| comi18n.cpp |
end of extern "C" |
1685 |
| comi18n.h |
__cplusplus |
1377 |
| components.conf |
|
3667 |
| jsmime.sys.mjs |
This file exports the JSMime code, polyfilling code as appropriate for use in
Gecko.
|
1615 |
| mime.def |
|
226 |
| mime_closure.cpp |
|
2387 |
| mime_closure.h |
|
2152 |
| mimebuf.cpp |
mimebuf.c - libmsg like buffer handling routines for libmime
|
6972 |
| mimebuf.h |
_MIMEBUF_H_ |
1417 |
| mimecms.cpp |
SEC_PKCS7DecoderContentCallback for SEC_PKCS7DecoderStart() |
30024 |
| mimecms.h |
The MimeEncryptedCMS class implements a type of MIME object where the
object is passed through a CMS decryption engine to decrypt or verify
signatures. That module returns a new MIME object, which is then presented
to the user. See mimecryp.h for details of the general mechanism on which
this is built.
|
1189 |
| mimecom.cpp |
MimeObject (abstract) |
1924 |
| mimecom.h |
XP-COM Bridges for C function calls
|
1306 |
| mimecont.cpp |
This is an abstract class; it shouldn't be directly instantiated. |
6693 |
| mimecont.h |
MimeContainer is the class for the objects representing all MIME
types which can contain other MIME objects within them. In addition
to the methods inherited from MimeObject, it provides one method:
int add_child (MimeObject *parent, MimeObject *child)
Given a parent (a subclass of MimeContainer) this method adds the
child (any MIME object) to the parent's list of children.
The MimeContainer `finalize' method will finalize the children as well.
|
1400 |
| mimecryp.cpp |
(Mostly duplicated from MimeLeaf, see comments in mimecryp.h.)
Initialize a decoder if necessary.
|
18428 |
| mimecryp.h |
The MimeEncrypted class implements a type of MIME object where the object
is passed to some other routine, which then returns a new MIME object.
This is the basis of a decryption module.
Oddly, this class behaves both as a container and as a leaf: it acts as a
container in that it parses out data in order to eventually present a
contained object; however, it acts as a leaf in that this container may
itself have a Content-Transfer-Encoding applied to its body. This violates
the cardinal rule of MIME containers, which is that encodings don't nest,
and therefore containers can't have encodings. But, the fact that the
S/MIME spec doesn't follow the groundwork laid down by previous MIME specs
isn't something we can do anything about at this point...
Therefore, this class duplicates some of the work done by the MimeLeaf
class, to meet its dual goals of container-hood and leaf-hood. (We could
alternately have made this class be a subclass of leaf, and had it duplicate
the effort of MimeContainer, but that seemed like the harder approach.)
The MimeEncrypted class provides the following methods:
void *crypto_init(MimeObject *obj,
int (*output_fn) (const char *data, int32 data_size,
void *output_closure),
void *output_closure)
This is called with the MimeObject representing the encrypted data.
The obj->headers should be used to initialize the decryption engine.
NULL indicates failure; otherwise, an opaque closure object should
be returned.
output_fn is what the decryption module should use to write a new MIME
object (the decrypted data.) output_closure should be passed along to
every call to the output routine.
The data sent to output_fn should begin with valid MIME headers indicating
the type of the data. For example, if decryption resulted in a text
document, the data fed through to the output_fn might minimally look like
Content-Type: text/plain
This is the decrypted data.
It is only two lines long.
Of course, the data may be of any MIME type, including multipart/mixed.
Any returned MIME object will be recursively processed and presented
appropriately. (This also imples that encrypted objects may nest, and
thus that the underlying decryption module must be reentrant.)
int crypto_write (const char *data, int32 data_size, void *crypto_closure)
This is called with the raw encrypted data. This data might not come
in line-based chunks: if there was a Content-Transfer-Encoding applied
to the data (base64 or quoted-printable) then it will have been decoded
first (handing binary data to the filter_fn.) `crypto_closure' is the
object that `crypto_init' returned. This may return negative on error.
