tvl-depot/third_party/lisp/mime4cl/mime.lisp

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

1050 lines
40 KiB
Common Lisp
Raw Permalink Normal View History

;;; mime4cl.lisp --- MIME primitives for Common Lisp
;;; Copyright (C) 2005-2008, 2010 by Walter C. Pelissero
;;; Copyright (C) 2021-2023 by the TVL Authors
;;; Author: Walter C. Pelissero <walter@pelissero.de>
;;; Project: mime4cl
;;; This library is free software; you can redistribute it and/or
;;; modify it under the terms of the GNU Lesser General Public License
;;; as published by the Free Software Foundation; either version 2.1
;;; of the License, or (at your option) any later version.
;;; This library is distributed in the hope that it will be useful,
;;; but WITHOUT ANY WARRANTY; without even the implied warranty of
;;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
;;; Lesser General Public License for more details.
;;; You should have received a copy of the GNU Lesser General Public
;;; License along with this library; if not, write to the Free
;;; Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
;;; 02111-1307 USA
(in-package :mime4cl)
(defclass mime-part ()
((subtype
:type (or string null)
:initarg :subtype
:accessor mime-subtype
;; some mime types don't require a subtype
:initform nil)
(type-parameters
:type list
:initarg :type-parameters
:initform '()
:accessor mime-type-parameters)
(version
:type (or string null)
:initarg :mime-version
:initform "1.0"
:accessor mime-version)
(id
:initform nil
:initarg :id
:reader mime-id)
(description
:initform nil
:initarg :description
:accessor mime-description)
(encoding
:initform :7bit
:initarg :encoding
:reader mime-encoding
:documentation
"It's supposed to be either:
:7BIT, :8BIT, :BINARY, :QUOTED-PRINTABLE, :BASE64, a
X-token or an ietf-token (whatever that means).")
(disposition
:type (or string null)
:initarg :disposition
:initform nil
:accessor mime-disposition)
(disposition-parameters
:type list
:initarg :disposition-parameters
:initform '()
:accessor mime-disposition-parameters))
(:documentation
"Abstract base class for all types of MIME parts."))
(defclass mime-bodily-part (mime-part)
((body
:initarg :body
:accessor mime-body))
(:documentation
"Abstract base class for MIME parts with a body."))
(defclass mime-unknown-part (mime-bodily-part)
((type
:initarg :type
:reader mime-type
:documentation
"The original type string from the MIME header."))
(:documentation
"MIME part unknown to this library. Accepted but not handled."))
(defclass mime-text (mime-bodily-part) ())
;; This turns out to be handy when making methods specialised
;; non-textual attachments.
(defclass mime-binary (mime-bodily-part) ())
(defclass mime-image (mime-binary) ())
(defclass mime-audio (mime-binary) ())
(defclass mime-video (mime-binary) ())
(defclass mime-application (mime-binary) ())
(defclass mime-multipart (mime-part)
((parts :initarg :parts
:accessor mime-parts)))
(defclass mime-message (mime-part)
((headers :initarg :headers
:initform '()
:type list
:accessor mime-message-headers)
(real-message :initarg :body
:accessor mime-body)))
(defun mime-part-p (object)
(typep object 'mime-part))
(defmethod initialize-instance ((part mime-multipart) &key &allow-other-keys)
(call-next-method)
;; The initialization argument of the PARTS slot of a mime-multipart
;; is expected to be a list of mime-parts. Thus, we implicitly
;; create the mime parts using the arguments found in this list.
(with-slots (parts) part
(when (slot-boundp part 'parts)
(setf parts
(mapcar #'(lambda (subpart)
(if (mime-part-p subpart)
subpart
(apply #'make-instance subpart)))
parts)))))
(defmethod initialize-instance ((part mime-message) &key &allow-other-keys)
(call-next-method)
;; Allow a list of mime parts to be specified as body of a
;; mime-message. In that case we implicitly create a mime-multipart
;; and assign to the body slot.
(with-slots (real-message) part
(when (and (slot-boundp part 'real-message)
(consp real-message))
(setf real-message
(make-instance 'mime-multipart :parts real-message)))))
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
(defun alist= (alist1 alist2 &key (test #'eql))
(null
(set-difference alist1 alist2
:test #'(lambda (x y)
(and (funcall test (car x) (car y))
(funcall test (cdr x) (cdr y)))))))
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
(defgeneric mime= (mime1 mime2)
(:documentation
"Return true if MIME1 and MIME2 have equivalent structure and identical bodies (as for EQ)."))
