Additions and Changes Made to the Netscape Messaging Access SDK
===============================================================
Additions
---------
base64.h
5/19/00 - Added base64 functions
11/10/00 - Added parameter to rfc822_binary() to control whether we want to
have CRLF sequences added to the encoded string.
md5.h
10/20/99 - Added basic MD5 algorithm.
base64.c
5/19/00 - Added abse64 functions
11/10/00 - Added parameter to rfc822_binary() to control whether we want to
have CRLF sequences added to the encoded string.
md5.c
10/20/99 - Added basic MD5 algorithm.
Changes
-------
imap4.h
6/5/99 - Added a new exportable function imap4_clearResponses() to clear any
pending responses.
10/20/99 - Added functions for authentication implementation.
10/26/00 - Added imap4_isConnected().
11/14/00 - Added imap4_processPLAIN().
3/2/01 - Added imap4_startTLS, imap4_getSocket.
5/16/01 - Added imap4_stopProcessResponses().
10/16/01 - imap4_auth() and smtp_auth() now take a username and password which
are used as initial client data for the AUTHENTICATE command. Also
removed all traces of the CDSI macro, since it is no longer useful.
5/27/02 - Added imap4_uidExpunge(). Added boolean flag to imap4_disconnect()
to allow us to force bytes to be sent out when disconnecting, in
case the caller doesn't want to wait for a response when shutting
their connection down.
11/5/02 - Added imap4_sendLocalCommand().
11/15/02 - The fetchdata callback now takes an extra parameter - the data item
we're getting data for.
12/1/02 - Added imap4_cleanupLocalCommand() and changed the type of the tag
parameter in imap4_sendLocalCommand().
12/10/02 - Added code to handle X-ORACLE-QUOTA command, and to handle COMMENT
data item of FETCH command. imap4_append() now takes an extra
annotations parameter. Added callbacks to handle these 2 commands.
12/12/02 - The IMAP4 library now supports the X-LIST command.
1/10/03 - Added entry points and initial implementations for imap4_xMove,
imap4_uidXMove, imap4_xDelete, imap4_xOraclePassword,
imap4_xGetPrefs, imap4_xSetPref. Fixed bug in parseMailboxInfo()
where we did not properly consider escape characters inside quoted
strings, so we would end up outputting too many characters.
1/14/03 - Renamed imap4_xSetPrefs() to imap4_xSetPref() to reflect its usage.
1/15/03 - Moved the majority of the logic to parse preferences into
XGetPrefs() callback so that it can use the services of
CParenParser. As a consequence we don't need xgetprefsstart() and
xgetprefsend() callbacks anymore. Also removed parseqstring() since
it wasn't really doing its job properly.
1/30/03 - Added support for X-RECOVER-MAIL command via imap4_xRecoverMail().
2/5/03 - Updated to reflect new consistent naming schemes of extended Oracle
commands.
2/27/03 - imap4_cleanupLocalCommand() now no longer takes a parameter for the
command since it is not needed - we only execute one local command
at a time.
5/28/03 - imap4_initialize() now takes an extra param that will be used to set
the chunk size as well as the read and write buffer sizes.
8/14/03 - Added yield callback to sink structure.
3/25/04 - Added initial support for saving and loading custom folder types on
the mail server. The code might need to be adjusted when the email
server actually delivers this functionality. For now the folder type
will only be saved in the current profile.
4/6/04 - Updated the name of the IMAP4 command to set the folder type. It's
now called imap4_xOracleSetFolderDesc(). Added extra optional param
to imap4_append() to allow the message type to be saved to the
server. Added ability to parse the message type response from the
server. This data is returned to the client through the
sink_fetchXMsgType() callback.
4/7/04 - Fixed the way we parse the folder description data from the server
in parseXOracleList(). Updated the code to reflect that the data is
a folder description and not a type.
11/2/04 - Modified code related to displaying the current quota usage to use
__int64 instead of unsigned longs, so that we can handle quotas
larger than 4GB.
12/7/04 - Now pass the literal size into imap4_append() rather than
calculating it by reading the whole input stream.
3/22/05 - Added a new callback that signals when all data for an APPEND
command has been sent.
imapdata.h
3/2/01 - Added support for STARTTLS command.
5/16/01 - Added field to imap4Client structure.
11/15/02 - The fetchdata callback now takes an extra parameter - the data item
we're getting data for.
12/10/02 - Added code to handle X-ORACLE-QUOTA command, and to handle COMMENT
data item of FETCH command. imap4_append() now takes an extra
annotations parameter. Added callbacks to handle these 2 commands.
12/12/02 - The IMAP4 library now supports the X-LIST command.
1/10/03 - Added entry points and initial implementations for imap4_xMove,
imap4_uidXMove, imap4_xDelete, imap4_xOraclePassword,
imap4_xGetPrefs, imap4_xSetPref. Fixed bug in parseMailboxInfo()
where we did not properly consider escape characters inside quoted
strings, so we would end up outputting too many characters.
1/15/03 - Moved the majority of the logic to parse preferences into
XGetPrefs() callback so that it can use the services of
CParenParser. As a consequence we don't need xgetprefsstart() and
xgetprefsend() callbacks anymore. Also removed parseqstring()
since it wasn't really doing its job properly.
1/30/03 - Added support for X-RECOVER-MAIL command via imap4_xRecoverMail().
2/5/03 - Updated to reflect new consistent naming schemes of extended
Oracle commands.
2/10/03 - Update variables to reflect new Oracle command naming. Fixed
getExtraLiteral() to handle literals that span multiple lines, and
that end in CRLF sequences. Previously, we did not get the end of
the last line of the response. Also optimized the way we looked for
the end of the literal - instead of reading one line at a time, we
now read the number of bytes in the literal.
2/10/03 - Updated to reflect correct naming of setpref command.
2/12/03 - Updated the string used for X-ORACLE-MAILRECOVER command.
3/25/04 - Added initial support for saving and loading custom folder types on
the mail server. The code might need to be adjusted when the email
server actually delivers this functionality. For now the folder
type will only be saved in the current profile.