int crypto_eof (void *crypto_closure, bool abort_p)
This is called when no more data remains. It may call `output_fn' again
to flush out any buffered data. If `abort_p' is true, then it may choose
to discard any data rather than processing it, as we're terminating
abnormally.
char * crypto_generate_html (void *crypto_closure)
This is called after `crypto_eof' but before `crypto_free'. The crypto
module should return a newly-allocated string of HTML code which
explains the status of the decryption to the user (whether the signature
checked out, etc.)
void crypto_free (void *crypto_closure)
This will be called when we're all done, after `crypto_eof' and
`crypto_emit_html'. It is intended to free any data represented
by the crypto_closure. output_fn may not be called.
int (*parse_decoded_buffer) (const char *buf, int32 size, MimeObject *obj)
This method, of the same name as one in MimeLeaf, is a part of the
afforementioned leaf/container hybridization. This method is invoked
with the content-transfer-decoded body of this part (without line
buffering.) The default behavior of this method is to simply invoke
`crypto_write' on the data with which it is called. It's unlikely that
a subclass will need to specialize this.
|
6418 |
| mimecth.cpp |
These calls are necessary to expose the object class hierarchy
to externally developed content type handlers.
|
1038 |
| mimecth.h |
This is the definitions for the Content Type Handler plugins for
libmime. This will allow developers the dynamically add the ability
for libmime to render new content types in the MHTML rendering of
HTML messages.
|
4349 |
| mimedrft.cpp |
588595FE-1ADA-11d3-A715-0060B0EB39B5 |
76010 |
| mimeebod.cpp |
0 |
14107 |
| mimeebod.h |
The MimeExternalBody class implements the message/external-body MIME type.
(This is not to be confused with MimeExternalObject, which implements the
handler for application/octet-stream and other types with no more specific
handlers.)
|
1272 |
| mimeenc.cpp |
Which encoding to use |
30974 |
| mimeeobj.cpp |
|
8135 |
| mimeeobj.h |
The MimeExternalObject class represents MIME parts which contain data
which cannot be displayed inline -- application/octet-stream and any
other type that is not otherwise specially handled. (This is not to
be confused with MimeExternalBody, which is the handler for the
message/external-object MIME type only.)
|
1547 |
| mimehdrs.cpp |
Don't try and feed me more data after having fed me a blank line... |
24378 |
| mimehdrs.h |
This file defines the interface to message-header parsing and formatting
code, including conversion to HTML. |
3218 |
| MimeHeaderParser.cpp |
|
7401 |
| mimei.cpp |
MimeObject (abstract) |
67410 |
| mimei.h |
|
16166 |
| mimeiimg.cpp |
Can't happen? Close enough. |
7853 |
| mimeiimg.h |
The MimeInlineImage class implements those MIME image types which can be
displayed inline.
|
1001 |
| MimeJSComponents.sys.mjs |
If we get XPConnect-wrapped objects for msgIAddressObjects, we will have
properties defined for 'group' that throws off jsmime. This function converts
the addresses into the form that jsmime expects.
|
16622 |
| mimeleaf.cpp |
Default `parse_buffer' method is one which line-buffers the now-decoded
data and passes it on to `parse_line'. (We snarf the implementation of
this method from our superclass's implementation of `parse_buffer', which
inherited it from MimeObject.)
|
6761 |
| mimeleaf.h |
MimeLeaf is the class for the objects representing all MIME types which
are not containers for other MIME objects. The implication of this is
that they are MIME types which can have Content-Transfer-Encodings
applied to their data. This class provides that service in its
parse_buffer() method:
int (*parse_decoded_buffer) (const char *buf, int32_t size, MimeObject
obj)
The `parse_buffer' method of MimeLeaf passes each block of data through
the appropriate decoder (if any) and then calls `parse_decoded_buffer'
on each block (not line) of output.
The default `parse_decoded_buffer' method of MimeLeaf line-buffers the
now-decoded data, handing each line to the `parse_line' method in turn.
If different behavior is desired (for example, if a class wants access
to the decoded data before it is line-buffered) the `parse_decoded_buffer'
method should be overridden. (MimeExternalObject does this.)
|
2200 |
| mimemalt.cpp |
BACKGROUND
----------
At the simplest level, multipart/alternative means "pick one of these and
display it." However, it's actually a lot more complicated than that.
