2009-11-10 21:03:05 +01:00
|
|
|
/* notmuch - Not much of an email program, (just index and search)
|
|
|
|
*
|
|
|
|
* Copyright © 2009 Carl Worth
|
|
|
|
*
|
|
|
|
* This program is free software: you can redistribute it and/or modify
|
|
|
|
* it under the terms of the GNU General Public License as published by
|
|
|
|
* the Free Software Foundation, either version 3 of the License, or
|
|
|
|
* (at your option) any later version.
|
|
|
|
*
|
|
|
|
* This program is distributed in the hope that it will be useful,
|
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
|
|
* GNU General Public License for more details.
|
|
|
|
*
|
|
|
|
* You should have received a copy of the GNU General Public License
|
|
|
|
* along with this program. If not, see http://www.gnu.org/licenses/ .
|
|
|
|
*
|
|
|
|
* Author: Carl Worth <cworth@cworth.org>
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include "notmuch-client.h"
|
|
|
|
|
2009-11-22 01:44:32 +01:00
|
|
|
#include <unistd.h>
|
|
|
|
|
2010-01-06 02:56:11 +01:00
|
|
|
typedef struct _filename_node {
|
|
|
|
char *filename;
|
2011-02-09 00:24:33 +01:00
|
|
|
time_t mtime;
|
2010-01-06 02:56:11 +01:00
|
|
|
struct _filename_node *next;
|
|
|
|
} _filename_node_t;
|
|
|
|
|
|
|
|
typedef struct _filename_list {
|
2011-01-21 10:59:37 +01:00
|
|
|
unsigned count;
|
2010-01-06 02:56:11 +01:00
|
|
|
_filename_node_t *head;
|
|
|
|
_filename_node_t **tail;
|
|
|
|
} _filename_list_t;
|
|
|
|
|
2010-01-06 00:11:21 +01:00
|
|
|
typedef struct {
|
|
|
|
int output_is_a_tty;
|
2012-12-03 21:56:13 +01:00
|
|
|
notmuch_bool_t verbose;
|
|
|
|
notmuch_bool_t debug;
|
2010-04-23 18:19:52 +02:00
|
|
|
const char **new_tags;
|
2010-03-12 23:13:26 +01:00
|
|
|
size_t new_tags_length;
|
2012-02-15 10:17:31 +01:00
|
|
|
const char **new_ignore;
|
|
|
|
size_t new_ignore_length;
|
2010-01-06 00:11:21 +01:00
|
|
|
|
|
|
|
int total_files;
|
|
|
|
int processed_files;
|
2011-06-11 18:54:05 +02:00
|
|
|
int added_messages, removed_messages, renamed_messages;
|
2010-01-06 00:11:21 +01:00
|
|
|
struct timeval tv_start;
|
2010-01-06 02:56:11 +01:00
|
|
|
|
|
|
|
_filename_list_t *removed_files;
|
|
|
|
_filename_list_t *removed_directories;
|
2011-02-09 00:24:33 +01:00
|
|
|
_filename_list_t *directory_mtimes;
|
2010-11-11 11:49:24 +01:00
|
|
|
|
2010-11-11 01:26:14 +01:00
|
|
|
notmuch_bool_t synchronize_flags;
|
2010-01-06 00:11:21 +01:00
|
|
|
} add_files_state_t;
|
|
|
|
|
2011-01-21 10:59:37 +01:00
|
|
|
static volatile sig_atomic_t do_print_progress = 0;
|
2009-11-12 04:50:15 +01:00
|
|
|
|
|
|
|
static void
|
|
|
|
handle_sigalrm (unused (int signal))
|
|
|
|
{
|
2011-01-21 10:59:37 +01:00
|
|
|
do_print_progress = 1;
|
2009-11-12 04:50:15 +01:00
|
|
|
}
|
|
|
|
|
2009-11-13 18:05:13 +01:00
|
|
|
static volatile sig_atomic_t interrupted;
|
|
|
|
|
|
|
|
static void
|
|
|
|
handle_sigint (unused (int sig))
|
|
|
|
{
|
|
|
|
static char msg[] = "Stopping... \n";
|
2009-11-23 07:03:35 +01:00
|
|
|
|
2012-01-19 23:29:19 +01:00
|
|
|
/* This write is "opportunistic", so it's okay to ignore the
|
|
|
|
* result. It is not required for correctness, and if it does
|
|
|
|
* fail or produce a short write, we want to get out of the signal
|
|
|
|
* handler as quickly as possible, not retry it. */
|
|
|
|
IGNORE_RESULT (write (2, msg, sizeof(msg)-1));
|
2009-11-13 18:05:13 +01:00
|
|
|
interrupted = 1;
|
|
|
|
}
|
|
|
|
|
2010-01-06 02:56:11 +01:00
|
|
|
static _filename_list_t *
|
|
|
|
_filename_list_create (const void *ctx)
|
|
|
|
{
|
|
|
|
_filename_list_t *list;
|
|
|
|
|
|
|
|
list = talloc (ctx, _filename_list_t);
|
|
|
|
if (list == NULL)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
list->head = NULL;
|
|
|
|
list->tail = &list->head;
|
2011-01-21 10:59:37 +01:00
|
|
|
list->count = 0;
|
2010-01-06 02:56:11 +01:00
|
|
|
|
|
|
|
return list;
|
|
|
|
}
|
|
|
|
|
2011-02-09 00:24:33 +01:00
|
|
|
static _filename_node_t *
|
2010-01-06 02:56:11 +01:00
|
|
|
_filename_list_add (_filename_list_t *list,
|
|
|
|
const char *filename)
|
|
|
|
{
|
|
|
|
_filename_node_t *node = talloc (list, _filename_node_t);
|
|
|
|
|
2011-01-21 10:59:37 +01:00
|
|
|
list->count++;
|
|
|
|
|
2010-01-06 02:56:11 +01:00
|
|
|
node->filename = talloc_strdup (list, filename);
|
|
|
|
node->next = NULL;
|
|
|
|
|
|
|
|
*(list->tail) = node;
|
|
|
|
list->tail = &node->next;
|
2011-02-09 00:24:33 +01:00
|
|
|
|
|
|
|
return node;
|
2010-01-06 02:56:11 +01:00
|
|
|
}
|
|
|
|
|
2009-11-12 04:50:15 +01:00
|
|
|
static void
|
2011-01-21 10:59:37 +01:00
|
|
|
generic_print_progress (const char *action, const char *object,
|
|
|
|
struct timeval tv_start, unsigned processed, unsigned total)
|
2009-11-12 04:50:15 +01:00
|
|
|
{
|
|
|
|
struct timeval tv_now;
|
|
|
|
double elapsed_overall, rate_overall;
|
|
|
|
|
|
|
|
gettimeofday (&tv_now, NULL);
|
|
|
|
|
2011-01-21 10:59:37 +01:00
|
|
|
elapsed_overall = notmuch_time_elapsed (tv_start, tv_now);
|
|
|
|
rate_overall = processed / elapsed_overall;
|
2009-11-12 04:50:15 +01:00
|
|
|
|
2011-01-21 10:59:37 +01:00
|
|
|
printf ("%s %d ", action, processed);
|
2009-11-12 04:50:15 +01:00
|
|
|
|
2011-01-21 10:59:37 +01:00
|
|
|
if (total) {
|
2011-01-26 14:06:57 +01:00
|
|
|
printf ("of %d %s", total, object);
|
|
|
|
if (processed > 0 && elapsed_overall > 0.5) {
|
|
|
|
double time_remaining = ((total - processed) / rate_overall);
|
|
|
|
printf (" (");
|
|
|
|
notmuch_time_print_formatted_seconds (time_remaining);
|
|
|
|
printf (" remaining)");
|
|
|
|
}
|
2009-11-12 04:50:15 +01:00
|
|
|
} else {
|
2011-01-26 14:06:57 +01:00
|
|
|
printf ("%s", object);
|
|
|
|
if (elapsed_overall > 0.5)
|
|
|
|
printf (" (%d %s/sec.)", (int) rate_overall, object);
|
2009-11-12 04:50:15 +01:00
|
|
|
}
|
2011-01-26 14:06:57 +01:00
|
|
|
printf (".\033[K\r");
|
2009-11-12 04:50:15 +01:00
|
|
|
|
|
|
|
fflush (stdout);
|
|
|
|
}
|
|
|
|
|
2010-01-06 02:43:03 +01:00
|
|
|
static int
|
|
|
|
dirent_sort_inode (const struct dirent **a, const struct dirent **b)
|
Read mail directory in inode number order
This gives a rather decent reduction in number of seeks required when
reading a Maildir that isn't in pagecache.
Most filesystems give some locality on disk based on inode numbers.
In ext[234] this is the inode tables, in XFS groups of sequential inode
numbers are together on disk and the most significant bits indicate
allocation group (i.e inode 1,000,000 is always after inode 1,000).
With this patch, we read in the whole directory, sort by inode number
before stat()ing the contents.
Ideally, directory is sequential and then we make one scan through the
file system stat()ing.