4/6/04 - Updated the name of the IMAP4 command to set the folder type. It's
now called imap4_xOracleSetFolderDesc(). Added extra optional param
to imap4_append() to allow the message type to be saved to the
server. Added ability to parse the message type response from the
server. This data is returned to the client through the
sink_fetchXMsgType() callback.
4/7/04 - Fixed the way we parse the folder description data from the server
in parseXOracleList(). Updated the code to reflect that the data
is a folder description and not a type.
11/2/04 - Modified code related to displaying the current quota usage to use
__int64 instead of unsigned longs, so that we can handle quotas
larger than 4GB.
3/22/05 - Added a new callback that signals when all data for an APPEND
command has been sent.
mime.h
3/10/99 - Added a parameter to mime_messagePart_setMessage() to control
whether to copy memory or not.
6/21/99 - Added exportable functions that enable the caller to decode base64
and quoted-printable buffers.
8/12/99 - Added mime_decodeModifiedBase64Buffer() and
mime_encodeModifiedBase64Buffer() functions.
8/26/99 - Added a parameter to mime_decodeModifiedBase64().
5/11/00 - Added mime_messagePart_getBasicPart() to retrieve the attachment
from unknown message content types.
11/30/00 - Added getDefaultFileExtFromContentType().
12/9/00 - getDefaultFileExtFromContentType() now takes a mime_content_type
instead of a string.
12/18/00 - Added MIME_ENCODING_UU to mime_encoding_type enumeration.
9/7/01 - Added mime_multiPart_getBasicPart() and parameter to
mime_message_putByteStream().
9/12/01 - Added mime_message_associateByteStream.
8/20/02 - Started adding support for i18n in mail messages. Fixed access of
hexmap structure for large char values. We need to treat the chars
as unsigned chars to ensure correct access into the array. Modified
mime_bufEncodeQ(), mime_bufEncodeB64(), and
mime_encodeHeaderString() to allow the user to pass in how many
bytes they can store in the output buffer.
12/10/02 - We want to export mime_header_clone() so that it can be called from
outside the library.
mime_internal.h
4/29/99 - Added a data file name field to the mime_basicPart_internal
structure so that we can delete the file associated with the
pTheDataStream field if need be.
5/11/00 - Added field to mime_messagePart_internal structure to handle
unknown message content subtypes.
9/7/01 - Added mime_basicPart * field to mime_multiPart_internal_t.
9/12/01 - Added pInputStream field to mime_message_internal.
mimeparser.h
11/29/00 - Added stristr().
mimeparser_internal.h
4/29/99 - Replaced the message data vector field with a message data output
stream in the mimeParser structure.
9/7/01 - Added boolean field to mimeParser_t.
5/12/04 - Added code to handle lines that are longer than 1024 characters.
Very long header lines and body lines are now handled.
4/7/05 - Added code to stop a buffer overrun when the content-type header
field is very long. Added a buffer size parameter to
mimeParser_nGetContentType() to make sure we don't overrun the
buffer.
4/8/05 - Updated parameter in mimeParser_nGetContentType() to reflect that
it needs to be the buffer length, rather than the buffer size. The
original fix was still crashing, but only in release. This stops
the crash.
nsio.h
3/23/04 - Added void * to the IO structure so that we can pass some context
into the IO callback functions.
nsmail.h
5/29/00 - Added logon error
3/2/01 - Added NSMAIL_ERR_TLS_INIT.
4/21/02 - Added NSMAIL_ERR_TLS_BADVERSION to detect when the SSL version
needed by the server doesn't match what we are trying to connect
with.
5/3/02 - Renamed NSMAIL_ERR_TLS_BADVERSION to NSMAIL_ERR_TLS_BADMETHOD to
be more consistent with the nomenclature used elsewhere.
12/1/02 - IO_read(), IO_readLine(), IO_readDLine() now make a callback call
(if it's registered) so that the client can know what data is coming
from the server. Also updated functions that set and reset the io
functions to reflect the new function pointer.
3/23/04 - Added void * to the IO structure and nsmail_io_fns structure so
that we can pass some context into the IO callback functions.
nssocket.h
3/23/04 - Added void * to the IO structure and nsmail_io_fns structure so
that we can pass some context into the IO callback functions.
nsstream.h
12/7/04 - Added yield callback function to the file_outputStream_tracker
structure that can be called when generating output streams. This
function pointer can be passed in via file_outputStream_create().
smtp.h
5/19/00 - Added the auth, LOGIN and CRAMMD5
11/1/00 - Added initial support for AUTH PLAIN mechanism. Added
smtp_processPLAIN().
3/2/01 - Added smtp_startTLS, smtp_getSocket.
10/16/01 - imap4_auth() and smtp_auth() now take a username and password
which are used as initial client data for the AUTHENTICATE
command. Also removed all traces of the CDSI macro, since it is no
longer useful.
5/29/03 - smtp_initialize() now takes an extra param that will be used to set
the chunk size as well as the read and write buffer sizes.
8/14/03 - Added yield callback to sink structure.
smtppriv.h
5/19/00 - Added the Auth Response
5/29/00 - ADDED AUTH END
6/12/00 - removed AuthEnd since it had no more use
3/2/01 - Added support for STARTTLS command.
util.h
8/12/99 - Added decodeModifiedBase64() function.
8/26/99 - Added decodeQPHeader() to handle RFC2047 style quoted printable.
3/19/01 - Changed definition of BASE64_NOFIT and BASE64_SOMELEFTto make them
unique error codes.
10/16/01 - Added generatePlainClientData().
7/14/03 - Added code to handle case where quoted-printable encoded text ends
line with only LF instead of CRLF. Also improved generic handling
of bad encoded byte sequences - now we print out the 2 bytes
following the '=' instead of printing out NUL chars (and thereby
stopping the buffer). nConvertHexToDec() now returns an errorcode,
and takes an out param for the decoded byte.