The alternatives are in preference order, and counterintuitively, they go
from *least* to *most* preferred rather than the reverse. Therefore, when
we're parsing, we can't just take the first one we like and throw the rest
away -- we have to parse through the whole thing, discarding the n'th part if
we are capable of displaying the n+1'th.
Adding a wrinkle to that is the fact that we give the user the option of
demanding the plain-text alternative even though we are perfectly capable of
displaying the HTML, and it is almost always the preferred format, i.e., it
almost always comes after the plain-text alternative.
Speaking of which, you can't assume that each of the alternatives is just a
basic text/[whatever]. There may be, for example, a text/plain followed by a
multipart/related which contains text/html and associated embedded
images. Yikes!
You also can't assume that there will be just two parts. There can be an
arbitrary number, and the ones we are capable of displaying and the ones we
aren't could be interspersed in any order by the producer of the MIME.
We can't just throw away the parts we're not displaying when we're processing
the MIME for display. If we were to do that, then the MIME parts that
remained wouldn't get numbered properly, and that would mean, for example,
that deleting attachments wouldn't work in some messages. Indeed, that very
problem is what prompted a rewrite of this file into its current
architecture.
ARCHITECTURE
------------
Parts are read and queued until we know whether we're going to display
them. If the first pending part is one we don't know how to display, then we
can add it to the MIME structure immediately, with output_p disabled. If the
first pending part is one we know how to display, then we can't add it to the
in-memory MIME structure until either (a) we encounter a later, more
preferred part we know how to display, or (b) we reach the end of the
parts. A display-capable part of the queue may be followed by one or more
display-incapable parts. We can't add them to the in-memory structure until
we figure out what to do with the first, display-capable pending part,
because otherwise the order and numbering will be wrong. All of the logic in
this paragraph is implemented in the flush_children function.
The display_cached_part function is what actually adds a MIME part to the
in-memory MIME structure. There is one complication there which forces us to
violate abstrations... Even if we set output_p on a child before adding it to
the parent, the parse_begin function resets it. The kluge I came up with to
prevent that was to give the child a separate options object and set
output_fn to nullptr in it, because that causes parse_begin to set output_p to
false. This seemed like the least onerous way to accomplish this, although I
can't say it's a solution I'm particularly fond of.
Another complication in display_cached_part is that if we were just a normal
multipart type, we could rely on MimeMultipart_parse_line to notify emitters
about content types, character sets, part numbers, etc. as our new children
get created. However, since we defer creation of some children, the
notification doesn't happen there, so we have to handle it
ourselves. Unfortunately, this requires a small abstraction violation in
MimeMultipart_parse_line -- we have to check there if the entity is
multipart/alternative and if so not notify emitters there because
MimeMultipartAlternative_create_child handles it.
- Jonathan Kamens, 2010-07-23
When the option prefer_plaintext is on, the last text/plain part
should be preferred over any other part that can be displayed. But
if no text/plain part is found, then the algorithm should go as
normal and convert any html part found to text. To achieve this I
found that the simplest way was to change the function display_part_p
into returning priority as an integer instead of boolean can/can't
display. Then I also changed the function flush_children so it selects
the last part with the highest priority. (Priority 0 means it cannot
be displayed and the part is never chosen.)
- Terje Bråten, 2013-02-16
|
21041 |
| mimemalt.h |
The MimeMultipartAlternative class implements the multipart/alternative
MIME container, which displays only one (the `best') of a set of enclosed
documents.
|
1554 |
| mimemapl.cpp |
#### This method is identical to MimeExternalObject_parse_begin
which kinda s#$%s...
|
5563 |
| mimemapl.h |
The MimeMultipartAppleDouble class implements the multipart/appledouble
MIME container, which provides a method of encapsulating and reconstructing
a two-forked Macintosh file.
|
1037 |
| mimemcms.cpp |
#### MimeEncryptedCMS and MimeMultipartSignedCMS have a sleazy,
incestuous, dysfunctional relationship. |
17425 |
| mimemcms.h |
The MimeMultipartSignedCMS class implements a multipart/signed MIME
container with protocol=application/x-CMS-signature, which passes the
signed object through CMS code to verify the signature. See mimemsig.h
for details of the general mechanism on which this is built.