Since the universe is not ideal, we'll probably seek during reading the
directory and a fair bit while reading the inodes themselves.
However... with readahead, and stat()ing in inode order, we should be
in the best place possible to hit the cache.
In a (not very good) benchmark of "how long does it take to find the first
15,000 messages in my Maildir after 'echo 3 > /proc/sys/vm/drop_caches'",
this patch consistently cut at least 8 seconds off the scan time.
Without patch: 50 seconds
With patch: 38-42 seconds.
(I did this in a previous maildir reading project and saw large improvements too)
2009-11-18 02:56:40 +01:00
|
|
|
{
|
2009-11-18 22:27:32 +01:00
|
|
|
return ((*a)->d_ino < (*b)->d_ino) ? -1 : 1;
|
Read mail directory in inode number order
This gives a rather decent reduction in number of seeks required when
reading a Maildir that isn't in pagecache.
Most filesystems give some locality on disk based on inode numbers.
In ext[234] this is the inode tables, in XFS groups of sequential inode
numbers are together on disk and the most significant bits indicate
allocation group (i.e inode 1,000,000 is always after inode 1,000).
With this patch, we read in the whole directory, sort by inode number
before stat()ing the contents.
Ideally, directory is sequential and then we make one scan through the
file system stat()ing.
Since the universe is not ideal, we'll probably seek during reading the
directory and a fair bit while reading the inodes themselves.
However... with readahead, and stat()ing in inode order, we should be
in the best place possible to hit the cache.
In a (not very good) benchmark of "how long does it take to find the first
15,000 messages in my Maildir after 'echo 3 > /proc/sys/vm/drop_caches'",
this patch consistently cut at least 8 seconds off the scan time.
Without patch: 50 seconds
With patch: 38-42 seconds.
(I did this in a previous maildir reading project and saw large improvements too)
2009-11-18 02:56:40 +01:00
|
|
|
}
|
|
|
|
|
2010-01-06 02:43:03 +01:00
|
|
|
static int
|
|
|
|
dirent_sort_strcmp_name (const struct dirent **a, const struct dirent **b)
|
|
|
|
{
|
|
|
|
return strcmp ((*a)->d_name, (*b)->d_name);
|
|
|
|
}
|
|
|
|
|
2012-05-25 00:01:11 +02:00
|
|
|
/* Return the type of a directory entry relative to path as a stat(2)
|
|
|
|
* mode. Like stat, this follows symlinks. Returns -1 and sets errno
|
|
|
|
* if the file's type cannot be determined (which includes dangling
|
|
|
|
* symlinks).
|
|
|
|
*/
|
|
|
|
static int
|
|
|
|
dirent_type (const char *path, const struct dirent *entry)
|
|
|
|
{
|
|
|
|
struct stat statbuf;
|
|
|
|
char *abspath;
|
|
|
|
int err, saved_errno;
|
|
|
|
|
|
|
|
#ifdef _DIRENT_HAVE_D_TYPE
|
|
|
|
/* Mapping from d_type to stat mode_t. We omit DT_LNK so that
|
|
|
|
* we'll fall through to stat and get the real file type. */
|
|
|
|
static const mode_t modes[] = {
|
|
|
|
[DT_BLK] = S_IFBLK,
|
|
|
|
[DT_CHR] = S_IFCHR,
|
|
|
|
[DT_DIR] = S_IFDIR,
|
|
|
|
[DT_FIFO] = S_IFIFO,
|
|
|
|
[DT_REG] = S_IFREG,
|
|
|
|
[DT_SOCK] = S_IFSOCK
|
|
|
|
};
|
|
|
|
if (entry->d_type < ARRAY_SIZE(modes) && modes[entry->d_type])
|
|
|
|
return modes[entry->d_type];
|
|
|
|
#endif
|
|
|
|
|
|
|
|
abspath = talloc_asprintf (NULL, "%s/%s", path, entry->d_name);
|
|
|
|
if (!abspath) {
|
|
|
|
errno = ENOMEM;
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
err = stat(abspath, &statbuf);
|
|
|
|
saved_errno = errno;
|
|
|
|
talloc_free (abspath);
|
|
|
|
if (err < 0) {
|
|
|
|
errno = saved_errno;
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
return statbuf.st_mode & S_IFMT;
|
|
|
|
}
|
|
|
|
|
2009-11-25 22:11:45 +01:00
|
|
|
/* Test if the directory looks like a Maildir directory.
|
|
|
|
*
|
|
|
|
* Search through the array of directory entries to see if we can find all
|
|
|
|
* three subdirectories typical for Maildir, that is "new", "cur", and "tmp".
|
|
|
|
*
|
|
|
|
* Return 1 if the directory looks like a Maildir and 0 otherwise.
|
|
|
|
*/
|
2009-11-28 04:38:46 +01:00
|
|
|
static int
|
2012-05-25 00:01:11 +02:00
|
|
|
_entries_resemble_maildir (const char *path, struct dirent **entries, int count)
|
2009-11-25 22:11:45 +01:00
|
|
|
{
|
|
|
|
int i, found = 0;
|
|
|
|
|
|
|
|
for (i = 0; i < count; i++) {
|
2012-05-25 00:01:11 +02:00
|
|
|
if (dirent_type (path, entries[i]) != S_IFDIR)
|
2010-01-06 01:42:14 +01:00
|
|
|
continue;
|
|
|
|
|
2009-11-25 22:11:45 +01:00
|
|
|
if (strcmp(entries[i]->d_name, "new") == 0 ||
|
|
|
|
strcmp(entries[i]->d_name, "cur") == 0 ||
|
|
|
|
strcmp(entries[i]->d_name, "tmp") == 0)
|
|
|
|
{
|
|
|
|
found++;
|
2009-11-28 04:38:46 +01:00
|
|
|
if (found == 3)
|
|
|
|
return 1;
|
2009-11-25 22:11:45 +01:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2012-02-15 10:17:31 +01:00
|
|
|
/* Test if the file/directory is to be ignored.
|
|
|
|
*/
|
|
|
|
static notmuch_bool_t
|
|
|
|
_entry_in_ignore_list (const char *entry, add_files_state_t *state)
|
|
|
|
{
|
|
|
|
size_t i;
|
|
|
|
|
|
|
|
for (i = 0; i < state->new_ignore_length; i++)
|
|
|
|
if (strcmp (entry, state->new_ignore[i]) == 0)
|
|
|
|
return TRUE;
|
|
|
|
|
|
|
|
return FALSE;
|
|
|
|
}
|
|
|
|
|
2009-11-12 04:50:15 +01:00
|
|
|
/* Examine 'path' recursively as follows:
|
|
|
|
*
|
2010-01-06 00:31:56 +01:00
|
|
|
* o Ask the filesystem for the mtime of 'path' (fs_mtime)
|
|
|
|
* o Ask the database for its timestamp of 'path' (db_mtime)
|
2009-11-12 04:50:15 +01:00
|
|
|
*
|
2010-01-06 02:43:03 +01:00
|
|
|
* o Ask the filesystem for files and directories within 'path'
|
|
|
|
* (via scandir and stored in fs_entries)
|
2010-01-06 01:35:02 +01:00
|
|
|
*
|
2010-01-06 02:43:03 +01:00
|
|
|
* o Pass 1: For each directory in fs_entries, recursively call into
|
|
|
|
* this same function.
|
2009-11-12 04:50:15 +01:00
|
|
|
*
|
2011-03-10 20:56:16 +01:00
|
|
|
* o Compare fs_mtime to db_mtime. If they are equivalent, terminate
|
|
|
|
* the algorithm at this point, (this directory has not been
|
|
|
|
* updated in the filesystem since the last database scan of PASS
|
|
|
|
* 2).
|
|
|
|
*
|
|
|
|
* o Ask the database for files and directories within 'path'
|
|
|
|
* (db_files and db_subdirs)
|
|
|
|
*
|
|
|
|
* o Pass 2: Walk fs_entries simultaneously with db_files and
|
|
|
|
* db_subdirs. Look for one of three interesting cases:
|
2009-11-12 04:50:15 +01:00
|
|
|
*
|
2010-01-06 02:43:03 +01:00
|
|
|
* 1. Regular file in fs_entries and not in db_files
|
|
|
|
* This is a new file to add_message into the database.
|
2009-11-12 04:50:15 +01:00
|
|
|
*
|
2010-01-06 02:43:03 +01:00
|
|
|
* 2. Filename in db_files not in fs_entries.
|
|
|
|
* This is a file that has been removed from the mail store.
|
|
|
|
*
|
|
|
|
* 3. Directory in db_subdirs not in fs_entries
|
|
|
|
* This is a directory that has been removed from the mail store.
|
|
|
|
*
|
|
|
|
* Note that the addition of a directory is not interesting here,
|
|
|
|
* since that will have been taken care of in pass 1. Also, we
|
|
|
|
* don't immediately act on file/directory removal since we must
|
|
|
|
* ensure that in the case of a rename that the new filename is
|
|
|
|
* added before the old filename is removed, (so that no
|
|
|
|
* information is lost from the database).
|
|
|
|
*
|
|
|
|
* o Tell the database to update its time of 'path' to 'fs_mtime'
|
2011-06-29 09:10:54 +02:00
|
|
|
* if fs_mtime isn't the current wall-clock time.
|
2009-11-12 04:50:15 +01:00
|
|
|
*/
|
|
|
|
static notmuch_status_t
|
2012-05-25 00:01:13 +02:00
|
|
|
add_files (notmuch_database_t *notmuch,
|
|
|
|
const char *path,
|
|
|
|
add_files_state_t *state)
|
2009-11-12 04:50:15 +01:00
|
|
|
{
|
|
|
|
DIR *dir = NULL;
|
Read mail directory in inode number order
This gives a rather decent reduction in number of seeks required when
reading a Maildir that isn't in pagecache.