4/6/05 - Added code to fix Bug 4287596. This bug was happening because we
were not gracefully handling badly encoded base64 streams. We
would decode the trailing byte(s) if the encoded stream suggested
there were further bytes. This is because the functionality to
decode base64 streams used the return value both for the number of
bytes decoded and to return error codes. T solve this, an extra
parameter was added to return the number of decoded bytes. This is
all we really care about in most cases, as we want to decode as
many bytes as possible.
imap4.c
3/12/99 - Commented out code that was causing the app to crash when it
encountered very long header fields. The individual header fields
do not need to be parsed as of right now.
3/29/99 - Increased the size of memory buffers by 1 to take the NULL
character into consideration when doing reads.
4/1/99 - Added code to parseMailboxInfo() so that the SDK no longer chokes
on folder names that are string literals.
4/6/99 - Did more cleanup of memory buffer sizes so that now both malloc()
and memset() calls are fixed.
4/20/99 - Added getExtraLiteral() function that retrieves the next line from
the server if we come across the start of a string literal. Added
code to handle parsing string literals when parsing the
BODYSTRUCTURE response as well as all non-FETCH responses.
5/6/99 - Added code to parseFetch() to handle fetching several body parts
at the same time. Added a new exportable function
imap4_clearResponses().
5/21/99 - parseMailboxInfo() can now handle parsing folder names that are
literals.
7/8/99 - Updated resolveFetchDataItem() so that it BODY responses are
handled the same as BODYSTRUCTURE responses.
7/15/99 - Added a check in parseStatus() to see if the folder name is
actually a string literal.
8/17/99 - Fixed a parsing problem in parseAcl().
8/23/99 - Updated ACL functions so that parse quoted strings, and string
literals properly.
8/25/99 - Fixed code in parseMyRights() and parseListRights() to handle end
of line parsing.
10/20/99 - Added functions for authentication implementation.
11/3/99 - Fixed parseStatus() to handle folder names with '(' characters in
them.
2/4/00 - Added code to parseNoBad() to handle getting a '*' at the beginning
of the response line. Added code to parseMailboxInfo() to handle
getting a NIL hierarchy delimiter.
5/19/00 - Put the rfc functions in base64.c. Removed some none functional
functions.
9/8/00 - Fixed matchBrackets() so that it now ignores parentheses that are
inside of quotes.
10/23/00 - getExtraLiteral() now takes an extra parameter - the number of
chars in the literal. We now get at least that many characters in
getExtraLiteral(). This fixes a bug that was caused by literals that
contained embedded CRLF sequences. Callers of getExtraLiteral() now
need to calculate the number of chars in the literal as well.
10/26/00 - Added imap4_isConnected(). It checks to see if the socket is still
valid.
10/27/00 - Improved imap4_isConnected() to recognize the case where the server
has disconnected us because of inactivity.
11/10/00 - Added parameter to rfc822_binary() to control whether we want to
have CRLF sequences added to the encoded string.
11/14/00 - Added imap4_processPLAIN(), which knows how to properly handle the
challenge and response for the AUTHENTICATE PLAIN command.
11/24/00 - Added code to matchBrackets() to ignore literal strings when trying
to find the correct right parentheses. This stops us from having
forever looking for string literals that do not exist.
11/27/00 - Added code to resolveFetchDataItem() to better recognize the case
where we are asking for header info, so that in_pimap4->bReadHeader
is set to TRUE. Although we don't currently care about the value of
this variable, we might at some point want to use it.
12/2/00 - Added code to make imap4_isConnected() more robust. We now check
for a SOCKET_ERROR return value from select(). Combined code to deal
with SOCKET_ERROR return values from select() and recv().
3/2/01 - Added workaround for compiler optimization bug in
imap4_processCRAMMD5(). Added imap4_startTLS, imap4_getSocket for
STARTTLS support.
4/19/01 - Now send a blank authorize-id for PLAIN authentication. This is
allowed by RFC2595 and this allows to login to Netscape 4.15 servers
that do not like us sending both and authorize-id and authenticate-
id. Also fixed a silly typo in smtp_startTLS().
5/3/01 - Fixed bug in parseFetch(). When parsing a BODYSTRUCTURE response
from the server, we were calculating the beginning of the response
under the assumption that BODYSTRUCTURE would be in the first block
of bytes read from the server.
5/10/01 - We now handle getting back an empty list of ACLs in parseAcl(). We
used to return a parse error.
5/16/01 - Added field to imap4Client structure that keeps track of when we
want to ignore responses from the server. We check for this boolean
being true in imap4_processResponses() and parseFetch(). Added check
in imap4_isConnected() to only bother checking the state of the
socket if we are actually connected. If we aren't connected, we
haven't initialized the socket properly. Added
imap4_stopProcessResponses() which sets the field in the
imap4Client structure.
7/18/01 - Added check to see if we want to ignore the server response to the
loop in parseFetch() for parsing the message body. This dramatically
improves the responsiveness of cancelling (or skipping) when
downloading an email with a large attachment.
8/14/01 - Removed code in parseEndingLine() and parseBye() that was freeing
memory that was never allocated. This code caused a crash.
10/12/01 - Added local function RemoveAllPendingCommands() to take care of
freeing items from the pending command list. It is now called in
imap4_free(), imap4_disconnect(), parseBye(). The pCommand field of
the pending command list needs to be freed whenever an element is
removed from the list. As a result we need to allocate memory for
that field in imap4_connect(). Fixed memory leak in parseMyrights()
and parseListrights().
10/16/01 - imap4_auth() now takes a username and password which are used as
initial client data for the AUTHENTICATE command. Also removed all
traces of the CDSI macro, since it is no longer useful.
imap4_processPLAIN() now calls generatePlainClientData() to generate
the data to send to the server. imap4_auth() now calls
generatePlainClientData() as well if we are trying to authenticate
using plain and both username and password were passed in. We are
allowed to send initial client data when using PLAIN without waiting
for an intermediate server response.
12/6/01 - parseMailboxInfo() now always passes an unquoted folder path to the
callback function. Now the callback function doesn't need to care
about the low-level IMAP details of how folder names need to be sent
across the wire. Also, we now remove quoted characters (e.g. '\\')
from the strings returned. This lets us handle the Domino IMAP4
server in a better way - it has "\\" as its hierarchy delimiter.