|
1254 |
| mimemdig.cpp |
|
749 |
| mimemdig.h |
The MimeMultipartDigest class implements the multipart/digest MIME
container, which is just like multipart/mixed, except that the default
type (for parts with no type explicitly specified) is message/rfc822
instead of text/plain.
|
1048 |
| mimemmix.cpp |
|
634 |
| mimemmix.h |
The MimeMultipartMixed class implements the multipart/mixed MIME container,
and is also used for any and all otherwise-unrecognised subparts of
multipart/.
|
962 |
| mimemoz2.cpp |
for moved MimeDisplayData struct |
62398 |
| mimemoz2.h |
__cplusplus |
8608 |
| mimempar.cpp |
|
646 |
| mimempar.h |
The MimeMultipartParallel class implements the multipart/parallel MIME
container, which is currently no different from multipart/mixed, since
it's not clear that there's anything useful it could do differently.
|
1044 |
| mimemrel.cpp |
Thoughts on how to implement this:
= if the type of this multipart/related is not text/html, then treat
it the same as multipart/mixed.
= For each part in this multipart/related
= if this part is not the "top" part
= then save this part to a tmp file or a memory object,
kind-of like what we do for multipart/alternative sub-parts.
If this is an object we're blocked on (see below) send its data along.
= else
= emit this part (remember, it's of type text/html)
= at some point, layout may load a URL for <IMG SRC="cid:xxxx">.
we intercept that.
= if one of our cached parts has that cid, return the data for it.
= else, "block", the same way the image library blocks layout when it
doesn't yet have the size of the image.
= at some point, layout may load a URL for <IMG SRC="relative/yyy">.
we need to intercept that too.
= expand the URL, and compare it to our cached objects.
if it matches, return it.
= else block on it.
= once we get to the end, if we have any sub-part references that we're
still blocked on, map over them:
= if they're cid: references, close them ("broken image" results.)
= if they're URLs, then load them in the normal way.
--------------------------------------------------
Ok, that's fairly complicated. How about an approach where we go through
all the parts first, and don't emit until the end?
= if the type of this multipart/related is not text/html, then treat
it the same as multipart/mixed.
= For each part in this multipart/related
= save this part to a tmp file or a memory object,
like what we do for multipart/alternative sub-parts.
= Emit the "top" part (the text/html one)
= intercept all calls to NET_GetURL, to allow us to rewrite the URL.
(hook into netlib, or only into imglib's calls to GetURL?)
(make sure we're behaving in a context-local way.)
= when a URL is loaded, look through our cached parts for a match.
= if we find one, map that URL to a "cid:" URL
= else, let it load normally
= at some point, layout may load a URL for <IMG SRC="cid:xxxx">.
it will do this either because that's what was in the HTML, or because
that's how we "rewrote" the URLs when we intercepted NET_GetURL.
= if one of our cached parts has the requested cid, return the data
for it.
= else, generate a "broken image"
= free all the cached data
--------------------------------------------------
How hard would be an approach where we rewrite the HTML?
(Looks like it's not much easier, and might be more error-prone.)
= if the type of this multipart/related is not text/html, then treat
it the same as multipart/mixed.
= For each part in this multipart/related
= save this part to a tmp file or a memory object,
like what we do for multipart/alternative sub-parts.
= Parse the "top" part, and emit slightly different HTML:
= for each <IMG SRC>, <IMG LOWSRC>, <A HREF>? Any others?
= look through our cached parts for a matching URL
= if we find one, map that URL to a "cid:" URL
= else, let it load normally
= at some point, layout may load a URL for <IMG SRC="cid:xxxx">.
= if one of our cached parts has the requested cid, return the data
for it.
