scmutil: rewrite dirs in C, use if available

This is over twice as fast as the Python dirs code. Upcoming changes
will nearly double its speed again.

perfdirs results for a working dir with 170,000 files:
  Python     638 msec
  C          244

mercurial/dirs.c

@@ -0,0 +1,298
+/*
+ dirs.c - dynamic directory diddling for dirstates
+
+ Copyright 2013 Facebook
+
+ This software may be used and distributed according to the terms of
+ the GNU General Public License, incorporated herein by reference.
+*/
+
+#define PY_SSIZE_T_CLEAN
+#include <Python.h>
+#include "util.h"
+
+/*
+ * This is a multiset of directory names, built from the files that
+ * appear in a dirstate or manifest.
+ */
+typedef struct {
+	PyObject_HEAD
+	PyObject *dict;
+} dirsObject;
+
+static inline Py_ssize_t _finddir(PyObject *path, Py_ssize_t pos)
+{
+	const char *s = PyString_AS_STRING(path);
+
+	while (pos != -1) {
+		if (s[pos] == '/')
+			break;
+		pos -= 1;
+	}
+
+	return pos;
+}
+
+static int _addpath(PyObject *dirs, PyObject *path)
+{
+	Py_ssize_t pos = PyString_GET_SIZE(path);
+	PyObject *newval = NULL, *key = NULL;
+	int ret = -1;
+
+	while ((pos = _finddir(path, pos - 1)) != -1) {
+		PyObject *val;
+		long v = 0;
+
+		key = PyString_FromStringAndSize(PyString_AS_STRING(path), pos);
+
+		if (key == NULL)
+			goto bail;
+
+		val = PyDict_GetItem(dirs, key);
+		if (val != NULL)
+			v = PyInt_AS_LONG(val);
+
+		newval = PyInt_FromLong(v + 1);
+
+		if (newval == NULL)
+			goto bail;
+
+		ret = PyDict_SetItem(dirs, key, newval);
+		if (ret == -1)
+			goto bail;
+		Py_CLEAR(key);
+		Py_CLEAR(newval);
+	}
+	ret = 0;
+
+bail:
+	Py_XDECREF(key);
+	Py_XDECREF(newval);
+
+	return ret;
+}
+
+static int _delpath(PyObject *dirs, PyObject *path)
+{
+	Py_ssize_t pos = PyString_GET_SIZE(path);
+	PyObject *newval = NULL, *key = NULL;
+	int ret = -1;
+
+	while ((pos = _finddir(path, pos - 1)) != -1) {
+		PyObject *val;
+		long v;
+
+		key = PyString_FromStringAndSize(PyString_AS_STRING(path), pos);
+
+		if (key == NULL)
+			goto bail;
+
+		val = PyDict_GetItem(dirs, key);
+		if (val == NULL) {
+			PyErr_SetString(PyExc_ValueError,
+					"expected a value, found none");
+			goto bail;
+		}
+		v = PyInt_AS_LONG(val);
+
+		if (v <= 1) {
+			if (PyDict_DelItem(dirs, key) == -1)
+				goto bail;
+			continue;
+		}
+		newval = PyInt_FromLong(v - 1);
+
+		if (newval == NULL)
+			goto bail;
+
+		ret = PyDict_SetItem(dirs, key, newval);
+		if (ret == -1)
+			goto bail;
+		Py_CLEAR(key);
+		Py_CLEAR(newval);
+	}
+	ret = 0;
+
+bail:
+	Py_XDECREF(key);
+	Py_XDECREF(newval);
+
+	return ret;
+}
+
+static int dirs_fromdict(PyObject *dirs, PyObject *source, char skipchar)
+{
+	PyObject *key, *value;
+	Py_ssize_t pos = 0;
+
+	while (PyDict_Next(source, &pos, &key, &value)) {
+		if (!PyString_Check(key)) {
+			PyErr_SetString(PyExc_TypeError, "expected string key");
+			return -1;
+		}
+		if (skipchar) {
+			PyObject *st;
+
+			if (!PyTuple_Check(value) ||
+			    PyTuple_GET_SIZE(value) == 0) {
+				PyErr_SetString(PyExc_TypeError,
+						"expected non-empty tuple");
+				return -1;
+			}
+
+			st = PyTuple_GET_ITEM(value, 0);
+
+			if (!PyString_Check(st) || PyString_GET_SIZE(st) == 0) {
+				PyErr_SetString(PyExc_TypeError,
+						"expected non-empty string "
+						"at tuple index 0");
+				return -1;
+			}
+
+			if (PyString_AS_STRING(st)[0] == skipchar)
+				continue;
+		}
+
+		if (_addpath(dirs, key) == -1)
+			return -1;
+	}
+
+	return 0;
+}
+
+static int dirs_fromiter(PyObject *dirs, PyObject *source)
+{
+	PyObject *iter, *item = NULL;
+	int ret;
+
+	iter = PyObject_GetIter(source);
+	if (iter == NULL)
+		return -1;
+
+	while ((item = PyIter_Next(iter)) != NULL) {
+		if (!PyString_Check(item)) {
+			PyErr_SetString(PyExc_TypeError, "expected string");
+			break;
+		}
+
+		if (_addpath(dirs, item) == -1)
+			break;
+		Py_CLEAR(item);
+	}
+
+	ret = PyErr_Occurred() ? -1 : 0;
+	Py_XDECREF(item);
+	return ret;
+}
+
+/*
+ * Calculate a refcounted set of directory names for the files in a
+ * dirstate.
+ */
+static int dirs_init(dirsObject *self, PyObject *args)
+{
+	PyObject *dirs = NULL, *source = NULL;
+	char skipchar = 0;
+	int ret = -1;
+
+	self->dict = NULL;
+
+	if (!PyArg_ParseTuple(args, "|Oc:__init__", &source, &skipchar))
+		return -1;
+
+	dirs = PyDict_New();
+
+	if (dirs == NULL)
+		return -1;
+
+	if (source == NULL)
+		ret = 0;
+	else if (PyDict_Check(source))
+		ret = dirs_fromdict(dirs, source, skipchar);
+	else if (skipchar)
+		PyErr_SetString(PyExc_ValueError,
+				"skip character is only supported "
+				"with a dict source");
+	else
+		ret = dirs_fromiter(dirs, source);
+
+	if (ret == -1)
+		Py_XDECREF(dirs);
+	else
+		self->dict = dirs;
+
+	return ret;
+}
+
+PyObject *dirs_addpath(dirsObject *self, PyObject *args)
+{
+	PyObject *path;
+
+	if (!PyArg_ParseTuple(args, "O!:addpath", &PyString_Type, &path))
+		return NULL;
+
+	if (_addpath(self->dict, path) == -1)
+		return NULL;
+
+	Py_RETURN_NONE;
+}
+
+static PyObject *dirs_delpath(dirsObject *self, PyObject *args)
+{
+	PyObject *path;
+
+	if (!PyArg_ParseTuple(args, "O!:delpath", &PyString_Type, &path))
+		return NULL;
+
+	if (_delpath(self->dict, path) == -1)
+		return NULL;
+
+	Py_RETURN_NONE;
+}
+
+static int dirs_contains(dirsObject *self, PyObject *value)
+{
+	return PyString_Check(value) ? PyDict_Contains(self->dict, value) : 0;
+}
+
+static void dirs_dealloc(dirsObject *self)
+{
+	Py_XDECREF(self->dict);
+	PyObject_Del(self);
+}
+
+static PyObject *dirs_iter(dirsObject *self)
+{
+	return PyObject_GetIter(self->dict);
+}
+
+static PySequenceMethods dirs_sequence_methods;
+
+static PyMethodDef dirs_methods[] = {
+	{"addpath", (PyCFunction)dirs_addpath, METH_VARARGS, "add a path"},
+	{"delpath", (PyCFunction)dirs_delpath, METH_VARARGS, "remove a path"},
+	{NULL} /* Sentinel */
+};
+
+static PyTypeObject dirsType = { PyObject_HEAD_INIT(NULL) };
+
+void dirs_module_init(PyObject *mod)
+{
+	dirs_sequence_methods.sq_contains = (objobjproc)dirs_contains;
+	dirsType.tp_name = "parsers.dirs";
+	dirsType.tp_new = PyType_GenericNew;
+	dirsType.tp_basicsize = sizeof(dirsObject);
+	dirsType.tp_dealloc = (destructor)dirs_dealloc;
+	dirsType.tp_as_sequence = &dirs_sequence_methods;
+	dirsType.tp_flags = Py_TPFLAGS_DEFAULT;
+	dirsType.tp_doc = "dirs";
+	dirsType.tp_iter = (getiterfunc)dirs_iter;
+	dirsType.tp_methods = dirs_methods;
+	dirsType.tp_init = (initproc)dirs_init;
+
+	if (PyType_Ready(&dirsType) < 0)
+		return;
+	Py_INCREF(&dirsType);
+
+	PyModule_AddObject(mod, "dirs", (PyObject *)&dirsType);
+}

mercurial/parsers.c

@@ -1,1569 +1,1573
 /*
  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 "util.h"
 
 static inline int hexdigit(const char *p, Py_ssize_t off)
 {
 	char c = p[off];
 
 	if (c >= '0' && c <= '9')
 		return c - '0';
 	if (c >= 'a' && c <= 'f')
 		return c - 'a' + 10;
 	if (c >= 'A' && c <= 'F')
 		return c - 'A' + 10;
 