(defmethod mime= ((part1 mime-part) (part2 mime-part))
(macrolet ((null-or (compare x y)
`(or (and (not ,x)
(not ,y))
(and ,x ,y
(,compare ,x ,y))))
(cmp-slot (compare reader)
`(null-or ,compare (,reader part1) (,reader part2))))
(and (eq (class-of part1) (class-of part2))
(cmp-slot string-equal mime-subtype)
(alist= (mime-type-parameters part1)
(mime-type-parameters part2)
:test #'string-equal)
(cmp-slot string= mime-id)
(cmp-slot string= mime-description)
(cmp-slot eq mime-encoding)
(cmp-slot equal mime-disposition)
(alist= (mime-disposition-parameters part1)
(mime-disposition-parameters part2)
:test #'string-equal))))
(defmethod mime= ((part1 mime-multipart) (part2 mime-multipart))
(and (call-next-method)
(every #'mime= (mime-parts part1) (mime-parts part2))))
(defmethod mime= ((part1 mime-message) (part2 mime-message))
(and (call-next-method)
(alist= (mime-message-headers part1) (mime-message-headers part2)
:test #'string=)
(mime= (mime-body part1) (mime-body part2))))
refactor(mime4cl): replace *-input-adapter-stream with flexi-streams The input adapter streams were input streams yielding either binary or character data that could be constructed from a variable data source. The stream would take care not to destroy the underlying data source (i.e. not close it if it was a stream), so similar to with FILE-PORTIONs, but simpler. Unfortunately, the implementation was quite inefficient: They are ultimately defined in terms of a function that retrieves the next character in the source. This only allows for an implementation of READ-CHAR (and READ-BYTE). Thanks to cl/8559, READ-SEQUENCE can be used on e.g. FILE-PORTION, but this was still negated by a input adapter based on one—then, READ-SEQUENCE would need to fall back on READ-CHAR or READ-BYTE again. Luckily, we can replace BINARY-INPUT-ADAPTER-STREAM and CHARACTER-INPUT-ADAPTER-STREAM with a much simpler abstraction: Instead of extra stream classes, we have a function, MAKE-INPUT-ADAPTER, which returns an appropriate instance of FLEXI-STREAM based on a given source. This way, the need for a distinction between binary and character input adapter is eliminated, since FLEXI-STREAMS supports both binary and character reads (external format is not yet handled, though). Consequently, the :binary keyword argument to MIME-BODY-STREAM can be dropped. flexi-streams provides stream classes for everything except a stream that doesn't close the underlying one. Since we have already implemented this in POSITIONED-FLEXI-INPUT-STREAM, we can split this functionality into a new superclass ADAPTER-FLEXI-INPUT-STREAM. This change also allows addressing the performance regression encountered in cl/8559: It seems that flexi-streams performs worse when we are reading byte by byte or char by char. (After this change mblog is still two times slower than on r/6150.) By eliminating the adapter streams, we can start utilizing READ-SEQUENCE via decoding code that supports it (i.e. qbase64) and bring performance on par with r/6150 again. Surely there are also ways to gain back even more performance which has to be determined using profiling. Buffering more aggressively seems like a sure bet, though. Switching to flexi-streams still seems like a no-brainer, as it allows us to drop a lot of code that was quite hacky (e.g. DELIMITED-INPUT- STREAM) and implements en/decoding handling we did not support before, but would need for improved correctness. Change-Id: Ie2d1f4e42b47512a5660a1ccc0deeec2bff9788d Reviewed-on: https://cl.tvl.fyi/c/depot/+/8581 Autosubmit: sterni <sternenseemann@systemli.org> Reviewed-by: sterni <sternenseemann@systemli.org> Tested-by: BuildkiteCI
2023-05-15 23:36:40 +02:00
(defun mime-body-stream (mime-part)
(make-input-adapter (mime-body mime-part)))
(defun mime-body-length (mime-part)
(let ((body (mime-body mime-part)))
;; here the stream type is missing on purpose, because we may not
;; be able to size the length of a stream
(etypecase body
(string
(length body))
(vector
(length body))
(pathname
(file-size body))
(file-portion
