indexObject *self)
 {
 	if (self->added == NULL)
 		return self->length;
 	return self->length + PyList_GET_SIZE(self->added);
 }
 
 static PyObject *nullentry;
 static const char nullid[20];
 
 static Py_ssize_t inline_scan(indexObject *self, const char **offsets);
 
 #if LONG_MAX == 0x7fffffffL
 static char *tuple_format = "Kiiiiiis#";
 #else
 static char *tuple_format = "kiiiiiis#";
 #endif
 
 /* A RevlogNG v1 index entry is 64 bytes long. */
 static const long v1_hdrsize = 64;
 
 /*
  * Return a pointer to the beginning of a RevlogNG record.
  */
 static const char *index_deref(indexObject *self, Py_ssize_t pos)
 {
 	if (self->inlined && pos > 0) {
 		if (self->offsets == NULL) {
 			self->offsets = PyMem_Malloc(self->raw_length *
 					             sizeof(*self->offsets));
 			if (self->offsets == NULL)
 				return (const char *)PyErr_NoMemory();
 			inline_scan(self, self->offsets);
 		}
 		return self->offsets[pos];
 	}
 
 	return (const char *)(self->buf.buf) + pos * v1_hdrsize;
 }
 
 static inline int index_get_parents(indexObject *self, Py_ssize_t rev,
 				    int *ps, int maxrev)
 {
 	if (rev >= self->length - 1) {
 		PyObject *tuple = PyList_GET_ITEM(self->added,
 						  rev - self->length + 1);
 		ps[0] = (int)PyInt_AS_LONG(PyTuple_GET_ITEM(tuple, 5));
 		ps[1] = (int)PyInt_AS_LONG(PyTuple_GET_ITEM(tuple, 6));
 	} else {
 		const char *data = index_deref(self, rev);
 		ps[0] = getbe32(data + 24);
 		ps[1] = getbe32(data + 28);
 	}
 	/* If index file is corrupted, ps[] may point to invalid revisions. So
 	 * there is a risk of buffer overflow to trust them unconditionally. */
 	if (ps[0] > maxrev || ps[1] > maxrev) {
 		PyErr_SetString(PyExc_ValueError, "parent out of range");
 		return -1;
 	}
 	return 0;
 }
 
 
 /*
  * RevlogNG format (all in big endian, data may be inlined):
  *    6 bytes: offset
  *    2 bytes: flags
  *    4 bytes: compressed length
  *    4 bytes: uncompressed length
  *    4 bytes: base revision
  *    4 bytes: link revision
  *    4 bytes: parent 1 revision
  *    4 bytes: parent 2 revision
  *   32 bytes: nodeid (only 20 bytes used)
  */
 static PyObject *index_get(indexObject *self, Py_ssize_t pos)
 {
 	uint64_t offset_flags;
 	int comp_len, uncomp_len, base_rev, link_rev, parent_1, parent_2;
 	const char *c_node_id;
 	const char *data;
 	Py_ssize_t length = index_length(self);
 	PyObject *entry;
 
 	if (pos < 0)
 		pos += length;
 
 	if (pos < 0 || pos >= length) {
 		PyErr_SetString(PyExc_IndexError, "revlog index out of range");
 		return NULL;
 	}
 
 	if (pos == length - 1) {
 		Py_INCREF(nullentry);
 		return nullentry;
 	}
 
 	if (pos >= self->length - 1) {
 		PyObject *obj;
 		obj = PyList_GET_ITEM(self->added, pos - self->length + 1);
 		Py_INCREF(obj);
 		return obj;
 	}
 
 	if (self->cache) {
 		if (self->cache[pos]) {
 			Py_INCREF(self->cache[pos]);
 			return self->cache[pos];
 		}
 	} else {
 		self->cache = calloc(self->raw_length, sizeof(PyObject *));
 		if (self->cache == NULL)
 			return PyErr_NoMemory();
 	}
 
 	data = index_deref(self, pos);
 	if (data == NULL)
 		return NULL;
 
 	offset_flags = getbe32(data + 4);
 	if (pos == 0) /* mask out version number for the first entry */
 		offset_flags &= 0xFFFF;
 	else {
 		uint32_t offset_high = getbe32(data);
 		offset_flags |= ((uint64_t)offset_high) << 32;
 	}
 
 	comp_len = getbe32(data + 8);
 	uncomp_len = getbe32(data + 12);
 	base_rev = getbe32(data + 16);
 	link_rev = getbe32(data + 20);
 	parent_1 = getbe32(data + 24);
 	parent_2 = getbe32(data + 28);
 	c_node_id = data + 32;
 
 	entry = Py_BuildValue(tuple_format, offset_flags, comp_len,
 			      uncomp_len, base_rev, link_rev,
 			      parent_1, parent_2, c_node_id, 20);
 
 	if (entry) {
 		PyObject_GC_UnTrack(entry);
 		Py_INCREF(entry);
 	}
 
 	self->cache[pos] = entry;
 
 	return entry;
 }
 
 /*
  * Return the 20-byte SHA of the node corresponding to the given rev.
  */
 static const char *index_node(indexObject *self, Py_ssize_t pos)
 {
 	Py_ssize_t length = index_length(self);
 	const char *data;
 
 	if (pos == length - 1 || pos == INT_MAX)
 		return nullid;
 
 	if (pos >= length)
 		return NULL;
 
 	if (pos >= self->length - 1) {
 		PyObject *tuple, *str;
 		tuple = PyList_GET_ITEM(self->added, pos - self->length + 1);
 		str = PyTuple_GetItem(tuple, 7);
 		return str ? PyBytes_AS_STRING(str) : NULL;
 	}
 
 	data = index_deref(self, pos);
 	return data ? data + 32 : NULL;
 }
 
 static int nt_insert(indexObject *self, const char *node, int rev);
 
 static int node_check(PyObject *obj, char **node, Py_ssize_t *nodelen)
 {
 	if (PyBytes_AsStringAndSize(obj, node, nodelen) == -1)
 		return -1;
 	if (*nodelen == 20)
 		return 0;
 	PyErr_SetString(PyExc_ValueError, "20-byte hash required");
 	return -1;
 }
 
 static PyObject *index_insert(indexObject *self, PyObject *args)
 {
 	PyObject *obj;
 	char *node;
 	int index;
 	Py_ssize_t len, nodelen;
 
 	if (!PyArg_ParseTuple(args, "iO", &index, &obj))
 		return NULL;
 
 	if (!PyTuple_Check(obj) || PyTuple_GET_SIZE(obj) != 8) {
 		PyErr_SetString(PyExc_TypeError, "8-tuple required");
 		return NULL;
 	}
 
 	if (node_check(PyTuple_GET_ITEM(obj, 7), &node, &nodelen) == -1)
 		return NULL;
 
 	len = index_length(self);
 
 	if (index < 0)
 		index += len;
 
 	if (index != len - 1) {
 		PyErr_SetString(PyExc_IndexError,
 				"insert only supported at index -1");
 		return NULL;
 	}
 
 	if (self->added == NULL) {
 		self->added = PyList_New(0);
 		if (self->added == NULL)
 			return NULL;
 	}
 
 	if (PyList_Append(self->added, obj) == -1)
 		return NULL;
 
 	if (self->nt)
 		nt_insert(self, node, index);
 
 	Py_CLEAR(self->headrevs);
 	Py_RETURN_NONE;
 }
 
 static void _index_clearcaches(indexObject *self)
 {
 	if (self->cache) {
 		Py_ssize_t i;
 
 		for (i = 0; i < self->raw_length; i++)
 			Py_CLEAR(self->cache[i]);
 		free(self->cache);
 		self->cache = NULL;
 	}
 	if (self->offsets) {
 		PyMem_Free(self->offsets);
 		self->offsets = NULL;
 	}
 	if (self->nt) {
 		free(self->nt);
 		self->nt = NULL;
 	}
 	Py_CLEAR(self->headrevs);
 }
 
 static PyObject *index_clearcaches(indexObject *self)
 {
 	_index_clearcaches(self);
 	self->ntlength = self->ntcapacity = 0;
 	self->ntdepth = self->ntsplits = 0;
 	self->ntrev = -1;
 	self->ntlookups = self->ntmisses = 0;
 	Py_RETURN_NONE;
 }
 
 static PyObject *index_stats(indexObject *self)
 {
 	PyObject *obj = PyDict_New();
 	PyObject *t = NULL;
 
 	if (obj == NULL)
 		return NULL;
 
 #define istat(__n, __d) \
 	do { \
 		t = PyInt_FromSsize_t(self->__n); \
 		if (!t) \
 			goto bail; \
 		if (PyDict_SetItemString(obj, __d, t) == -1) \
 			goto bail; \
 		Py_DECREF(t); \
 	} while (0)
 
 	if (self->added) {
 		Py_ssize_t len = PyList_GET_SIZE(self->added);
 		t = PyInt_FromSsize_t(len);
 		if (!t)
 			goto bail;
 		if (PyDict_SetItemString(obj, "index entries added", t) == -1)
 			goto bail;
 		Py_DECREF(t);
 	}
 
 	if (self->raw_length != self->length - 1)
 		istat(raw_length, "revs on disk");
 	istat(length, "revs in memory");
 	istat(ntcapacity, "node trie capacity");
 	istat(ntdepth, "node trie depth");
 	istat(ntlength, "node trie count");
 	istat(ntlookups, "node trie lookups");
 	istat(ntmisses, "node trie misses");
 	istat(ntrev, "node trie last rev scanned");
 	istat(ntsplits, "node trie splits");
 
 #undef istat
 
 	return obj;
 
 bail:
 	Py_XDECREF(obj);
 	Py_XDECREF(t);
 	return NULL;
 }
 
 /*
  * When we cache a list, we want to be sure the caller can't mutate
  * the cached copy.
  */
 static PyObject *list_copy(PyObject *list)
 {
 	Py_ssize_t len = PyList_GET_SIZE(list);
 	PyObject *newlist = PyList_New(len);
 	Py_ssize_t i;
 
 	if (newlist == NULL)
 		return NULL;
 
 	for (i = 0; i < len; i++) {
 		PyObject *obj = PyList_GET_ITEM(list, i);
 		Py_INCREF(obj);
 		PyList_SET_ITEM(newlist, i, obj);
 	}
 
 	return newlist;
 }
 
 static int check_filter(PyObject *filter, Py_ssize_t arg) {
 	if (filter) {
 		PyObject *arglist, *result;
 		int isfiltered;
 
 		arglist = Py_BuildValue("(n)", arg);
 		if (!arglist) {
 			return -1;
 		}
 
 		result = PyEval_CallObject(filter, arglist);
 		Py_DECREF(arglist);
 		if (!result) {
 			return -1;
 		}
 
 		/* PyObject_IsTrue returns 1 if true, 0 if false, -1 if error,
 		 * same as this function, so we can just return it directly.*/
 		isfiltered = PyObject_IsTrue(result);
 		Py_DECREF(result);
 		return isfiltered;
 	} else {
 		return 0;
 	}
 }
 
 static Py_ssize_t add_roots_get_min(indexObject *self, PyObject *list,
                                     Py_ssize_t marker, char *phases)
 {
 	PyObject *iter = NULL;
 	PyObject *iter_item = NULL;
 	Py_ssize_t min_idx = index_length(self) + 1;
 	long iter_item_long;
 
 	if (PyList_GET_SIZE(list) != 0) {
 		iter = PyObject_GetIter(list);
 		if (iter == NULL)
 			return -2;
 		while ((iter_item = PyIter_Next(iter)))
 		{
 			iter_item_long = PyInt_AS_LONG(iter_item);
 			Py_DECREF(iter_item);
 			if (iter_item_long < min_idx)
 				min_idx = iter_item_long;
 			phases[iter_item_long] = marker;
 		}
 		Py_DECREF(iter);
 	}
 
 	return min_idx;
 }
 
 static inline void set_phase_from_parents(char *phases, int parent_1,
                                           int parent_2, Py_ssize_t i)
 {
 	if (parent_1 >= 0 && phases[parent_1] > phases[i])
 		phases[i] = phases[parent_1];
 	if (parent_2 >= 0 && phases[parent_2] > phases[i])
 		phases[i] = phases[parent_2];
 }
 
 static PyObject *reachableroots2(indexObject *self, PyObject *args)
 {
 
 	/* Input */
 	long minroot;
 	PyObject *includepatharg = NULL;
 	int includepath = 0;
 	/* heads and roots are lists */
 	PyObject *heads = NULL;
 	PyObject *roots = NULL;
 	PyObject *reachable = NULL;
 