Most filesystems give some locality on disk based on inode numbers.
In ext[234] this is the inode tables, in XFS groups of sequential inode
numbers are together on disk and the most significant bits indicate
allocation group (i.e inode 1,000,000 is always after inode 1,000).
With this patch, we read in the whole directory, sort by inode number
before stat()ing the contents.
Ideally, directory is sequential and then we make one scan through the
file system stat()ing.
Since the universe is not ideal, we'll probably seek during reading the
directory and a fair bit while reading the inodes themselves.
However... with readahead, and stat()ing in inode order, we should be
in the best place possible to hit the cache.
In a (not very good) benchmark of "how long does it take to find the first
15,000 messages in my Maildir after 'echo 3 > /proc/sys/vm/drop_caches'",
this patch consistently cut at least 8 seconds off the scan time.
Without patch: 50 seconds
With patch: 38-42 seconds.
(I did this in a previous maildir reading project and saw large improvements too)
2009-11-18 02:56:40 +01:00
|
|
|
struct dirent *entry = NULL;
|
2009-11-12 04:50:15 +01:00
|
|
|
char *next = NULL;
|
2010-01-06 00:31:56 +01:00
|
|
|
time_t fs_mtime, db_mtime;
|
2009-11-12 04:50:15 +01:00
|
|
|
notmuch_status_t status, ret = NOTMUCH_STATUS_SUCCESS;
|
|
|
|
notmuch_message_t *message = NULL;
|
2010-01-06 01:15:43 +01:00
|
|
|
struct dirent **fs_entries = NULL;
|
2012-05-25 00:01:11 +02:00
|
|
|
int i, num_fs_entries = 0, entry_type;
|
2010-01-05 22:29:23 +01:00
|
|
|
notmuch_directory_t *directory;
|
2010-01-06 02:43:03 +01:00
|
|
|
notmuch_filenames_t *db_files = NULL;
|
|
|
|
notmuch_filenames_t *db_subdirs = NULL;
|
2011-06-29 09:10:54 +02:00
|
|
|
time_t stat_time;
|
2010-01-06 01:06:46 +01:00
|
|
|
struct stat st;
|
2012-05-18 06:13:38 +02:00
|
|
|
notmuch_bool_t is_maildir;
|
2010-03-12 23:13:26 +01:00
|
|
|
const char **tag;
|
2009-11-12 04:50:15 +01:00
|
|
|
|
2010-01-06 01:06:46 +01:00
|
|
|
if (stat (path, &st)) {
|
|
|
|
fprintf (stderr, "Error reading directory %s: %s\n",
|
|
|
|
path, strerror (errno));
|
|
|
|
return NOTMUCH_STATUS_FILE_ERROR;
|
|
|
|
}
|
2011-06-29 09:10:54 +02:00
|
|
|
stat_time = time (NULL);
|
2010-01-06 01:06:46 +01:00
|
|
|
|
2012-05-25 00:01:12 +02:00
|
|
|
if (! S_ISDIR (st.st_mode)) {
|
|
|
|
fprintf (stderr, "Error: %s is not a directory.\n", path);
|
|
|
|
return NOTMUCH_STATUS_FILE_ERROR;
|
|
|
|
}
|
2010-01-06 01:06:46 +01:00
|
|
|
|
|
|
|
fs_mtime = st.st_mtime;
|
2009-11-12 04:50:15 +01:00
|
|
|
|
2012-05-14 01:36:09 +02:00
|
|
|
status = notmuch_database_get_directory (notmuch, path, &directory);
|
|
|
|
if (status) {
|
|
|
|
ret = status;
|
|
|
|
goto DONE;
|
|
|
|
}
|
|
|
|
db_mtime = directory ? notmuch_directory_get_mtime (directory) : 0;
|
2010-01-06 23:35:56 +01:00
|
|
|
|
2010-01-06 02:43:03 +01:00
|
|
|
/* If the database knows about this directory, then we sort based
|
|
|
|
* on strcmp to match the database sorting. Otherwise, we can do
|
|
|
|
* inode-based sorting for faster filesystem operation. */
|
|
|
|
num_fs_entries = scandir (path, &fs_entries, 0,
|
2012-05-18 06:13:38 +02:00
|
|
|
directory ?
|
|
|
|
dirent_sort_strcmp_name : dirent_sort_inode);
|
2009-11-12 04:50:15 +01:00
|
|
|
|
2010-01-06 01:15:43 +01:00
|
|
|
if (num_fs_entries == -1) {
|
2009-11-12 04:50:15 +01:00
|
|
|
fprintf (stderr, "Error opening directory %s: %s\n",
|
|
|
|
path, strerror (errno));
|
2012-04-22 17:50:49 +02:00
|
|
|
/* We consider this a fatal error because, if a user moved a
|
|
|
|
* message from another directory that we were able to scan
|
|
|
|
* into this directory, skipping this directory will cause
|
|
|
|
* that message to be lost. */
|
2009-11-12 04:50:15 +01:00
|
|
|
ret = NOTMUCH_STATUS_FILE_ERROR;
|
|
|
|
goto DONE;
|
|
|
|
}
|
|
|
|
|
2010-01-06 02:43:03 +01:00
|
|
|
/* Pass 1: Recurse into all sub-directories. */
|
2012-05-25 00:01:11 +02:00
|
|
|
is_maildir = _entries_resemble_maildir (path, fs_entries, num_fs_entries);
|
2010-01-06 01:42:14 +01:00
|
|
|
|
2010-01-06 01:35:02 +01:00
|
|
|
for (i = 0; i < num_fs_entries; i++) {
|
|
|
|
if (interrupted)
|
2009-11-12 04:50:15 +01:00
|
|
|
break;
|
|
|
|
|
2010-01-06 01:35:02 +01:00
|
|
|
entry = fs_entries[i];
|
Read mail directory in inode number order
This gives a rather decent reduction in number of seeks required when
reading a Maildir that isn't in pagecache.
Most filesystems give some locality on disk based on inode numbers.
In ext[234] this is the inode tables, in XFS groups of sequential inode
numbers are together on disk and the most significant bits indicate
allocation group (i.e inode 1,000,000 is always after inode 1,000).
With this patch, we read in the whole directory, sort by inode number
before stat()ing the contents.
Ideally, directory is sequential and then we make one scan through the
file system stat()ing.
Since the universe is not ideal, we'll probably seek during reading the
directory and a fair bit while reading the inodes themselves.
However... with readahead, and stat()ing in inode order, we should be
in the best place possible to hit the cache.
In a (not very good) benchmark of "how long does it take to find the first
15,000 messages in my Maildir after 'echo 3 > /proc/sys/vm/drop_caches'",
this patch consistently cut at least 8 seconds off the scan time.
Without patch: 50 seconds
With patch: 38-42 seconds.
(I did this in a previous maildir reading project and saw large improvements too)
2009-11-18 02:56:40 +01:00
|
|
|
|
2012-11-25 06:25:45 +01:00
|
|
|
/* Ignore any files/directories the user has configured to
|
|
|
|
* ignore. We do this before dirent_type both for performance
|
|
|
|
* and because we don't care if dirent_type fails on entries
|
|
|
|
* that are explicitly ignored.
|
|
|
|
*/
|
|
|
|
if (_entry_in_ignore_list (entry->d_name, state)) {
|
|
|
|
if (state->debug)
|
|
|
|
printf ("(D) add_files_recursive, pass 1: explicitly ignoring %s/%s\n",
|
|
|
|
path, entry->d_name);
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
2012-05-25 00:01:11 +02:00
|
|
|
/* We only want to descend into directories (and symlinks to
|
|
|
|
* directories). */
|
|
|
|
entry_type = dirent_type (path, entry);
|
|
|
|
if (entry_type == -1) {
|
|
|
|
/* Be pessimistic, e.g. so we don't lose lots of mail just
|
|
|
|
* because a user broke a symlink. */
|
|
|
|
fprintf (stderr, "Error reading file %s/%s: %s\n",
|
|
|
|
path, entry->d_name, strerror (errno));
|
|
|
|
return NOTMUCH_STATUS_FILE_ERROR;
|
|
|
|
} else if (entry_type != S_IFDIR) {
|
2009-11-12 04:50:15 +01:00
|
|
|
continue;
|
2010-01-23 06:58:30 +01:00
|
|
|
}
|
2009-11-12 04:50:15 +01:00
|
|
|
|
|
|
|
/* Ignore special directories to avoid infinite recursion.