(with-open-stream (in (open-decoded-file-portion body))
(loop
for byte = (read-byte in nil)
while byte
count byte))))))
refactor(mime4cl): replace *-input-adapter-stream with flexi-streams The input adapter streams were input streams yielding either binary or character data that could be constructed from a variable data source. The stream would take care not to destroy the underlying data source (i.e. not close it if it was a stream), so similar to with FILE-PORTIONs, but simpler. Unfortunately, the implementation was quite inefficient: They are ultimately defined in terms of a function that retrieves the next character in the source. This only allows for an implementation of READ-CHAR (and READ-BYTE). Thanks to cl/8559, READ-SEQUENCE can be used on e.g. FILE-PORTION, but this was still negated by a input adapter based on one—then, READ-SEQUENCE would need to fall back on READ-CHAR or READ-BYTE again. Luckily, we can replace BINARY-INPUT-ADAPTER-STREAM and CHARACTER-INPUT-ADAPTER-STREAM with a much simpler abstraction: Instead of extra stream classes, we have a function, MAKE-INPUT-ADAPTER, which returns an appropriate instance of FLEXI-STREAM based on a given source. This way, the need for a distinction between binary and character input adapter is eliminated, since FLEXI-STREAMS supports both binary and character reads (external format is not yet handled, though). Consequently, the :binary keyword argument to MIME-BODY-STREAM can be dropped. flexi-streams provides stream classes for everything except a stream that doesn't close the underlying one. Since we have already implemented this in POSITIONED-FLEXI-INPUT-STREAM, we can split this functionality into a new superclass ADAPTER-FLEXI-INPUT-STREAM. This change also allows addressing the performance regression encountered in cl/8559: It seems that flexi-streams performs worse when we are reading byte by byte or char by char. (After this change mblog is still two times slower than on r/6150.) By eliminating the adapter streams, we can start utilizing READ-SEQUENCE via decoding code that supports it (i.e. qbase64) and bring performance on par with r/6150 again. Surely there are also ways to gain back even more performance which has to be determined using profiling. Buffering more aggressively seems like a sure bet, though. Switching to flexi-streams still seems like a no-brainer, as it allows us to drop a lot of code that was quite hacky (e.g. DELIMITED-INPUT- STREAM) and implements en/decoding handling we did not support before, but would need for improved correctness. Change-Id: Ie2d1f4e42b47512a5660a1ccc0deeec2bff9788d Reviewed-on: https://cl.tvl.fyi/c/depot/+/8581 Autosubmit: sterni <sternenseemann@systemli.org> Reviewed-by: sterni <sternenseemann@systemli.org> Tested-by: BuildkiteCI
2023-05-15 23:36:40 +02:00
(defmacro with-input-from-mime-body-stream ((stream part) &body forms)
`(with-open-stream (,stream (mime-body-stream ,part))
,@forms))
(defmethod mime= ((part1 mime-bodily-part) (part2 mime-bodily-part))
(and (call-next-method)
(with-input-from-mime-body-stream (in1 part1)
(with-input-from-mime-body-stream (in2 part2)
(loop
for b1 = (read-byte in1 nil)
for b2 = (read-byte in2 nil)
always (eq b1 b2)
while (and b1 b2))))))
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
(defgeneric get-mime-type-parameter (part name)
(:documentation
"Return the MIME type parameter associated to NAME of PART."))
(defgeneric (setf get-mime-type-parameter) (value part name)
(:documentation
"Set the MIME type parameter associated to NAME of PART."))
(defmethod get-mime-type-parameter ((part mime-part) name)
(cdr (assoc name (mime-type-parameters part) :test #'string-equal)))
(defmethod (setf get-mime-type-parameter) (value part name)
(aif (assoc name (mime-type-parameters part) :test #'string-equal)
(setf (cdr it) value)
(push (cons name value)
(mime-type-parameters part)))
value)
(defgeneric get-mime-disposition-parameter (part name)
(:documentation
"Return the MIME disposition parameter associated to NAME of PART."))