 	PyErr_SetString(PyExc_ValueError, "input contains non-hex character");
 	return 0;
 }
 
 /*
  * Turn a hex-encoded string into binary.
  */
 static 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;
 }
 
 /*
  * 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, *cur, *start, *zero;
 	int len;
 
 	if (!PyArg_ParseTuple(args, "O!O!s#:parse_manifest",
 			      &PyDict_Type, &mfdict,
 			      &PyDict_Type, &fdict,
 			      &str, &len))
 		goto quit;
 
 	for (start = cur = str, zero = NULL; cur < str + len; cur++) {
 		PyObject *file = NULL, *node = NULL;
 		PyObject *flags = NULL;
 		ptrdiff_t nlen;
 
 		if (!*cur) {
 			zero = cur;
 			continue;
 		}
 		else if (*cur != '\n')
 			continue;
 
 		if (!zero) {
 			PyErr_SetString(PyExc_ValueError,
 					"manifest entry has no separator");
 			goto quit;
 		}
 
 		file = PyBytes_FromStringAndSize(start, zero - start);
 
 		if (!file)
 			goto bail;
 
 		nlen = cur - 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 = cur + 1;
 		zero = NULL;
 
 		Py_XDECREF(flags);
 		Py_XDECREF(node);
 		Py_XDECREF(file);
 		continue;
 	bail:
 		Py_XDECREF(flags);
 		Py_XDECREF(node);
 		Py_XDECREF(file);
 		goto quit;
 	}
 
 	if (len > 0 && *(cur - 1) != '\n') {
 		PyErr_SetString(PyExc_ValueError,
 				"manifest contains trailing garbage");
 		goto quit;
 	}
 
 	Py_INCREF(Py_None);
 	return Py_None;
 quit:
 	return NULL;
 }
 
 static PyObject *parse_dirstate(PyObject *self, PyObject *args)
 {
 	PyObject *dmap, *cmap, *parents = NULL, *ret = NULL;
 	PyObject *fname = NULL, *cname = NULL, *entry = NULL;
 	char *str, *cur, *end, *cpos;
 	int state, mode, size, mtime;
 	unsigned int flen;
 	int len;
 
 	if (!PyArg_ParseTuple(args, "O!O!s#:parse_dirstate",
 			      &PyDict_Type, &dmap,
 			      &PyDict_Type, &cmap,
 			      &str, &len))
 		goto quit;
 
 	/* read parents */
 	if (len < 40)
 		goto quit;
 
 	parents = Py_BuildValue("s#s#", str, 20, str + 20, 20);
 	if (!parents)
 		goto quit;
 
 	/* read filenames */
 	cur = str + 40;
 	end = str + len;
 
 	while (cur < end - 17) {
 		/* unpack header */
 		state = *cur;
 		mode = getbe32(cur + 1);
 		size = getbe32(cur + 5);
 		mtime = getbe32(cur + 9);
 		flen = getbe32(cur + 13);
 		cur += 17;
 		if (cur + flen > end || cur + flen < cur) {
 			PyErr_SetString(PyExc_ValueError, "overflow in dirstate");
 			goto quit;
 		}
 
 		entry = Py_BuildValue("ciii", state, mode, size, mtime);
 		if (!entry)
 			goto quit;
 		PyObject_GC_UnTrack(entry); /* don't waste time with this */
 
 		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;
 		}
 		cur += flen;
 		Py_DECREF(fname);
 		Py_DECREF(entry);
 		fname = cname = entry = NULL;
 	}
 
 	ret = parents;
 	Py_INCREF(ret);
 quit:
 	Py_XDECREF(fname);
 	Py_XDECREF(cname);
 	Py_XDECREF(entry);
 	Py_XDECREF(parents);
 	return ret;
 }
 
 static inline int getintat(PyObject *tuple, int off, uint32_t *v)
 {
 	PyObject *o = PyTuple_GET_ITEM(tuple, off);
 	long val;
 
 	if (PyInt_Check(o))
 		val = PyInt_AS_LONG(o);
 	else if (PyLong_Check(o)) {
 		val = PyLong_AsLong(o);
 		if (val == -1 && PyErr_Occurred())
 			return -1;
 	} else {
 		PyErr_SetString(PyExc_TypeError, "expected an int or long");
 		return -1;
 	}
 	if (LONG_MAX > INT_MAX && (val > INT_MAX || val < INT_MIN)) {
 		PyErr_SetString(PyExc_OverflowError,
 				"Python value to large to convert to uint32_t");
 		return -1;
 	}
 	*v = (uint32_t)val;
 	return 0;
 }
 
 static PyObject *dirstate_unset;
 
 /*
  * 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;
 	Py_ssize_t nbytes, pos, l;
 	PyObject *k, *v, *pn;
 	char *p, *s;
 	double now;
 
 	if (!PyArg_ParseTuple(args, "O!O!Od: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 (!PyString_Check(k)) {
 			PyErr_SetString(PyExc_TypeError, "expected string key");
 			goto bail;
 		}
 		nbytes += PyString_GET_SIZE(k) + 17;
 		c = PyDict_GetItem(copymap, k);
 		if (c) {
 			if (!PyString_Check(c)) {
 				PyErr_SetString(PyExc_TypeError,
 						"expected string key");
 				goto bail;
 			}
 			nbytes += PyString_GET_SIZE(c) + 1;
 		}
 	}
 
 	packobj = PyString_FromStringAndSize(NULL, nbytes);
 	if (packobj == NULL)
 		goto bail;
 
 	p = PyString_AS_STRING(packobj);
 
 	pn = PySequence_ITEM(pl, 0);
 	if (PyString_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 (PyString_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); ) {
 		uint32_t mode, size, mtime;
 		Py_ssize_t len, l;
 		PyObject *o;
 		char *s, *t;
 
 		if (!PyTuple_Check(v) || PyTuple_GET_SIZE(v) != 4) {
 			PyErr_SetString(PyExc_TypeError, "expected a 4-tuple");
 			goto bail;
 		}
 		o = PyTuple_GET_ITEM(v, 0);
 		if (PyString_AsStringAndSize(o, &s, &l) == -1 || l != 1) {
 			PyErr_SetString(PyExc_TypeError, "expected one byte");
 			goto bail;
 		}
 		*p++ = *s;
 		if (getintat(v, 1, &mode) == -1)
 			goto bail;
 		if (getintat(v, 2, &size) == -1)
 			goto bail;
 		if (getintat(v, 3, &mtime) == -1)
 			goto bail;
 		if (*s == 'n' && mtime == (uint32_t)now) {
 			/* See pure/parsers.py:pack_dirstate for why we do
 			 * this. */
 			if (PyDict_SetItem(map, k, dirstate_unset) == -1)
 				goto bail;
 			mode = 0, size = -1, mtime = -1;
 		}
 		putbe32(mode, p);
 		putbe32(size, p + 4);
 		putbe32(mtime, p + 8);
 		t = p + 12;
 		p += 16;
 		len = PyString_GET_SIZE(k);
 		memcpy(p, PyString_AS_STRING(k), len);
 		p += len;
 		o = PyDict_GetItem(copymap, k);
 		if (o) {
 			*p++ = '\0';
 			l = PyString_GET_SIZE(o);
 			memcpy(p, PyString_AS_STRING(o), l);
 			p += l;
 			len += l + 1;
 		}
 		putbe32((uint32_t)len, t);
 	}
 
 	pos = p - PyString_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(packobj);
 	return NULL;
 }
 