|
2012-11-25 06:25:45 +01:00
|
|
|
* Also ignore the .notmuch directory and any "tmp" directory
|
|
|
|
* that appears within a maildir.
|
2009-11-12 04:50:15 +01:00
|
|
|
*/
|
|
|
|
if (strcmp (entry->d_name, ".") == 0 ||
|
|
|
|
strcmp (entry->d_name, "..") == 0 ||
|
2010-01-06 01:42:14 +01:00
|
|
|
(is_maildir && strcmp (entry->d_name, "tmp") == 0) ||
|
2012-11-25 06:25:45 +01:00
|
|
|
strcmp (entry->d_name, ".notmuch") == 0)
|
2009-11-12 04:50:15 +01:00
|
|
|
continue;
|
|
|
|
|
|
|
|
next = talloc_asprintf (notmuch, "%s/%s", path, entry->d_name);
|
2012-05-25 00:01:13 +02:00
|
|
|
status = add_files (notmuch, next, state);
|
2012-04-22 17:50:49 +02:00
|
|
|
if (status) {
|
2010-01-06 01:35:02 +01:00
|
|
|
ret = status;
|
2012-04-22 17:50:49 +02:00
|
|
|
goto DONE;
|
|
|
|
}
|
2010-01-06 01:35:02 +01:00
|
|
|
talloc_free (next);
|
|
|
|
next = NULL;
|
|
|
|
}
|
|
|
|
|
2010-12-05 10:40:16 +01:00
|
|
|
/* If the directory's modification time in the filesystem is the
|
|
|
|
* same as what we recorded in the database the last time we
|
|
|
|
* scanned it, then we can skip the second pass entirely.
|
|
|
|
*
|
|
|
|
* We test for strict equality here to avoid a bug that can happen
|
|
|
|
* if the system clock jumps backward, (preventing new mail from
|
|
|
|
* being discovered until the clock catches up and the directory
|
|
|
|
* is modified again).
|
|
|
|
*/
|
2012-05-18 06:13:38 +02:00
|
|
|
if (directory && fs_mtime == db_mtime)
|
2010-01-06 01:35:02 +01:00
|
|
|
goto DONE;
|
|
|
|
|
2012-05-18 06:13:38 +02:00
|
|
|
/* If the database has never seen this directory before, we can
|
|
|
|
* simply leave db_files and db_subdirs NULL. */
|
|
|
|
if (directory) {
|
2011-02-04 22:44:31 +01:00
|
|
|
db_files = notmuch_directory_get_child_files (directory);
|
|
|
|
db_subdirs = notmuch_directory_get_child_directories (directory);
|
|
|
|
}
|
|
|
|
|
2010-01-06 02:43:03 +01:00
|
|
|
/* Pass 2: Scan for new files, removed files, and removed directories. */
|
|
|
|
for (i = 0; i < num_fs_entries; i++)
|
|
|
|
{
|
2010-01-06 01:35:02 +01:00
|
|
|
if (interrupted)
|
|
|
|
break;
|
|
|
|
|
|
|
|
entry = fs_entries[i];
|
|
|
|
|
2012-02-15 10:17:31 +01:00
|
|
|
/* Ignore files & directories user has configured to be ignored */
|
2012-02-19 21:47:54 +01:00
|
|
|
if (_entry_in_ignore_list (entry->d_name, state)) {
|
|
|
|
if (state->debug)
|
|
|
|
printf ("(D) add_files_recursive, pass 2: explicitly ignoring %s/%s\n",
|
|
|
|
path,
|
|
|
|
entry->d_name);
|
2012-02-15 10:17:31 +01:00
|
|
|
continue;
|
2012-02-19 21:47:54 +01:00
|
|
|
}
|
2012-02-15 10:17:31 +01:00
|
|
|
|
2010-01-06 02:43:03 +01:00
|
|
|
/* Check if we've walked past any names in db_files or
|
|
|
|
* db_subdirs. If so, these have been deleted. */
|
2010-03-09 18:22:29 +01:00
|
|
|
while (notmuch_filenames_valid (db_files) &&
|
2010-01-06 02:43:03 +01:00
|
|
|
strcmp (notmuch_filenames_get (db_files), entry->d_name) < 0)
|
|
|
|
{
|
2010-01-06 02:56:11 +01:00
|
|
|
char *absolute = talloc_asprintf (state->removed_files,
|
|
|
|
"%s/%s", path,
|
|
|
|
notmuch_filenames_get (db_files));
|
|
|
|
|
|
|
|
_filename_list_add (state->removed_files, absolute);
|
2010-01-06 02:43:03 +01:00
|
|
|
|
2010-03-09 18:22:29 +01:00
|
|
|
notmuch_filenames_move_to_next (db_files);
|
2010-01-06 02:43:03 +01:00
|
|
|
}
|
|
|
|
|
2010-03-09 18:22:29 +01:00
|
|
|
while (notmuch_filenames_valid (db_subdirs) &&
|
2010-01-06 02:43:03 +01:00
|
|
|
strcmp (notmuch_filenames_get (db_subdirs), entry->d_name) <= 0)
|
|
|
|
{
|
2010-01-06 02:56:11 +01:00
|
|
|
const char *filename = notmuch_filenames_get (db_subdirs);
|
|
|
|
|
|
|
|
if (strcmp (filename, entry->d_name) < 0)
|
|
|
|
{
|
|
|
|
char *absolute = talloc_asprintf (state->removed_directories,
|
|
|
|
"%s/%s", path, filename);
|
|
|
|
|
|
|
|
_filename_list_add (state->removed_directories, absolute);
|
|
|
|
}
|
2010-01-06 02:43:03 +01:00
|
|
|
|
2010-03-09 18:22:29 +01:00
|
|
|
notmuch_filenames_move_to_next (db_subdirs);
|
2010-01-06 02:43:03 +01:00
|
|
|
}
|
|
|
|
|
2012-05-25 00:01:11 +02:00
|
|
|
/* Only add regular files (and symlinks to regular files). */
|
|
|
|
entry_type = dirent_type (path, entry);
|
|
|
|
if (entry_type == -1) {
|
|
|
|
fprintf (stderr, "Error reading file %s/%s: %s\n",
|
|
|
|
path, entry->d_name, strerror (errno));
|
|
|
|
return NOTMUCH_STATUS_FILE_ERROR;
|
|
|
|
} else if (entry_type != S_IFREG) {
|
2010-01-06 01:35:02 +01:00
|
|
|
continue;
|
2010-01-06 19:30:08 +01:00
|
|
|
}
|
2010-01-06 01:35:02 +01:00
|
|
|
|
2010-01-06 02:43:03 +01:00
|
|
|
/* Don't add a file that we've added before. */
|
2010-03-09 18:22:29 +01:00
|
|
|
if (notmuch_filenames_valid (db_files) &&
|
2010-01-06 02:43:03 +01:00
|
|
|
strcmp (notmuch_filenames_get (db_files), entry->d_name) == 0)
|
|
|
|
{
|
2010-03-09 18:22:29 +01:00
|
|
|
notmuch_filenames_move_to_next (db_files);
|
2010-01-06 02:43:03 +01:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
2010-01-06 19:08:51 +01:00
|
|
|
/* We're now looking at a regular file that doesn't yet exist
|
2010-01-06 02:43:03 +01:00
|
|
|
* in the database, so add it. */
|
2010-01-06 01:35:02 +01:00
|
|
|
next = talloc_asprintf (notmuch, "%s/%s", path, entry->d_name);
|
|
|
|
|
|
|
|
state->processed_files++;
|
|
|
|
|
|
|
|
if (state->verbose) {
|
|
|
|
if (state->output_is_a_tty)
|
|
|
|
printf("\r\033[K");
|
|
|
|
|
|
|
|
printf ("%i/%i: %s",
|
|
|
|
state->processed_files,
|
|
|
|
state->total_files,
|
|
|
|
next);
|
|
|
|
|
|
|
|
putchar((state->output_is_a_tty) ? '\r' : '\n');
|
|
|
|
fflush (stdout);
|
|
|
|
}
|
|
|
|
|
2011-01-29 17:25:56 +01:00
|
|
|
status = notmuch_database_begin_atomic (notmuch);
|
|
|
|
if (status) {
|
|
|
|
ret = status;
|
|
|
|
goto DONE;
|
|
|
|
}
|
|
|
|
|
2010-01-06 01:35:02 +01:00
|
|
|
status = notmuch_database_add_message (notmuch, next, &message);
|
|
|
|
switch (status) {
|
|
|
|
/* success */
|
|
|
|
case NOTMUCH_STATUS_SUCCESS:
|
|
|
|
state->added_messages++;
|
2011-01-21 10:59:36 +01:00
|
|
|
notmuch_message_freeze (message);
|
2010-03-12 23:13:26 +01:00
|
|
|
for (tag=state->new_tags; *tag != NULL; tag++)
|
|
|
|
notmuch_message_add_tag (message, *tag);
|
2010-11-11 01:26:14 +01:00
|
|
|
if (state->synchronize_flags == TRUE)
|
2011-01-26 12:52:54 +01:00
|
|
|
notmuch_message_maildir_flags_to_tags (message);
|
2011-01-21 10:59:36 +01:00
|
|
|
notmuch_message_thaw (message);
|
2010-01-06 01:35:02 +01:00
|
|
|
break;
|
|
|
|
/* Non-fatal issues (go on to next file) */
|
|
|
|
case NOTMUCH_STATUS_DUPLICATE_MESSAGE_ID:
|
2010-11-11 01:26:14 +01:00
|
|
|
if (state->synchronize_flags == TRUE)
|
2011-01-28 02:51:04 +01:00
|
|
|
notmuch_message_maildir_flags_to_tags (message);
|
2010-01-06 01:35:02 +01:00
|
|
|
break;
|
|
|
|
case NOTMUCH_STATUS_FILE_NOT_EMAIL:
|
|
|
|
fprintf (stderr, "Note: Ignoring non-mail file: %s\n",