(defmethod get-mime-disposition-parameter ((part mime-part) name)
(cdr (assoc name (mime-disposition-parameters part) :test #'string-equal)))
(defmethod (setf get-mime-disposition-parameter) (value part name)
(aif (assoc name (mime-disposition-parameters part) :test #'string-equal)
(setf (cdr it) value)
(push (cons name value)
(mime-disposition-parameters part))))
(defmethod mime-part-file-name ((part mime-part))
"Return the filename associated to mime PART or NIL if the mime
part doesn't have a file name."
(or (get-mime-disposition-parameter part :filename)
(get-mime-type-parameter part :name)))
(defmethod (setf mime-part-file-name) (value (part mime-part))
"Set the filename associated to mime PART."
(setf (get-mime-disposition-parameter part :filename) value
(get-mime-type-parameter part :name) value))
(defun mime-text-charset (part)
(get-mime-type-parameter part :charset))
(defun split-header-parts (string)
"Split parts of a MIME headers. These are divided by
semi-colons not within strings or comments."
(labels ((skip-comment (pos)
(loop
while (< pos (length string))
do (case (elt string pos)
(#\( (setf pos (skip-comment (1+ pos))))
(#\\ (incf pos 2))
(#\) (return (1+ pos)))
(otherwise (incf pos)))
finally (return pos)))
(skip-string (pos)
(loop
while (< pos (length string))
do (case (elt string pos)
(#\\ (incf pos 2))
(#\" (return (1+ pos)))
(otherwise (incf pos)))
finally (return pos))))
(loop
with start = 0 and i = 0 and parts = '()
while (< i (length string))
do (case (elt string i)
(#\; (push (subseq string start i) parts)
(setf start (incf i)))
(#\" (setf i (skip-string i)))
(#\( (setf i (skip-comment (1+ i))))
(otherwise (incf i)))
finally (return (mapcar #'string-trim-whitespace (nreverse (cons (subseq string start) parts)))))))
(defun parse-parameter (string)
"Given a string like \"foo=bar\" return a pair (\"foo\" .
\"bar\"). Return NIL if string is not parsable."
;; TODO(sterni): when-let
(let ((equal-position (position #\= string)))
(when equal-position
(let ((key (subseq string 0 equal-position)))
(if (= equal-position (1- (length string)))
(cons key "")
(let ((value (string-trim-whitespace (subseq string (1+ equal-position)))))
(cons key
(if (and (> (length value) 1)
(char= #\" (elt value 0)))
;; the syntax of a RFC822 string is more or
;; less the same as the Lisp one: use the Lisp
;; reader
(or (ignore-errors (read-from-string value))
(subseq value 1))
(let ((end (or (position-if #'whitespace-p value)
(length value))))
(subseq value 0 end))))))))))
(defun parse-content-type (string)
"Parse string as a Content-Type MIME header and return a list
of three elements. The first is the type, the second is the
subtype and the third is an alist of parameters and their values.
Example: (\"text\" \"plain\" ((\"charset\" . \"us-ascii\")...))."
(let* ((parts (split-header-parts string))
(content-type-string (car parts))
(slash (position #\/ content-type-string)))
;; You'd be amazed to know how many MUA can't produce an RFC
;; compliant message.
(when slash
(let ((type (subseq content-type-string 0 slash))
(subtype (subseq content-type-string (1+ slash))))
(list type subtype (remove nil (mapcar #'parse-parameter (cdr parts))))))))
(defun parse-content-disposition (string)
"Parse string as a Content-Disposition MIME header and return a
list. The first element is the layout, the other elements are
the optional parameters alist.
Example: (\"inline\" (\"filename\" . \"doggy.jpg\"))."
(let ((parts (split-header-parts string)))
(cons (car parts) (mapcan #'(lambda (parameter-string)
(awhen (parse-parameter parameter-string)
(list it)))
(cdr parts)))))
(defun parse-RFC822-header (string)
"Parse STRING which should be a valid RFC822 message header and
return two values: a string of the header name and a string of
the header value."
(let ((colon (position #\: string)))
(when colon
(values (string-trim-whitespace (subseq string 0 colon))
(string-trim-whitespace (subseq string (1+ colon)))))))
(defvar *default-type* '("text" "plain" (("charset" . "us-ascii")))
"Internal special variable that contains the default MIME type at
any given time of the parsing phase. There are MIME container parts
that may change this.")