 /*
  * 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 behaviours.
  *
  * 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 */
 	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 */
 	nodetree *nt;          /* base-16 trie */
 	int ntlength;          /* # nodes in use */
 	int 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 long 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 = 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 PyString_AS_STRING(self->data) + pos * v1_hdrsize;
 }
 
 /*
  * 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);
 
 	self->cache[pos] = entry;
 	Py_INCREF(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 ? PyString_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 (PyString_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;
 	long offset;
 	Py_ssize_t len, nodelen;
 
 	if (!PyArg_ParseTuple(args, "lO", &offset, &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 (offset < 0)
 		offset += len;
 
 	if (offset != len - 1) {
 		PyErr_SetString(PyExc_IndexError,
 				"insert only supported at index -1");
 		return NULL;
 	}
 
 	if (offset > INT_MAX) {
 		PyErr_SetString(PyExc_ValueError,
 				"currently only 2**31 revs supported");
 		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, (int)offset);
 
 	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) {
 		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();
 
 	if (obj == NULL)
 		return NULL;
 
 #define istat(__n, __d) \
 	if (PyDict_SetItemString(obj, __d, PyInt_FromSsize_t(self->__n)) == -1) \
 		goto bail;
 
 	if (self->added) {
 		Py_ssize_t len = PyList_GET_SIZE(self->added);
 		if (PyDict_SetItemString(obj, "index entries added",
 					 PyInt_FromSsize_t(len)) == -1)
 			goto bail;
 	}
 
 	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);
 	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 PyObject *index_headrevs(indexObject *self)
 {
 	Py_ssize_t i, len, addlen;
 	char *nothead = NULL;
 	PyObject *heads;
 
 	if (self->headrevs)
 		return list_copy(self->headrevs);
 
 	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)
 		goto bail;
 
 	for (i = 0; i < self->raw_length; i++) {
 		const char *data = index_deref(self, i);
 		int parent_1 = getbe32(data + 24);
 		int parent_2 = getbe32(data + 28);
 		if (parent_1 >= 0)
 			nothead[parent_1] = 1;
 		if (parent_2 >= 0)
 			nothead[parent_2] = 1;
 	}
 
 	addlen = self->added ? PyList_GET_SIZE(self->added) : 0;
 
 	for (i = 0; i < addlen; i++) {
 		PyObject *rev = PyList_GET_ITEM(self->added, i);
 		PyObject *p1 = PyTuple_GET_ITEM(rev, 5);
 		PyObject *p2 = PyTuple_GET_ITEM(rev, 6);
 		long parent_1, parent_2;
 
 		if (!PyInt_Check(p1) || !PyInt_Check(p2)) {
 			PyErr_SetString(PyExc_TypeError,
 					"revlog parents are invalid");
 			goto bail;
 		}
 		parent_1 = PyInt_AS_LONG(p1);
 		parent_2 = PyInt_AS_LONG(p2);
 		if (parent_1 >= 0)
 			nothead[parent_1] = 1;
 		if (parent_2 >= 0)
 			nothead[parent_2] = 1;
 	}
 
 	for (i = 0; i < len; i++) {
 		PyObject *head;
 
 		if (nothead[i])
 			continue;
 		head = PyInt_FromLong(i);
 		if (head == NULL || PyList_Append(heads, head) == -1) {
 			Py_XDECREF(head);
 			goto bail;
 		}
 	}
 
 done:
 	self->headrevs = heads;
 	free(nothead);
 	return list_copy(self->headrevs);
 bail:
 	Py_XDECREF(heads);
 	free(nothead);
 	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) {
 		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) {
 		self->ntcapacity = self->raw_length < 4
 			? 4 : 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 PyObject *raise_revlog_error(void)
 {
 	static PyObject *errclass;
 	PyObject *mod = NULL, *errobj;
 
 	if (errclass == NULL) {
 		PyObject *dict;
 
 		mod = PyImport_ImportModule("mercurial.error");
 		if (mod == NULL)
 			goto classfail;
 
 		dict = PyModule_GetDict(mod);
 		if (dict == NULL)
 			goto classfail;
 
 		errclass = PyDict_GetItemString(dict, "RevlogError");
 		if (errclass == NULL) {
 			PyErr_SetString(PyExc_SystemError,
 					"could not find RevlogError");
 			goto classfail;
 		}
 		Py_INCREF(errclass);
 	}
 
 	errobj = PyObject_CallFunction(errclass, NULL);
 	if (errobj == NULL)
 		return NULL;
 	PyErr_SetObject(errclass, errobj);
 	return errobj;
 
 classfail:
 	Py_XDECREF(mod);
 	return NULL;
 }
 
 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 PyString_FromStringAndSize(nullid, 20);
 	}
 
 	fullnode = index_node(self, rev);
 	if (fullnode == NULL) {
 		PyErr_Format(PyExc_IndexError,
 			     "could not access rev %d", rev);
 		return NULL;
 	}
 	return PyString_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;
 	}
 }
 
 /*
  * 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, PyString_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;
 
 	if (PySlice_GetIndicesEx((PySliceObject*)item, length,
 				 &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;
 	}
 	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 long inline_scan(indexObject *self, const char **offsets)
 {
 	const char *data = PyString_AS_STRING(self->data);
 	const char *end = data + PyString_GET_SIZE(self->data);
 	long incr = v1_hdrsize;
 	Py_ssize_t len = 0;
 
 	while (data + v1_hdrsize <= end) {
 		uint32_t comp_len;
 		const char *old_data;
 		/* 3rd element of header is length of compressed inline data */
 		comp_len = getbe32(data + 8);
 		incr = v1_hdrsize + comp_len;
 		if (incr < v1_hdrsize)
 			break;
 		if (offsets)
 			offsets[len] = data;
 		len++;
 		old_data = data;
 		data += incr;
 		if (data <= old_data)
 			break;
 	}
 
 	if (data != end && data + v1_hdrsize != 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;
 
 	if (!PyArg_ParseTuple(args, "OO", &data_obj, &inlined_obj))
 		return -1;
 	if (!PyString_Check(data_obj)) {
 		PyErr_SetString(PyExc_TypeError, "data is not a string");
 		return -1;
 	}
 	size = PyString_GET_SIZE(data_obj);
 
 	self->inlined = inlined_obj && PyObject_IsTrue(inlined_obj);
 	self->data = data_obj;
 	self->cache = NULL;
 
 	self->added = NULL;
 	self->headrevs = NULL;
 	self->offsets = NULL;
 	self->nt = NULL;
 	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) {
 		long 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_DECREF(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[] = {
 	{"clearcaches", (PyCFunction)index_clearcaches, METH_NOARGS,
 	 "clear the index caches"},
 	{"get", (PyCFunction)index_m_get, METH_VARARGS,
 	 "get an index entry"},
 	{"headrevs", (PyCFunction)index_headrevs, METH_NOARGS,
 	 "get head revisions"},
 	{"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 = {
 	PyObject_HEAD_INIT(NULL)
 	0,                         /* ob_size */
 	"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 (index_file_content, 0), else None
  *
  * added complications are for backwards compatibility
  */
 static 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;
 }
 
 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);
 
 static PyMethodDef methods[] = {
 	{"pack_dirstate", pack_dirstate, METH_VARARGS, "pack a 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"},
 	{"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"},
 	{NULL, NULL}
 };
 
+void dirs_module_init(PyObject *mod);
+
 static void module_init(PyObject *mod)
 {
+	dirs_module_init(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);
 
 	dirstate_unset = Py_BuildValue("ciii", 'n', 0, -1, -1);
 }
 
 #ifdef IS_PY3K
 static struct PyModuleDef parsers_module = {
 	PyModuleDef_HEAD_INIT,
 	"parsers",
 	parsers_doc,
 	-1,
 	methods
 };
 