5/27/02 - Added imap4_uidExpunge(). Added boolean flag to imap4_disconnect()
to allow us to force bytes to be sent out when disconnecting, in
case the caller doesn't want to wait for a response when shutting
their connection down.
7/18/02 - Added code to parseFetch() to handle parsing body responses that
were not sent with a string literal. RFC2060 states that the string
for this response could also be either a simple or quoted string. We
came across this situation when trying to open an empty message on
the Oracle IMAP4 server. We were returning a parse error that
resulted in an infinite loop when trying to open a newmail that was
essentially empty. Now we don't return an error and parse as much
info as is available.
7/24/02 - Fixed the way we calculate the beginning and end of the quoted
string in parseFetch(). We used to call sink_fetchData() with a
string length of -1. This fixes a crash when trying to open an
embedded message on the Oracle webmail IMAP4 server that got
returned to us as being empty.
7/24/02 - Changed code in parseFetch() so that it executes the way it used to
in most cases - only in exceptional cases does it try to do more
complicated parsing. This fixes a problem we were getting opening
signed mail (because the response contained a list of header fields
too, and we weren't prepared for that extra info). So now we only do
the extra parsing work when the string returned isn't a literal.
Right now, the only response we get like that is an empty string
response from the Oracle server, but I added code that should work
with more complicated responses, too.
11/5/02 - Added imap4_sendLocalCommand(), which lets a caller simulate passing
a command to the IMAP4 server without actually doing so. This allows
a client to have a client-side cache for frequently requested data.
11/15/02 - The fetchdata callback now takes an extra parameter - the data item
we're getting data for.
12/1/02 - Added imap4_cleanupLocalCommand() and changed the type of the tag
parameter in imap4_sendLocalCommand(). Added code to parseFetch() to
call the read callback function, if it's registered. This lets us
associate data with a particular message, by tracking the fetchstart
and fetchend callbacks too. imap4_cleanupLocalCommand() looks for
the local command in the command list and removes it. We need to do
this because we no longer cache the tagged response from the server,
and so we need to manually remove the local command from the list.
12/10/02 - Added code to handle X-ORACLE-QUOTA command, and to handle COMMENT
data item of FETCH command. imap4_append() now takes an extra
annotations parameter. Added callbacks to handle these 2 commands.
12/12/02 - The IMAP4 library now supports the X-LIST command. The callback
function returns the extra info from this command: uidvalidity,
uidnext, last-modified-time for each folder.
12/18/02 - Fixed parsing bug when parsing a RFC822TEXT data block that is
encoded in a quoted string, and that is following by another data
block that is encoded in a literal. We were looking for a '{' and
finding the literal block from the second data block.
1/10/03 - Added entry points and initial implementations for imap4_xMove,
imap4_uidXMove, imap4_xDelete, imap4_xOraclePassword,
imap4_xGetPrefs, imap4_xSetPref. Fixed bug in parseMailboxInfo()
where we did not properly consider escape characters inside quoted
strings, so we would end up outputting too many characters.
1/15/03 - Moved the majority of the logic to parse preferences into
XGetPrefs() callback so that it can use the services of
CParenParser. As a consequence we don't need xgetprefsstart() and
xgetprefsend() callbacks anymore. Also removed parseqstring() since
it wasn't really doing its job properly.
1/30/03 - Added support for X-RECOVER-MAIL command via imap4_xRecoverMail().
2/5/03 - Updated to reflect new consistent naming schemes of extended Oracle
commands.
2/10/03 - Update variables to reflect new Oracle command naming. Fixed
getExtraLiteral() to handle literals that span multiple lines, and
that end in CRLF sequences. Previously, we did not get the end of
the last line of the response. Also optimized the way we looked for
the end of the literal - instead of reading one line at a time, we
now read the number of bytes in the literal.
2/19/03 - Added 2 new errors that mean we've been disconnected from the server
in imap4_isConnected().
2/25/03 - Fixed broken ParseFetch() logic when trying to parse a response that
contained BODY[] section info. We were finding the literal properly,
and this was messing up all subsequent interaction with the IMAP4
server. Now we recognize when the dataitem in question has a section
and we skip directly to the end of the section before trying to find
the literal. Due to this change, we don't need to look for a list of
header fields later on in ParseFetch().
2/27/03 - imap4_sendLocalCommand() now makes sure the bConnected and
pIO->socket fields are set to values that will not cause us
problems. Since the connection passed into this function is now used
only for local commands, we can reset these fields without worrying
about the potential side-effects.imap4_cleanupLocalCommand() now no
longer takes a parameter for the command since it is not needed - we
only execute one local command at a time.
imap4_cleanupLocalCommand() now resets the pIO->readBufferSize to 0
on successful completion of parseEndingLine(). This is to work
around the case where there is extra info in the local connection
that we want to make sure isn't parsed.
5/28/03 - imap4_initialize() now takes an extra param that will be used to set
the chunk size as well as the read and write buffer sizes.
8/14/03 - Added yield callback to sink structure. It is called (if it is non-
NULL) inside parsePlus() so that large emails that are being
appended to the server don't cause the app to block.
9/5/03 - Added code to parseXOracleList() and matchBrackets() to handle
quoted strings that contain backslashes. We need to ignore escaped
" characters when looking for the quote character. This fixes Bug
3104734.
3/23/04 - Added void * to the IO structure and nsmail_io_fns structure so
that we can pass some context into the IO callback functions. The
IO context can now be get and set via imap4_set_option() and
imap4_get_option().
3/25/04 - Added initial support for saving and loading custom folder types on
the mail server. The code might need to be adjusted when the email
server actually delivers this functionality. For now the folder
type will only be saved in the current profile.
4/6/04 - Updated the name of the IMAP4 command to set the folder type. It's
now called imap4_xOracleSetFolderDesc(). Added extra optional param
to imap4_append() to allow the message type to be saved to the
server. Added ability to parse the message type response from the
server. This data is returned to the client through the
sink_fetchXMsgType() callback.
4/7/04 - Fixed the way we parse the folder description data from the server
in parseXOracleList(). Updated the code to reflect that the data
is a folder description and not a type.