 	PyObject *val;
 	Py_ssize_t len = index_length(self) - 1;
 	long revnum;
 	Py_ssize_t k;
 	Py_ssize_t i;
 	Py_ssize_t l;
 	int r;
 	int parents[2];
 
 	/* Internal data structure:
 	 * tovisit: array of length len+1 (all revs + nullrev), filled upto lentovisit
 	 * revstates: array of length len+1 (all revs + nullrev) */
 	int *tovisit = NULL;
 	long lentovisit = 0;
 	enum { RS_SEEN = 1, RS_ROOT = 2, RS_REACHABLE = 4 };
 	char *revstates = NULL;
 
 	/* Get arguments */
 	if (!PyArg_ParseTuple(args, "lO!O!O!", &minroot, &PyList_Type, &heads,
 			      &PyList_Type, &roots,
 			      &PyBool_Type, &includepatharg))
 		goto bail;
 
 	if (includepatharg == Py_True)
 		includepath = 1;
 
 	/* Initialize return set */
 	reachable = PyList_New(0);
 	if (reachable == NULL)
 		goto bail;
 
 	/* Initialize internal datastructures */
 	tovisit = (int *)malloc((len + 1) * sizeof(int));
 	if (tovisit == NULL) {
 		PyErr_NoMemory();
 		goto bail;
 	}
 
 	revstates = (char *)calloc(len + 1, 1);
 	if (revstates == NULL) {
 		PyErr_NoMemory();
 		goto bail;
 	}
 
 	l = PyList_GET_SIZE(roots);
 	for (i = 0; i < l; i++) {
 		revnum = PyInt_AsLong(PyList_GET_ITEM(roots, i));
 		if (revnum == -1 && PyErr_Occurred())
 			goto bail;
 		/* If root is out of range, e.g. wdir(), it must be unreachable
 		 * from heads. So we can just ignore it. */
 		if (revnum + 1 < 0 || revnum + 1 >= len + 1)
 			continue;
 		revstates[revnum + 1] |= RS_ROOT;
 	}
 
 	/* Populate tovisit with all the heads */
 	l = PyList_GET_SIZE(heads);
 	for (i = 0; i < l; i++) {
 		revnum = PyInt_AsLong(PyList_GET_ITEM(heads, i));
 		if (revnum == -1 && PyErr_Occurred())
 			goto bail;
 		if (revnum + 1 < 0 || revnum + 1 >= len + 1) {
 			PyErr_SetString(PyExc_IndexError, "head out of range");
 			goto bail;
 		}
 		if (!(revstates[revnum + 1] & RS_SEEN)) {
 			tovisit[lentovisit++] = (int)revnum;
 			revstates[revnum + 1] |= RS_SEEN;
 		}
 	}
 
 	/* Visit the tovisit list and find the reachable roots */
 	k = 0;
 	while (k < lentovisit) {
 		/* Add the node to reachable if it is a root*/
 		revnum = tovisit[k++];
 		if (revstates[revnum + 1] & RS_ROOT) {
 			revstates[revnum + 1] |= RS_REACHABLE;
 			val = PyInt_FromLong(revnum);
 			if (val == NULL)
 				goto bail;
 			r = PyList_Append(reachable, val);
 			Py_DECREF(val);
 			if (r < 0)
 				goto bail;
 			if (includepath == 0)
 				continue;
 		}
 
 		/* Add its parents to the list of nodes to visit */
 		if (revnum == -1)
 			continue;
 		r = index_get_parents(self, revnum, parents, (int)len - 1);
 		if (r < 0)
 			goto bail;
 		for (i = 0; i < 2; i++) {
 			if (!(revstates[parents[i] + 1] & RS_SEEN)
 			    && parents[i] >= minroot) {
 				tovisit[lentovisit++] = parents[i];
 				revstates[parents[i] + 1] |= RS_SEEN;
 			}
 		}
 	}
 
 	/* Find all the nodes in between the roots we found and the heads
 	 * and add them to the reachable set */
 	if (includepath == 1) {
 		long minidx = minroot;
 		if (minidx < 0)
 			minidx = 0;
 		for (i = minidx; i < len; i++) {
 			if (!(revstates[i + 1] & RS_SEEN))
 				continue;
 			r = index_get_parents(self, i, parents, (int)len - 1);
 			/* Corrupted index file, error is set from
 			 * index_get_parents */
 			if (r < 0)
 				goto bail;
 			if (((revstates[parents[0] + 1] |
 			      revstates[parents[1] + 1]) & RS_REACHABLE)
 			    && !(revstates[i + 1] & RS_REACHABLE)) {
 				revstates[i + 1] |= RS_REACHABLE;
 				val = PyInt_FromLong(i);
 				if (val == NULL)
 					goto bail;
 				r = PyList_Append(reachable, val);
 				Py_DECREF(val);
 				if (r < 0)
 					goto bail;
 			}
 		}
 	}
 
 	free(revstates);
 	free(tovisit);
 	return reachable;
 bail:
 	Py_XDECREF(reachable);
 	free(revstates);
 	free(tovisit);
 	return NULL;
 }
 
 static PyObject *compute_phases_map_sets(indexObject *self, PyObject *args)
 {
 	PyObject *roots = Py_None;
 	PyObject *ret = NULL;
 	PyObject *phaseslist = NULL;
 	PyObject *phaseroots = NULL;
 	PyObject *phaseset = NULL;
 	PyObject *phasessetlist = NULL;
 	PyObject *rev = NULL;
 	Py_ssize_t len = index_length(self) - 1;
 	Py_ssize_t numphase = 0;
 	Py_ssize_t minrevallphases = 0;
 	Py_ssize_t minrevphase = 0;
 	Py_ssize_t i = 0;
 	char *phases = NULL;
 	long phase;
 
 	if (!PyArg_ParseTuple(args, "O", &roots))
 		goto done;
 	if (roots == NULL || !PyList_Check(roots))
 		goto done;
 
 	phases = calloc(len, 1); /* phase per rev: {0: public, 1: draft, 2: secret} */
 	if (phases == NULL) {
 		PyErr_NoMemory();
 		goto done;
 	}
 	/* Put the phase information of all the roots in phases */
 	numphase = PyList_GET_SIZE(roots)+1;
 	minrevallphases = len + 1;
 	phasessetlist = PyList_New(numphase);
 	if (phasessetlist == NULL)
 		goto done;
 
 	PyList_SET_ITEM(phasessetlist, 0, Py_None);
 	Py_INCREF(Py_None);
 
 	for (i = 0; i < numphase-1; i++) {
 		phaseroots = PyList_GET_ITEM(roots, i);
 		phaseset = PySet_New(NULL);
 		if (phaseset == NULL)
 			goto release;
 		PyList_SET_ITEM(phasessetlist, i+1, phaseset);
 		if (!PyList_Check(phaseroots))
 			goto release;
 		minrevphase = add_roots_get_min(self, phaseroots, i+1, phases);
 		if (minrevphase == -2) /* Error from add_roots_get_min */
 			goto release;
 		minrevallphases = MIN(minrevallphases, minrevphase);
 	}
 	/* Propagate the phase information from the roots to the revs */
 	if (minrevallphases != -1) {
 		int parents[2];
 		for (i = minrevallphases; i < len; i++) {
 			if (index_get_parents(self, i, parents,
 					      (int)len - 1) < 0)
 				goto release;
 			set_phase_from_parents(phases, parents[0], parents[1], i);
 		}
 	}
 	/* Transform phase list to a python list */
 	phaseslist = PyList_New(len);
 	if (phaseslist == NULL)
 		goto release;
 	for (i = 0; i < len; i++) {
 		PyObject *phaseval;
 
 		phase = phases[i];
 		/* We only store the sets of phase for non public phase, the public phase
 		 * is computed as a difference */
 		if (phase != 0) {
 			phaseset = PyList_GET_ITEM(phasessetlist, phase);
 			rev = PyInt_FromLong(i);
 			if (rev == NULL)
 				goto release;
 			PySet_Add(phaseset, rev);
 			Py_XDECREF(rev);
 		}
 		phaseval = PyInt_FromLong(phase);
 		if (phaseval == NULL)
 			goto release;
 		PyList_SET_ITEM(phaseslist, i, phaseval);
 	}
 	ret = PyTuple_Pack(2, phaseslist, phasessetlist);
 
 release:
 	Py_XDECREF(phaseslist);
 	Py_XDECREF(phasessetlist);
 done:
 	free(phases);
 	return ret;
 }
 
 static PyObject *index_headrevs(indexObject *self, PyObject *args)
 {
 	Py_ssize_t i, j, len;
 	char *nothead = NULL;
 	PyObject *heads = NULL;
 	PyObject *filter = NULL;
 	PyObject *filteredrevs = Py_None;
 
 	if (!PyArg_ParseTuple(args, "|O", &filteredrevs)) {
 		return NULL;
 	}
 
 	if (self->headrevs && filteredrevs == self->filteredrevs)
 		return list_copy(self->headrevs);
 
 	Py_DECREF(self->filteredrevs);
 	self->filteredrevs = filteredrevs;
 	Py_INCREF(filteredrevs);
 
 	if (filteredrevs != Py_None) {
 		filter = PyObject_GetAttrString(filteredrevs, "__contains__");
 		if (!filter) {
 			PyErr_SetString(PyExc_TypeError,
 				"filteredrevs has no attribute __contains__");
 			goto bail;
 		}
 	}
 
 	len = index_length(self) - 1;
 	heads = PyList_New(0);
 	if (heads == NULL)
 		goto bail;
 	if (len == 0) {
 		PyObject *nullid = PyInt_FromLong(-1);
 		if (nullid == NULL || PyList_Append(heads, nullid) == -1) {
 			Py_XDECREF(nullid);
 			goto bail;
 		}
 		goto done;
 	}
 
 	nothead = calloc(len, 1);
 	if (nothead == NULL) {
 		PyErr_NoMemory();
 		goto bail;
 	}
 
 	for (i = len - 1; i >= 0; i--) {
 		int isfiltered;
 		int parents[2];
 
 		/* If nothead[i] == 1, it means we've seen an unfiltered child of this
 		 * node already, and therefore this node is not filtered. So we can skip
 		 * the expensive check_filter step.
 		 */
 		if (nothead[i] != 1) {
 			isfiltered = check_filter(filter, i);
 			if (isfiltered == -1) {
 				PyErr_SetString(PyExc_TypeError,
 					"unable to check filter");
 				goto bail;
 			}
 
 			if (isfiltered) {
 				nothead[i] = 1;
 				continue;
 			}
 		}
 
 		if (index_get_parents(self, i, parents, (int)len - 1) < 0)
 			goto bail;
 		for (j = 0; j < 2; j++) {
 			if (parents[j] >= 0)
 				nothead[parents[j]] = 1;
 		}
 	}
 
 	for (i = 0; i < len; i++) {
 		PyObject *head;
 
 		if (nothead[i])
 			continue;
 		head = PyInt_FromSsize_t(i);
 		if (head == NULL || PyList_Append(heads, head) == -1) {
 			Py_XDECREF(head);
 			goto bail;
 		}
 	}
 
 done:
 	self->headrevs = heads;
 	Py_XDECREF(filter);
 	free(nothead);
 	return list_copy(self->headrevs);
 bail:
 	Py_XDECREF(filter);
 	Py_XDECREF(heads);
 	free(nothead);
 	return NULL;
 }
 