 PyMODINIT_FUNC PyInit_parsers(void)
 {
 	PyObject *mod = PyModule_Create(&parsers_module);
 	module_init(mod);
 	return mod;
 }
 #else
 PyMODINIT_FUNC initparsers(void)
 {
 	PyObject *mod = Py_InitModule3("parsers", methods, parsers_doc);
 	module_init(mod);
 }
 #endif

mercurial/scmutil.py

@@ -1,934 +1,937
 # scmutil.py - Mercurial core utility functions
 #
 #  Copyright Matt Mackall <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 from i18n import _
 from mercurial.node import nullrev
-import util, error, osutil, revset, similar, encoding, phases
+import util, error, osutil, revset, similar, encoding, phases, parsers
 import match as matchmod
 import os, errno, re, stat, glob
 
 if os.name == 'nt':
     import scmwindows as scmplatform
 else:
     import scmposix as scmplatform
 
 systemrcpath = scmplatform.systemrcpath
 userrcpath = scmplatform.userrcpath
 
 def nochangesfound(ui, repo, excluded=None):
     '''Report no changes for push/pull, excluded is None or a list of
     nodes excluded from the push/pull.
     '''
     secretlist = []
     if excluded:
         for n in excluded:
             if n not in repo:
                 # discovery should not have included the filtered revision,
                 # we have to explicitly exclude it until discovery is cleanup.
                 continue
             ctx = repo[n]
             if ctx.phase() >= phases.secret and not ctx.extinct():
                 secretlist.append(n)
 
     if secretlist:
         ui.status(_("no changes found (ignored %d secret changesets)\n")
                   % len(secretlist))
     else:
         ui.status(_("no changes found\n"))
 
 def checknewlabel(repo, lbl, kind):
     if lbl in ['tip', '.', 'null']:
         raise util.Abort(_("the name '%s' is reserved") % lbl)
     for c in (':', '\0', '\n', '\r'):
         if c in lbl:
             raise util.Abort(_("%r cannot be used in a name") % c)
     try:
         int(lbl)
         raise util.Abort(_("a %s cannot have an integer as its name") % kind)
     except ValueError:
         pass
 
 def checkfilename(f):
     '''Check that the filename f is an acceptable filename for a tracked file'''
     if '\r' in f or '\n' in f:
         raise util.Abort(_("'\\n' and '\\r' disallowed in filenames: %r") % f)
 
 def checkportable(ui, f):
     '''Check if filename f is portable and warn or abort depending on config'''
     checkfilename(f)
     abort, warn = checkportabilityalert(ui)
     if abort or warn:
         msg = util.checkwinfilename(f)
         if msg:
             msg = "%s: %r" % (msg, f)
             if abort:
                 raise util.Abort(msg)
             ui.warn(_("warning: %s\n") % msg)
 
 def checkportabilityalert(ui):
     '''check if the user's config requests nothing, a warning, or abort for
     non-portable filenames'''
     val = ui.config('ui', 'portablefilenames', 'warn')
     lval = val.lower()
     bval = util.parsebool(val)
     abort = os.name == 'nt' or lval == 'abort'
     warn = bval or lval == 'warn'
     if bval is None and not (warn or abort or lval == 'ignore'):
         raise error.ConfigError(
             _("ui.portablefilenames value is invalid ('%s')") % val)
     return abort, warn
 
 class casecollisionauditor(object):
     def __init__(self, ui, abort, dirstate):
         self._ui = ui
         self._abort = abort
         allfiles = '\0'.join(dirstate._map)
         self._loweredfiles = set(encoding.lower(allfiles).split('\0'))
         self._dirstate = dirstate
         # The purpose of _newfiles is so that we don't complain about
         # case collisions if someone were to call this object with the
         # same filename twice.
         self._newfiles = set()
 
     def __call__(self, f):
         fl = encoding.lower(f)
         if (fl in self._loweredfiles and f not in self._dirstate and
             f not in self._newfiles):
             msg = _('possible case-folding collision for %s') % f
             if self._abort:
                 raise util.Abort(msg)
             self._ui.warn(_("warning: %s\n") % msg)
         self._loweredfiles.add(fl)
         self._newfiles.add(f)
 
 class pathauditor(object):
     '''ensure that a filesystem path contains no banned components.
     the following properties of a path are checked:
 
     - ends with a directory separator
     - under top-level .hg
     - starts at the root of a windows drive
     - contains ".."
     - traverses a symlink (e.g. a/symlink_here/b)
     - inside a nested repository (a callback can be used to approve
       some nested repositories, e.g., subrepositories)
     '''
 
     def __init__(self, root, callback=None):
         self.audited = set()
         self.auditeddir = set()
         self.root = root
         self.callback = callback
         if os.path.lexists(root) and not util.checkcase(root):
             self.normcase = util.normcase
         else:
             self.normcase = lambda x: x
 
     def __call__(self, path):
         '''Check the relative path.
         path may contain a pattern (e.g. foodir/**.txt)'''
 
         path = util.localpath(path)
         normpath = self.normcase(path)
         if normpath in self.audited:
             return
         # AIX ignores "/" at end of path, others raise EISDIR.
         if util.endswithsep(path):
             raise util.Abort(_("path ends in directory separator: %s") % path)
         parts = util.splitpath(path)
         if (os.path.splitdrive(path)[0]
             or parts[0].lower() in ('.hg', '.hg.', '')
             or os.pardir in parts):
             raise util.Abort(_("path contains illegal component: %s") % path)
         if '.hg' in path.lower():
             lparts = [p.lower() for p in parts]
             for p in '.hg', '.hg.':
                 if p in lparts[1:]:
                     pos = lparts.index(p)
                     base = os.path.join(*parts[:pos])
                     raise util.Abort(_("path '%s' is inside nested repo %r")
                                      % (path, base))
 
         normparts = util.splitpath(normpath)
         assert len(parts) == len(normparts)
 
         parts.pop()
         normparts.pop()
         prefixes = []
         while parts:
             prefix = os.sep.join(parts)
             normprefix = os.sep.join(normparts)
             if normprefix in self.auditeddir:
                 break
             curpath = os.path.join(self.root, prefix)
             try:
                 st = os.lstat(curpath)
             except OSError, err:
                 # EINVAL can be raised as invalid path syntax under win32.
                 # They must be ignored for patterns can be checked too.
                 if err.errno not in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL):
                     raise
             else:
                 if stat.S_ISLNK(st.st_mode):
                     raise util.Abort(
                         _('path %r traverses symbolic link %r')
                         % (path, prefix))
                 elif (stat.S_ISDIR(st.st_mode) and
                       os.path.isdir(os.path.join(curpath, '.hg'))):
                     if not self.callback or not self.callback(curpath):
                         raise util.Abort(_("path '%s' is inside nested "
                                            "repo %r")
                                          % (path, prefix))
             prefixes.append(normprefix)
             parts.pop()
             normparts.pop()
 
         self.audited.add(normpath)
         # only add prefixes to the cache after checking everything: we don't
         # want to add "foo/bar/baz" before checking if there's a "foo/.hg"
         self.auditeddir.update(prefixes)
 
     def check(self, path):
         try:
             self(path)
             return True
         except (OSError, util.Abort):
             return False
 
 class abstractvfs(object):
     """Abstract base class; cannot be instantiated"""
 
     def __init__(self, *args, **kwargs):
         '''Prevent instantiation; don't call this from subclasses.'''
         raise NotImplementedError('attempted instantiating ' + str(type(self)))
 
     def tryread(self, path):
         '''gracefully return an empty string for missing files'''
         try:
             return self.read(path)
         except IOError, inst:
             if inst.errno != errno.ENOENT:
                 raise
         return ""
 
     def read(self, path):
         fp = self(path, 'rb')
         try:
             return fp.read()
         finally:
             fp.close()
 
     def write(self, path, data):
         fp = self(path, 'wb')
         try:
             return fp.write(data)
         finally:
             fp.close()
 
     def append(self, path, data):
         fp = self(path, 'ab')
         try:
             return fp.write(data)
         finally:
             fp.close()
 
     def exists(self, path=None):
         return os.path.exists(self.join(path))
 
     def isdir(self, path=None):
         return os.path.isdir(self.join(path))
 
     def makedir(self, path=None, notindexed=True):
         return util.makedir(self.join(path), notindexed)
 
     def makedirs(self, path=None, mode=None):
         return util.makedirs(self.join(path), mode)
 
     def mkdir(self, path=None):
         return os.mkdir(self.join(path))
 
     def readdir(self, path=None, stat=None, skip=None):
         return osutil.listdir(self.join(path), stat, skip)
 
     def stat(self, path=None):
         return os.stat(self.join(path))
 
 class vfs(abstractvfs):
     '''Operate files relative to a base directory
 
     This class is used to hide the details of COW semantics and
     remote file access from higher level code.
     '''
     def __init__(self, base, audit=True, expand=False):
         if expand:
             base = os.path.realpath(util.expandpath(base))
         self.base = base
         self._setmustaudit(audit)
         self.createmode = None
         self._trustnlink = None
 
     def _getmustaudit(self):
         return self._audit
 
     def _setmustaudit(self, onoff):
         self._audit = onoff
         if onoff:
             self.audit = pathauditor(self.base)
         else:
             self.audit = util.always
 
     mustaudit = property(_getmustaudit, _setmustaudit)
 