11/2/04 - Modified code related to displaying the current quota usage to use
__int64 instead of unsigned longs, so that we can handle quotas
larger than 4GB.
12/7/04 - Now pass the literal size into imap4_append() rather than
calculating it by reading the whole input stream.
1/10/05 - Call to IO_read() in parseFetch() did not check error code
returned, and this resulted in an infinite loop when there were
network problems during this call.
3/22/05 - Added a new callback that signals when all data for an APPEND
command has been sent.
mime.c
4/6/99 - Previous changes made to this file seem to have been rolledback.
Did cleanup of memory buffer sizes so that now both malloc() and
memset() calls are fixed. Added an extra parameter to
mime_messagePart_setMessage() to handle shallow copying.
4/7/99 - Added a fix to prevent a buffer overrun when dealing with very long
header field values.
4/29/99 - Made changes to mime_decodeBase64() to better handle the last few
bytes of a stream we're trying to decode.
6/21/99 - Added exportable functions that enable the caller to decode base64
and quoted-printable buffers. mime_decodeQP() now calls decodeQP().
8/12/99 - Added mime_decodeModifiedBase64Buffer(),
mime_encodeModifiedBase64Buffer() and mime_bufEncodeModifiedB64()
functions. Fixed code to prevent warnings.
8/26/99 - Added a parameter to mime_decodeModifiedBase64(). Now all of the
encoding and decoding functions return an nsResult.
8/30/99 - Fixed encodeHeaderString().
9/8/99 - Added code to getFileMIMEType() to handle Excel files.
9/9/99 - Update getFileMIMEType() to handle RTF and to return the correct
type for excel documents.
9/10/99 - In decodeQP(), increated the size of achWriteBuffer by 1 to prevent
croak!!!
11/3/99 - Added a fix to mime_decodeQP() to handle the case where a read
buffer doesn't have all of the data required in its buffer (it
needs bytes that will only be in the next read buffer).
3/7/00 - Modified mime_decodeQP() so that it only copies CRLF sequences, not
arbitrary CR of LF characters. Modified getFileMIMEType() so that
the content type parameter is no longer hard-coded to us-ascii, and
the encoding used is quoted printable instead of 7bit (i.e. no
encoding). This should allow attachments with foreign characters to
be encoded properly.
3/23/00 - mime_basicPart_setDataStream() now handles zero length streams
better when we only to save a reference to the stream.
5/11/00 - Now return MIME_ERR_NOT_FOUND in mime_messagePart_getMessage() if
there's no embedded message. Added mime_messagePart_getBasicPart()
for unknown message content subtypes.
11/14/00 - Handle ics extension by setting subtype to calendar.
11/30/00 - Added getDefaultFileExtFromContentType(), which gets the default
extension for a MIME content type (e.g. text/plain -> "txt")
12/9/00 - getDefaultFileExtFromContentType() now takes a mime_content_type
instead of a string.
12/15/00 - Fixed a bug in mime_decodeQP(). We weren't handling the case where
there were stray bytes at the end of a read buffer properly. We now
no longer lose those stray bytes.
12/18/00 - Fixed warning in getDefaultFileExtFromContentType().
3/16/01 - Now ignore BASE64_SOMELEFT error code coming out of decodeBase64()
in mime_decodeBase64Buffer(), so that we don't return an error
when there are stray bytes at the end of a buffer that we could
ignore.
6/5/01 - Added support for MS Powerpoint and MS Access content types to
getFileMIMEType() and getDefaultFileExtFromContentType().
9/7/01 - Added code to mime_multiPart_putByteStream() to ensure that we
don't output the boundary parameter in the Content-Type field twice.
Added mime_multiPart_getBasicPart(). Added parameter to
mime_message_putByteStream() so that we can output only top-level
headers.
9/11/01 - Removed stray fprintf() call that was being output when running
exmigrate.
9/12/01 - Added mime_message_associateByteStream, which associates an already
built byte stream with a mime_message. This stream is now used if
present when mime_message_putByteStream() is called.
10/12/01 - In mime_basicPart_putByteStream(), we need to duplicate the memory
before passing it into buf_instream_create() so that when we free it
we don't free the szMessageBody field of mime_basicPart. Also fixed
a memory leak to do with the buf_instream.
10/12/01 - Added code to mime_basicPart_deleteData() to properly free the
pTheDataStream's memory. This function is currently not called by
anyone.
12/7/01 - Modified mime_decodeQP() to handle .vcf file
4/28/02 - Added code to algorithms that read from a stream into a buffer to
ensure that we don't try to read data into the buffer beyond the
bounds of the buffer's memory. We do this by making sure that if we
right at the end of the stream, that we only try to read however
many bytes are left to be read - not MIME_BUFSIZE bytes. Otherwise
we could crash.
8/20/02 - Started adding support for i18n in mail messages. Fixed access of
hexmap structure for large char values. We need to treat the chars
as unsigned chars to ensure correct access into the array. Modified
mime_bufEncodeQ(), mime_bufEncodeB64(), and
mime_encodeHeaderString() to allow the user to pass in how many
bytes they can store in the output buffer.
2/18/03 - Fixed bug in mime_bufEncodeB64() where did not encode the last 3
byte sequence of the buffer, even though there's just enough space
in the output buffer.
7/14/03 - Added code to handle case where quoted-printable encoded text ends
line with only LF instead of CRLF. Also improved generic handling
of bad encoded byte sequences - now we print out the 2 bytes
following the '=' instead of printing out NUL chars (and thereby
stopping the buffer).
10/28/04 - Fixed several bugs related to multipart/signed emails. We did not
handle mime_message_putByteStream() being called twice for the
same mime_message for multipart/signed emails. The second it was
called, we would output nothing. This is because the input stream
wasn't being rewound to the beginning. We do so now. In
nNewMessageStructure(), we now always use the content_type_params
that have been parsed - before they were being discarded if they
happened to start with "boundary". This didn't handle the case
where some other parameter was first. Also in
mimeParser_parseLine(), we now make sure that the
pMessageDataOutStream is reset to NULL if we encounter a start
or end boundary. Before we were not prepending the necessary
multipart/signed header when extra unexpected blank lines were
present between multiparts.