(defvar *mime-types*
'((:text mime-text)
(:image mime-image)
(:audio mime-audio)
(:video mime-video)
(:application mime-application)
(:multipart mime-multipart)
(:message mime-message)))
(defgeneric mime-part-size (part)
(:documentation
"Return the size in bytes of the body of a MIME part."))
(defgeneric print-mime-part (part stream)
(:documentation
"Output to STREAM one of the possible human-readable representation
of mime PART. Binary parts are omitted. This function can be used to
quote messages, for instance."))
(defun do-multipart-parts (body-stream part-boundary contents-function end-part-function)
"Read through BODY-STREAM. Call CONTENTS-FUNCTION at
each (non-boundary) line or END-PART-FUNCTION at each PART-BOUNDARY."
(let* ((boundary (s+ "--" part-boundary))
(boundary-length (length boundary)))
(labels ((output-line (line)
(funcall contents-function line))
(end-part ()
(funcall end-part-function))
(last-part ()
(end-part)
(return-from do-multipart-parts))
(process-line (line)
(cond ((not (string-starts-with boundary line))
;; normal line
(output-line line))
((and (= (length (string-trim-whitespace line))
(+ 2 boundary-length))
(string= "--" line :start2 boundary-length))
;; end of the last part
(last-part))
;; according to RFC2046 "the boundary may be followed
;; by zero or more characters of linear whitespace"
((= (length (string-trim-whitespace line)) boundary-length)
;; beginning of the next part
(end-part))
(t
;; the line boundary is followed by some
;; garbage; we treat it as a normal line
(output-line line)))))
(loop
for line = (read-line body-stream nil)
;; we should never reach the end of a proper multipart MIME
;; stream, but we don't want to be fooled by corrupted ones,
;; so we check for EOF
unless line
do (last-part)
do (process-line line)))))
(defun index-multipart-parts (body-stream part-boundary)
"Read from BODY-STREAM and return the file offset of the MIME parts
separated by PART-BOUNDARY."
(let ((parts '())
(start 0)
(len 0)
(beginning-of-part-p t))
(flet ((sum-chars (line)
(incf len (length line))
;; account for the #\newline
(if beginning-of-part-p
(setf beginning-of-part-p nil)
(incf len)))
(end-part ()
(setf beginning-of-part-p t)
(push (cons start (+ start len)) parts)
(setf start (file-position body-stream)
len 0)))
(do-multipart-parts body-stream part-boundary #'sum-chars #'end-part)
;; the first part is all the stuff up to the first boundary;
;; just junk
(cdr (nreverse parts)))))
(defgeneric encode-mime-part (part stream))
(defgeneric encode-mime-body (part stream))
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
(defun write-mime-header (part stream)
(when (mime-version part)
(format stream "~&MIME-Version: ~A~%" (mime-version part)))
(format stream "~&Content-Type: ~A~:{; ~A=~S~}~%" (mime-type-string part)
(mapcar #'(lambda (pair)
(list (car pair) (cdr pair)))
(mime-type-parameters part)))
(awhen (mime-encoding part)
(format stream "Content-Transfer-Encoding: ~A~%" it))
(awhen (mime-description part)
(format stream "Content-Description: ~A~%" it))
(when (mime-disposition part)
(format stream "Content-Disposition: ~A~:{; ~A=~S~}~%"
(mime-disposition part)
(mapcar #'(lambda (pair)
(list (car pair) (cdr pair)))
(mime-disposition-parameters part))))
(awhen (mime-id part)
(format stream "Content-ID: ~A~%" it))
(terpri stream))
(defmethod encode-mime-part ((part mime-part) stream)
(write-mime-header part stream)
(encode-mime-body part stream))
(defmethod encode-mime-part ((part mime-message) stream)
;; tricky: we have to mix the MIME headers with the message headers
(dolist (h (mime-message-headers part))
(unless (stringp (car h))
(setf (car h)
(string-capitalize (car h))))
(unless (or (string-starts-with "content-" (car h) #'string-equal)
(string-equal "mime-version" (car h)))
(format stream "~A: ~A~%"
(car h) (cdr h))))
(encode-mime-part (mime-body part) stream))
(defmethod encode-mime-part ((part mime-multipart) stream)
;; choose a boundary if not already set
(let* ((original-boundary (get-mime-type-parameter part :boundary))
(boundary (choose-boundary (mime-parts part) original-boundary)))
(unless (and original-boundary
(string= boundary original-boundary))
(setf (get-mime-type-parameter part :boundary) boundary))
(call-next-method)))
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
(defmethod encode-mime-body ((part mime-part) stream)
(with-input-from-mime-body-stream (in part)
(encode-stream in stream (mime-encoding part))))
(defmethod encode-mime-body ((part mime-message) stream)
(encode-mime-body (mime-body part) stream))
(defmethod encode-mime-body ((part mime-multipart) stream)
(let ((boundary (or (get-mime-type-parameter part :boundary)
(setf (get-mime-type-parameter part :boundary)
(choose-boundary (mime-parts part))))))
(dolist (p (mime-parts part))
(format stream "~%--~A~%" boundary)
(encode-mime-part p stream))
(format stream "~%--~A--~%" boundary)))
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
(defun time-RFC822-string (&optional (epoch (get-universal-time)))
"Return a string describing the current time according to
the RFC822."