     @util.propertycache
     def _cansymlink(self):
         return util.checklink(self.base)
 
     @util.propertycache
     def _chmod(self):
         return util.checkexec(self.base)
 
     def _fixfilemode(self, name):
         if self.createmode is None or not self._chmod:
             return
         os.chmod(name, self.createmode & 0666)
 
     def __call__(self, path, mode="r", text=False, atomictemp=False):
         if self._audit:
             r = util.checkosfilename(path)
             if r:
                 raise util.Abort("%s: %r" % (r, path))
         self.audit(path)
         f = self.join(path)
 
         if not text and "b" not in mode:
             mode += "b" # for that other OS
 
         nlink = -1
         if mode not in ('r', 'rb'):
             dirname, basename = util.split(f)
             # If basename is empty, then the path is malformed because it points
             # to a directory. Let the posixfile() call below raise IOError.
             if basename:
                 if atomictemp:
                     util.ensuredirs(dirname, self.createmode)
                     return util.atomictempfile(f, mode, self.createmode)
                 try:
                     if 'w' in mode:
                         util.unlink(f)
                         nlink = 0
                     else:
                         # nlinks() may behave differently for files on Windows
                         # shares if the file is open.
                         fd = util.posixfile(f)
                         nlink = util.nlinks(f)
                         if nlink < 1:
                             nlink = 2 # force mktempcopy (issue1922)
                         fd.close()
                 except (OSError, IOError), e:
                     if e.errno != errno.ENOENT:
                         raise
                     nlink = 0
                     util.ensuredirs(dirname, self.createmode)
                 if nlink > 0:
                     if self._trustnlink is None:
                         self._trustnlink = nlink > 1 or util.checknlink(f)
                     if nlink > 1 or not self._trustnlink:
                         util.rename(util.mktempcopy(f), f)
         fp = util.posixfile(f, mode)
         if nlink == 0:
             self._fixfilemode(f)
         return fp
 
     def symlink(self, src, dst):
         self.audit(dst)
         linkname = self.join(dst)
         try:
             os.unlink(linkname)
         except OSError:
             pass
 
         util.ensuredirs(os.path.dirname(linkname), self.createmode)
 
         if self._cansymlink:
             try:
                 os.symlink(src, linkname)
             except OSError, err:
                 raise OSError(err.errno, _('could not symlink to %r: %s') %
                               (src, err.strerror), linkname)
         else:
             self.write(dst, src)
 
     def join(self, path):
         if path:
             return os.path.join(self.base, path)
         else:
             return self.base
 
 opener = vfs
 
 class auditvfs(object):
     def __init__(self, vfs):
         self.vfs = vfs
 
     def _getmustaudit(self):
         return self.vfs.mustaudit
 
     def _setmustaudit(self, onoff):
         self.vfs.mustaudit = onoff
 
     mustaudit = property(_getmustaudit, _setmustaudit)
 
 class filtervfs(abstractvfs, auditvfs):
     '''Wrapper vfs for filtering filenames with a function.'''
 
     def __init__(self, vfs, filter):
         auditvfs.__init__(self, vfs)
         self._filter = filter
 
     def __call__(self, path, *args, **kwargs):
         return self.vfs(self._filter(path), *args, **kwargs)
 
     def join(self, path):
         if path:
             return self.vfs.join(self._filter(path))
         else:
             return self.vfs.join(path)
 
 filteropener = filtervfs
 
 class readonlyvfs(abstractvfs, auditvfs):
     '''Wrapper vfs preventing any writing.'''
 
     def __init__(self, vfs):
         auditvfs.__init__(self, vfs)
 
     def __call__(self, path, mode='r', *args, **kw):
         if mode not in ('r', 'rb'):
             raise util.Abort('this vfs is read only')
         return self.vfs(path, mode, *args, **kw)
 
 
 def canonpath(root, cwd, myname, auditor=None):
     '''return the canonical path of myname, given cwd and root'''
     if util.endswithsep(root):
         rootsep = root
     else:
         rootsep = root + os.sep
     name = myname
     if not os.path.isabs(name):
         name = os.path.join(root, cwd, name)
     name = os.path.normpath(name)
     if auditor is None:
         auditor = pathauditor(root)
     if name != rootsep and name.startswith(rootsep):
         name = name[len(rootsep):]
         auditor(name)
         return util.pconvert(name)
     elif name == root:
         return ''
     else:
         # Determine whether `name' is in the hierarchy at or beneath `root',
         # by iterating name=dirname(name) until that causes no change (can't
         # check name == '/', because that doesn't work on windows). The list
         # `rel' holds the reversed list of components making up the relative
         # file name we want.
         rel = []
         while True:
             try:
                 s = util.samefile(name, root)
             except OSError:
                 s = False
             if s:
                 if not rel:
                     # name was actually the same as root (maybe a symlink)
                     return ''
                 rel.reverse()
                 name = os.path.join(*rel)
                 auditor(name)
                 return util.pconvert(name)
             dirname, basename = util.split(name)
             rel.append(basename)
             if dirname == name:
                 break
             name = dirname
 
         raise util.Abort(_("%s not under root '%s'") % (myname, root))
 
 def walkrepos(path, followsym=False, seen_dirs=None, recurse=False):
     '''yield every hg repository under path, always recursively.
     The recurse flag will only control recursion into repo working dirs'''
     def errhandler(err):
         if err.filename == path:
             raise err
     samestat = getattr(os.path, 'samestat', None)
     if followsym and samestat is not None:
         def adddir(dirlst, dirname):
             match = False
             dirstat = os.stat(dirname)
             for lstdirstat in dirlst:
                 if samestat(dirstat, lstdirstat):
                     match = True
                     break
             if not match:
                 dirlst.append(dirstat)
             return not match
     else:
         followsym = False
 
     if (seen_dirs is None) and followsym:
         seen_dirs = []
         adddir(seen_dirs, path)
     for root, dirs, files in os.walk(path, topdown=True, onerror=errhandler):
         dirs.sort()
         if '.hg' in dirs:
             yield root # found a repository
             qroot = os.path.join(root, '.hg', 'patches')
             if os.path.isdir(os.path.join(qroot, '.hg')):
                 yield qroot # we have a patch queue repo here
             if recurse:
                 # avoid recursing inside the .hg directory
                 dirs.remove('.hg')
             else:
                 dirs[:] = [] # don't descend further
         elif followsym:
             newdirs = []
             for d in dirs:
                 fname = os.path.join(root, d)
                 if adddir(seen_dirs, fname):
                     if os.path.islink(fname):
                         for hgname in walkrepos(fname, True, seen_dirs):
                             yield hgname
                     else:
                         newdirs.append(d)
             dirs[:] = newdirs
 
 def osrcpath():
     '''return default os-specific hgrc search path'''
     path = systemrcpath()
     path.extend(userrcpath())
     path = [os.path.normpath(f) for f in path]
     return path
 
