Kevin McCarthy [Sun, 23 Oct 2016 21:46:37 +0000 (14:46 -0700)]
Fix pager segfault when lineInfo.chunks overflows. (closes #3888)
The reporter had an html attachment with extremely long lines,
combined with a color pattern of "color body default default ."
This overflowed the lineInfo.chunks, causing a segfault.
Kevin McCarthy [Mon, 17 Oct 2016 18:22:13 +0000 (11:22 -0700)]
Actually fix gpgme segfault in create_recipient_set().
Changeset 6e44bfa16096 did not fix the segv. (Sorry, I made the fix
based off a report on IRC but didn't trigger the segv myself: it was
caused by an out-of-tree patch).
The actual problem was that the rset was only resized on a successful
gpgme_get_key(). However, on error, the array still needs to be
NULL-terminated before calling free_recipient_set().
Move the resize so it always takes place. This obviates the need for
the NULL check added in 6e44bfa16096.
Kevin McCarthy [Sun, 16 Oct 2016 21:16:47 +0000 (14:16 -0700)]
Fix gpgme segfault in create_recipient_set().
If gpgme_get_key() errors on the first key, the rset will not be
allocated yet. Attempting to null-terminate (and then free) the array
causes a segfault.
Kevin McCarthy [Tue, 4 Oct 2016 18:13:24 +0000 (11:13 -0700)]
Preserve forwarded attachment names in d_filename.
When forwarding an attachment with an non-ascii name, mutt_copy_body()
mangles the filename when calling mutt_adv_mktemp. Preserve
the original attachment filename in d_filename.
Remove the double copy of b->filename, which is a memory leak.
Kevin McCarthy [Sun, 25 Sep 2016 20:11:03 +0000 (13:11 -0700)]
Reset invalid parsed received dates to 0. (closes #3878)
The actual problem in the ticket would be solved by d3f31cf9239e (see
#3798). However there is still the bug that Mutt considers a
(hdr->received != 0) to be set and usable, despite not checking the
return value of mutt_parse_date().
Change mutt_read_rfc822_header() to unset an invalid received value
back to 0. We don't do this inside mutt_read_rfc822_line() because
that would cause the next received line to be parsed.
Antonio Radici [Thu, 22 Sep 2016 01:10:33 +0000 (18:10 -0700)]
Mark some gpgme pgp menu keybinding translations as fuzzy. (closes #3874)
Some translations for crypt-gpgme.c are marked as fuzzy but the keybindings
attached to these translations are not, this creates confusions for the users
who see the english message but have the keybindings for a message in their own
language available.
As long as the translations are fuzzy, the keybindings should stay fuzzy.
Kevin McCarthy [Wed, 21 Sep 2016 20:51:01 +0000 (22:51 +0200)]
Check for NULL mx_ops in mx.c
Eike Rathke reported this happening when in an IMAP index view the
underlying connection was terminated, ctx->mx_ops was NULL and thus
accessing ctx->mx_ops->check segfaulted.
Thanks also to Eike Rathke for the initial patch, for which I expanded
the checks to other functions.
Kevin McCarthy [Mon, 5 Sep 2016 19:35:19 +0000 (12:35 -0700)]
Stub out getdnsdomainname() unless HAVE_GETADDRINFO.
It seems unlikely there are systems without it (given that this
mistake has been in since 1.6.0), but for correctness we should stub
out the function for those without it.
Kevin McCarthy [Sat, 3 Sep 2016 02:32:39 +0000 (19:32 -0700)]
Disable ~X when message scoring. (closes #3861)
mutt_score_message() purposely passes a NULL context to
mutt_pattern_exec(). The idea was to block slow patterns, and the
scoring documentation notes this by saying:
"For efficiency reasons, patterns which scan information not
available in the index, such as ~b, ~B or ~h, may not be used"
~X needs the context to parse the messages (during message scoring at
least), and thus isn't suitable for message scoring either.
Block ~X from being used when the context is NULL. Add ~X to the list
of patterns noted as unusable in the message scoring documentation.
Kevin McCarthy [Mon, 8 Aug 2016 20:13:30 +0000 (13:13 -0700)]
Clear pager position upon returning to the index menu.
This fixes a bug where opening a message sometimes shows it scrolled
down.
The easiest way to trigger this is to open a long message in a mailbox
with no new messages, scroll down, hit <next-new-then-unread> and then
reopen the message. The pager will "remember" the position and scroll
down. If you simply exit the message and re-enter it, the position is
not saved.