 /**
  * Obtain the base revision index entry.
  *
  * Callers must ensure that rev >= 0 or illegal memory access may occur.
  */
 static inline int index_baserev(indexObject *self, int rev)
 {
 	const char *data;
 
 	if (rev >= self->length - 1) {
 		PyObject *tuple = PyList_GET_ITEM(self->added,
 			rev - self->length + 1);
 		return (int)PyInt_AS_LONG(PyTuple_GET_ITEM(tuple, 3));
 	}
 	else {
 		data = index_deref(self, rev);
 		if (data == NULL) {
 			return -2;
 		}
 
 		return getbe32(data + 16);
 	}
 }
 
 static PyObject *index_deltachain(indexObject *self, PyObject *args)
 {
 	int rev, generaldelta;
 	PyObject *stoparg;
 	int stoprev, iterrev, baserev = -1;
 	int stopped;
 	PyObject *chain = NULL, *result = NULL;
 	const Py_ssize_t length = index_length(self);
 
 	if (!PyArg_ParseTuple(args, "iOi", &rev, &stoparg, &generaldelta)) {
 		return NULL;
 	}
 
 	if (PyInt_Check(stoparg)) {
 		stoprev = (int)PyInt_AsLong(stoparg);
 		if (stoprev == -1 && PyErr_Occurred()) {
 			return NULL;
 		}
 	}
 	else if (stoparg == Py_None) {
 		stoprev = -2;
 	}
 	else {
 		PyErr_SetString(PyExc_ValueError,
 			"stoprev must be integer or None");
 		return NULL;
 	}
 
 	if (rev < 0 || rev >= length - 1) {
 		PyErr_SetString(PyExc_ValueError, "revlog index out of range");
 		return NULL;
 	}
 
 	chain = PyList_New(0);
 	if (chain == NULL) {
 		return NULL;
 	}
 
 	baserev = index_baserev(self, rev);
 
 	/* This should never happen. */
 	if (baserev <= -2) {
 		/* Error should be set by index_deref() */
 		assert(PyErr_Occurred());
 		goto bail;
 	}
 
 	iterrev = rev;
 
 	while (iterrev != baserev && iterrev != stoprev) {
 		PyObject *value = PyInt_FromLong(iterrev);
 		if (value == NULL) {
 			goto bail;
 		}
 		if (PyList_Append(chain, value)) {
 			Py_DECREF(value);
 			goto bail;
 		}
 		Py_DECREF(value);
 
 		if (generaldelta) {
 			iterrev = baserev;
 		}
 		else {
 			iterrev--;
 		}
 
 		if (iterrev < 0) {
 			break;
 		}
 
 		if (iterrev >= length - 1) {
 			PyErr_SetString(PyExc_IndexError, "revision outside index");
 			return NULL;
 		}
 
 		baserev = index_baserev(self, iterrev);
 
 		/* This should never happen. */
 		if (baserev <= -2) {
 			/* Error should be set by index_deref() */
 			assert(PyErr_Occurred());
 			goto bail;
 		}
 	}
 
 	if (iterrev == stoprev) {
 		stopped = 1;
 	}
 	else {
 		PyObject *value = PyInt_FromLong(iterrev);
 		if (value == NULL) {
 			goto bail;
 		}
 		if (PyList_Append(chain, value)) {
 			Py_DECREF(value);
 			goto bail;
 		}
 		Py_DECREF(value);
 
 		stopped = 0;
 	}
 
 	if (PyList_Reverse(chain)) {
 		goto bail;
 	}
 
 	result = Py_BuildValue("OO", chain, stopped ? Py_True : Py_False);
 	Py_DECREF(chain);
 	return result;
 
 bail:
 	Py_DECREF(chain);
 	return NULL;
 }
 
 static inline int nt_level(const char *node, Py_ssize_t level)
 {
 	int v = node[level>>1];
 	if (!(level & 1))
 		v >>= 4;
 	return v & 0xf;
 }
 
 /*
  * Return values:
  *
  *   -4: match is ambiguous (multiple candidates)
  *   -2: not found
  * rest: valid rev
  */
 static int nt_find(indexObject *self, const char *node, Py_ssize_t nodelen,
 		   int hex)
 {
 	int (*getnybble)(const char *, Py_ssize_t) = hex ? hexdigit : nt_level;
 	int level, maxlevel, off;
 
 	if (nodelen == 20 && node[0] == '\0' && memcmp(node, nullid, 20) == 0)
 		return -1;
 
 	if (self->nt == NULL)
 		return -2;
 
 	if (hex)
 		maxlevel = nodelen > 40 ? 40 : (int)nodelen;
 	else
 		maxlevel = nodelen > 20 ? 40 : ((int)nodelen * 2);
 
 	for (level = off = 0; level < maxlevel; level++) {
 		int k = getnybble(node, level);
 		nodetree *n = &self->nt[off];
 		int v = n->children[k];
 
 		if (v < 0) {
 			const char *n;
 			Py_ssize_t i;
 
 			v = -(v + 1);
 			n = index_node(self, v);
 			if (n == NULL)
 				return -2;
 			for (i = level; i < maxlevel; i++)
 				if (getnybble(node, i) != nt_level(n, i))
 					return -2;
 			return v;
 		}
 		if (v == 0)
 			return -2;
 		off = v;
 	}
 	/* multiple matches against an ambiguous prefix */
 	return -4;
 }
 
 static int nt_new(indexObject *self)
 {
 	if (self->ntlength == self->ntcapacity) {
 		if (self->ntcapacity >= INT_MAX / (sizeof(nodetree) * 2)) {
 			PyErr_SetString(PyExc_MemoryError,
 					"overflow in nt_new");
 			return -1;
 		}
 		self->ntcapacity *= 2;
 		self->nt = realloc(self->nt,
 				   self->ntcapacity * sizeof(nodetree));
 		if (self->nt == NULL) {
 			PyErr_SetString(PyExc_MemoryError, "out of memory");
 			return -1;
 		}
 		memset(&self->nt[self->ntlength], 0,
 		       sizeof(nodetree) * (self->ntcapacity - self->ntlength));
 	}
 	return self->ntlength++;
 }
 
 static int nt_insert(indexObject *self, const char *node, int rev)
 {
 	int level = 0;
 	int off = 0;
 
 	while (level < 40) {
 		int k = nt_level(node, level);
 		nodetree *n;
 		int v;
 
 		n = &self->nt[off];
 		v = n->children[k];
 
 		if (v == 0) {
 			n->children[k] = -rev - 1;
 			return 0;
 		}
 		if (v < 0) {
 			const char *oldnode = index_node(self, -(v + 1));
 			int noff;
 
 			if (!oldnode || !memcmp(oldnode, node, 20)) {
 				n->children[k] = -rev - 1;
 				return 0;
 			}
 			noff = nt_new(self);
 			if (noff == -1)
 				return -1;
 			/* self->nt may have been changed by realloc */
 			self->nt[off].children[k] = noff;
 			off = noff;
 			n = &self->nt[off];
 			n->children[nt_level(oldnode, ++level)] = v;
 			if (level > self->ntdepth)
 				self->ntdepth = level;
 			self->ntsplits += 1;
 		} else {
 			level += 1;
 			off = v;
 		}
 	}
 
 	return -1;
 }
 
 static int nt_init(indexObject *self)
 {
 	if (self->nt == NULL) {
 		if ((size_t)self->raw_length > INT_MAX / sizeof(nodetree)) {
 			PyErr_SetString(PyExc_ValueError, "overflow in nt_init");
 			return -1;
 		}
 		self->ntcapacity = self->raw_length < 4
 			? 4 : (int)self->raw_length / 2;
 
 		self->nt = calloc(self->ntcapacity, sizeof(nodetree));
 		if (self->nt == NULL) {
 			PyErr_NoMemory();
 			return -1;
 		}
 		self->ntlength = 1;
 		self->ntrev = (int)index_length(self) - 1;
 		self->ntlookups = 1;
 		self->ntmisses = 0;
 		if (nt_insert(self, nullid, INT_MAX) == -1)
 			return -1;
 	}
 	return 0;
 }
 
 /*
  * Return values:
  *
  *   -3: error (exception set)
  *   -2: not found (no exception set)
  * rest: valid rev
  */
 static int index_find_node(indexObject *self,
 			   const char *node, Py_ssize_t nodelen)
 {
 	int rev;
 
 	self->ntlookups++;
 	rev = nt_find(self, node, nodelen, 0);
 	if (rev >= -1)
 		return rev;
 
 	if (nt_init(self) == -1)
 		return -3;
 
 	/*
 	 * For the first handful of lookups, we scan the entire index,
 	 * and cache only the matching nodes. This optimizes for cases
 	 * like "hg tip", where only a few nodes are accessed.
 	 *
 	 * After that, we cache every node we visit, using a single
 	 * scan amortized over multiple lookups.  This gives the best
 	 * bulk performance, e.g. for "hg log".
 	 */
 	if (self->ntmisses++ < 4) {
 		for (rev = self->ntrev - 1; rev >= 0; rev--) {
 			const char *n = index_node(self, rev);
 			if (n == NULL)
 				return -2;
 			if (memcmp(node, n, nodelen > 20 ? 20 : nodelen) == 0) {
 				if (nt_insert(self, n, rev) == -1)
 					return -3;
 				break;
 			}
 		}
 	} else {
 		for (rev = self->ntrev - 1; rev >= 0; rev--) {
 			const char *n = index_node(self, rev);
 			if (n == NULL) {
 				self->ntrev = rev + 1;
 				return -2;
 			}
 			if (nt_insert(self, n, rev) == -1) {
 				self->ntrev = rev + 1;
 				return -3;
 			}
 			if (memcmp(node, n, nodelen > 20 ? 20 : nodelen) == 0) {
 				break;
 			}
 		}
 		self->ntrev = rev;
 	}
 
 	if (rev >= 0)
 		return rev;
 	return -2;
 }
 
 static void raise_revlog_error(void)
 {
 	PyObject *mod = NULL, *dict = NULL, *errclass = NULL;
 
 	mod = PyImport_ImportModule("mercurial.error");
 	if (mod == NULL) {
 		goto cleanup;
 	}
 
 	dict = PyModule_GetDict(mod);
 	if (dict == NULL) {
 		goto cleanup;
 	}
 	Py_INCREF(dict);
 
 	errclass = PyDict_GetItemString(dict, "RevlogError");
 	if (errclass == NULL) {
 		PyErr_SetString(PyExc_SystemError,
 				"could not find RevlogError");
 		goto cleanup;
 	}
 
 	/* value of exception is ignored by callers */
 	PyErr_SetString(errclass, "RevlogError");
 
 cleanup:
 	Py_XDECREF(dict);
 	Py_XDECREF(mod);
 }
 
 static PyObject *index_getitem(indexObject *self, PyObject *value)
 {
 	char *node;
 	Py_ssize_t nodelen;
 	int rev;
 
 	if (PyInt_Check(value))
 		return index_get(self, PyInt_AS_LONG(value));
 
 	if (node_check(value, &node, &nodelen) == -1)
 		return NULL;
 	rev = index_find_node(self, node, nodelen);
 	if (rev >= -1)
 		return PyInt_FromLong(rev);
 	if (rev == -2)
 		raise_revlog_error();
 	return NULL;
 }
 
 static int nt_partialmatch(indexObject *self, const char *node,
 			   Py_ssize_t nodelen)
 {
 	int rev;
 
 	if (nt_init(self) == -1)
 		return -3;
 
 	if (self->ntrev > 0) {
 		/* ensure that the radix tree is fully populated */
 		for (rev = self->ntrev - 1; rev >= 0; rev--) {
 			const char *n = index_node(self, rev);
 			if (n == NULL)
 				return -2;
 			if (nt_insert(self, n, rev) == -1)
 				return -3;
 		}
 		self->ntrev = rev;
 	}
 
 	return nt_find(self, node, nodelen, 1);
 }
 
 static PyObject *index_partialmatch(indexObject *self, PyObject *args)
 {
 	const char *fullnode;
 	int nodelen;
 	char *node;
 	int rev, i;
 
 	if (!PyArg_ParseTuple(args, "s#", &node, &nodelen))
 		return NULL;
 
 	if (nodelen < 4) {
 		PyErr_SetString(PyExc_ValueError, "key too short");
 		return NULL;
 	}
 
 	if (nodelen > 40) {
 		PyErr_SetString(PyExc_ValueError, "key too long");
 		return NULL;
 	}
 
 	for (i = 0; i < nodelen; i++)
 		hexdigit(node, i);
 	if (PyErr_Occurred()) {
 		/* input contains non-hex characters */
 		PyErr_Clear();
 		Py_RETURN_NONE;
 	}
 
 	rev = nt_partialmatch(self, node, nodelen);
 
 	switch (rev) {
 	case -4:
 		raise_revlog_error();
 	case -3:
 		return NULL;
 	case -2:
 		Py_RETURN_NONE;
 	case -1:
 		return PyBytes_FromStringAndSize(nullid, 20);
 	}
 
 	fullnode = index_node(self, rev);
 	if (fullnode == NULL) {
 		PyErr_Format(PyExc_IndexError,
 			     "could not access rev %d", rev);
 		return NULL;
 	}
 	return PyBytes_FromStringAndSize(fullnode, 20);
 }
 
 static PyObject *index_m_get(indexObject *self, PyObject *args)
 {
 	Py_ssize_t nodelen;
 	PyObject *val;
 	char *node;
 	int rev;
 
 	if (!PyArg_ParseTuple(args, "O", &val))
 		return NULL;
 	if (node_check(val, &node, &nodelen) == -1)
 		return NULL;
 	rev = index_find_node(self, node, nodelen);
 	if (rev == -3)
 		return NULL;
 	if (rev == -2)
 		Py_RETURN_NONE;
 	return PyInt_FromLong(rev);
 }
 
 static int index_contains(indexObject *self, PyObject *value)
 {
 	char *node;
 	Py_ssize_t nodelen;
 
 	if (PyInt_Check(value)) {
 		long rev = PyInt_AS_LONG(value);
 		return rev >= -1 && rev < index_length(self);
 	}
 
 	if (node_check(value, &node, &nodelen) == -1)
 		return -1;
 
 	switch (index_find_node(self, node, nodelen)) {
 	case -3:
 		return -1;
 	case -2:
 		return 0;
 	default:
 		return 1;
 	}
 }
 
 typedef uint64_t bitmask;
 