 _rcpath = None
 
 def rcpath():
     '''return hgrc search path. if env var HGRCPATH is set, use it.
     for each item in path, if directory, use files ending in .rc,
     else use item.
     make HGRCPATH empty to only look in .hg/hgrc of current repo.
     if no HGRCPATH, use default os-specific path.'''
     global _rcpath
     if _rcpath is None:
         if 'HGRCPATH' in os.environ:
             _rcpath = []
             for p in os.environ['HGRCPATH'].split(os.pathsep):
                 if not p:
                     continue
                 p = util.expandpath(p)
                 if os.path.isdir(p):
                     for f, kind in osutil.listdir(p):
                         if f.endswith('.rc'):
                             _rcpath.append(os.path.join(p, f))
                 else:
                     _rcpath.append(p)
         else:
             _rcpath = osrcpath()
     return _rcpath
 
 def revsingle(repo, revspec, default='.'):
     if not revspec:
         return repo[default]
 
     l = revrange(repo, [revspec])
     if len(l) < 1:
         raise util.Abort(_('empty revision set'))
     return repo[l[-1]]
 
 def revpair(repo, revs):
     if not revs:
         return repo.dirstate.p1(), None
 
     l = revrange(repo, revs)
 
     if len(l) == 0:
         if revs:
             raise util.Abort(_('empty revision range'))
         return repo.dirstate.p1(), None
 
     if len(l) == 1 and len(revs) == 1 and _revrangesep not in revs[0]:
         return repo.lookup(l[0]), None
 
     return repo.lookup(l[0]), repo.lookup(l[-1])
 
 _revrangesep = ':'
 
 def revrange(repo, revs):
     """Yield revision as strings from a list of revision specifications."""
 
     def revfix(repo, val, defval):
         if not val and val != 0 and defval is not None:
             return defval
         return repo[val].rev()
 
     seen, l = set(), []
     for spec in revs:
         if l and not seen:
             seen = set(l)
         # attempt to parse old-style ranges first to deal with
         # things like old-tag which contain query metacharacters
         try:
             if isinstance(spec, int):
                 seen.add(spec)
                 l.append(spec)
                 continue
 
             if _revrangesep in spec:
                 start, end = spec.split(_revrangesep, 1)
                 start = revfix(repo, start, 0)
                 end = revfix(repo, end, len(repo) - 1)
                 if end == nullrev and start <= 0:
                     start = nullrev
                 rangeiter = repo.changelog.revs(start, end)
                 if not seen and not l:
                     # by far the most common case: revs = ["-1:0"]
                     l = list(rangeiter)
                     # defer syncing seen until next iteration
                     continue
                 newrevs = set(rangeiter)
                 if seen:
                     newrevs.difference_update(seen)
                     seen.update(newrevs)
                 else:
                     seen = newrevs
                 l.extend(sorted(newrevs, reverse=start > end))
                 continue
             elif spec and spec in repo: # single unquoted rev
                 rev = revfix(repo, spec, None)
                 if rev in seen:
                     continue
                 seen.add(rev)
                 l.append(rev)
                 continue
         except error.RepoLookupError:
             pass
 
         # fall through to new-style queries if old-style fails
         m = revset.match(repo.ui, spec)
         dl = [r for r in m(repo, list(repo)) if r not in seen]
         l.extend(dl)
         seen.update(dl)
 
     return l
 
 def expandpats(pats):
     if not util.expandglobs:
         return list(pats)
     ret = []
     for p in pats:
         kind, name = matchmod._patsplit(p, None)
         if kind is None:
             try:
                 globbed = glob.glob(name)
             except re.error:
                 globbed = [name]
             if globbed:
                 ret.extend(globbed)
                 continue
         ret.append(p)
     return ret
 
 def matchandpats(ctx, pats=[], opts={}, globbed=False, default='relpath'):
     if pats == ("",):
         pats = []
     if not globbed and default == 'relpath':
         pats = expandpats(pats or [])
 
     m = ctx.match(pats, opts.get('include'), opts.get('exclude'),
                          default)
     def badfn(f, msg):
         ctx._repo.ui.warn("%s: %s\n" % (m.rel(f), msg))
     m.bad = badfn
     return m, pats
 
 def match(ctx, pats=[], opts={}, globbed=False, default='relpath'):
     return matchandpats(ctx, pats, opts, globbed, default)[0]
 
 def matchall(repo):
     return matchmod.always(repo.root, repo.getcwd())
 
 def matchfiles(repo, files):
     return matchmod.exact(repo.root, repo.getcwd(), files)
 
 def addremove(repo, pats=[], opts={}, dry_run=None, similarity=None):
     if dry_run is None:
         dry_run = opts.get('dry_run')
     if similarity is None:
         similarity = float(opts.get('similarity') or 0)
     # we'd use status here, except handling of symlinks and ignore is tricky
     added, unknown, deleted, removed = [], [], [], []
     audit_path = pathauditor(repo.root)
     m = match(repo[None], pats, opts)
     rejected = []
     m.bad = lambda x, y: rejected.append(x)
 
     ctx = repo[None]
     dirstate = repo.dirstate
     walkresults = dirstate.walk(m, sorted(ctx.substate), True, False)
     for abs, st in walkresults.iteritems():
         dstate = dirstate[abs]
         if dstate == '?' and audit_path.check(abs):
             unknown.append(abs)
         elif dstate != 'r' and not st:
             deleted.append(abs)
         # for finding renames
         elif dstate == 'r':
             removed.append(abs)
         elif dstate == 'a':
             added.append(abs)
 
     unknownset = set(unknown)
     toprint = unknownset.copy()
     toprint.update(deleted)
     for abs in sorted(toprint):
         if repo.ui.verbose or not m.exact(abs):
             rel = m.rel(abs)
             if abs in unknownset:
                 status = _('adding %s\n') % ((pats and rel) or abs)
             else:
                 status = _('removing %s\n') % ((pats and rel) or abs)
             repo.ui.status(status)
 
     copies = {}
     if similarity > 0:
         for old, new, score in similar.findrenames(repo,
                 added + unknown, removed + deleted, similarity):
             if repo.ui.verbose or not m.exact(old) or not m.exact(new):
                 repo.ui.status(_('recording removal of %s as rename to %s '
                                  '(%d%% similar)\n') %
                                (m.rel(old), m.rel(new), score * 100))
             copies[new] = old
 
     if not dry_run:
         wctx = repo[None]
         wlock = repo.wlock()
         try:
             wctx.forget(deleted)
             wctx.add(unknown)
             for new, old in copies.iteritems():
                 wctx.copy(old, new)
         finally:
             wlock.release()
 
     for f in rejected:
         if f in m.files():
             return 1
     return 0
 
 def dirstatecopy(ui, repo, wctx, src, dst, dryrun=False, cwd=None):
     """Update the dirstate to reflect the intent of copying src to dst. For
     different reasons it might not end with dst being marked as copied from src.
     """
     origsrc = repo.dirstate.copied(src) or src
     if dst == origsrc: # copying back a copy?
         if repo.dirstate[dst] not in 'mn' and not dryrun:
             repo.dirstate.normallookup(dst)
     else:
         if repo.dirstate[origsrc] == 'a' and origsrc == src:
             if not ui.quiet:
                 ui.warn(_("%s has not been committed yet, so no copy "
                           "data will be stored for %s.\n")
                         % (repo.pathto(origsrc, cwd), repo.pathto(dst, cwd)))
             if repo.dirstate[dst] in '?r' and not dryrun:
                 wctx.add([dst])
         elif not dryrun:
             wctx.copy(origsrc, dst)
 
 def readrequires(opener, supported):
     '''Reads and parses .hg/requires and checks if all entries found
     are in the list of supported features.'''
     requirements = set(opener.read("requires").splitlines())
     missings = []
     for r in requirements:
         if r not in supported:
             if not r or not r[0].isalnum():
                 raise error.RequirementError(_(".hg/requires file is corrupt"))
             missings.append(r)
     missings.sort()
     if missings:
         raise error.RequirementError(
             _("unknown repository format: requires features '%s' (upgrade "
               "Mercurial)") % "', '".join(missings))
     return requirements
 
 class filecacheentry(object):
     def __init__(self, path, stat=True):
         self.path = path
         self.cachestat = None
         self._cacheable = None
 
         if stat:
             self.cachestat = filecacheentry.stat(self.path)
 
             if self.cachestat:
                 self._cacheable = self.cachestat.cacheable()
             else:
                 # None means we don't know yet
                 self._cacheable = None
 
     def refresh(self):
         if self.cacheable():
             self.cachestat = filecacheentry.stat(self.path)
 
     def cacheable(self):
         if self._cacheable is not None:
             return self._cacheable
 
         # we don't know yet, assume it is for now
         return True
 
     def changed(self):
         # no point in going further if we can't cache it
         if not self.cacheable():
             return True
 
         newstat = filecacheentry.stat(self.path)
 
         # we may not know if it's cacheable yet, check again now
         if newstat and self._cacheable is None:
             self._cacheable = newstat.cacheable()
 
             # check again
             if not self._cacheable:
                 return True
 
         if self.cachestat != newstat:
             self.cachestat = newstat
             return True
         else:
             return False
 
     @staticmethod
     def stat(path):
         try:
             return util.cachestat(path)
         except OSError, e:
             if e.errno != errno.ENOENT:
                 raise
 
 class filecache(object):
     '''A property like decorator that tracks a file under .hg/ for updates.
 