More annoyingly, once in a very great while I will open a new message
and have the message scrolled down partway. I believe this is due to
OldHdr happening to point to the same address as the Header of the new
message.
This is somewhat easily triggered on a mailbox with one message, which
you open, scroll down in, hit <next-new-then-unread> to exit, and then
delete/sync. In another mutt instance, copy the same messsage back to
the mailbox, then open the message in the original mutt instance. At
least some of the time, the OldHdr pointer matches the new message,
and so it opens scrolled down.
This patch solves the problem by clearing the pager position once all
redirections through the index menu are done.
Kevin McCarthy [Fri, 5 Aug 2016 21:33:39 +0000 (14:33 -0700)]
Remove nonsensical size check in mutt_choose_charset()
The charsets parameter is being tokenized by the : delimeter.
The checks against ENCWORD_LEN_MAX and ENCWORD_LEN_MIN make no sense,
and appear to be the result of a large merge a very long time ago
(changeset cb27682966d5).
I can only guess where this check was supposed to be, but it certainly
doesn't belong here.
Kevin McCarthy [Tue, 2 Aug 2016 01:25:28 +0000 (18:25 -0700)]
Use a different flag in mx_open_mailbox_append() when mailbox doesn't exist.
The previous commit re-used MUTT_NEWFOLDER, but the meaning of that
flag is slightly different: it causes mbox to use fopen with mode "w",
and is used only for the case of a brand-new mktemp-generated mbox.
Setting it for other non-existing mbox files leads to a race condition
between the stat and the fopen/lock, and so could end up truncating an
existing mailbox created in-between!
Create a different flag, MUTT_APPENDNEW to notify the open_append()
functions that the mailbox doesn't exist. Change maildir and mh to
check for that flag to create their directory structures.
Kevin McCarthy [Sat, 30 Jul 2016 18:11:07 +0000 (11:11 -0700)]
Add unread and total message count format strings to $folder_format.
%n will show the unread message count in the mailbox. %m will show
total message count. Except for %n with IMAP, these both require
$mail_check_stats to be set, which puts these counts inside BUFFY.
Since the imap_mboxcache is never fresher than the value in BUFFY,
remove the special imap_mailbox_state() call. Instead, just update
from the current Context for all mailboxes.
Remove the logic that overrode the %N format string to show unread
count for IMAP mailboxes. If they want to see unread count, they will
have to change the $folder_format.
Although it doesn't look possible to reuse browser_state.entry slots,
change the OP_DELETE_MAILBOX to memset(0) the deleted slot. Then, we
can change to logic in add_folder() so it doesn't need to zero-out
unset values.
Anton Lindqvist [Wed, 20 Jul 2016 01:56:13 +0000 (18:56 -0700)]
Fix arithmetic exception due to menu->pagelen being negative.
Resizing the terminal window down to two lines when in an empty
mailbox causes mutt to crash due to division by zero since menu->max
equals 0 and menu->pagelen < 0 in status.c:205.
Fixing the problem at this specific line felt wrong since I did notice
menu->pagelen being negative. The pagelen is inherited from the rows
calculation in mutt_reflow_windows. Since the number of lines can
potentially be smaller than the accumulated number of rows acquired by
the status, help and message window, make sure the calculation does
not turn negative.
Kevin McCarthy [Mon, 18 Jul 2016 02:31:16 +0000 (19:31 -0700)]
Add imap-fast-trash patch.
This is based on the patch by Paul Miller.
Modifications are:
* Create a new flag, MUTT_TRASH for imap_make_msg_set(), rather than
use MUTT_EXPIRED.
* Change imap_make_msg_set(MUTT_TRASH) to only look at
hdrs[n]->deleted && !hdrs[n]->purge, behaving like MUTT_TAG, rather
than looking at the HEADER_DATA.
* Reimplement imap_fast_trash() based on imap_copy_message().
It looks the old version was too, but it lacked handling of TRYCREATE
and also queued the UID COPY but didn't exec it. (Presumably this
happened in the subsequent sync).
* Move the Context magic and mx_is_imap() checks outside of
imap_fast_trash()
Kevin McCarthy [Mon, 18 Jul 2016 02:31:09 +0000 (19:31 -0700)]
Add the trash folder patch.
This is based on the trash folder patch by Cedric Duval.
Modifications to the original patch are:
* Use a flag called M_PURGE instead of M_APPENDED. The same flag is
then used in the following "purge" patch instead of adding a
different flag.