12/3/04 - Modified getDefaultFileExtFromContentType() so that we now return
an extension for message content types. If the subtype is rfc822,
we return .eml. Otherwise, we return .txt, so that we can let the
user open unknown message types as simple text files.
1/20/05 - Fixed a bug in mime_decodeQP() where CRLF sequences would be lost
if they staddled a buffer block boundary.
3/8/05 - Modified mime_decodeBase64() so that we if we no longer return an
error if we successfully decode at least some data.
4/6/05 - Added code to fix Bug 4287596. This bug was happening because we
were not gracefully handling badly encoded base64 streams. We
would decode the trailing byte(s) if the encoded stream suggested
there were further bytes. This is because the functionality to
decode base64 streams used the return value both for the number
of bytes decoded and to return error codes. T solve this, an
extra parameter was added to return the number of decoded bytes.
This is all we really care about in most cases, as we want to
decode as many bytes as possible.
mime_internal.c
3/12/99 - mime_header_apend() now handles values that are more than 512 bytes
long.
4/13/99 - Added code to mime_basicPart_internal_clone() to handle cloning a
basic part with a data stream set. It now reads from the stream and
fills a buffer. This needs to be improved in the case of large
files.
4/23/99 - In mime_header_new() when szName == NULL. If szValue == NULL
however assume empty string.
4/29/99 - Added code to initialize and free the pDataFileName field of the
mime_basicPart_internal structure.
5/11/99 - Added a fix to mime_header_new() to prevent a crash. Now
mime_header_add() calls mime_header_new().
5/11/00 - Added handling for extra field in mime_messagePart_internal_t
structure.
12/13/00 - Now return MIME_ERR_MAX_NESTED_PARTS_REACHED in
mime_multiPart_addPart_clonable() when we try to add more parts that
is allowed to a multipart. This stops a crash when downloading a
message with "too many" multipart parts.
12/18/00 - Added code to recognize "x-uuencode" to
mime_translateMimeEncodingType().
1/20/01 - We do not want to add header fields whose name or value is empty in
mime_header_add(). This stops a crash when cloning the extra_headers
field.
9/7/01 - Added mime_basicPart * field to mime_multiPart_internal_t. So added
code to mime_multiPart_internal_new(),
mime_multiPart_internal_free(), mime_multiPart_internal_clone() to
manage it.
9/12/01 - Added pInputStream field to mime_message_internal, and added code to
mime_message_internal_new(), mime_message_internal_free() and
mime_message_internal_clone() to support it.
10/12/01 - Now free the memory allocated for the pTheDataStream field in
mime_basicPart_internal_free(). This could cause problems if the
caller tries to free the same stream again.
4/28/02 - In mime_basicPart_internal_clone(), added check to optimize how we
deal with very large text parts. If the size of the stream is above
a certain threshold, we create a stream - otherwise we fall back on
the old code of just allocating a buffer. Added code to algorithms
that read from a stream into a buffer to ensure that we don't try to
read data into the buffer beyond the bounds of the buffer's memory.
We do this by making sure that if we right at the end of the stream,
that we only try to read however many bytes are left to be read -
not MIME_BUFSIZE bytes. Otherwise we could crash. Added code to all
internal clone functions to stop allocated memory being overwritten
and thereby creating a memory leak.
9/5/03 - Added code to mime_header_apend() to handle the case where we're
trying to append to a header that couldn't be found. Rather than
return an error, we add it. This solves the problem where a folded
header line contains only CRLF (Bug 3121863).
mimeparser.c
1/14/99 - Fixed a bug in decodeDataBuffer() that was causing problems with
very long lines in message bodies.
3/10/99 - Added a fix to mimeParser_parseMimeMessage() that stops it from
choking on messages with lines over 1024 characters long.
3/11/99 - Fixed code that was assigning static text strings that were later
being changed. Now use strdup() instead.
3/24/99 - Now zero out a buffer that had garbage at the end of it sometimes
in mimeParser_parseMimeMessage().
3/25/99 - Now only zero out of the buffer in mimeParser_parseMimeMessage() at
the beginning of the call, not in the middle of the loop.
3/29/99 - Increased the size of memory buffers by 1 to take the NULL
character into consideration when doing reads.
4/6/99 - Did cleanup of memory buffer sizes so that now both malloc() and
memset() calls are fixed.
4/7/99 - Added a fix to prevent a buffer overrun when dealing with very long
header field values in mimeParser_parseLine(). Added code to deal
with long lines in mimeParser_parseMimeMessage().
4/23/99 - Fixed a memory leak in addHeader().
4/29/99 - Added code to store the encoded and decoded message data bytes in
streams as opposed to memory. This dramatically speeds up parsing
long attachments. Made a correction to a fix dealing with mime
headers. Added a fix to properly calculate the offset in
mimeParser_parseMimeMessage(). Fixed more memory leaks.
5/20/99 - Added code to fix a parse error that caused problems with parsing
several embedded message attachments at once.
6/23/99 - Now add the proper CRLF to folded mime header field values in
mimeParse_parseLine().
8/17/99 - Fixed a bug that was causing the preamble to show up as part of the
message body for deeply nested embedded messages.
10/18/99 - Commented out some code that resulted in a '\n\r\n' sequence being
appended to every line that is parsed in a quoted printable message
body. Now append '\r\n' to each line instead.
4/21/00 - Fixed a bug in mimeParser_parseLine() that was not outputting lines
of message data that were empty (i.e. only had \r\n).
5/4/00 - Added code to mimeDynamicParser_free() and decodeDataBuffer() to
better clean up the pMessageDataOutStream field of the mimeParser_t
structure.
5/11/00 - Added handling for unknown message content subtypes, especially
message/delivery-status and message/disposition-notification.
5/12/00 - Fixed mimeParser_parseForBoundary() so that it handles long
boundaries properly. Now as many bytes as possible are stored, up
to 70 (as per RFC 2046).
10/20/00 - Commented out some lines that were causing emails not to show up if
the content- headers were ordered badly (i.e.
Content-Transfer-Encoding before Content-Type).