 /*
  * Given a disjoint set of revs, return all candidates for the
  * greatest common ancestor. In revset notation, this is the set
  * "heads(::a and ::b and ...)"
  */
 static PyObject *find_gca_candidates(indexObject *self, const int *revs,
 				     int revcount)
 {
 	const bitmask allseen = (1ull << revcount) - 1;
 	const bitmask poison = 1ull << revcount;
 	PyObject *gca = PyList_New(0);
 	int i, v, interesting;
 	int maxrev = -1;
 	bitmask sp;
 	bitmask *seen;
 
 	if (gca == NULL)
 		return PyErr_NoMemory();
 
 	for (i = 0; i < revcount; i++) {
 		if (revs[i] > maxrev)
 			maxrev = revs[i];
 	}
 
 	seen = calloc(sizeof(*seen), maxrev + 1);
 	if (seen == NULL) {
 		Py_DECREF(gca);
 		return PyErr_NoMemory();
 	}
 
 	for (i = 0; i < revcount; i++)
 		seen[revs[i]] = 1ull << i;
 
 	interesting = revcount;
 
 	for (v = maxrev; v >= 0 && interesting; v--) {
 		bitmask sv = seen[v];
 		int parents[2];
 
 		if (!sv)
 			continue;
 
 		if (sv < poison) {
 			interesting -= 1;
 			if (sv == allseen) {
 				PyObject *obj = PyInt_FromLong(v);
 				if (obj == NULL)
 					goto bail;
 				if (PyList_Append(gca, obj) == -1) {
 					Py_DECREF(obj);
 					goto bail;
 				}
 				sv |= poison;
 				for (i = 0; i < revcount; i++) {
 					if (revs[i] == v)
 						goto done;
 				}
 			}
 		}
 		if (index_get_parents(self, v, parents, maxrev) < 0)
 			goto bail;
 
 		for (i = 0; i < 2; i++) {
 			int p = parents[i];
 			if (p == -1)
 				continue;
 			sp = seen[p];
 			if (sv < poison) {
 				if (sp == 0) {
 					seen[p] = sv;
 					interesting++;
 				}
 				else if (sp != sv)
 					seen[p] |= sv;
 			} else {
 				if (sp && sp < poison)
 					interesting--;
 				seen[p] = sv;
 			}
 		}
 	}
 
 done:
 	free(seen);
 	return gca;
 bail:
 	free(seen);
 	Py_XDECREF(gca);
 	return NULL;
 }
 
 /*
  * Given a disjoint set of revs, return the subset with the longest
  * path to the root.
  */
 static PyObject *find_deepest(indexObject *self, PyObject *revs)
 {
 	const Py_ssize_t revcount = PyList_GET_SIZE(revs);
 	static const Py_ssize_t capacity = 24;
 	int *depth, *interesting = NULL;
 	int i, j, v, ninteresting;
 	PyObject *dict = NULL, *keys = NULL;
 	long *seen = NULL;
 	int maxrev = -1;
 	long final;
 
 	if (revcount > capacity) {
 		PyErr_Format(PyExc_OverflowError,
 			     "bitset size (%ld) > capacity (%ld)",
 			     (long)revcount, (long)capacity);
 		return NULL;
 	}
 
 	for (i = 0; i < revcount; i++) {
 		int n = (int)PyInt_AsLong(PyList_GET_ITEM(revs, i));
 		if (n > maxrev)
 			maxrev = n;
 	}
 
 	depth = calloc(sizeof(*depth), maxrev + 1);
 	if (depth == NULL)
 		return PyErr_NoMemory();
 
 	seen = calloc(sizeof(*seen), maxrev + 1);
 	if (seen == NULL) {
 		PyErr_NoMemory();
 		goto bail;
 	}
 
 	interesting = calloc(sizeof(*interesting), 1 << revcount);
 	if (interesting == NULL) {
 		PyErr_NoMemory();
 		goto bail;
 	}
 
 	if (PyList_Sort(revs) == -1)
 		goto bail;
 
 	for (i = 0; i < revcount; i++) {
 		int n = (int)PyInt_AsLong(PyList_GET_ITEM(revs, i));
 		long b = 1l << i;
 		depth[n] = 1;
 		seen[n] = b;
 		interesting[b] = 1;
 	}
 
 	/* invariant: ninteresting is the number of non-zero entries in
 	 * interesting. */
 	ninteresting = (int)revcount;
 
 	for (v = maxrev; v >= 0 && ninteresting > 1; v--) {
 		int dv = depth[v];
 		int parents[2];
 		long sv;
 
 		if (dv == 0)
 			continue;
 
 		sv = seen[v];
 		if (index_get_parents(self, v, parents, maxrev) < 0)
 			goto bail;
 
 		for (i = 0; i < 2; i++) {
 			int p = parents[i];
 			long sp;
 			int dp;
 
 			if (p == -1)
 				continue;
 
 			dp = depth[p];
 			sp = seen[p];
 			if (dp <= dv) {
 				depth[p] = dv + 1;
 				if (sp != sv) {
 					interesting[sv] += 1;
 					seen[p] = sv;
 					if (sp) {
 						interesting[sp] -= 1;
 						if (interesting[sp] == 0)
 							ninteresting -= 1;
 					}
 				}
 			}
 			else if (dv == dp - 1) {
 				long nsp = sp | sv;
 				if (nsp == sp)
 					continue;
 				seen[p] = nsp;
 				interesting[sp] -= 1;
 				if (interesting[sp] == 0)
 					ninteresting -= 1;
 				if (interesting[nsp] == 0)
 					ninteresting += 1;
 				interesting[nsp] += 1;
 			}
 		}
 		interesting[sv] -= 1;
 		if (interesting[sv] == 0)
 			ninteresting -= 1;
 	}
 
 	final = 0;
 	j = ninteresting;
 	for (i = 0; i < (int)(2 << revcount) && j > 0; i++) {
 		if (interesting[i] == 0)
 			continue;
 		final |= i;
 		j -= 1;
 	}
 	if (final == 0) {
 		keys = PyList_New(0);
 		goto bail;
 	}
 
 	dict = PyDict_New();
 	if (dict == NULL)
 		goto bail;
 
 	for (i = 0; i < revcount; i++) {
 		PyObject *key;
 
 		if ((final & (1 << i)) == 0)
 			continue;
 
 		key = PyList_GET_ITEM(revs, i);
 		Py_INCREF(key);
 		Py_INCREF(Py_None);
 		if (PyDict_SetItem(dict, key, Py_None) == -1) {
 			Py_DECREF(key);
 			Py_DECREF(Py_None);
 			goto bail;
 		}
 	}
 
 	keys = PyDict_Keys(dict);
 
 bail:
 	free(depth);
 	free(seen);
 	free(interesting);
 	Py_XDECREF(dict);
 
 	return keys;
 }
 
 /*
  * Given a (possibly overlapping) set of revs, return all the
  * common ancestors heads: heads(::args[0] and ::a[1] and ...)
  */
 static PyObject *index_commonancestorsheads(indexObject *self, PyObject *args)
 {
 	PyObject *ret = NULL;
 	Py_ssize_t argcount, i, len;
 	bitmask repeat = 0;
 	int revcount = 0;
 	int *revs;
 
 	argcount = PySequence_Length(args);
 	revs = PyMem_Malloc(argcount * sizeof(*revs));
 	if (argcount > 0 && revs == NULL)
 		return PyErr_NoMemory();
 	len = index_length(self) - 1;
 
 	for (i = 0; i < argcount; i++) {
 		static const int capacity = 24;
 		PyObject *obj = PySequence_GetItem(args, i);
 		bitmask x;
 		long val;
 
 		if (!PyInt_Check(obj)) {
 			PyErr_SetString(PyExc_TypeError,
 					"arguments must all be ints");
 			Py_DECREF(obj);
 			goto bail;
 		}
 		val = PyInt_AsLong(obj);
 		Py_DECREF(obj);
 		if (val == -1) {
 			ret = PyList_New(0);
 			goto done;
 		}
 		if (val < 0 || val >= len) {
 			PyErr_SetString(PyExc_IndexError,
 					"index out of range");
 			goto bail;
 		}
 		/* this cheesy bloom filter lets us avoid some more
 		 * expensive duplicate checks in the common set-is-disjoint
 		 * case */
 		x = 1ull << (val & 0x3f);
 		if (repeat & x) {
 			int k;
 			for (k = 0; k < revcount; k++) {
 				if (val == revs[k])
 					goto duplicate;
 			}
 		}
 		else repeat |= x;
 		if (revcount >= capacity) {
 			PyErr_Format(PyExc_OverflowError,
 				     "bitset size (%d) > capacity (%d)",
 				     revcount, capacity);
 			goto bail;
 		}
 		revs[revcount++] = (int)val;
 	duplicate:;
 	}
 
 	if (revcount == 0) {
 		ret = PyList_New(0);
 		goto done;
 	}
 	if (revcount == 1) {
 		PyObject *obj;
 		ret = PyList_New(1);
 		if (ret == NULL)
 			goto bail;
 		obj = PyInt_FromLong(revs[0]);
 		if (obj == NULL)
 			goto bail;
 		PyList_SET_ITEM(ret, 0, obj);
 		goto done;
 	}
 
 	ret = find_gca_candidates(self, revs, revcount);
 	if (ret == NULL)
 		goto bail;
 
 done:
 	PyMem_Free(revs);
 	return ret;
 
 bail:
 	PyMem_Free(revs);
 	Py_XDECREF(ret);
 	return NULL;
 }
 
 /*
  * Given a (possibly overlapping) set of revs, return the greatest
  * common ancestors: those with the longest path to the root.
  */
 static PyObject *index_ancestors(indexObject *self, PyObject *args)
 {
 	PyObject *ret;
 	PyObject *gca = index_commonancestorsheads(self, args);
 	if (gca == NULL)
 		return NULL;
 
 	if (PyList_GET_SIZE(gca) <= 1) {
 		return gca;
 	}
 
 	ret = find_deepest(self, gca);
 	Py_DECREF(gca);
 	return ret;
 }
 
 /*
  * Invalidate any trie entries introduced by added revs.
  */
 static void nt_invalidate_added(indexObject *self, Py_ssize_t start)
 {
 	Py_ssize_t i, len = PyList_GET_SIZE(self->added);
 
 	for (i = start; i < len; i++) {
 		PyObject *tuple = PyList_GET_ITEM(self->added, i);
 		PyObject *node = PyTuple_GET_ITEM(tuple, 7);
 
 		nt_insert(self, PyBytes_AS_STRING(node), -1);
 	}
 
 	if (start == 0)
 		Py_CLEAR(self->added);
 }
 
 /*
  * Delete a numeric range of revs, which must be at the end of the
  * range, but exclude the sentinel nullid entry.
  */
 static int index_slice_del(indexObject *self, PyObject *item)
 {
 	Py_ssize_t start, stop, step, slicelength;
 	Py_ssize_t length = index_length(self);
 	int ret = 0;
 