     Records stat info when called in _filecache.
 
     On subsequent calls, compares old stat info with new info, and recreates
     the object when needed, updating the new stat info in _filecache.
 
     Mercurial either atomic renames or appends for files under .hg,
     so to ensure the cache is reliable we need the filesystem to be able
     to tell us if a file has been replaced. If it can't, we fallback to
     recreating the object on every call (essentially the same behaviour as
     propertycache).'''
     def __init__(self, path):
         self.path = path
 
     def join(self, obj, fname):
         """Used to compute the runtime path of the cached file.
 
         Users should subclass filecache and provide their own version of this
         function to call the appropriate join function on 'obj' (an instance
         of the class that its member function was decorated).
         """
         return obj.join(fname)
 
     def __call__(self, func):
         self.func = func
         self.name = func.__name__
         return self
 
     def __get__(self, obj, type=None):
         # do we need to check if the file changed?
         if self.name in obj.__dict__:
             assert self.name in obj._filecache, self.name
             return obj.__dict__[self.name]
 
         entry = obj._filecache.get(self.name)
 
         if entry:
             if entry.changed():
                 entry.obj = self.func(obj)
         else:
             path = self.join(obj, self.path)
 
             # We stat -before- creating the object so our cache doesn't lie if
             # a writer modified between the time we read and stat
             entry = filecacheentry(path)
             entry.obj = self.func(obj)
 
             obj._filecache[self.name] = entry
 
         obj.__dict__[self.name] = entry.obj
         return entry.obj
 
     def __set__(self, obj, value):
         if self.name not in obj._filecache:
             # we add an entry for the missing value because X in __dict__
             # implies X in _filecache
             ce = filecacheentry(self.join(obj, self.path), False)
             obj._filecache[self.name] = ce
         else:
             ce = obj._filecache[self.name]
 
         ce.obj = value # update cached copy
         obj.__dict__[self.name] = value # update copy returned by obj.x
 
     def __delete__(self, obj):
         try:
             del obj.__dict__[self.name]
         except KeyError:
             raise AttributeError(self.name)
 
 class dirs(object):
     '''a multiset of directory names from a dirstate or manifest'''
 
     def __init__(self, map, skip=None):
         self._dirs = {}
         addpath = self.addpath
         if util.safehasattr(map, 'iteritems') and skip is not None:
             for f, s in map.iteritems():
                 if s[0] != skip:
                     addpath(f)
         else:
             for f in map:
                 addpath(f)
 
     def addpath(self, path):
         dirs = self._dirs
         for base in finddirs(path):
             if base in dirs:
                 dirs[base] += 1
                 return
             dirs[base] = 1
 
     def delpath(self, path):
         dirs = self._dirs
         for base in finddirs(path):
             if dirs[base] > 1:
                 dirs[base] -= 1
                 return
             del dirs[base]
 
     def __iter__(self):
         return self._dirs.iterkeys()
 
     def __contains__(self, d):
         return d in self._dirs
 
+if util.safehasattr(parsers, 'dirs'):
+    dirs = parsers.dirs
+
 def finddirs(path):
     pos = path.rfind('/')
     while pos != -1:
         yield path[:pos]
         pos = path.rfind('/', 0, pos)

setup.py

@@ -1,566 +1,567
 #
 # This is the mercurial setup script.
 #
 # 'python setup.py install', or
 # 'python setup.py --help' for more options
 
 import sys, platform
 if getattr(sys, 'version_info', (0, 0, 0)) < (2, 4, 0, 'final'):
     raise SystemExit("Mercurial requires Python 2.4 or later.")
 
 if sys.version_info[0] >= 3:
     def b(s):
         '''A helper function to emulate 2.6+ bytes literals using string
         literals.'''
         return s.encode('latin1')
 else:
     def b(s):
         '''A helper function to emulate 2.6+ bytes literals using string
         literals.'''
         return s
 
 # Solaris Python packaging brain damage
 try:
     import hashlib
     sha = hashlib.sha1()
 except ImportError:
     try:
         import sha
     except ImportError:
         raise SystemExit(
             "Couldn't import standard hashlib (incomplete Python install).")
 
 try:
     import zlib
 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
     except ImportError:
         raise SystemExit(
             "Couldn't import standard bz2 (incomplete Python install).")
 
 import os, subprocess, time
 import shutil
 import tempfile
 from distutils import log
 from distutils.core import setup, 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.install_scripts import install_scripts
 from distutils.spawn import spawn, find_executable
 from distutils.ccompiler import new_compiler
 from distutils import cygwinccompiler
 from distutils.errors import CCompilerError, DistutilsExecError
 from distutils.sysconfig import get_python_inc
 from distutils.version import StrictVersion
 
 convert2to3 = '--c2to3' in sys.argv
 if convert2to3:
     try:
         from distutils.command.build_py import build_py_2to3 as build_py
         from lib2to3.refactor import get_fixers_from_package as getfixers
     except ImportError:
         if sys.version_info[0] < 3:
             raise SystemExit("--c2to3 is only compatible with python3.")
         raise
     sys.path.append('contrib')
 elif sys.version_info[0] >= 3:
     raise SystemExit("setup.py with python3 needs --c2to3 (experimental)")
 
 scripts = ['hg']
 if os.name == 'nt':
     scripts.append('contrib/win32/hg.bat')
 
 # simplified version of distutils.ccompiler.CCompiler.has_function
 # that actually removes its temporary files.
 def hasfunction(cc, funcname):
     tmpdir = tempfile.mkdtemp(prefix='hg-install-')
     devnull = oldstderr = None
     try:
         try:
             fname = os.path.join(tmpdir, 'funcname.c')
             f = open(fname, 'w')
             f.write('int main(void) {\n')
             f.write('    %s();\n' % funcname)
             f.write('}\n')
             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"))
         except Exception:
             return False
         return True
     finally:
         if oldstderr is not None:
             os.dup2(oldstderr, sys.stderr.fileno())
         if devnull is not None:
             devnull.close()
         shutil.rmtree(tmpdir)
 
 # py2exe needs to be installed to work
 try:
     import py2exe
     py2exeloaded = True
     # import py2exe's patched Distribution class
     from distutils.core import Distribution
 except ImportError:
     py2exeloaded = False
 
 def runcmd(cmd, env):
     if sys.platform == 'plan9':
         # subprocess kludge to work around issues in half-baked Python
         # ports, notably bichued/python:
         _, out, err = os.popen3(cmd)
         return str(out), str(err)
     else:
         p = subprocess.Popen(cmd, stdout=subprocess.PIPE,
                              stderr=subprocess.PIPE, env=env)
         out, err = p.communicate()
         return out, err
 
 def runhg(cmd, env):
     out, err = runcmd(cmd, env)
     # 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'))]
     if err:
         print >> sys.stderr, "stderr from '%s':" % (' '.join(cmd))
         print >> sys.stderr, '\n'.join(['  ' + e for e in err])
         return ''
     return out
 
 version = ''
 