11/29/00 - Added stristr(), which we now call in mimeParser_parseForBoundary()
since we need to be able handle when "boundary" is not always in
lowercase. stristr() does a case-insensitive substring search.
12/7/00 - Added code to properly set the content type in
nNewMessageStructure(). It was getting lost.
12/18/00 - Fixed warnings in mimeParser_setData() and stristr().
12/19/00 - Added code to mimeParser_parseLine() to add MIME header fields to
the parent message part as well as the in usual message part. This
fixes a bug where we were not able to display "Content-" header
fields when trying to show all Internet headers.
1/15/01 - Commented out code in nNewMessageStructure() so that we do not
choke on message/partial content type. Put fix in call to addHeader()
in mimeParser_parseLine() so that we do not add headers to a queue in
all cases. Now return MIME_CONTENT_APPLICATION as default content
type when we come accross an unknown unknown media type in
mimeParser_nGetContentType().
1/20/01 - Rearranged code again in mimeParser_parseLine() to do with adding
Content-xxx header fields to the list as well. Condit
still missing around the call to addHeader() that could
change the logic path, which we do not want. Now ther
simple added call to addHeader() when we're d
"Content-xxx" headers.
2/20/01 - Fixed a bug in mimeParser_parseMimeMessage() in the case where we
did not have enough bytes at the end of input buffer to determine
whether we were dealing with a wrapped line.
3/26/01 - Fixed buffer overrun bug in mimeParser_parseForBoundary() that
would happen when the boundary string in an email was longer than
the RFC allows. We would write a NULL terminating character in
memory not associated with a buffer. Fixed it by allocating a new
buffer to copy the long boundary string into in the case when the
buffer would be too long for our static character array.
9/7/01 - Added boolean field to mimeParser_t to control whether we should be
treating the data that is being parsed as opaque. It's set to TRUE
when parsing multipart/signed parts. When we parsing opaque data,
we cache the data in a basic part that was added to multipart
structure. Added code to correctly format that stream for
multipart/signed parts - we need MIME header fields at the beginning
of the stream and the trailing boundary at the end of it. Fixed code
in mimeParser_parseLine() to put CRLF at the end of the preamble
lines as required by the RFC. Also added code to
mimeParser_parseMimeMessage() to ensure we put CRLF when folding
header fields. mimeParser_checkForLineType() now ignores start
boundaries when we are parsing opaque data - it only cares about
trailing boundaries.
1/29/02 - Improved the logic of the MIME parser so that we detect when we the
header fields are finished (i.e. when we come across one blank line
in the message). This fixes the case where we were treating all of
the lines of a message as if they were header fields (because they
contain ':' characters in them).
4/14/02 - Fixed stupid mistake when building the opaque file for a
multipart/signed part. We used to add the parameters for the
Content-Type field to the MIME-Version instead. It w
matter of putting the two fields in the wrong order. Thi
#1211.
4/28/02 - Added code to prevent small memory leak in nNewMessageStructure().
7/16/02 - Added fix to stop a crash when trying to parse a message where the
first line of the message is a header field that starts with a space.
By definition header lines in a message cannot start with a space
unless it is a folder header field (see RFC2822 for details). This
fix is being done for the Oracle Email server.
3/19/03 - Added code to support parsing message/external-body parts. We treat
such parts as unknown message parts, and we render them as file
attachments (like application/octet-stream).
4/29/03 - Added support for parsing multipart/digest messages. Parts under
this are by default rfc822 messages. Changed code to recognize the
start of headers if we haven't gotten any headers yet, and we have
come across 1 or 2 blank lines - this lets us handle the case where
there are no headers for a nested part.
5/12/04 - Added code to handle lines that are longer than 1024 characters.
Very long header lines and body lines are now handled.
6/9/04 - Fixed bug where we messages without a content-type header field
would not get parsed properly when the whole message was parsed at
once. Now when message data is encountered and we haven't seen a
content-type header field, a default content-type of
will be assumed. This will mean that the whole messag
show up as the body of the message.
6/23/04 - Fixed boundary case bug when parsing very long lines. When looking
for the last space character in a line, we didn't handle the case
where the last space would be the first character of the buffer.
This was resulting in an infinite loop because we were not
updating loop variables properly. Now we need handle this case in
the same way we do when we can't find a space character in the
buffer - we parse the whole buffer as a line. This fixes Bug
3717640.
10/13/04 - Fixed a bug in mimeParser_parseMimeMessage() that could lead to a
crash when dealing with very long header lines when the data is
passed in via a buffer rather than a stream. We now ensure that
the left over buffer is zeroed out also when we are passed a
buffer. Also we now make sure that we add a leading space when we
come across a very long line that only has one space, the first
character. We were already handling the case where the space
wasn't the first character. This fixes an issue found when
investigating Bug 3920779.
10/28/04 - Fixed several bugs related to multipart/signed emails. We did
not handle mime_message_putByteStream() being called twice for
the same mime_message for multipart/signed emails. The second
it was called, we would output nothing. This is because the
input stream wasn't being rewound to the beginning. We do so now.
In nNewMessageStructure(), we now always use the
content_type_params that have been parsed - before they were
being discarded if they happened to start with "boundary". This
didn't handle the case where some other parameter was first.
Also in mimeParser_parseLine(), we now make sure that the
pMessageDataOutStream is reset to NULL if we encounter a start
or end boundary. Before we were not prepending the necessary
multipart/signed header when extra unexpected blank lines
were present between multiparts.
11/29/04 - In mimeParser_parseLine() we only want to hard-code the content-
type to be text/plain if if this is not an unknown message part,
where we've decided to treat it as a basic part (actually
application/octet-stream). This stops us from losing data when
parsing parts with unknown message subtypes.
12/3/04 - Rather that treat unknown unknown message types as
application/octet-stream, we now treat them as text/plain, so
that the user can open them as simple text files.
3/8/05 - Added extra check in decodeDataBuffer() to make sure that there
is a data stream available to decode from.
4/7/05 - Added code to stop a buffer overrun when the content-type header
field is very long. Added a buffer size parameter to
mimeParser_nGetContentType() to make sure we don't overrun the
buffer.