 /* Argument changed from PySliceObject* to PyObject* in Python 3. */
 #ifdef IS_PY3K
 	if (PySlice_GetIndicesEx(item, length,
 #else
 	if (PySlice_GetIndicesEx((PySliceObject*)item, length,
 #endif
 				 &start, &stop, &step, &slicelength) < 0)
 		return -1;
 
 	if (slicelength <= 0)
 		return 0;
 
 	if ((step < 0 && start < stop) || (step > 0 && start > stop))
 		stop = start;
 
 	if (step < 0) {
 		stop = start + 1;
 		start = stop + step*(slicelength - 1) - 1;
 		step = -step;
 	}
 
 	if (step != 1) {
 		PyErr_SetString(PyExc_ValueError,
 				"revlog index delete requires step size of 1");
 		return -1;
 	}
 
 	if (stop != length - 1) {
 		PyErr_SetString(PyExc_IndexError,
 				"revlog index deletion indices are invalid");
 		return -1;
 	}
 
 	if (start < self->length - 1) {
 		if (self->nt) {
 			Py_ssize_t i;
 
 			for (i = start + 1; i < self->length - 1; i++) {
 				const char *node = index_node(self, i);
 
 				if (node)
 					nt_insert(self, node, -1);
 			}
 			if (self->added)
 				nt_invalidate_added(self, 0);
 			if (self->ntrev > start)
 				self->ntrev = (int)start;
 		}
 		self->length = start + 1;
 		if (start < self->raw_length) {
 			if (self->cache) {
 				Py_ssize_t i;
 				for (i = start; i < self->raw_length; i++)
 					Py_CLEAR(self->cache[i]);
 			}
 			self->raw_length = start;
 		}
 		goto done;
 	}
 
 	if (self->nt) {
 		nt_invalidate_added(self, start - self->length + 1);
 		if (self->ntrev > start)
 			self->ntrev = (int)start;
 	}
 	if (self->added)
 		ret = PyList_SetSlice(self->added, start - self->length + 1,
 				      PyList_GET_SIZE(self->added), NULL);
 done:
 	Py_CLEAR(self->headrevs);
 	return ret;
 }
 
 /*
  * Supported ops:
  *
  * slice deletion
  * string assignment (extend node->rev mapping)
  * string deletion (shrink node->rev mapping)
  */
 static int index_assign_subscript(indexObject *self, PyObject *item,
 				  PyObject *value)
 {
 	char *node;
 	Py_ssize_t nodelen;
 	long rev;
 
 	if (PySlice_Check(item) && value == NULL)
 		return index_slice_del(self, item);
 
 	if (node_check(item, &node, &nodelen) == -1)
 		return -1;
 
 	if (value == NULL)
 		return self->nt ? nt_insert(self, node, -1) : 0;
 	rev = PyInt_AsLong(value);
 	if (rev > INT_MAX || rev < 0) {
 		if (!PyErr_Occurred())
 			PyErr_SetString(PyExc_ValueError, "rev out of range");
 		return -1;
 	}
 
 	if (nt_init(self) == -1)
 		return -1;
 	return nt_insert(self, node, (int)rev);
 }
 
 /*
  * Find all RevlogNG entries in an index that has inline data. Update
  * the optional "offsets" table with those entries.
  */
 static Py_ssize_t inline_scan(indexObject *self, const char **offsets)
 {
 	const char *data = (const char *)self->buf.buf;
 	Py_ssize_t pos = 0;
 	Py_ssize_t end = self->buf.len;
 	long incr = v1_hdrsize;
 	Py_ssize_t len = 0;
 
 	while (pos + v1_hdrsize <= end && pos >= 0) {
 		uint32_t comp_len;
 		/* 3rd element of header is length of compressed inline data */
 		comp_len = getbe32(data + pos + 8);
 		incr = v1_hdrsize + comp_len;
 		if (offsets)
 			offsets[len] = data + pos;
 		len++;
 		pos += incr;
 	}
 
 	if (pos != end) {
 		if (!PyErr_Occurred())
 			PyErr_SetString(PyExc_ValueError, "corrupt index file");
 		return -1;
 	}
 
 	return len;
 }
 
 static int index_init(indexObject *self, PyObject *args)
 {
 	PyObject *data_obj, *inlined_obj;
 	Py_ssize_t size;
 
 	/* Initialize before argument-checking to avoid index_dealloc() crash. */
 	self->raw_length = 0;
 	self->added = NULL;
 	self->cache = NULL;
 	self->data = NULL;
 	memset(&self->buf, 0, sizeof(self->buf));
 	self->headrevs = NULL;
 	self->filteredrevs = Py_None;
 	Py_INCREF(Py_None);
 	self->nt = NULL;
 	self->offsets = NULL;
 
 	if (!PyArg_ParseTuple(args, "OO", &data_obj, &inlined_obj))
 		return -1;
 	if (!PyObject_CheckBuffer(data_obj)) {
 		PyErr_SetString(PyExc_TypeError,
 				"data does not support buffer interface");
 		return -1;
 	}
 
 	if (PyObject_GetBuffer(data_obj, &self->buf, PyBUF_SIMPLE) == -1)
 		return -1;
 	size = self->buf.len;
 
 	self->inlined = inlined_obj && PyObject_IsTrue(inlined_obj);
 	self->data = data_obj;
 
 	self->ntlength = self->ntcapacity = 0;
 	self->ntdepth = self->ntsplits = 0;
 	self->ntlookups = self->ntmisses = 0;
 	self->ntrev = -1;
 	Py_INCREF(self->data);
 
 	if (self->inlined) {
 		Py_ssize_t len = inline_scan(self, NULL);
 		if (len == -1)
 			goto bail;
 		self->raw_length = len;
 		self->length = len + 1;
 	} else {
 		if (size % v1_hdrsize) {
 			PyErr_SetString(PyExc_ValueError, "corrupt index file");
 			goto bail;
 		}
 		self->raw_length = size / v1_hdrsize;
 		self->length = self->raw_length + 1;
 	}
 
 	return 0;
 bail:
 	return -1;
 }
 
 static PyObject *index_nodemap(indexObject *self)
 {
 	Py_INCREF(self);
 	return (PyObject *)self;
 }
 
 static void index_dealloc(indexObject *self)
 {
 	_index_clearcaches(self);
 	Py_XDECREF(self->filteredrevs);
 	if (self->buf.buf) {
 		PyBuffer_Release(&self->buf);
 		memset(&self->buf, 0, sizeof(self->buf));
 	}
 	Py_XDECREF(self->data);
 	Py_XDECREF(self->added);
 	PyObject_Del(self);
 }
 
 static PySequenceMethods index_sequence_methods = {
 	(lenfunc)index_length,   /* sq_length */
 	0,                       /* sq_concat */
 	0,                       /* sq_repeat */
 	(ssizeargfunc)index_get, /* sq_item */
 	0,                       /* sq_slice */
 	0,                       /* sq_ass_item */
 	0,                       /* sq_ass_slice */
 	(objobjproc)index_contains, /* sq_contains */
 };
 
 static PyMappingMethods index_mapping_methods = {
 	(lenfunc)index_length,                 /* mp_length */
 	(binaryfunc)index_getitem,             /* mp_subscript */
 	(objobjargproc)index_assign_subscript, /* mp_ass_subscript */
 };
 
 static PyMethodDef index_methods[] = {
 	{"ancestors", (PyCFunction)index_ancestors, METH_VARARGS,
 	 "return the gca set of the given revs"},
 	{"commonancestorsheads", (PyCFunction)index_commonancestorsheads,
 	  METH_VARARGS,
 	  "return the heads of the common ancestors of the given revs"},
 	{"clearcaches", (PyCFunction)index_clearcaches, METH_NOARGS,
 	 "clear the index caches"},
 	{"get", (PyCFunction)index_m_get, METH_VARARGS,
 	 "get an index entry"},
 	{"computephasesmapsets", (PyCFunction)compute_phases_map_sets,
 			METH_VARARGS, "compute phases"},
 	{"reachableroots2", (PyCFunction)reachableroots2, METH_VARARGS,
 		"reachableroots"},
 	{"headrevs", (PyCFunction)index_headrevs, METH_VARARGS,
 	 "get head revisions"}, /* Can do filtering since 3.2 */
 	{"headrevsfiltered", (PyCFunction)index_headrevs, METH_VARARGS,
 	 "get filtered head revisions"}, /* Can always do filtering */
 	{"deltachain", (PyCFunction)index_deltachain, METH_VARARGS,
 	 "determine revisions with deltas to reconstruct fulltext"},
 	{"insert", (PyCFunction)index_insert, METH_VARARGS,
 	 "insert an index entry"},
 	{"partialmatch", (PyCFunction)index_partialmatch, METH_VARARGS,
 	 "match a potentially ambiguous node ID"},
 	{"stats", (PyCFunction)index_stats, METH_NOARGS,
 	 "stats for the index"},
 	{NULL} /* Sentinel */
 };
 
 static PyGetSetDef index_getset[] = {
 	{"nodemap", (getter)index_nodemap, NULL, "nodemap", NULL},
 	{NULL} /* Sentinel */
 };
 
 static PyTypeObject indexType = {
 	PyVarObject_HEAD_INIT(NULL, 0)
 	"parsers.index",           /* tp_name */
 	sizeof(indexObject),       /* tp_basicsize */
 	0,                         /* tp_itemsize */
 	(destructor)index_dealloc, /* tp_dealloc */
 	0,                         /* tp_print */
 	0,                         /* tp_getattr */
 	0,                         /* tp_setattr */
 	0,                         /* tp_compare */
 	0,                         /* tp_repr */
 	0,                         /* tp_as_number */
 	&index_sequence_methods,   /* tp_as_sequence */
 	&index_mapping_methods,    /* 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 */
 	"revlog index",            /* tp_doc */
 	0,                         /* tp_traverse */
 	0,                         /* tp_clear */
 	0,                         /* tp_richcompare */
 	0,                         /* tp_weaklistoffset */
 	0,                         /* tp_iter */
 	0,                         /* tp_iternext */
 	index_methods,             /* tp_methods */
 	0,                         /* tp_members */
 	index_getset,              /* tp_getset */
 	0,                         /* tp_base */
 	0,                         /* tp_dict */
 	0,                         /* tp_descr_get */
 	0,                         /* tp_descr_set */
 	0,                         /* tp_dictoffset */
 	(initproc)index_init,      /* tp_init */
 	0,                         /* tp_alloc */
 };
 
 /*
  * returns a tuple of the form (index, index, cache) with elements as
  * follows:
  *
  * index: an index object that lazily parses RevlogNG records
  * cache: if data is inlined, a tuple (0, index_file_content), else None
  *        index_file_content could be a string, or a buffer
  *
  * added complications are for backwards compatibility
  */
 PyObject *parse_index2(PyObject *self, PyObject *args)
 {
 	PyObject *tuple = NULL, *cache = NULL;
 	indexObject *idx;
 	int ret;
 
 	idx = PyObject_New(indexObject, &indexType);
 	if (idx == NULL)
 		goto bail;
 
 	ret = index_init(idx, args);
 	if (ret == -1)
 		goto bail;
 
 	if (idx->inlined) {
 		cache = Py_BuildValue("iO", 0, idx->data);
 		if (cache == NULL)
 			goto bail;
 	} else {
 		cache = Py_None;
 		Py_INCREF(cache);
 	}
 
 	tuple = Py_BuildValue("NN", idx, cache);
 	if (!tuple)
 		goto bail;
 	return tuple;
 
 bail:
 	Py_XDECREF(idx);
 	Py_XDECREF(cache);
 	Py_XDECREF(tuple);
 	return NULL;
 }
 
 void revlog_module_init(PyObject *mod)
 {
 	indexType.tp_new = PyType_GenericNew;
 	if (PyType_Ready(&indexType) < 0)
 		return;
 	Py_INCREF(&indexType);
 	PyModule_AddObject(mod, "index", (PyObject *)&indexType);
 
 	nullentry = Py_BuildValue("iiiiiiis#", 0, 0, 0,
 				  -1, -1, -1, -1, nullid, 20);
 	if (nullentry)
 		PyObject_GC_UnTrack(nullentry);
 }

mercurial/cext/util.h

@@ -1,88 +1,50 b''
 /*
  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
 
 typedef struct {
 	PyObject_HEAD
 	char state;
 	int mode;
 	int size;
 	int mtime;
 } dirstateTupleObject;
 
 extern PyTypeObject dirstateTupleType;
 #define dirstate_tuple_check(op) (Py_TYPE(op) == &dirstateTupleType)
 