|
|
|
|
next);
|
|
|
|
break;
|
|
|
|
/* Fatal issues. Don't process anymore. */
|
2010-01-07 19:29:05 +01:00
|
|
|
case NOTMUCH_STATUS_READ_ONLY_DATABASE:
|
2010-01-06 01:35:02 +01:00
|
|
|
case NOTMUCH_STATUS_XAPIAN_EXCEPTION:
|
|
|
|
case NOTMUCH_STATUS_OUT_OF_MEMORY:
|
|
|
|
fprintf (stderr, "Error: %s. Halting processing.\n",
|
|
|
|
notmuch_status_to_string (status));
|
|
|
|
ret = status;
|
|
|
|
goto DONE;
|
|
|
|
default:
|
|
|
|
case NOTMUCH_STATUS_FILE_ERROR:
|
|
|
|
case NOTMUCH_STATUS_NULL_POINTER:
|
|
|
|
case NOTMUCH_STATUS_TAG_TOO_LONG:
|
|
|
|
case NOTMUCH_STATUS_UNBALANCED_FREEZE_THAW:
|
2011-06-11 05:35:06 +02:00
|
|
|
case NOTMUCH_STATUS_UNBALANCED_ATOMIC:
|
2010-01-06 01:35:02 +01:00
|
|
|
case NOTMUCH_STATUS_LAST_STATUS:
|
|
|
|
INTERNAL_ERROR ("add_message returned unexpected value: %d", status);
|
|
|
|
goto DONE;
|
|
|
|
}
|
|
|
|
|
2011-01-29 17:25:56 +01:00
|
|
|
status = notmuch_database_end_atomic (notmuch);
|
|
|
|
if (status) {
|
|
|
|
ret = status;
|
|
|
|
goto DONE;
|
|
|
|
}
|
|
|
|
|
2010-01-06 01:35:02 +01:00
|
|
|
if (message) {
|
|
|
|
notmuch_message_destroy (message);
|
|
|
|
message = NULL;
|
|
|
|
}
|
2009-11-12 04:50:15 +01:00
|
|
|
|
2011-01-21 10:59:37 +01:00
|
|
|
if (do_print_progress) {
|
|
|
|
do_print_progress = 0;
|
|
|
|
generic_print_progress ("Processed", "files", state->tv_start,
|
|
|
|
state->processed_files, state->total_files);
|
2009-11-12 04:50:15 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
talloc_free (next);
|
|
|
|
next = NULL;
|
|
|
|
}
|
|
|
|
|
2010-03-30 10:23:23 +02:00
|
|
|
if (interrupted)
|
|
|
|
goto DONE;
|
|
|
|
|
2010-01-06 22:26:47 +01:00
|
|
|
/* Now that we've walked the whole filesystem list, anything left
|
|
|
|
* over in the database lists has been deleted. */
|
2010-03-09 18:22:29 +01:00
|
|
|
while (notmuch_filenames_valid (db_files))
|
2010-01-06 22:26:47 +01:00
|
|
|
{
|
|
|
|
char *absolute = talloc_asprintf (state->removed_files,
|
|
|
|
"%s/%s", path,
|
|
|
|
notmuch_filenames_get (db_files));
|
|
|
|
|
|
|
|
_filename_list_add (state->removed_files, absolute);
|
|
|
|
|
2010-03-09 18:22:29 +01:00
|
|
|
notmuch_filenames_move_to_next (db_files);
|
2010-01-06 22:26:47 +01:00
|
|
|
}
|
|
|
|
|
2010-03-09 18:22:29 +01:00
|
|
|
while (notmuch_filenames_valid (db_subdirs))
|
2010-01-06 22:26:47 +01:00
|
|
|
{
|
|
|
|
char *absolute = talloc_asprintf (state->removed_directories,
|
|
|
|
"%s/%s", path,
|
|
|
|
notmuch_filenames_get (db_subdirs));
|
|
|
|
|
|
|
|
_filename_list_add (state->removed_directories, absolute);
|
|
|
|
|
2010-03-09 18:22:29 +01:00
|
|
|
notmuch_filenames_move_to_next (db_subdirs);
|
2010-01-06 22:26:47 +01:00
|
|
|
}
|
|
|
|
|
2011-06-29 09:10:54 +02:00
|
|
|
/* If the directory's mtime is the same as the wall-clock time
|
|
|
|
* when we stat'ed the directory, we skip updating the mtime in
|
|
|
|
* the database because a message could be delivered later in this
|
|
|
|
* same second. This may lead to unnecessary re-scans, but it
|
2011-06-30 01:00:01 +02:00
|
|
|
* avoids overlooking messages. */
|
2011-02-09 00:24:33 +01:00
|
|
|
if (fs_mtime != stat_time)
|
|
|
|
_filename_list_add (state->directory_mtimes, path)->mtime = fs_mtime;
|
2009-11-12 04:50:15 +01:00
|
|
|
|
|
|
|
DONE:
|
|
|
|
if (next)
|
|
|
|
talloc_free (next);
|
|
|
|
if (dir)
|
|
|
|
closedir (dir);
|
2012-02-07 11:05:03 +01:00
|
|
|
if (fs_entries) {
|
|
|
|
for (i = 0; i < num_fs_entries; i++)
|
|
|
|
free (fs_entries[i]);
|
|
|
|
|
2010-01-06 01:15:43 +01:00
|
|
|
free (fs_entries);
|
2012-02-07 11:05:03 +01:00
|
|
|
}
|
2010-01-06 02:43:03 +01:00
|
|
|
if (db_subdirs)
|
|
|
|
notmuch_filenames_destroy (db_subdirs);
|
|
|
|
if (db_files)
|
|
|
|
notmuch_filenames_destroy (db_files);
|
|
|
|
if (directory)
|
|
|
|
notmuch_directory_destroy (directory);
|
2009-11-12 04:50:15 +01:00
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2011-01-21 10:59:37 +01:00
|
|
|
static void
|
|
|
|
setup_progress_printing_timer (void)
|
|
|
|
{
|
|
|
|
struct sigaction action;
|
|
|
|
struct itimerval timerval;
|
|
|
|
|
|
|
|
/* Setup our handler for SIGALRM */
|
|
|
|
memset (&action, 0, sizeof (struct sigaction));
|
|
|
|
action.sa_handler = handle_sigalrm;
|
|
|
|
sigemptyset (&action.sa_mask);
|
|
|
|
action.sa_flags = SA_RESTART;
|
|
|
|
sigaction (SIGALRM, &action, NULL);
|
|
|
|
|
|
|
|
/* Then start a timer to send SIGALRM once per second. */
|
|
|
|
timerval.it_interval.tv_sec = 1;
|
|
|
|
timerval.it_interval.tv_usec = 0;
|
|
|
|
timerval.it_value.tv_sec = 1;
|
|
|
|
timerval.it_value.tv_usec = 0;
|
|
|
|
setitimer (ITIMER_REAL, &timerval, NULL);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
stop_progress_printing_timer (void)
|
|
|
|
{
|
|
|
|
struct sigaction action;
|
|
|
|
struct itimerval timerval;
|
|
|
|
|
|
|
|
/* Now stop the timer. */
|
|
|
|
timerval.it_interval.tv_sec = 0;
|
|
|
|
timerval.it_interval.tv_usec = 0;
|
|
|
|
timerval.it_value.tv_sec = 0;
|
|
|
|
timerval.it_value.tv_usec = 0;
|
|
|
|
setitimer (ITIMER_REAL, &timerval, NULL);
|
|
|
|
|
|
|
|
/* And disable the signal handler. */
|
|
|
|
action.sa_handler = SIG_IGN;
|
|
|
|
sigaction (SIGALRM, &action, NULL);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2009-11-12 04:50:15 +01:00
|
|
|
/* XXX: This should be merged with the add_files function since it
|
|
|
|
* shares a lot of logic with it. */
|
2009-11-18 00:23:42 +01:00
|
|
|
/* Recursively count all regular files in path and all sub-directories
|
2009-11-12 04:50:15 +01:00
|
|
|
* of path. The result is added to *count (which should be
|
|
|
|
* initialized to zero by the top-level caller before calling
|
|
|
|
* count_files). */
|
|
|
|
static void
|
2012-02-15 10:17:31 +01:00
|
|
|
count_files (const char *path, int *count, add_files_state_t *state)
|
2009-11-12 04:50:15 +01:00
|
|
|
{
|
2009-11-18 03:22:20 +01:00
|
|
|
struct dirent *entry = NULL;
|
2009-11-12 04:50:15 +01:00
|
|
|
char *next;
|
|
|
|
struct stat st;
|
2010-01-06 01:15:43 +01:00
|
|
|
struct dirent **fs_entries = NULL;
|
2010-01-06 02:43:03 +01:00
|
|
|
int num_fs_entries = scandir (path, &fs_entries, 0, dirent_sort_inode);
|
2009-11-18 03:22:20 +01:00
|
|
|
int i = 0;
|
2009-11-12 04:50:15 +01:00
|
|
|
|
2010-01-06 01:15:43 +01:00
|
|
|
if (num_fs_entries == -1) {
|
2009-11-12 04:50:15 +01:00
|
|
|
fprintf (stderr, "Warning: failed to open directory %s: %s\n",
|
|
|
|
path, strerror (errno));
|
|
|
|
goto DONE;
|
|
|
|
}
|
|
|
|
|
2009-11-13 18:05:13 +01:00
|
|
|
while (!interrupted) {
|
2010-01-06 01:15:43 +01:00
|
|
|
if (i == num_fs_entries)
|
2009-11-12 04:50:15 +01:00
|
|
|
break;
|
|
|
|
|
2010-01-06 01:15:43 +01:00
|
|
|
entry = fs_entries[i++];
|
2009-11-18 03:22:20 +01:00
|
|
|
|
2009-11-12 04:50:15 +01:00
|
|
|
/* Ignore special directories to avoid infinite recursion.