* Removed the counter in context. The existing context->deleted
is all that's needed.
* Removed the "auto unset M_PURGE" when M_DELETED is unset inside
_mutt_set_flag(), although this is convenient, it easily leads to
header->purge not being reset in a few situations.
* Reset purge flag along with the deleted flag if $delete is answered
no.
* Set M_PURGE on an edited message. (edit_one_message())
* Preserve purge flag in mutt_reopen_mailbox()
* Turn off OPTCONFIRMAPPEND when saving to the trash, rather than
hardcoding it off in mutt_save_confirm(). That way, normal save to the
folder will respect the option.
Kevin McCarthy [Tue, 12 Jul 2016 01:36:21 +0000 (18:36 -0700)]
Initialize mutt windows even in batch mode. (closes #3853)
mutt_select_fcc() calls mutt_addr_hook() -> mutt_make_string() which
refers to MuttIndexWindow->cols when calling mutt_FormatString(). In
batch mode, MuttIndexWindow hasn't been initialized, leading to a
segfault.
This might be the only overlap, but it seems wiser to just initialize
the mutt windows in case there are other references (now or in the
future) when processing format strings in batch mode.
Kevin McCarthy [Sat, 9 Jul 2016 01:52:56 +0000 (18:52 -0700)]
Change sidebar highlighted mailbox behavior.
Delay selecting the highlighted mailbox until prepare_mailbox(), to
avoid a hidden mailbox being selected during the Buffy list population
(in mutt_sb_notify_mailbox()).
Change update_entries_visibility() to not automatically make the
highlighted mailbox visible.
Change prepare_sidebar() to (re)set the highlighted mailbox when the
current highlighted mailbox is hidden.
Kevin McCarthy [Sat, 9 Jul 2016 01:47:57 +0000 (18:47 -0700)]
Don't overwrite imap_status->uidnext with a lower value. (closes #3771)
imap_read_headers() updates the idata and imap_status uidnext after
reading through all the new headers.
The idata is updated properly (only if its uidnext is below maxuid+1),
but the imap_status was always being set to maxuid.
This causes a problem with new mail checking if the most recent
messages are deleted. Then the uidnext will be greater than maxuid+1
in the mailbox, and if there are any other unread messages it will
*always* report new mail even if $mail_check_recent is set.
Kevin McCarthy [Sat, 2 Jul 2016 19:24:25 +0000 (12:24 -0700)]
Fix sidebar "unsorted" order to match Buffy list order.
Since the previous commit decoupled the sidebar from the Buffy list,
we can now restore the order to match the buffy list when
sidebar_sort_method is set (back) to "unsorted".
Kevin McCarthy [Sat, 2 Jul 2016 19:22:51 +0000 (12:22 -0700)]
Decouple the sidebar from the Buffy list.
Change the sidebar to use an array of SBENTRY* instead. Move the
"is_hidden" into SBENTRY. Remove the added "prev" pointer from BUFFY.
This way, sorting the sidebar doesn't affect the BUFFY list order, and
we don't need elements inside BUFFY just for the sidebar presentation.
Fix sidebar-next for the case where the mailboxes are unsorted and
$sidebar_new_mail_only is set. Change sorting not to clump hidden
mailboxes at the bottom, instead simply skip over hidden mailboxes in
sidebar-next/prev.
Kevin McCarthy [Fri, 1 Jul 2016 20:33:20 +0000 (13:33 -0700)]
Check $pgp_autoinline and $pgp_replyinline if oppenc is set. (closes #3846)
The first oppenc call takes place after the initial checks of
$pgp_autoline and $pgp_replyinline, and doesn't go through the pgp
menu.
Therefore, check for $pgp_autoline and $pgp_replyinline if oppenc is
set too, to avoid oppenc enabling encryption without INLINE being set
in those cases.
a6a4d6ed0f19 previously cleaned things up so that it is safe
to set INLINE even if encryption isn't enabled.
USE_SIDEBAR needed to be set in doc/makedoc-defs.h so that the sidebar
option documentation is generated whether mutt is configured with the
sidebar or not.
Kevin McCarthy [Thu, 30 Jun 2016 01:58:53 +0000 (18:58 -0700)]
Fix cppflags and muttlibs duplication with --with-gss.
When krb5-config was found, MUTT_AM_PATH_GSSAPI included CFLAGS in
GSSAPI_CFLAGS and MUTTLIBS in GSSAPI_LIBS. However, configure.ac
afterwards sets:
CPPFLAGS="$CPPFLAGS $GSSAPI_CFLAGS"
MUTTLIBS="$MUTTLIBS $GSSAPI_LIBS"
This caused the flags and libs to be duplicated.