-/* This should be kept in sync with normcasespecs in encoding.py. */
-enum normcase_spec {
-	NORMCASE_LOWER = -1,
-	NORMCASE_UPPER = 1,
-	NORMCASE_OTHER = 0
-};
-
 #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);
 }
 
-static const int8_t hextable[256] = {
-	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
-	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
-	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
-	 0,  1,  2,  3,  4,  5,  6,  7,  8,  9, -1, -1, -1, -1, -1, -1, /* 0-9 */
-	-1, 10, 11, 12, 13, 14, 15, -1, -1, -1, -1, -1, -1, -1, -1, -1, /* A-F */
-	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
-	-1, 10, 11, 12, 13, 14, 15, -1, -1, -1, -1, -1, -1, -1, -1, -1, /* a-f */
-	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
-	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
-	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
-	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
-	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
-	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
-	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
-	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
-	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1
-};
-
-static inline int hexdigit(const char *p, Py_ssize_t off)
-{
-	int8_t val = hextable[(unsigned char)p[off]];
-
-	if (val >= 0) {
-		return val;
-	}
-
-	PyErr_SetString(PyExc_ValueError, "input contains non-hex character");
-	return 0;
-}
-
 #endif /* _HG_UTIL_H_ */

setup.py

@@ -1,953 +1,953 b''
 #
 # This is the mercurial setup script.
 #
 # 'python setup.py install', or
 # 'python setup.py --help' for more options
 
 import os
 
 supportedpy = '~= 2.7'
 if os.environ.get('HGALLOWPYTHON3', ''):
     # Mercurial will never work on Python 3 before 3.5 due to a lack
     # of % formatting on bytestrings, and can't work on 3.6.0 or 3.6.1
     # due to a bug in % formatting in bytestrings.
     #
     # TODO: when we actually work on Python 3, use this string as the
     # actual supportedpy string.
     supportedpy = ','.join([
         '>=2.7',
         '!=3.0.*',
         '!=3.1.*',
         '!=3.2.*',
         '!=3.3.*',
         '!=3.4.*',
         '!=3.6.0',
         '!=3.6.1',
     ])
 
 import sys, platform
 if sys.version_info[0] >= 3:
     printf = eval('print')
     libdir_escape = 'unicode_escape'
 else:
     libdir_escape = 'string_escape'
     def printf(*args, **kwargs):
         f = kwargs.get('file', sys.stdout)
         end = kwargs.get('end', '\n')
         f.write(b' '.join(args) + end)
 
 # Attempt to guide users to a modern pip - this means that 2.6 users
 # should have a chance of getting a 4.2 release, and when we ratchet
 # the version requirement forward again hopefully everyone will get
 # something that works for them.
 if sys.version_info < (2, 7, 0, 'final'):
     pip_message = ('This may be due to an out of date pip. '
                    'Make sure you have pip >= 9.0.1.')
     try:
         import pip
         pip_version = tuple([int(x) for x in pip.__version__.split('.')[:3]])
         if pip_version < (9, 0, 1) :
             pip_message = (
                 'Your pip version is out of date, please install '
                 'pip >= 9.0.1. pip {} detected.'.format(pip.__version__))
         else:
             # pip is new enough - it must be something else
             pip_message = ''
     except Exception:
         pass
     error = """
 Mercurial does not support Python older than 2.7.
 Python {py} detected.
 {pip}
 """.format(py=sys.version_info, pip=pip_message)
     printf(error, file=sys.stderr)
     sys.exit(1)
 
 # Solaris Python packaging brain damage
 try:
     import hashlib
     sha = hashlib.sha1()
 except ImportError:
     try:
         import sha
         sha.sha # silence unused import warning
     except ImportError:
         raise SystemExit(
             "Couldn't import standard hashlib (incomplete Python install).")
 
 try:
     import zlib
     zlib.compressobj # silence unused import warning
 except ImportError:
     raise SystemExit(
         "Couldn't import standard zlib (incomplete Python install).")
 
 # The base IronPython distribution (as of 2.7.1) doesn't support bz2
 isironpython = False
 try:
     isironpython = (platform.python_implementation()
                     .lower().find("ironpython") != -1)
 except AttributeError:
     pass
 
 if isironpython:
     sys.stderr.write("warning: IronPython detected (no bz2 support)\n")
 else:
     try:
         import bz2
         bz2.BZ2Compressor # silence unused import warning
     except ImportError:
         raise SystemExit(
             "Couldn't import standard bz2 (incomplete Python install).")
 
 ispypy = "PyPy" in sys.version
 
 import ctypes
 import stat, subprocess, time
 import re
 import shutil
 import tempfile
 from distutils import log
 # We have issues with setuptools on some platforms and builders. Until
 # those are resolved, setuptools is opt-in except for platforms where
 # we don't have issues.
 issetuptools = (os.name == 'nt' or 'FORCE_SETUPTOOLS' in os.environ)
 if issetuptools:
     from setuptools import setup
 else:
     from distutils.core import setup
 from distutils.ccompiler import new_compiler
 from distutils.core import Command, Extension
 from distutils.dist import Distribution
 from distutils.command.build import build
 from distutils.command.build_ext import build_ext
 from distutils.command.build_py import build_py
 from distutils.command.build_scripts import build_scripts
 from distutils.command.install import install
 from distutils.command.install_lib import install_lib
 from distutils.command.install_scripts import install_scripts
 from distutils.spawn import spawn, find_executable
 from distutils import file_util
 from distutils.errors import (
     CCompilerError,
     DistutilsError,
     DistutilsExecError,
 )
 from distutils.sysconfig import get_python_inc, get_config_var
 from distutils.version import StrictVersion
 
 scripts = ['hg']
 if os.name == 'nt':
     # We remove hg.bat if we are able to build hg.exe.
     scripts.append('contrib/win32/hg.bat')
 
 def cancompile(cc, code):
     tmpdir = tempfile.mkdtemp(prefix='hg-install-')
     devnull = oldstderr = None
     try:
         fname = os.path.join(tmpdir, 'testcomp.c')
         f = open(fname, 'w')
         f.write(code)
         f.close()
         # Redirect stderr to /dev/null to hide any error messages
         # from the compiler.
         # This will have to be changed if we ever have to check
         # for a function on Windows.
         devnull = open('/dev/null', 'w')
         oldstderr = os.dup(sys.stderr.fileno())
         os.dup2(devnull.fileno(), sys.stderr.fileno())
         objects = cc.compile([fname], output_dir=tmpdir)
         cc.link_executable(objects, os.path.join(tmpdir, "a.out"))
         return True
     except Exception:
         return False
     finally:
         if oldstderr is not None:
             os.dup2(oldstderr, sys.stderr.fileno())
         if devnull is not None:
             devnull.close()
         shutil.rmtree(tmpdir)
 
 # simplified version of distutils.ccompiler.CCompiler.has_function
 # that actually removes its temporary files.
 def hasfunction(cc, funcname):
     code = 'int main(void) { %s(); }\n' % funcname
     return cancompile(cc, code)
 
 def hasheader(cc, headername):
     code = '#include <%s>\nint main(void) { return 0; }\n' % headername
     return cancompile(cc, code)
 
 # py2exe needs to be installed to work
 try:
     import py2exe
     py2exe.Distribution # silence unused import warning
     py2exeloaded = True
     # import py2exe's patched Distribution class
     from distutils.core import Distribution
 except ImportError:
     py2exeloaded = False
 
 def runcmd(cmd, env):
     p = subprocess.Popen(cmd, stdout=subprocess.PIPE,
                          stderr=subprocess.PIPE, env=env)
     out, err = p.communicate()
     return p.returncode, out, err
 
 class hgcommand(object):
     def __init__(self, cmd, env):
         self.cmd = cmd
         self.env = env
 
     def run(self, args):
         cmd = self.cmd + args
         returncode, out, err = runcmd(cmd, self.env)
         err = filterhgerr(err)
         if err or returncode != 0:
             printf("stderr from '%s':" % (' '.join(cmd)), file=sys.stderr)
             printf(err, file=sys.stderr)
             return ''
         return out
 
 def filterhgerr(err):
     # If root is executing setup.py, but the repository is owned by
     # another user (as in "sudo python setup.py install") we will get
     # trust warnings since the .hg/hgrc file is untrusted. That is
     # fine, we don't want to load it anyway.  Python may warn about
     # a missing __init__.py in mercurial/locale, we also ignore that.
     err = [e for e in err.splitlines()
            if (not e.startswith(b'not trusting file')
                and not e.startswith(b'warning: Not importing')
                and not e.startswith(b'obsolete feature not enabled'))]
     return b'\n'.join(b'  ' + e for e in err)
 
 def findhg():
     """Try to figure out how we should invoke hg for examining the local
     repository contents.
 
     Returns an hgcommand object."""
     # By default, prefer the "hg" command in the user's path.  This was
     # presumably the hg command that the user used to create this repository.
     #
     # This repository may require extensions or other settings that would not
     # be enabled by running the hg script directly from this local repository.
     hgenv = os.environ.copy()
     # Use HGPLAIN to disable hgrc settings that would change output formatting,
     # and disable localization for the same reasons.
     hgenv['HGPLAIN'] = '1'
     hgenv['LANGUAGE'] = 'C'
     hgcmd = ['hg']
     # Run a simple "hg log" command just to see if using hg from the user's
     # path works and can successfully interact with this repository.
     check_cmd = ['log', '-r.', '-Ttest']
     try:
         retcode, out, err = runcmd(hgcmd + check_cmd, hgenv)
     except EnvironmentError:
         retcode = -1
     if retcode == 0 and not filterhgerr(err):
         return hgcommand(hgcmd, hgenv)
 