|
2012-02-15 10:17:31 +01:00
|
|
|
* Also ignore the .notmuch directory and files/directories
|
|
|
|
* the user has configured to be ignored.
|
2009-11-12 04:50:15 +01:00
|
|
|
*/
|
|
|
|
if (strcmp (entry->d_name, ".") == 0 ||
|
|
|
|
strcmp (entry->d_name, "..") == 0 ||
|
2012-02-15 10:17:31 +01:00
|
|
|
strcmp (entry->d_name, ".notmuch") == 0 ||
|
|
|
|
_entry_in_ignore_list (entry->d_name, state))
|
2009-11-12 04:50:15 +01:00
|
|
|
{
|
2012-02-19 21:47:54 +01:00
|
|
|
if (_entry_in_ignore_list (entry->d_name, state) && state->debug)
|
|
|
|
printf ("(D) count_files: explicitly ignoring %s/%s\n",
|
|
|
|
path,
|
|
|
|
entry->d_name);
|
2009-11-12 04:50:15 +01:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (asprintf (&next, "%s/%s", path, entry->d_name) == -1) {
|
|
|
|
next = NULL;
|
|
|
|
fprintf (stderr, "Error descending from %s to %s: Out of memory\n",
|
|
|
|
path, entry->d_name);
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
stat (next, &st);
|
|
|
|
|
|
|
|
if (S_ISREG (st.st_mode)) {
|
|
|
|
*count = *count + 1;
|
|
|
|
if (*count % 1000 == 0) {
|
|
|
|
printf ("Found %d files so far.\r", *count);
|
|
|
|
fflush (stdout);
|
|
|
|
}
|
|
|
|
} else if (S_ISDIR (st.st_mode)) {
|
2012-02-15 10:17:31 +01:00
|
|
|
count_files (next, count, state);
|
2009-11-12 04:50:15 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
free (next);
|
|
|
|
}
|
|
|
|
|
|
|
|
DONE:
|
2012-02-07 11:05:03 +01:00
|
|
|
if (fs_entries) {
|
|
|
|
for (i = 0; i < num_fs_entries; i++)
|
|
|
|
free (fs_entries[i]);
|
|
|
|
|
2010-01-06 01:15:43 +01:00
|
|
|
free (fs_entries);
|
2012-02-07 11:05:03 +01:00
|
|
|
}
|
2009-11-12 04:50:15 +01:00
|
|
|
}
|
|
|
|
|
2010-01-08 03:30:32 +01:00
|
|
|
static void
|
|
|
|
upgrade_print_progress (void *closure,
|
2010-01-10 02:38:23 +01:00
|
|
|
double progress)
|
2010-01-08 03:30:32 +01:00
|
|
|
{
|
|
|
|
add_files_state_t *state = closure;
|
|
|
|
|
2010-01-10 02:38:23 +01:00
|
|
|
printf ("Upgrading database: %.2f%% complete", progress * 100.0);
|
2010-01-08 03:30:32 +01:00
|
|
|
|
2010-01-10 02:38:23 +01:00
|
|
|
if (progress > 0) {
|
|
|
|
struct timeval tv_now;
|
|
|
|
double elapsed, time_remaining;
|
|
|
|
|
|
|
|
gettimeofday (&tv_now, NULL);
|
|
|
|
|
|
|
|
elapsed = notmuch_time_elapsed (state->tv_start, tv_now);
|
|
|
|
time_remaining = (elapsed / progress) * (1.0 - progress);
|
|
|
|
printf (" (");
|
|
|
|
notmuch_time_print_formatted_seconds (time_remaining);
|
|
|
|
printf (" remaining)");
|
|
|
|
}
|
2010-01-08 03:30:32 +01:00
|
|
|
|
2010-01-10 02:38:23 +01:00
|
|
|
printf (". \r");
|
2010-01-08 03:30:32 +01:00
|
|
|
|
|
|
|
fflush (stdout);
|
|
|
|
}
|
|
|
|
|
2011-02-06 10:28:01 +01:00
|
|
|
/* Remove one message filename from the database. */
|
|
|
|
static notmuch_status_t
|
|
|
|
remove_filename (notmuch_database_t *notmuch,
|
|
|
|
const char *path,
|
|
|
|
add_files_state_t *add_files_state)
|
|
|
|
{
|
|
|
|
notmuch_status_t status;
|
2011-01-28 02:51:04 +01:00
|
|
|
notmuch_message_t *message;
|
2011-01-29 17:25:56 +01:00
|
|
|
status = notmuch_database_begin_atomic (notmuch);
|
|
|
|
if (status)
|
|
|
|
return status;
|
2011-10-04 06:55:29 +02:00
|
|
|
status = notmuch_database_find_message_by_filename (notmuch, path, &message);
|
|
|
|
if (status || message == NULL)
|
2012-04-22 17:50:52 +02:00
|
|
|
goto DONE;
|
|
|
|
|
2011-02-06 10:28:01 +01:00
|
|
|
status = notmuch_database_remove_message (notmuch, path);
|
2011-01-28 02:51:04 +01:00
|
|
|
if (status == NOTMUCH_STATUS_DUPLICATE_MESSAGE_ID) {
|
2011-02-06 10:28:01 +01:00
|
|
|
add_files_state->renamed_messages++;
|
2011-01-28 02:51:04 +01:00
|
|
|
if (add_files_state->synchronize_flags == TRUE)
|
|
|
|
notmuch_message_maildir_flags_to_tags (message);
|
2012-04-22 17:50:50 +02:00
|
|
|
status = NOTMUCH_STATUS_SUCCESS;
|
|
|
|
} else if (status == NOTMUCH_STATUS_SUCCESS) {
|
2011-02-06 10:28:01 +01:00
|
|
|
add_files_state->removed_messages++;
|
2012-04-22 17:50:50 +02:00
|
|
|
}
|
2011-01-28 02:51:04 +01:00
|
|
|
notmuch_message_destroy (message);
|
2012-04-22 17:50:52 +02:00
|
|
|
|
|
|
|
DONE:
|
2011-01-29 17:25:56 +01:00
|
|
|
notmuch_database_end_atomic (notmuch);
|
2011-02-06 10:28:01 +01:00
|
|
|
return status;
|
|
|
|
}
|
|
|
|
|
2010-01-08 03:20:28 +01:00
|
|
|
/* Recursively remove all filenames from the database referring to
|
|
|
|
* 'path' (or to any of its children). */
|
2012-04-22 17:50:50 +02:00
|
|
|
static notmuch_status_t
|
2010-01-08 03:20:28 +01:00
|
|
|
_remove_directory (void *ctx,
|
|
|
|
notmuch_database_t *notmuch,
|
|
|
|
const char *path,
|
2011-06-11 18:54:05 +02:00
|
|
|
add_files_state_t *add_files_state)
|
2010-01-08 03:20:28 +01:00
|
|
|
{
|
2012-04-22 17:50:50 +02:00
|
|
|
notmuch_status_t status = NOTMUCH_STATUS_SUCCESS;
|
2010-01-08 03:20:28 +01:00
|
|
|
notmuch_directory_t *directory;
|
|
|
|
notmuch_filenames_t *files, *subdirs;
|
|
|
|
char *absolute;
|
|
|
|
|
2012-05-14 01:36:09 +02:00
|
|
|
status = notmuch_database_get_directory (notmuch, path, &directory);
|
|
|
|
if (status || !directory)
|
|
|
|
return status;
|
2010-01-08 03:20:28 +01:00
|
|
|
|
|
|
|
for (files = notmuch_directory_get_child_files (directory);
|
2010-03-09 18:22:29 +01:00
|
|
|
notmuch_filenames_valid (files);
|
|
|
|
notmuch_filenames_move_to_next (files))
|
2010-01-08 03:20:28 +01:00
|
|
|
{
|
|
|
|
absolute = talloc_asprintf (ctx, "%s/%s", path,
|
|
|
|
notmuch_filenames_get (files));
|
2012-04-22 17:50:50 +02:00
|
|
|
status = remove_filename (notmuch, absolute, add_files_state);
|
2010-01-08 03:20:28 +01:00
|
|
|
talloc_free (absolute);
|
2012-04-22 17:50:50 +02:00
|
|
|
if (status)
|
|
|
|
goto DONE;
|
2010-01-08 03:20:28 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