4/8/05 - Updated parameter in mimeParser_nGetContentType() to reflect that
it needs to be the buffer length, rather than the buffer size.
The original fix was still crashing, but only in release. This
stops the crash.
nsio.c
12/1/02 - IO_read(), IO_readLine(), IO_readDLine() now make a callback call
(if it's registered) so that the client can know what data is
coming from the server. Also updated functions that set and reset
the io functions to reflect the new function pointer.
3/23/04 - Added void * to the IO structure and nsmail_io_fns structure so
that we can pass some context into the IO callback functions.
nssocket.c
10/27/00 - We only call WSAStartup() and WSACleanup() if DISABLE_NSSOCK_INIT
is not defined. This is because we now call these functions when
cstmsg32.dll is loaded and unloaded.
3/23/04 - Added void * to the IO structure and nsmail_io_fns structure so
that we can pass some context into the IO callback functions.
nsstream.c
12/7/04 - Added yield callback function to the file_outputStream_tracker
structure that can be called when generating output streams.
This function pointer can be passed in via
file_outputStream_create().
smtp.c
5/19/00 - Added the Auth Response
5/19/00 - Added the Auth with LOGIN mechanism (Cram-Md5 not tested)
5/19/00 - Auth and login working
6/12/00 - Changed how the processResponse from the authentication is made.
Now, an Auth Element is put in the pending list at the end of
smtp_auth. At the begining of the smtp_processLogin, we check the
pending list for Auth and return an error if it isn't. But the auth
element is not removed from the list. It will be used in the
smtp_processResponses call afterwards. AUTHEND is not put in the
list and may be eliminated.
10/31/00 - Cleaned up smtp_processCRAMMD5() so that it actually works, and to
properly check for error conditions. Replaced // comments with /* */
comments, since they are not normally recognized by C compilers.
11/1/00 - Added initial support for AUTH PLAIN mechanism. Added
smtp_processPLAIN().
11/10/00 - Added parameter to rfc822_binary() to control whether we want to
have CRLF sequences added to the encoded string. Since IO_send()
appends a CRLF sequence itself, we don't want to send two of them
sequentially. This was causing a problem with AUTH LOGIN on the
mirapoint SMTP server. Also added code after rfc822_binary() calls
to make sure that we do not send any non-base64 characters to the
server.
11/25/00 - Fixed busted smtp_processPLAIN() so that it now works.
3/2/01 - Added workaround for compiler optimization bug in
smtp_processCRAMMD5(). Added smtp_startTLS, smtp_getSocket for
STARTTLS support.
4/19/01 - Now send a blank authorize-id for PLAIN authentication. This is
allowed by RFC2595 and this allows to login to Netscape 4.15 servers
that do not like us sending both and authorize-id and authenticate-
id. Also fixed a silly typo in smtp_startTLS().
10/16/01 - smtp_auth() now calls generatePlainClientData() if we are trying to
authenticate using plain and both username and password were passed
in. We are allowed to send initial client data when using PLAIN
without waiting for an intermediate server response.
smtp_processPLAIN() now calls generatePlainClientData() to generate
the data to send to the server.
1/22/02 - We can now properly handle multi-line responses to all commands in
smtp_processResponses(). This fixes a problem where we were not
processing server responses properly when the server gave us a
multiline welcome message.
5/29/03 - smtp_initialize() now takes an extra param that will be used to set
the chunk size as well as the read and write buffer sizes.
8/14/03 - Added yield callback to sink structure. It is called (if it is non-
NULL) inside smtp_sendStream() so that large emails that are being
sent to the server don't cause the app to block.
util.c
4/29/99 - Added an extra check to allow callers to specify that they want to
handle any bits still in the queue in decodeBase64().
6/21/99 - Fixed the implementation of decodeHeader() and decodeQP().
8/12/99 - Added decodeModifiedBase64(), decodeBase64WithTable() functions.
Fixed code to prevent warnings.
8/26/99 - Added decodeQPHeader() to handle RFC2047 style quoted printable.
8/30/99 - Initialize ch = 1 in decodeQPxx functions
3/7/00 - Modified decodeQP() so that it only copies CRLF sequences, not
arbitrary CR or LF characters.
2/9/01 - Modified decodeHeader() to better conform to RFC822. We now ignore
whitespace characters between encoded words.
2/10/01 - Stopped a crash in decodeHeader() when we don't find the end of an
encoded word. This crash would only manifest itself when downloading
mail, and it would be caught by our exception handling.
6/8/01 - Modified the behavior of szLTrim() and szRTrim() so that if the
string they are passed is composed completely of whitespace
characters, that they both leave one whitespace character in the
resulting string. This works around a bug in
mimeParser_parseMimeMessage() where folded headers lines were not
being handled properly if they contained only a space.
10/16/01 - Added generatePlainClientData() which takes a username and password
and generates the data string that needs to be sent to the server
when authenticating using the PLAIN mechanism. This code was
extracted from the IMAP4 and SMTP functionality and put here since
it is called now from many spots.
12/5/01 - Added code to decodeqp() and decodeqpheader() to ensure that we
don't read beyond the end of the input buffer. This was causing a
very hard to reproduce crash. Also added an assert() at the end of
each function to ensure that we haven't gone beyond the input buffer.
7/14/03 - Added code to handle case where quoted-printable encoded text ends
line with only LF instead of CRLF. Also improved generic handling of
bad encoded byte sequences - now we print out the 2 bytes following
the '=' instead of printing out NUL chars (and thereby stopping the
buffer). nConvertHexToDec() now returns an errorcode, and takes an
out param for the decoded byte.
4/6/05 - Added code to fix Bug 4287596. This bug was happening because we
were not gracefully handling badly encoded base64 streams. We would
decode the trailing byte(s) if the encoded stream suggested there
were further bytes. This is because the functionality to decode
base64 streams used the return value both for the number of bytes
decoded and to return error codes. T solve this, an extra parameter
was added to return the number of decoded bytes. This is all we
really care about in most cases, as we want to decode as many bytes
as possible. |
Bookmark