Kevin McCarthy [Thu, 30 Jun 2016 01:58:48 +0000 (18:58 -0700)]
Include ncurses tinfo library if found.
Thanks to Fabian Groffen for reporting this issue and providing a
couple possible patches. From Fabian's report:
For some time now, ncurses can be built in a mode where the low level
terminfo functionality lives in a separate lib called libtinfo.
Because some people do, this means Mutt needs to include this
library in that case to avoid linking errors [...]
Remove $sidebar_refresh_time from Sample Sidebar Config
The $sidebar_refresh_time option was removed with the changeset 1f840760e6e0. Remove it from the sample sidebar configuration as well,
so that anyone using that file in its entirety won't have to deal with "unknown
variable" errors.
Kevin McCarthy [Thu, 23 Jun 2016 19:38:07 +0000 (12:38 -0700)]
Change sidebar to consistently use realpath for context and buffy comparison.
The original sidebar patch contained a half-implemented attempt to use
realpath() mailbox paths for comparison. (Presumably so the open mailbox
remains highlighted despite symlink issues).
Add realpath to the Context, and set it when opening a mailbox.
Remove sidebar ifdef for the buffy member, and always set it there too.
Change the sidebar to use the realpath for comparison everywhere.
mutt_buffy_check() is using stat device and inode for comparison.
Perhaps this can be changed to use realpath instead, but that's beyond
the scope of this patch.
Damien Riegel [Sat, 18 Jun 2016 20:36:18 +0000 (13:36 -0700)]
prepend maildir_commit_message function name with an underscore
Basically, rename maildir_commit_message to _maildir_commit_message.
This commit is preparatory to make the maildir_commit_message symbol
available for further use.
Symbols starting with underscore should be avoided but this one is long
enough to prevent collision.
Damien Riegel [Sat, 18 Jun 2016 20:36:12 +0000 (13:36 -0700)]
remove unused HEADER parameter in mh_commit_message
mh_commit_message is only called in one place with the header parameter
set to NULL. To make the commit function consistent with other
mailboxes, which only takes ctx and msg as parameters, remove this
unused parameter.
Damien Riegel [Sat, 18 Jun 2016 19:41:43 +0000 (12:41 -0700)]
Remove magic member in MESSAGE structure
The "magic" was copied from the context to the message structure to be
able to determine which close function had to be called in
mx_close_message. Now that this function is context aware, there is no
need to store the magic in the MESSAGE structure and it can be safely
removed.
Damien Riegel [Sat, 18 Jun 2016 02:01:31 +0000 (19:01 -0700)]
Add open_msg to struct mx_ops
Add the callback to open an existing message to struct mx_ops. For mbox,
mmdf, maildir, and mh, the code was implemented directly into
mx_open_message, so it is moved in their respective source files. For
imap and pop, there were already <mailbox>_fetch_message functions, but
their argument order has been changed to pass the context as a first
argument.
Olaf Hering [Fri, 17 Jun 2016 17:33:56 +0000 (10:33 -0700)]
folder_file: remove struct stat
Add and use new flag to indicate the folder is on local filesystem.
Add and use new gid, uid and nlink member.
Use existing ->mode member instead of stat->st_mode.
Use existing ->size member instead of stat->st_size.
Use existing ->mtime member instead of stat->st_mtime.
Remove struct stat, the used values were already duplicated in the struct.
This reduces memory usage.
Kevin McCarthy [Fri, 17 Jun 2016 17:30:30 +0000 (10:30 -0700)]
Combine the basic and extended buffy functions.
Add a check_stats parameter to the mbox, maildir, and mh buffy
functions. Use that parameter to determine whether to also count
total, new, and flagged messages.
This makes the functions a bit more complicated, but improves
efficiency (for maildir and mh).
Also includes the following cleanup/fixes:
* Move the orig-value counter reset to the beginnining of the loop,
(before tmp->new is set to 0).
* Change trashed maildir messages to not be counted in msg_count
* Remove an incorrect setting of mailbox->new based on msg_count in
maildir. (I missed this one for 1f840760e6e0)
* Change mbox to use the context->mtime for stats_last_checked,
removing a race condition.
* Fix mh to actually count the messages in order to generate msg_count.
mh_sequences only covers the range of messages with some sort of
flag.