for (subdirs = notmuch_directory_get_child_directories (directory);
|
2010-03-09 18:22:29 +01:00
|
|
|
notmuch_filenames_valid (subdirs);
|
|
|
|
notmuch_filenames_move_to_next (subdirs))
|
2010-01-08 03:20:28 +01:00
|
|
|
{
|
|
|
|
absolute = talloc_asprintf (ctx, "%s/%s", path,
|
|
|
|
notmuch_filenames_get (subdirs));
|
2012-04-22 17:50:50 +02:00
|
|
|
status = _remove_directory (ctx, notmuch, absolute, add_files_state);
|
2010-01-08 03:20:28 +01:00
|
|
|
talloc_free (absolute);
|
2012-04-22 17:50:50 +02:00
|
|
|
if (status)
|
|
|
|
goto DONE;
|
2010-01-08 03:20:28 +01:00
|
|
|
}
|
|
|
|
|
2012-04-22 17:50:50 +02:00
|
|
|
DONE:
|
2010-01-08 03:20:28 +01:00
|
|
|
notmuch_directory_destroy (directory);
|
2012-04-22 17:50:50 +02:00
|
|
|
return status;
|
2010-01-08 03:20:28 +01:00
|
|
|
}
|
|
|
|
|
2009-11-10 21:03:05 +01:00
|
|
|
int
|
2009-11-21 01:17:18 +01:00
|
|
|
notmuch_new_command (void *ctx, int argc, char *argv[])
|
2009-11-10 21:03:05 +01:00
|
|
|
{
|
2009-11-12 04:50:15 +01:00
|
|
|
notmuch_config_t *config;
|
2009-11-10 21:03:05 +01:00
|
|
|
notmuch_database_t *notmuch;
|
|
|
|
add_files_state_t add_files_state;
|
|
|
|
double elapsed;
|
2011-01-21 10:59:37 +01:00
|
|
|
struct timeval tv_now, tv_start;
|
2009-11-10 21:03:05 +01:00
|
|
|
int ret = 0;
|
2009-11-12 04:50:15 +01:00
|
|
|
struct stat st;
|
|
|
|
const char *db_path;
|
|
|
|
char *dot_notmuch_path;
|
2009-11-13 18:05:13 +01:00
|
|
|
struct sigaction action;
|
2010-01-06 02:56:11 +01:00
|
|
|
_filename_node_t *f;
|
2012-12-03 21:56:13 +01:00
|
|
|
int opt_index;
|
2009-11-21 01:17:18 +01:00
|
|
|
int i;
|
2011-01-21 10:59:37 +01:00
|
|
|
notmuch_bool_t timer_is_active = FALSE;
|
2012-12-03 21:56:13 +01:00
|
|
|
notmuch_bool_t no_hooks = FALSE;
|
2009-11-21 01:17:18 +01:00
|
|
|
|
2012-12-03 21:56:13 +01:00
|
|
|
add_files_state.verbose = FALSE;
|
|
|
|
add_files_state.debug = FALSE;
|
2009-11-23 01:54:35 +01:00
|
|
|
add_files_state.output_is_a_tty = isatty (fileno (stdout));
|
2009-11-21 01:17:18 +01:00
|
|
|
|
2012-12-03 21:56:13 +01:00
|
|
|
notmuch_opt_desc_t options[] = {
|
|
|
|
{ NOTMUCH_OPT_BOOLEAN, &add_files_state.verbose, "verbose", 'v', 0 },
|
|
|
|
{ NOTMUCH_OPT_BOOLEAN, &add_files_state.debug, "debug", 'd', 0 },
|
|
|
|
{ NOTMUCH_OPT_BOOLEAN, &no_hooks, "no-hooks", 'n', 0 },
|
|
|
|
{ 0, 0, 0, 0, 0 }
|
|
|
|
};
|
2011-10-21 14:19:17 +02:00
|
|
|
|
2012-12-03 21:56:13 +01:00
|
|
|
opt_index = parse_arguments (argc, argv, options, 1);
|
|
|
|
if (opt_index < 0) {
|
|
|
|
/* diagnostics already printed */
|
|
|
|
return 1;
|
2009-11-21 01:17:18 +01:00
|
|
|
}
|
2012-12-03 21:56:13 +01:00
|
|
|
|
2009-11-12 04:50:15 +01:00
|
|
|
config = notmuch_config_open (ctx, NULL, NULL);
|
|
|
|
if (config == NULL)
|
|
|
|
return 1;
|
|
|
|
|
2010-03-12 23:13:26 +01:00
|
|
|
add_files_state.new_tags = notmuch_config_get_new_tags (config, &add_files_state.new_tags_length);
|
2012-02-15 10:17:31 +01:00
|
|
|
add_files_state.new_ignore = notmuch_config_get_new_ignore (config, &add_files_state.new_ignore_length);
|
2010-11-11 01:26:14 +01:00
|
|
|
add_files_state.synchronize_flags = notmuch_config_get_maildir_synchronize_flags (config);
|
2009-11-12 04:50:15 +01:00
|
|
|
db_path = notmuch_config_get_database_path (config);
|
|
|
|
|
2012-12-03 21:56:13 +01:00
|
|
|
if (!no_hooks) {
|
2011-12-08 23:48:30 +01:00
|
|
|
ret = notmuch_run_hook (db_path, "pre-new");
|
|
|
|
if (ret)
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2009-11-12 04:50:15 +01:00
|
|
|
dot_notmuch_path = talloc_asprintf (ctx, "%s/%s", db_path, ".notmuch");
|
|
|
|
|
|
|
|
if (stat (dot_notmuch_path, &st)) {
|
2009-11-13 18:05:13 +01:00
|
|
|
int count;
|
|
|
|
|
|
|
|
count = 0;
|
2012-02-15 10:17:31 +01:00
|
|
|
count_files (db_path, &count, &add_files_state);
|
2009-11-13 18:05:13 +01:00
|
|
|
if (interrupted)
|
|
|
|
return 1;
|
|
|
|
|
2009-12-15 01:06:37 +01:00
|
|
|
printf ("Found %d total files (that's not much mail).\n", count);
|
2012-04-30 18:25:34 +02:00
|
|
|
if (notmuch_database_create (db_path, ¬much))
|
|
|
|
return 1;
|
2009-11-13 18:05:13 +01:00
|
|
|
add_files_state.total_files = count;
|
2009-11-12 04:50:15 +01:00
|
|
|
} else {
|
2012-04-30 18:25:33 +02:00
|
|
|
if (notmuch_database_open (db_path, NOTMUCH_DATABASE_MODE_READ_WRITE,
|
|
|
|
¬much))
|
2010-01-08 03:30:32 +01:00
|
|
|
return 1;
|
|
|
|
|
|
|
|
if (notmuch_database_needs_upgrade (notmuch)) {
|
|
|
|
printf ("Welcome to a new version of notmuch! Your database will now be upgraded.\n");
|
|
|
|
gettimeofday (&add_files_state.tv_start, NULL);
|
|
|
|
notmuch_database_upgrade (notmuch, upgrade_print_progress,
|
|
|
|
&add_files_state);
|
|
|
|
printf ("Your notmuch database has now been upgraded to database format version %u.\n",
|
|
|
|
notmuch_database_get_version (notmuch));
|
|
|
|
}
|
|
|
|
|
2009-11-13 18:05:13 +01:00
|
|
|
add_files_state.total_files = 0;
|
2009-11-10 21:03:05 +01:00
|
|
|
}
|
|
|
|
|
2009-11-12 04:50:15 +01:00
|
|
|
if (notmuch == NULL)
|
|
|
|
return 1;
|
|
|
|
|
2010-01-08 17:45:16 +01:00
|
|
|
/* Setup our handler for SIGINT. We do this after having
|
|
|
|
* potentially done a database upgrade we this interrupt handler
|
|
|
|
* won't support. */
|
|
|
|
memset (&action, 0, sizeof (struct sigaction));
|
|
|
|
action.sa_handler = handle_sigint;
|
|
|
|
sigemptyset (&action.sa_mask);
|
|
|
|
action.sa_flags = SA_RESTART;
|
|
|
|
sigaction (SIGINT, &action, NULL);
|
|
|
|
|
2009-11-12 04:50:15 +01:00
|
|
|
talloc_free (dot_notmuch_path);
|
|
|
|
dot_notmuch_path = NULL;
|
|
|
|
|
2009-11-10 21:03:05 +01:00
|
|
|
add_files_state.processed_files = 0;
|
|
|
|
add_files_state.added_messages = 0;
|
2011-06-11 18:54:05 +02:00
|
|
|