= else, generate a "broken image"
= free all the cached data
|
37721 |
| mimemrel.h |
The MimeMultipartRelated class implements the multipart/related MIME
container, which allows `sibling' sub-parts to refer to each other.
|
2053 |
| mimemsg.cpp |
XP_UNIX |
29376 |
| mimemsg.h |
The MimeMessage class implements the message/rfc822 and message/news
MIME containers, which is to say, mail and news messages.
|
1296 |
| mimemsig.cpp |
This is an abstract class; it shouldn't be directly instantiated. |
27220 |
| mimemsig.h |
The MimeMultipartSigned class implements the multipart/signed MIME
container, which provides a general method of associating a cryptographic
signature to an arbitrary MIME object.
The MimeMultipartSigned class provides the following methods:
void *crypto_init (MimeObject *multipart_object)
This is called with the object, the object->headers of which should be
used to initialize the dexlateion engine. NULL indicates failure;
otherwise, an opaque closure object should be returned.
int crypto_data_hash (const char *data, int32_t data_size,
void *crypto_closure)
This is called with the raw data, for which a signature has been computed.
The crypto module should examine this, and compute a signature for it.
int crypto_data_eof (void *crypto_closure, bool abort_p)
This is called when no more data remains. If `abort_p' is true, then the
crypto module may choose to discard any data rather than processing it,
as we're terminating abnormally.
int crypto_signature_init (void *crypto_closure,
MimeObject *multipart_object,
MimeHeaders *signature_hdrs)
This is called after crypto_data_eof() and just before the first call to
crypto_signature_hash(). The crypto module may wish to do some
initialization here, or may wish to examine the actual headers of the
signature object itself.
The object should completely ignore this signature, and skip all
usual activity.
int crypto_signature_ignore(void *crypto_closure);
int crypto_signature_hash (const char *data, int32_t data_size,
void *crypto_closure)
This is called with the raw data of the detached signature block. It will
be called after crypto_data_eof() has been called to signify the end of
the data which is signed. This data is the data of the signature itself.
int crypto_signature_eof (void *crypto_closure, bool abort_p)
This is called when no more signature data remains. If `abort_p' is true,
then the crypto module may choose to discard any data rather than
processing it, as we're terminating abnormally.
char * crypto_generate_html (void *crypto_closure)
This is called after `crypto_signature_eof' but before `crypto_free'.
The crypto module should return a newly-allocated string of HTML code
which explains the status of the dexlateion to the user (whether the
signature checks out, etc.)
void crypto_free (void *crypto_closure)
This will be called when we're all done, after `crypto_signature_eof' and
`crypto_emit_html'. It is intended to free any data represented by the
crypto_closure.
|
5464 |
| mimemult.cpp |
This is an abstract class; it shouldn't be directly instantiated. |
27201 |
| mimemult.h |
The MimeMultipart class class implements the objects representing all of
the "multipart/" MIME types. In addition to the methods inherited from
MimeContainer, it provides the following methods and class variables:
int create_child (MimeObject *obj)
When it has been determined that a new sub-part should be created,
this method is called to do that. The default value for this method
does it in the usual multipart/mixed way. The headers of the object-
to-be-created may be found in the `hdrs' slot of the `MimeMultipart'
object.
bool output_child_p (MimeObject *parent, MimeObject *child)
Whether this child should be output. Default method always says `yes'.
int parse_child_line (MimeObject *obj, const char *line, int32_t length,
bool first_line_p)
When we have a line which should be handed off to the currently-active
child object, this method is called to do that. The `first_line_p'
variable will be true only for the very first line handed off to this
sub-part. The default method simply passes the line to the most-
recently-added child object.
int close_child (MimeObject *self)
When we reach the end of a sub-part (a separator line) this method is
called to shut down the currently-active child. The default method
simply calls `parse_eof' on the most-recently-added child object.
MimeMultipartBoundaryType check_boundary (MimeObject *obj,
const char *line, int32_t length)
This method is used to examine a line and determine whether it is a
part boundary, and if so, what kind. It should return a member of
the MimeMultipartBoundaryType describing the line.
const char *default_part_type
This is the type which should be assumed for sub-parts which have
no explicit type specified. The default is "text/plain", but the
"multipart/digest" subclass overrides this to "message/rfc822".
|
3804 |
| mimeobj.cpp |
This is an abstract class; it shouldn't be directly instantiated. |
10552 |
| mimeobj.h |
MimeObject is the base-class for the objects representing all other
MIME types. It provides several methods:
int initialize (MimeObject *obj)
This is called from mime_new() when a new instance is allocated.
