delete attachment cache files if these are inline images.
The purpose of original logic of loadAttachmnet() is to keep
at most one attachment cache, probably to limit the size of
cached file. But it also purges all inline images.
Integrates CL 149551 from DocomoEmail.
BUG=1884385,1860250
Automated import of CL 149714
Delete the original source message only when the message is in Drafts folder.
If the original message is in the other folder than Drafts, it should be
retained.
Integrates 149194,149306,149459 from DocomoEmail.
BUG=1860250,1887835
Automated import of CL 149607
Integrates 149422,149510,149516 from DocomoEmail.
In some situation, email header From: might be empty or have
illegal email address. The assumption that From: must exist
will be violated and causes ArrayIndexOutOfBoundsException.
BUG=1741969,1890770,1860250
Automated import of CL 149603
Integrates 149300 from DocomoEmail.
This is work around for
- Regex.WEB_URL_PATTERN would not handle protocol part case
insensitive correctly.
- WebView only could follow the href link that protocol part
is lower case.
BUG=1860250,1842508
Automated import of CL 149600
Integrate CL 144586,145919 from imode.
Remove the Emoji escaping/workaround for WebView.
Also integrate MimeHeader.java from CL 143064 from imode.
BUG=1785319,1860250
Automated import of CL 149564
This extension is necessary to support some vendor specific
extended header, such as X-NoRtn-Addr of i-mode.net mail.
This change will update LocalStore database version to 24.
And it should be consistent between donutburger Email and DocomoEmail.
BUG=1741969
Automated import of CL 149328
* changes:
Fix the java.lang.StringIndexOutOfBoundsException that occurs when an SMTP server closes the connection early or returns an empty line. Fix the same error in the Pop3 UIDL parser.
*** Reason for rollback ***
Rollback global lock because bug (now fixed) was not caused by
threading/concurrency.
*** Original change description ***
Evidence from failures, and inspection of source, leads me to believe
that SharedPreferences has some non-thread-safe paths. As a quick,
brute-force workaround, I'm putting a global lock around our use of it.
This is a bit inefficient, but cases of multiple threads writing to it
should be very rare.
Note, we don't have an explicit test for this (I will think about
finding a way to write one), but the evidence of this failure is that
after some amount of activity in the Email app, we see corrup
... description truncated by g4 rollback ...
Automated import of CL 149140
*** Reason for rollback ***
Problem found (bug in ICU encoder/decoder) so instrumentation no
longer required here.
*** Original change description ***
Heavily-instrumented Account.java that's looking for the precise moment
when an Account string gets corrupted. Looks for bad base64 strings
and bad store Uri's. Logs the error, and (optionally/disabled) throws
an exception (good for debugging).
BUG=1822859
Automated import of CL 149088
when an Account string gets corrupted. Looks for bad base64 strings
and bad store Uri's. Logs the error, and (optionally/disabled) throws
an exception (good for debugging).
BUG=1822859
Automated import of CL 148488
READ_PHONE_STATE so a push mode service can hold an active
network connection, monitor roaming status, etc. (2) Refresh
push mode status on Refresh(), not just on Check, so UI
(settings) changes, which call refresh(), propagate immediately
to the stores.
BUG=1776149
Automated import of CL 148377
for new accounts to "1 day". (2) Remove "all" from the list
of available lookback window sizes, to prevent runaway
during our simplistic sync model.
BUG=1789913
Automated import of CL 148376
indicate that the message of interest no longer exists. This is used
in remoteFolder.copyMessages, for example because the message was
already deleted by another client.
BUG=1807499
Automated import of CL 148334
that SharedPreferences has some non-thread-safe paths. As a quick,
brute-force workaround, I'm putting a global lock around our use of it.
This is a bit inefficient, but cases of multiple threads writing to it
should be very rare.
Note, we don't have an explicit test for this (I will think about
finding a way to write one), but the evidence of this failure is that
after some amount of activity in the Email app, we see corruption in
the string mSenderUri.
BUG=1822859
Automated import of CL 148333
the original filters. This affects preview, not download.
This does not preflight the launch at all, it is very simple
as it simply downloads the attachment and then tries to open it.
If no activity is found, it uses a toast to say "couldn't do
that." A better fix will be to pre-check the mime type to see
if there are any intent filters to handle it.
BUG=1650669
Automated import of CL 148257
At this point, *all* MessageListener callbacks should have their
work routed through the handler for deeper processing (especially
anything that affects the lists) to be in the UI thread only.
Also fixed a minor regression introduced in CL 148096 (on
configurations with a single account, the inbox was not
automatically preopening - this affected desktop shortcuts
as well.)
BUG=1812798
Automated import of CL 148256