     # Fall back to trying the local hg installation.
     hgenv = localhgenv()
     # Don't source any system hgrc files when using the local hg.
     hgenv['HGRCPATH'] = ''
     hgcmd = [sys.executable, 'hg']
     try:
         retcode, out, err = runcmd(hgcmd + check_cmd, hgenv)
     except EnvironmentError:
         retcode = -1
     if retcode == 0 and not filterhgerr(err):
         return hgcommand(hgcmd, hgenv)
 
     raise SystemExit('Unable to find a working hg binary to extract the '
                      'version from the repository tags')
 
 def localhgenv():
     """Get an environment dictionary to use for invoking or importing
     mercurial from the local repository."""
     # Execute hg out of this directory with a custom environment which takes
     # care to not use any hgrc files and do no localization.
     env = {'HGMODULEPOLICY': 'py',
            'HGRCPATH': '',
            'LANGUAGE': 'C',
            'PATH': ''} # make pypi modules that use os.environ['PATH'] happy
     if 'LD_LIBRARY_PATH' in os.environ:
         env['LD_LIBRARY_PATH'] = os.environ['LD_LIBRARY_PATH']
     if 'SystemRoot' in os.environ:
         # SystemRoot is required by Windows to load various DLLs.  See:
         # https://bugs.python.org/issue13524#msg148850
         env['SystemRoot'] = os.environ['SystemRoot']
     return env
 
 version = ''
 
 if os.path.isdir('.hg'):
     hg = findhg()
     cmd = ['log', '-r', '.', '--template', '{tags}\n']
     numerictags = [t for t in hg.run(cmd).split() if t[0:1].isdigit()]
     hgid = hg.run(['id', '-i']).strip()
     if not hgid:
         # Bail out if hg is having problems interacting with this repository,
         # rather than falling through and producing a bogus version number.
         # Continuing with an invalid version number will break extensions
         # that define minimumhgversion.
         raise SystemExit('Unable to determine hg version from local repository')
     if numerictags: # tag(s) found
         version = numerictags[-1]
         if hgid.endswith('+'): # propagate the dirty status to the tag
             version += '+'
     else: # no tag found
         ltagcmd = ['parents', '--template', '{latesttag}']
         ltag = hg.run(ltagcmd)
         changessincecmd = ['log', '-T', 'x\n', '-r', "only(.,'%s')" % ltag]
         changessince = len(hg.run(changessincecmd).splitlines())
         version = '%s+%s-%s' % (ltag, changessince, hgid)
     if version.endswith('+'):
         version += time.strftime('%Y%m%d')
 elif os.path.exists('.hg_archival.txt'):
     kw = dict([[t.strip() for t in l.split(':', 1)]
                for l in open('.hg_archival.txt')])
     if 'tag' in kw:
         version = kw['tag']
     elif 'latesttag' in kw:
         if 'changessincelatesttag' in kw:
             version = '%(latesttag)s+%(changessincelatesttag)s-%(node).12s' % kw
         else:
             version = '%(latesttag)s+%(latesttagdistance)s-%(node).12s' % kw
     else:
         version = kw.get('node', '')[:12]
 
 if version:
     with open("mercurial/__version__.py", "w") as f:
         f.write('# this file is autogenerated by setup.py\n')
         f.write('version = "%s"\n' % version)
 
 try:
     oldpolicy = os.environ.get('HGMODULEPOLICY', None)
     os.environ['HGMODULEPOLICY'] = 'py'
     from mercurial import __version__
     version = __version__.version
 except ImportError:
     version = 'unknown'
 finally:
     if oldpolicy is None:
         del os.environ['HGMODULEPOLICY']
     else:
         os.environ['HGMODULEPOLICY'] = oldpolicy
 
 class hgbuild(build):
     # Insert hgbuildmo first so that files in mercurial/locale/ are found
     # when build_py is run next.
     sub_commands = [('build_mo', None)] + build.sub_commands
 
 class hgbuildmo(build):
 
     description = "build translations (.mo files)"
 
     def run(self):
         if not find_executable('msgfmt'):
             self.warn("could not find msgfmt executable, no translations "
                      "will be built")
             return
 
         podir = 'i18n'
         if not os.path.isdir(podir):
             self.warn("could not find %s/ directory" % podir)
             return
 
         join = os.path.join
         for po in os.listdir(podir):
             if not po.endswith('.po'):
                 continue
             pofile = join(podir, po)
             modir = join('locale', po[:-3], 'LC_MESSAGES')
             mofile = join(modir, 'hg.mo')
             mobuildfile = join('mercurial', mofile)
             cmd = ['msgfmt', '-v', '-o', mobuildfile, pofile]
             if sys.platform != 'sunos5':
                 # msgfmt on Solaris does not know about -c
                 cmd.append('-c')
             self.mkpath(join('mercurial', modir))
             self.make_file([pofile], mobuildfile, spawn, (cmd,))
 
 
 class hgdist(Distribution):
     pure = False
     cffi = ispypy
 
     global_options = Distribution.global_options + \
                      [('pure', None, "use pure (slow) Python "
                         "code instead of C extensions"),
                      ]
 
     def has_ext_modules(self):
         # self.ext_modules is emptied in hgbuildpy.finalize_options which is
         # too late for some cases
         return not self.pure and Distribution.has_ext_modules(self)
 
 # This is ugly as a one-liner. So use a variable.
 buildextnegops = dict(getattr(build_ext, 'negative_options', {}))
 buildextnegops['no-zstd'] = 'zstd'
 
 class hgbuildext(build_ext):
     user_options = build_ext.user_options + [
         ('zstd', None, 'compile zstd bindings [default]'),
         ('no-zstd', None, 'do not compile zstd bindings'),
     ]
 
     boolean_options = build_ext.boolean_options + ['zstd']
     negative_opt = buildextnegops
 
     def initialize_options(self):
         self.zstd = True
         return build_ext.initialize_options(self)
 
     def build_extensions(self):
         # Filter out zstd if disabled via argument.
         if not self.zstd:
             self.extensions = [e for e in self.extensions
                                if e.name != 'mercurial.zstd']
 
         return build_ext.build_extensions(self)
 
     def build_extension(self, ext):
         try:
             build_ext.build_extension(self, ext)
         except CCompilerError:
             if not getattr(ext, 'optional', False):
                 raise
             log.warn("Failed to build optional extension '%s' (skipping)",
                      ext.name)
 
 class hgbuildscripts(build_scripts):
     def run(self):
         if os.name != 'nt' or self.distribution.pure:
             return build_scripts.run(self)
 
         exebuilt = False
         try:
             self.run_command('build_hgexe')
             exebuilt = True
         except (DistutilsError, CCompilerError):
             log.warn('failed to build optional hg.exe')
 
         if exebuilt:
             # Copying hg.exe to the scripts build directory ensures it is
             # installed by the install_scripts command.
             hgexecommand = self.get_finalized_command('build_hgexe')
             dest = os.path.join(self.build_dir, 'hg.exe')
             self.mkpath(self.build_dir)
             self.copy_file(hgexecommand.hgexepath, dest)
 
             # Remove hg.bat because it is redundant with hg.exe.
             self.scripts.remove('contrib/win32/hg.bat')
 
         return build_scripts.run(self)
 
 class hgbuildpy(build_py):
     def finalize_options(self):
         build_py.finalize_options(self)
 
         if self.distribution.pure:
             self.distribution.ext_modules = []
         elif self.distribution.cffi:
             from mercurial.cffi import (
                 bdiffbuild,
                 mpatchbuild,
             )
             exts = [mpatchbuild.ffi.distutils_extension(),
                     bdiffbuild.ffi.distutils_extension()]
             # cffi modules go here
             if sys.platform == 'darwin':
                 from mercurial.cffi import osutilbuild
                 exts.append(osutilbuild.ffi.distutils_extension())
             self.distribution.ext_modules = exts
         else:
             h = os.path.join(get_python_inc(), 'Python.h')
             if not os.path.exists(h):
                 raise SystemExit('Python headers are required to build '
                                  'Mercurial but weren\'t found in %s' % h)
 
     def run(self):
         basepath = os.path.join(self.build_lib, 'mercurial')
         self.mkpath(basepath)
 
         if self.distribution.pure:
             modulepolicy = 'py'
         elif self.build_lib == '.':
             # in-place build should run without rebuilding C extensions
             modulepolicy = 'allow'
         else:
             modulepolicy = 'c'
         with open(os.path.join(basepath, '__modulepolicy__.py'), "w") as f:
             f.write('# this file is autogenerated by setup.py\n')
             f.write('modulepolicy = b"%s"\n' % modulepolicy)
 
         build_py.run(self)
 
 class buildhgextindex(Command):
     description = 'generate prebuilt index of hgext (for frozen package)'
     user_options = []
     _indexfilename = 'hgext/__index__.py'
 
     def initialize_options(self):
         pass
 
     def finalize_options(self):
         pass
 
     def run(self):
         if os.path.exists(self._indexfilename):
             with open(self._indexfilename, 'w') as f:
                 f.write('# empty\n')
 
         # here no extension enabled, disabled() lists up everything
         code = ('import pprint; from mercurial import extensions; '
                 'pprint.pprint(extensions.disabled())')
         returncode, out, err = runcmd([sys.executable, '-c', code],
                                       localhgenv())
         if err or returncode != 0:
             raise DistutilsExecError(err)
 
         with open(self._indexfilename, 'w') as f:
             f.write('# this file is autogenerated by setup.py\n')
             f.write('docs = ')
             f.write(out)
 
 class buildhgexe(build_ext):
     description = 'compile hg.exe from mercurial/exewrapper.c'
 
     def build_extensions(self):
         if os.name != 'nt':
             return
         if isinstance(self.compiler, HackedMingw32CCompiler):
             self.compiler.compiler_so = self.compiler.compiler # no -mdll
             self.compiler.dll_libraries = [] # no -lmsrvc90
 
         # Different Python installs can have different Python library
         # names. e.g. the official CPython distribution uses pythonXY.dll
         # and MinGW uses libpythonX.Y.dll.
         _kernel32 = ctypes.windll.kernel32
         _kernel32.GetModuleFileNameA.argtypes = [ctypes.c_void_p,
                                                  ctypes.c_void_p,
                                                  ctypes.c_ulong]
         _kernel32.GetModuleFileNameA.restype = ctypes.c_ulong
         size = 1000
         buf = ctypes.create_string_buffer(size + 1)
         filelen = _kernel32.GetModuleFileNameA(sys.dllhandle, ctypes.byref(buf),
                                                size)
 
         if filelen > 0 and filelen != size:
             dllbasename = os.path.basename(buf.value)
             if not dllbasename.lower().endswith('.dll'):
                 raise SystemExit('Python DLL does not end with .dll: %s' %
                                  dllbasename)
             pythonlib = dllbasename[:-4]
         else:
             log.warn('could not determine Python DLL filename; '
                      'assuming pythonXY')
 
             hv = sys.hexversion
             pythonlib = 'python%d%d' % (hv >> 24, (hv >> 16) & 0xff)
 
         log.info('using %s as Python library name' % pythonlib)
         with open('mercurial/hgpythonlib.h', 'wb') as f:
             f.write('/* this file is autogenerated by setup.py */\n')
             f.write('#define HGPYTHONLIB "%s"\n' % pythonlib)
         objects = self.compiler.compile(['mercurial/exewrapper.c'],
                                          output_dir=self.build_temp)
         dir = os.path.dirname(self.get_ext_fullpath('dummy'))
         target = os.path.join(dir, 'hg')
         self.compiler.link_executable(objects, target,
                                       libraries=[],
                                       output_dir=self.build_temp)
 
     @property
     def hgexepath(self):
         dir = os.path.dirname(self.get_ext_fullpath('dummy'))
         return os.path.join(self.build_temp, dir, 'hg.exe')
 
 class hginstall(install):
 
     user_options = install.user_options + [
         ('old-and-unmanageable', None,
          'noop, present for eggless setuptools compat'),
         ('single-version-externally-managed', None,
          'noop, present for eggless setuptools compat'),
     ]
 
     # Also helps setuptools not be sad while we refuse to create eggs.
     single_version_externally_managed = True
 
     def get_sub_commands(self):
         # Screen out egg related commands to prevent egg generation.  But allow
         # mercurial.egg-info generation, since that is part of modern
         # packaging.
         excl = set(['bdist_egg'])
         return filter(lambda x: x not in excl, install.get_sub_commands(self))
 
 class hginstalllib(install_lib):
     '''
     This is a specialization of install_lib that replaces the copy_file used
     there so that it supports setting the mode of files after copying them,
     instead of just preserving the mode that the files originally had.  If your
     system has a umask of something like 027, preserving the permissions when
     copying will lead to a broken install.
 