Subclasses should do whatever setup is necessary from this method,
and should call the superclass's initialize method, unless there's
a specific reason not to.
void finalize (MimeObject *obj)
This is called from mime_free() and should free all data associated
with the object. If the object points to other MIME objects, they
should be finalized as well (by calling mime_free(), not by calling
their finalize() methods directly.)
int parse_buffer (const char *buf, int32_t size, MimeObject *obj)
This is the method by which you feed arbitrary data into the parser
for this object. Most subclasses will probably inherit this method
from the MimeObject base-class, which line-buffers the data and then
hands it off to the parse_line() method.
If this object uses a Content-Transfer-Encoding (base64, qp, uue)
then the data may be decoded by parse_buffer() before parse_line()
is called. (The MimeLeaf class provides this functionality.)
int parse_begin (MimeObject *obj)
Called after `init' but before `parse_line' or `parse_buffer'.
Can be used to initialize various parsing machinery.
int parse_line (const char *line, int32_t length, MimeObject *obj)
This method is called (by parse_buffer()) for each complete line of
data handed to the parser, and is the method which most subclasses
will override to implement their parsers.
When handing data off to a MIME object for parsing, one should always
call the parse_buffer() method, and not call the parse_line() method
directly, since the parse_buffer() method may do other transformations
on the data (like base64 decoding.)
One should generally not call parse_line() directly, since that could
bypass decoding. One should call parse_buffer() instead.
int parse_eof (MimeObject *obj, bool abort_p)
This is called when there is no more data to be handed to the object:
when the parent object is done feeding data to an object being parsed.
Implementors of this method should be sure to also call the parse_eof()
methods of any sub-objects to which they have pointers.
This is also called by the finalize() method, just before object
destruction, if it has not already been called.
The `closed_p' instance variable is used to prevent multiple calls to
`parse_eof'.
int parse_end (MimeObject *obj)
Called after `parse_eof' but before `finalize'.
This can be used to free up any memory no longer needed now that parsing
is done (to avoid surprises due to unexpected method combination, it's
best to free things in this method in preference to `parse_eof'.)
Implementors of this method should be sure to also call the parse_end()
methods of any sub-objects to which they have pointers.
This is also called by the finalize() method, just before object
destruction, if it has not already been called.
The `parsed_p' instance variable is used to prevent multiple calls to
`parse_end'.
bool displayable_inline_p (MimeObjectClass *class, MimeHeaders *hdrs)
This method should return true if this class of object will be displayed
directly, as opposed to being displayed as a link. This information is
used by the "multipart/alternative" parser to decide which of its children
is the ``best'' one to display. Note that this is a class method, not
an object method -- there is not yet an instance of this class at the time
that it is called. The `hdrs' provided are the headers of the object that
might be instantiated -- from this, the method may extract additional
information that it might need to make its decision.
|
8295 |
| mimeParser.sys.mjs |
Determine an arbitrary "parameter" part of a mail header.
@param {string} headerStr - The string containing all parts of the header.
@param {string} parameter - The parameter we are looking for.
'multipart/signed; protocol="xyz"', 'protocol' --> returns "xyz"
@return {string} String containing the value of the parameter; or "".
|
9916 |
| mimepbuf.cpp |
See mimepbuf.h for a description of the mission of this file.
Implementation:
When asked to buffer an object, we first try to malloc() a buffer to
hold the upcoming part. First we try to allocate a 50k buffer, and
then back off by 5k until we are able to complete the allocation,
or are unable to allocate anything.
As data is handed to us, we store it in the memory buffer, until the
size of the memory buffer is exceeded (including the case where no
memory buffer was able to be allocated at all.)
Once we've filled the memory buffer, we open a temp file on disk.
Anything that is currently in the memory buffer is then flushed out
to the disk file (and the memory buffer is discarded.) Subsequent
data that is passed in is appended to the file.
Thus only one of the memory buffer or the disk buffer ever exist at
the same time; and small parts tend to live completely in memory
while large parts tend to live on disk.