 # Execute hg out of this directory with a custom environment which
 # includes the pure Python modules in mercurial/pure. We also take
 # care to not use any hgrc files and do no localization.
 pypath = ['mercurial', os.path.join('mercurial', 'pure')]
 env = {'PYTHONPATH': os.pathsep.join(pypath),
        'HGRCPATH': '',
        'LANGUAGE': 'C'}
 if 'LD_LIBRARY_PATH' in os.environ:
     env['LD_LIBRARY_PATH'] = os.environ['LD_LIBRARY_PATH']
 if 'SystemRoot' in os.environ:
     # Copy SystemRoot into the custom environment for Python 2.6
     # under Windows. Otherwise, the subprocess will fail with
     # error 0xc0150004. See: http://bugs.python.org/issue3440
     env['SystemRoot'] = os.environ['SystemRoot']
 
 if os.path.isdir('.hg'):
     cmd = [sys.executable, 'hg', 'log', '-r', '.', '--template', '{tags}\n']
     numerictags = [t for t in runhg(cmd, env).split() if t[0].isdigit()]
     hgid = runhg([sys.executable, 'hg', 'id', '-i'], env).strip()
     if numerictags: # tag(s) found
         version = numerictags[-1]
         if hgid.endswith('+'): # propagate the dirty status to the tag
             version += '+'
     else: # no tag found
         cmd = [sys.executable, 'hg', 'parents', '--template',
                '{latesttag}+{latesttagdistance}-']
         version = runhg(cmd, env) + 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:
         version = '%(latesttag)s+%(latesttagdistance)s-%(node).12s' % kw
     else:
         version = kw.get('node', '')[:12]
 
 if version:
     f = open("mercurial/__version__.py", "w")
     f.write('# this file is autogenerated by setup.py\n')
     f.write('version = "%s"\n' % version)
     f.close()
 
 
 try:
     from mercurial import __version__
     version = __version__.version
 except ImportError:
     version = 'unknown'
 
 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),
 
     # We also need build_ext before build_py. Otherwise, when 2to3 is
     # called (in build_py), it will not find osutil & friends,
     # thinking that those modules are global and, consequently, making
     # a mess, now that all module imports are global.
 
                     ('build_ext', build.has_ext_modules),
                    ] + 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 = 0
 
     global_options = Distribution.global_options + \
                      [('pure', None, "use pure (slow) Python "
                         "code instead of C extensions"),
                       ('c2to3', None, "(experimental!) convert "
                         "code with 2to3"),
                      ]
 
     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)
 
 class hgbuildext(build_ext):
 
     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 hgbuildpy(build_py):
     if convert2to3:
         fixer_names = sorted(set(getfixers("lib2to3.fixes") +
                                  getfixers("hgfixes")))
 
     def finalize_options(self):
         build_py.finalize_options(self)
 
         if self.distribution.pure:
             if self.py_modules is None:
                 self.py_modules = []
             for ext in self.distribution.ext_modules:
                 if ext.name.startswith("mercurial."):
                     self.py_modules.append("mercurial.pure.%s" % ext.name[10:])
             self.distribution.ext_modules = []
         else:
             if not os.path.exists(os.path.join(get_python_inc(), 'Python.h')):
                 raise SystemExit('Python headers are required to build '
                                  'Mercurial')
 
     def find_modules(self):
         modules = build_py.find_modules(self)
         for module in modules:
             if module[0] == "mercurial.pure":
                 if module[1] != "__init__":
                     yield ("mercurial", module[1], module[2])
             else:
                 yield module
 
 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):
             os.unlink(self._indexfilename)
 
         # here no extension enabled, disabled() lists up everything
         code = ('import pprint; from mercurial import extensions; '
                 'pprint.pprint(extensions.disabled())')
         out, err = runcmd([sys.executable, '-c', code], env)
         if err:
             raise DistutilsExecError(err)
 
         f = open(self._indexfilename, 'w')
         f.write('# this file is autogenerated by setup.py\n')
         f.write('docs = ')
         f.write(out)
         f.close()
 
 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
         hv = sys.hexversion
         pythonlib = 'python%d%d' % (hv >> 24, (hv >> 16) & 0xff)
         f = open('mercurial/hgpythonlib.h', 'wb')
         f.write('/* this file is autogenerated by setup.py */\n')
         f.write('#define HGPYTHONLIB "%s"\n' % pythonlib)
         f.close()
         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)
 
 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)
 
         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:
             fp = open(outfile, 'rb')
             data = fp.read()
             fp.close()
 
             # skip binary files
             if b('\0') in data:
                 continue
 
             data = data.replace('@LIBDIR@', libdir.encode('string_escape'))
             fp = open(outfile, 'wb')
             fp.write(data)
             fp.close()
 
 cmdclass = {'build': hgbuild,
             'build_mo': hgbuildmo,
             'build_ext': hgbuildext,
             'build_py': hgbuildpy,
             'build_hgextindex': buildhgextindex,
             'install_scripts': hginstallscripts,
             'build_hgexe': buildhgexe,
             }
 
 packages = ['mercurial', 'mercurial.hgweb', 'mercurial.httpclient',
             'hgext', 'hgext.convert', 'hgext.highlight', 'hgext.zeroconf',
             'hgext.largefiles']
 
 pymodules = []
 
 extmodules = [
     Extension('mercurial.base85', ['mercurial/base85.c']),
     Extension('mercurial.bdiff', ['mercurial/bdiff.c']),
     Extension('mercurial.diffhelpers', ['mercurial/diffhelpers.c']),
     Extension('mercurial.mpatch', ['mercurial/mpatch.c']),
-    Extension('mercurial.parsers', ['mercurial/parsers.c',
+    Extension('mercurial.parsers', ['mercurial/dirs.c',
+                                    'mercurial/parsers.c',
                                     'mercurial/pathencode.c']),
     ]
 
 osutil_ldflags = []
 
 if sys.platform == 'darwin':
     osutil_ldflags += ['-framework', 'ApplicationServices']
 
 # disable osutil.c under windows + python 2.4 (issue1364)
 if sys.platform == 'win32' and sys.version_info < (2, 5, 0, 'final'):
     pymodules.append('mercurial.pure.osutil')
 else:
     extmodules.append(Extension('mercurial.osutil', ['mercurial/osutil.c'],
                                 extra_link_args=osutil_ldflags))
 
 # the -mno-cygwin option has been deprecated for years
 Mingw32CCompiler = cygwinccompiler.Mingw32CCompiler
 
 class HackedMingw32CCompiler(cygwinccompiler.Mingw32CCompiler):
     def __init__(self, *args, **kwargs):
         Mingw32CCompiler.__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
 
 if sys.platform.startswith('linux') and os.uname()[2] > '2.6':
     # The inotify extension is only usable with Linux 2.6 kernels.
     # You also need a reasonably recent C library.
     # In any case, if it fails to build the error will be skipped ('optional').
     cc = new_compiler()
     if hasfunction(cc, 'inotify_add_watch'):
         inotify = Extension('hgext.inotify.linux._inotify',
                             ['hgext/inotify/linux/_inotify.c'],
                             ['mercurial'])
         inotify.optional = True
         extmodules.append(inotify)
         packages.extend(['hgext.inotify', 'hgext.inotify.linux'])
 
 packagedata = {'mercurial': ['locale/*/LC_MESSAGES/hg.mo',
                              'help/*.txt']}
 
 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 = []
 setupversion = version
 extra = {}
 
 if py2exeloaded:
     extra['console'] = [
         {'script':'hg',
          'copyright':'Copyright (C) 2005-2010 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))
 
 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'):
     # XCode 4.0 dropped support for ppc architecture, which is hardcoded in
     # distutils.sysconfig
     version = runcmd(['/usr/bin/xcodebuild', '-version'], {})[0].splitlines()
     if version:
         version = version[0]
         xcode4 = (version.startswith('Xcode') and
                   StrictVersion(version.split()[1]) >= StrictVersion('4.0'))
     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
 
     if xcode4:
         os.environ['ARCHFLAGS'] = ''
 
 setup(name='mercurial',
       version=setupversion,
       author='Matt Mackall and many others',
       author_email='[e-mail]',
       url='http://mercurial.selenic.com/',
       download_url='http://mercurial.selenic.com/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,
       py_modules=pymodules,
       ext_modules=extmodules,
       data_files=datafiles,
       package_data=packagedata,
       cmdclass=cmdclass,
       distclass=hgdist,
       options=dict(py2exe=dict(packages=['hgext', 'email']),
                    bdist_mpkg=dict(zipdist=True,
                                    license='COPYING',
                                    readme='contrib/macosx/Readme.html',
                                    welcome='contrib/macosx/Welcome.html')),
       **extra)