     Note that just passing keep_permissions=False to copy_file would be
     insufficient, as it might still be applying a umask.
     '''
 
     def run(self):
         realcopyfile = file_util.copy_file
         def copyfileandsetmode(*args, **kwargs):
             src, dst = args[0], args[1]
             dst, copied = realcopyfile(*args, **kwargs)
             if copied:
                 st = os.stat(src)
                 # Persist executable bit (apply it to group and other if user
                 # has it)
                 if st[stat.ST_MODE] & stat.S_IXUSR:
                     setmode = int('0755', 8)
                 else:
                     setmode = int('0644', 8)
                 m = stat.S_IMODE(st[stat.ST_MODE])
                 m = (m & ~int('0777', 8)) | setmode
                 os.chmod(dst, m)
         file_util.copy_file = copyfileandsetmode
         try:
             install_lib.run(self)
         finally:
             file_util.copy_file = realcopyfile
 
 class hginstallscripts(install_scripts):
     '''
     This is a specialization of install_scripts that replaces the @LIBDIR@ with
     the configured directory for modules. If possible, the path is made relative
     to the directory for scripts.
     '''
 
     def initialize_options(self):
         install_scripts.initialize_options(self)
 
         self.install_lib = None
 
     def finalize_options(self):
         install_scripts.finalize_options(self)
         self.set_undefined_options('install',
                                    ('install_lib', 'install_lib'))
 
     def run(self):
         install_scripts.run(self)
 
         # It only makes sense to replace @LIBDIR@ with the install path if
         # the install path is known. For wheels, the logic below calculates
         # the libdir to be "../..". This is because the internal layout of a
         # wheel archive looks like:
         #
         #   mercurial-3.6.1.data/scripts/hg
         #   mercurial/__init__.py
         #
         # When installing wheels, the subdirectories of the "<pkg>.data"
         # directory are translated to system local paths and files therein
         # are copied in place. The mercurial/* files are installed into the
         # site-packages directory. However, the site-packages directory
         # isn't known until wheel install time. This means we have no clue
         # at wheel generation time what the installed site-packages directory
         # will be. And, wheels don't appear to provide the ability to register
         # custom code to run during wheel installation. This all means that
         # we can't reliably set the libdir in wheels: the default behavior
         # of looking in sys.path must do.
 
         if (os.path.splitdrive(self.install_dir)[0] !=
             os.path.splitdrive(self.install_lib)[0]):
             # can't make relative paths from one drive to another, so use an
             # absolute path instead
             libdir = self.install_lib
         else:
             common = os.path.commonprefix((self.install_dir, self.install_lib))
             rest = self.install_dir[len(common):]
             uplevel = len([n for n in os.path.split(rest) if n])
 
             libdir = uplevel * ('..' + os.sep) + self.install_lib[len(common):]
 
         for outfile in self.outfiles:
             with open(outfile, 'rb') as fp:
                 data = fp.read()
 
             # skip binary files
             if b'\0' in data:
                 continue
 
             # During local installs, the shebang will be rewritten to the final
             # install path. During wheel packaging, the shebang has a special
             # value.
             if data.startswith(b'#!python'):
                 log.info('not rewriting @LIBDIR@ in %s because install path '
                          'not known' % outfile)
                 continue
 
             data = data.replace(b'@LIBDIR@', libdir.encode(libdir_escape))
             with open(outfile, 'wb') as fp:
                 fp.write(data)
 
 cmdclass = {'build': hgbuild,
             'build_mo': hgbuildmo,
             'build_ext': hgbuildext,
             'build_py': hgbuildpy,
             'build_scripts': hgbuildscripts,
             'build_hgextindex': buildhgextindex,
             'install': hginstall,
             'install_lib': hginstalllib,
             'install_scripts': hginstallscripts,
             'build_hgexe': buildhgexe,
             }
 
 packages = ['mercurial',
             'mercurial.cext',
             'mercurial.cffi',
             'mercurial.hgweb',
             'mercurial.httpclient',
             'mercurial.pure',
             'hgext', 'hgext.convert', 'hgext.fsmonitor',
             'hgext.fsmonitor.pywatchman', 'hgext.highlight',
             'hgext.largefiles', 'hgext.zeroconf', 'hgext3rd',
             'hgdemandimport']
 
 common_depends = ['mercurial/bitmanipulation.h',
                   'mercurial/compat.h',
                   'mercurial/cext/util.h']
 common_include_dirs = ['mercurial']
 
 osutil_cflags = []
 osutil_ldflags = []
 
 # platform specific macros
 for plat, func in [('bsd', 'setproctitle')]:
     if re.search(plat, sys.platform) and hasfunction(new_compiler(), func):
         osutil_cflags.append('-DHAVE_%s' % func.upper())
 
 for plat, macro, code in [
     ('bsd|darwin', 'BSD_STATFS', '''
      #include <sys/param.h>
      #include <sys/mount.h>
      int main() { struct statfs s; return sizeof(s.f_fstypename); }
      '''),
     ('linux', 'LINUX_STATFS', '''
      #include <linux/magic.h>
      #include <sys/vfs.h>
      int main() { struct statfs s; return sizeof(s.f_type); }
      '''),
 ]:
     if re.search(plat, sys.platform) and cancompile(new_compiler(), code):
         osutil_cflags.append('-DHAVE_%s' % macro)
 
 if sys.platform == 'darwin':
     osutil_ldflags += ['-framework', 'ApplicationServices']
 
 extmodules = [
     Extension('mercurial.cext.base85', ['mercurial/cext/base85.c'],
               include_dirs=common_include_dirs,
               depends=common_depends),
     Extension('mercurial.cext.bdiff', ['mercurial/bdiff.c',
                                        'mercurial/cext/bdiff.c'],
               include_dirs=common_include_dirs,
               depends=common_depends + ['mercurial/bdiff.h']),
     Extension('mercurial.cext.diffhelpers', ['mercurial/cext/diffhelpers.c'],
               include_dirs=common_include_dirs,
               depends=common_depends),
     Extension('mercurial.cext.mpatch', ['mercurial/mpatch.c',
                                         '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),
+              depends=common_depends + ['mercurial/cext/charencode.h']),
     Extension('mercurial.cext.osutil', ['mercurial/cext/osutil.c'],
               include_dirs=common_include_dirs,
               extra_compile_args=osutil_cflags,
               extra_link_args=osutil_ldflags,
               depends=common_depends),
     Extension('hgext.fsmonitor.pywatchman.bser',
               ['hgext/fsmonitor/pywatchman/bser.c']),
     ]
 
 sys.path.insert(0, 'contrib/python-zstandard')
 import setup_zstd
 extmodules.append(setup_zstd.get_c_extension(name='mercurial.zstd'))
 
 try:
     from distutils import cygwinccompiler
 
     # the -mno-cygwin option has been deprecated for years
     compiler = cygwinccompiler.Mingw32CCompiler
 
     class HackedMingw32CCompiler(cygwinccompiler.Mingw32CCompiler):
         def __init__(self, *args, **kwargs):
             compiler.__init__(self, *args, **kwargs)
             for i in 'compiler compiler_so linker_exe linker_so'.split():
                 try:
                     getattr(self, i).remove('-mno-cygwin')
                 except ValueError:
                     pass
 
     cygwinccompiler.Mingw32CCompiler = HackedMingw32CCompiler
 except ImportError:
     # the cygwinccompiler package is not available on some Python
     # distributions like the ones from the optware project for Synology
     # DiskStation boxes
     class HackedMingw32CCompiler(object):
         pass
 
 if os.name == 'nt':
     # Allow compiler/linker flags to be added to Visual Studio builds.  Passing
     # extra_link_args to distutils.extensions.Extension() doesn't have any
     # effect.
     from distutils import msvccompiler
 
     compiler = msvccompiler.MSVCCompiler
 
     class HackedMSVCCompiler(msvccompiler.MSVCCompiler):
         def initialize(self):
             compiler.initialize(self)
             # "warning LNK4197: export 'func' specified multiple times"
             self.ldflags_shared.append('/ignore:4197')
             self.ldflags_shared_debug.append('/ignore:4197')
 
     msvccompiler.MSVCCompiler = HackedMSVCCompiler
 
 packagedata = {'mercurial': ['locale/*/LC_MESSAGES/hg.mo',
                              'help/*.txt',
                              'help/internals/*.txt',
                              'default.d/*.rc',
                              'dummycert.pem']}
 
 def ordinarypath(p):
     return p and p[0] != '.' and p[-1] != '~'
 
 for root in ('templates',):
     for curdir, dirs, files in os.walk(os.path.join('mercurial', root)):
         curdir = curdir.split(os.sep, 1)[1]
         dirs[:] = filter(ordinarypath, dirs)
         for f in filter(ordinarypath, files):
             f = os.path.join(curdir, f)
             packagedata['mercurial'].append(f)
 
 datafiles = []
 
 # distutils expects version to be str/unicode. Converting it to
 # unicode on Python 2 still works because it won't contain any
 # non-ascii bytes and will be implicitly converted back to bytes
 # when operated on.
 assert isinstance(version, bytes)
 setupversion = version.decode('ascii')
 
 extra = {}
 
 if issetuptools:
     extra['python_requires'] = supportedpy
 if py2exeloaded:
     extra['console'] = [
         {'script':'hg',
          'copyright':'Copyright (C) 2005-2017 Matt Mackall and others',
          'product_version':version}]
     # sub command of 'build' because 'py2exe' does not handle sub_commands
     build.sub_commands.insert(0, ('build_hgextindex', None))
     # put dlls in sub directory so that they won't pollute PATH
     extra['zipfile'] = 'lib/library.zip'
 
 if os.name == 'nt':
     # Windows binary file versions for exe/dll files must have the
     # form W.X.Y.Z, where W,X,Y,Z are numbers in the range 0..65535
     setupversion = version.split('+', 1)[0]
 
 if sys.platform == 'darwin' and os.path.exists('/usr/bin/xcodebuild'):
     version = runcmd(['/usr/bin/xcodebuild', '-version'], {})[1].splitlines()
     if version:
         version = version[0]
         if sys.version_info[0] == 3:
             version = version.decode('utf-8')
         xcode4 = (version.startswith('Xcode') and
                   StrictVersion(version.split()[1]) >= StrictVersion('4.0'))
         xcode51 = re.match(r'^Xcode\s+5\.1', version) is not None
     else:
         # xcodebuild returns empty on OS X Lion with XCode 4.3 not
         # installed, but instead with only command-line tools. Assume
         # that only happens on >= Lion, thus no PPC support.
         xcode4 = True
         xcode51 = False
 
     # XCode 4.0 dropped support for ppc architecture, which is hardcoded in
     # distutils.sysconfig
     if xcode4:
         os.environ['ARCHFLAGS'] = ''
 
     # XCode 5.1 changes clang such that it now fails to compile if the
     # -mno-fused-madd flag is passed, but the version of Python shipped with
     # OS X 10.9 Mavericks includes this flag. This causes problems in all
     # C extension modules, and a bug has been filed upstream at
     # http://bugs.python.org/issue21244. We also need to patch this here
     # so Mercurial can continue to compile in the meantime.
     if xcode51:
         cflags = get_config_var('CFLAGS')
         if cflags and re.search(r'-mno-fused-madd\b', cflags) is not None:
             os.environ['CFLAGS'] = (
                 os.environ.get('CFLAGS', '') + ' -Qunused-arguments')
 
 setup(name='mercurial',
       version=setupversion,
       author='Matt Mackall and many others',
       author_email='[e-mail]',
       url='https://mercurial-scm.org/',
       download_url='https://mercurial-scm.org/release/',
       description=('Fast scalable distributed SCM (revision control, version '
                    'control) system'),
       long_description=('Mercurial is a distributed SCM tool written in Python.'
                         ' It is used by a number of large projects that require'
                         ' fast, reliable distributed revision control, such as '
                         'Mozilla.'),
       license='GNU GPLv2 or any later version',
       classifiers=[
           'Development Status :: 6 - Mature',
           'Environment :: Console',
           'Intended Audience :: Developers',
           'Intended Audience :: System Administrators',
           'License :: OSI Approved :: GNU General Public License (GPL)',
           'Natural Language :: Danish',
           'Natural Language :: English',
           'Natural Language :: German',
           'Natural Language :: Italian',
           'Natural Language :: Japanese',
           'Natural Language :: Portuguese (Brazilian)',
           'Operating System :: Microsoft :: Windows',
           'Operating System :: OS Independent',
           'Operating System :: POSIX',
           'Programming Language :: C',
           'Programming Language :: Python',
           'Topic :: Software Development :: Version Control',
       ],
       scripts=scripts,
       packages=packages,
       ext_modules=extmodules,
       data_files=datafiles,
       package_data=packagedata,
       cmdclass=cmdclass,
       distclass=hgdist,
       options={'py2exe': {'packages': ['hgdemandimport', 'hgext', 'email',
                                        # implicitly imported per module policy
                                        # (cffi wouldn't be used as a frozen exe)
                                        'mercurial.cext',
                                        #'mercurial.cffi',
                                        'mercurial.pure']},
                'bdist_mpkg': {'zipdist': False,
                               'license': 'COPYING',
                               'readme': 'contrib/macosx/Readme.html',
                               'welcome': 'contrib/macosx/Welcome.html',
                               },
                },
       **extra)