When we are asked to read the data back out of the buffer, we call
the provided read-function with either: the contents of the memory
buffer; or blocks read from the disk file.
|
8328 |
| mimepbuf.h |
This file provides the ability to save up the entire contents of a MIME
object (of arbitrary size), and then emit it all at once later. The
buffering is done in an efficient way that works well for both very large
and very small objects.
This is used in two places:
= The implementation of multipart/alternative uses this code to do a
one-part-lookahead. As it traverses its children, it moves forward
until it finds a part which cannot be displayed; and then it displays
the *previous* part (the last which *could* be displayed.) This code
is used to hold the previous part until it is needed.
|
2432 |
| mimesun.cpp |
Sun messages always have separators at the beginning. |
11334 |
| mimesun.h |
MimeSunAttachment is the class for X-Sun-Attachment message contents, which
is the Content-Type assigned by that pile of garbage called MailTool. This
is not a MIME type per se, but it's very similar to multipart/mixed, so it's
easy to parse. Lots of people use MailTool, so what the hell.
The format is this:
= Content-Type is X-Sun-Attachment
= parts are separated by lines of exactly ten dashes
= just after the dashes comes a block of headers, including:
X-Sun-Data-Type: (manditory)
Values are Text, Postscript, Scribe, SGML, TeX, Troff, DVI,
and Message.
X-Sun-Encoding-Info: (optional)
Ordered, comma-separated values, including Compress and Uuencode.
X-Sun-Data-Name: (optional)
File name, maybe.
X-Sun-Data-Description: (optional)
Longer text.
X-Sun-Content-Lines: (manditory, unless Length is present)
Number of lines in the body, not counting headers and the blank
line that follows them.
X-Sun-Content-Length: (manditory, unless Lines is present)
Bytes, presumably using Unix line terminators.
|
1914 |
| mimetenr.cpp |
All the magic for this class is in mimetric.c; since text/enriched and
text/richtext are so similar, it was easiest to implement them in the
same method (but this is a subclass anyway just for general goodness.)
|
992 |
| mimetenr.h |
The MimeInlineTextEnriched class implements the text/enriched MIME content
type, as defined in RFC 1563. It does this largely by virtue of being a
subclass of the MimeInlineTextRichtext class.
|
1057 |
| mimetext.cpp |
|
14528 |
| mimetext.h |
The MimeInlineText class is the superclass of all handlers for the
MIME text/ content types (which convert various text formats to HTML,
in one form or another.)
It provides two services:
= if ROT13 decoding is desired, the text will be rotated before
the `parse_line' method it called;
= text will be converted from the message's charset to the "target"
charset before the `parse_line' method is called.
The contract with charset-conversion is that the converted data will
be such that one may interpret any octets (8-bit bytes) in the data
which are in the range of the ASCII characters (0-127) as ASCII
characters. It is explicitly legal, for example, to scan through
the string for "<" and replace it with "<", and to search for things
that look like URLs and to wrap them with interesting HTML tags.
The charset to which we convert will probably be UTF-8 (an encoding of
the Unicode character set, with the feature that all octets with the
high bit off have the same interpretations as ASCII.)
#### NOTE: if it turns out that we use JIS (ISO-2022-JP) as the target
encoding, then this is not quite true; it is safe to search for the
low ASCII values (under hex 0x40, octal 0100, which is '@') but it
is NOT safe to search for values higher than that -- they may be
being used as the subsequent bytes in a multi-byte escape sequence.
It's a nice coincidence that HTML's critical characters ("<", ">",
and "&") have values under 0x40...
|
3007 |
| mimeTextHTMLParsed.cpp |
Most of this code is copied from mimethsa. If you find a bug here, check that
class, too. |
6487 |
| mimeTextHTMLParsed.h |
_MIMETEXTHTMLPARSED_H_ |
948 |
| mimethpl.cpp |
TODO:
- If you Save As File .html with this mode, you get a total mess.
- Print is untested (crashes in all modes).
|
6225 |
| mimethpl.h |
The MimeInlineTextHTMLAsPlaintext class converts HTML->TXT->HTML, i.e.