add_files_state.removed_messages = add_files_state.renamed_messages = 0;
|
2009-11-10 21:03:05 +01:00
|
|
|
gettimeofday (&add_files_state.tv_start, NULL);
|
|
|
|
|
2010-01-06 02:56:11 +01:00
|
|
|
add_files_state.removed_files = _filename_list_create (ctx);
|
|
|
|
add_files_state.removed_directories = _filename_list_create (ctx);
|
2011-02-09 00:24:33 +01:00
|
|
|
add_files_state.directory_mtimes = _filename_list_create (ctx);
|
2010-01-06 02:56:11 +01:00
|
|
|
|
2011-01-21 10:59:37 +01:00
|
|
|
if (! debugger_is_active () && add_files_state.output_is_a_tty
|
|
|
|
&& ! add_files_state.verbose) {
|
|
|
|
setup_progress_printing_timer ();
|
|
|
|
timer_is_active = TRUE;
|
|
|
|
}
|
|
|
|
|
2009-11-12 04:50:15 +01:00
|
|
|
ret = add_files (notmuch, db_path, &add_files_state);
|
2012-04-22 17:50:49 +02:00
|
|
|
if (ret)
|
|
|
|
goto DONE;
|
2009-11-10 21:03:05 +01:00
|
|
|
|
2011-01-21 10:59:37 +01:00
|
|
|
gettimeofday (&tv_start, NULL);
|
2011-02-09 02:56:31 +01:00
|
|
|
for (f = add_files_state.removed_files->head; f && !interrupted; f = f->next) {
|
2012-04-22 17:50:50 +02:00
|
|
|
ret = remove_filename (notmuch, f->filename, &add_files_state);
|
|
|
|
if (ret)
|
|
|
|
goto DONE;
|
2011-01-21 10:59:37 +01:00
|
|
|
if (do_print_progress) {
|
|
|
|
do_print_progress = 0;
|
|
|
|
generic_print_progress ("Cleaned up", "messages",
|
2011-06-11 18:54:05 +02:00
|
|
|
tv_start, add_files_state.removed_messages + add_files_state.renamed_messages,
|
2011-01-21 10:59:37 +01:00
|
|
|
add_files_state.removed_files->count);
|
|
|
|
}
|
2010-01-06 02:56:11 +01:00
|
|
|
}
|
|
|
|
|
2011-01-21 10:59:37 +01:00
|
|
|
gettimeofday (&tv_start, NULL);
|
2011-02-09 02:56:31 +01:00
|
|
|
for (f = add_files_state.removed_directories->head, i = 0; f && !interrupted; f = f->next, i++) {
|
2012-04-22 17:50:50 +02:00
|
|
|
ret = _remove_directory (ctx, notmuch, f->filename, &add_files_state);
|
|
|
|
if (ret)
|
|
|
|
goto DONE;
|
2011-01-21 10:59:37 +01:00
|
|
|
if (do_print_progress) {
|
|
|
|
do_print_progress = 0;
|
|
|
|
generic_print_progress ("Cleaned up", "directories",
|
|
|
|
tv_start, i,
|
|
|
|
add_files_state.removed_directories->count);
|
|
|
|
}
|
2010-01-06 02:56:11 +01:00
|
|
|
}
|
|
|
|
|
2011-02-09 00:24:33 +01:00
|
|
|
for (f = add_files_state.directory_mtimes->head; f && !interrupted; f = f->next) {
|
2012-05-14 01:36:09 +02:00
|
|
|
notmuch_status_t status;
|
2011-02-09 00:24:33 +01:00
|
|
|
notmuch_directory_t *directory;
|
2012-05-14 01:36:09 +02:00
|
|
|
status = notmuch_database_get_directory (notmuch, f->filename, &directory);
|
|
|
|
if (status == NOTMUCH_STATUS_SUCCESS && directory) {
|
2011-02-09 00:24:33 +01:00
|
|
|
notmuch_directory_set_mtime (directory, f->mtime);
|
|
|
|
notmuch_directory_destroy (directory);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2012-04-22 17:50:49 +02:00
|
|
|
DONE:
|
2010-01-06 02:56:11 +01:00
|
|
|
talloc_free (add_files_state.removed_files);
|
|
|
|
talloc_free (add_files_state.removed_directories);
|
2011-02-09 00:24:33 +01:00
|
|
|
talloc_free (add_files_state.directory_mtimes);
|
2010-01-06 02:56:11 +01:00
|
|
|
|
2011-01-21 10:59:37 +01:00
|
|
|
if (timer_is_active)
|
|
|
|
stop_progress_printing_timer ();
|
|
|
|
|
2009-11-10 21:03:05 +01:00
|
|
|
gettimeofday (&tv_now, NULL);
|
|
|
|
elapsed = notmuch_time_elapsed (add_files_state.tv_start,
|
|
|
|
tv_now);
|
2010-01-06 19:07:49 +01:00
|
|
|
|
2009-11-10 21:03:05 +01:00
|
|
|
if (add_files_state.processed_files) {
|
|
|
|
printf ("Processed %d %s in ", add_files_state.processed_files,
|
|
|
|
add_files_state.processed_files == 1 ?
|
|
|
|
"file" : "total files");
|
|
|
|
notmuch_time_print_formatted_seconds (elapsed);
|
|
|
|
if (elapsed > 1) {
|
2011-01-21 10:59:37 +01:00
|
|
|
printf (" (%d files/sec.).\033[K\n",
|
2009-11-10 21:03:05 +01:00
|
|
|
(int) (add_files_state.processed_files / elapsed));
|
|
|
|
} else {
|
2011-01-21 10:59:37 +01:00
|
|
|
printf (".\033[K\n");
|
2009-11-10 21:03:05 +01:00
|
|
|
}
|
|
|
|
}
|
2010-01-06 19:07:49 +01:00
|
|
|
|
2009-11-10 21:03:05 +01:00
|
|
|
if (add_files_state.added_messages) {
|
2010-01-06 19:07:49 +01:00
|
|
|
printf ("Added %d new %s to the database.",
|
2009-11-10 21:03:05 +01:00
|
|
|
add_files_state.added_messages,
|
|
|
|
add_files_state.added_messages == 1 ?
|
|
|
|
"message" : "messages");
|
|
|
|
} else {
|
2010-01-06 19:07:49 +01:00
|
|
|
printf ("No new mail.");
|
|
|
|
}
|
|
|
|
|
2011-06-11 18:54:05 +02:00
|
|
|
if (add_files_state.removed_messages) {
|
2010-01-06 19:07:49 +01:00
|
|
|
printf (" Removed %d %s.",
|
2011-06-11 18:54:05 +02:00
|
|
|
add_files_state.removed_messages,
|
|
|
|
add_files_state.removed_messages == 1 ? "message" : "messages");
|
2010-01-06 19:07:49 +01:00
|
|
|
}
|
|
|
|
|
2011-06-11 18:54:05 +02:00
|
|
|
if (add_files_state.renamed_messages) {
|
2010-01-06 19:07:49 +01:00
|
|
|
printf (" Detected %d file %s.",
|
2011-06-11 18:54:05 +02:00
|
|
|
add_files_state.renamed_messages,
|
|
|
|
add_files_state.renamed_messages == 1 ? "rename" : "renames");
|
2009-11-10 21:03:05 +01:00
|
|
|
}
|
|
|
|
|
2010-01-06 19:07:49 +01:00
|
|
|
printf ("\n");
|
|
|
|
|
2012-04-22 17:50:49 +02:00
|
|
|
if (ret)
|
2012-04-22 17:50:51 +02:00
|
|
|
fprintf (stderr, "Note: A fatal error was encountered: %s\n",
|
|
|
|
notmuch_status_to_string (ret));
|
2009-11-10 21:03:05 +01:00
|
|
|
|
2012-04-22 14:07:53 +02:00
|
|
|
notmuch_database_destroy (notmuch);
|
2009-11-10 21:03:05 +01:00
|
|
|
|
2012-12-03 21:56:13 +01:00
|
|
|
if (!no_hooks && !ret && !interrupted)
|
2011-12-08 23:48:30 +01:00
|
|
|
ret = notmuch_run_hook (db_path, "post-new");
|
|
|
|
|
2009-11-13 18:05:13 +01:00
|
|
|
return ret || interrupted;
|
2009-11-10 21:03:05 +01:00
|
|
|
}
|