HTML to Plaintext and the result to HTML again.
This might sound crazy, maybe it is, but it is for the "View as Plaintext"
option, if the sender didn't supply a plaintext alternative (bah!).
|
1281 |
| mimethsa.cpp |
Most of this code is copied from mimethpl; see there for source comments.
If you find a bug here, check that class, too.
|
4915 |
| mimethsa.h |
_MIMETHPL_H_ |
971 |
| mimethtm.cpp |
If this HTML part has a Content-Base header, and if we're displaying
to the screen (that is, not writing this part "raw") then translate
that Content-Base header into a <BASE> tag in the HTML.
|
8245 |
| mimethtm.h |
The MimeInlineTextHTML class implements the text/html MIME content type.
|
1140 |
| mimetpfl.cpp |
force out any separators... |
20922 |
| mimetpfl.h |
The MimeInlineTextPlainFlowed class implements the
text/plain MIME content type for the special case of a supplied
format=flowed. See
ftp://ftp.ietf.org/internet-drafts/draft-gellens-format-06.txt for
more information.
|
1874 |
| mimetpla.cpp |
|
14958 |
| mimetpla.h |
The MimeInlineTextPlain class implements the text/plain MIME content type,
and is also used for all otherwise-unknown text/ subtypes.
|
1253 |
| mimetric.cpp |
This function has this clunky interface because it needs to be called
from outside this module (no MimeObject, etc.)
|
11554 |
| mimetric.h |
The MimeInlineTextRichtext class implements the (obsolete and deprecated)
text/richtext MIME content type, as defined in RFC 1341, and also the
text/enriched MIME content type, as defined in RFC 1563.
|
1111 |
| mimeunty.cpp |
Oops, those shouldn't still be here... |
17098 |
| mimeunty.h |
The MimeUntypedText class is used for untyped message contents, that is,
it is the class used for the body of a message/rfc822 object which had
no* Content-Type header, as opposed to an unrecognized content-type.
Such a message, technically, does not contain MIME data (it follows only
RFC 822, not RFC 1521.)
This is a container class, and the reason for that is that it loosely
parses the body of the message looking for ``sub-parts'' and then
creates appropriate containers for them.
More specifically, it looks for uuencoded data. It may do more than that
some day.
Basically, the algorithm followed is:
if line is "begin 644 foo.gif"
if there is an open sub-part, close it
add a sub-part with type: image/gif; encoding: x-uue
hand this line to it
and hand subsequent lines to that subpart
else if there is an open uuencoded sub-part, and line is "end"
hand this line to it
close off the uuencoded sub-part
else if there is an open sub-part
hand this line to it
else
open a text/plain subpart
hand this line to it
Adding other types than uuencode to this (for example, PGP) would be
pretty straightforward.
|
2505 |
| modlmime.h |
Opaque object describing a block of message headers, and a couple of
routines for extracting data from one.
|
17573 |
| modmimee.h |
mimeenc.c --- MIME encoders and decoders, version 2 (see mimei.h)
Copyright (c) 1996 Netscape Communications Corporation, all rights reserved.
Created: Jamie Zawinski <jwz@netscape.com>, 15-May-96.
|
2215 |
| moz.build |
|
1725 |
| nsMimeObjectClassAccess.cpp |
The following macros actually implement addref, release and
query interface for our component.
|
2261 |
| nsMimeObjectClassAccess.h |
This interface is implemented by libmime. This interface is used by
a Content-Type handler "Plug In" (i.e. vCard) for accessing various
internal information about the object class system of libmime. When
libmime progresses to a C++ object class, this would probably change.
|
1947 |
| nsMimeStringResources.h |
Note that the negative values are not actually strings: they are error
codes masquerading as strings. Do not pass them to MimeGetStringByID()
expecting to get anything back for your trouble.
|
1477 |
| nsSimpleMimeConverterStub.cpp |
|
5882 |
| nsSimpleMimeConverterStub.h |
NS_SIMPLE_MIME_CONVERTER_STUB_H_ |
603 |
| nsStreamConverter.cpp |
|
32516 |
| nsStreamConverter.h |
nsStreamConverter_h_ |
3240 |