(multiple-value-bind (ss mm hh day month year week-day dst tz) (decode-universal-time epoch)
(declare (ignore dst))
(format nil "~A, ~A ~A ~2,'0D ~2,'0D:~2,'0D:~2,'0D ~:[-~;+~]~2,'0D~2,'0D"
(subseq (week-day->string week-day) 0 3)
day (subseq (month->string month) 0 3) (mod year 100) hh mm ss
(plusp tz) (abs (truncate tz)) (mod (* 60 tz) 60))))
(defun parse-RFC822-date (date-string)
"Parse a RFC822 compliant date string and return an universal
time."
;; if we can't parse it, just return NIL
(ignore-errors
;; skip the optional DoW
(awhen (position #\, date-string)
(setf date-string (subseq date-string (1+ it))))
(destructuring-bind (day month year time &optional tz &rest rubbish)
(split-at '(#\space #\tab) date-string)
(declare (ignore rubbish))
(destructuring-bind (hh mm &optional ss) (split-string-at-char time #\:)
(encode-universal-time
(if ss
(read-from-string ss)
0)
(read-from-string mm)
(read-from-string hh)
(read-from-string day)
(1+ (position month
'("Jan" "Feb" "Mar" "Apr" "May" "Jun"
"Jul" "Aug" "Sep" "Oct" "Nov" "Dec")
:test #'string-equal))
(read-from-string year)
(when (and tz (or (char= #\+ (elt tz 0))
(char= #\- (elt tz 0))))
(/ (read-from-string tz) 100)))))))
(defun read-RFC822-headers (stream &optional required-headers)
"Read RFC822 compliant headers from STREAM and return them in a
alist of keyword and string pairs. REQUIRED-HEADERS is a list of
header names we are interested in; if NIL return all headers
found in STREAM."
;; the skip-header variable is to avoid the mistake of appending a
;; continuation line of a header we don't want to a header we want
(loop
with headers = '() and skip-header = nil
for line = (let ((line (read-line stream nil)))
;; skip the Unix "From " header if present
(if (string-starts-with "From " line)
(read-line stream nil)
line))
then (read-line stream nil)
while (and line
(not (zerop (length line))))
do (if (whitespace-p (elt line 0))
(unless (or skip-header
(null headers))
(setf (cdar headers) (s+ (cdar headers) '(#\newline) line)))
(multiple-value-bind (name value) (parse-RFC822-header line)
;; the line contained rubbish instead of an header: we
;; play nice and return as we were at the end of the
;; headers
(unless name
(return (nreverse headers)))
(if (or (null required-headers)
(member name required-headers :test #'string-equal))
(progn
(push (cons name value) headers)
(setf skip-header nil))
(setf skip-header t))))
finally (return (nreverse headers))))
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
(defgeneric mime-message (thing)
(:documentation
"Convert THING to a MIME-MESSAGE object."))
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
(defun mime-message-header-values (name message &key decode)
"Return all values of the header with NAME in MESSAGE, optionally decoding
it according to RFC2047 if :DECODE is T."
(loop ;; A header may occur multiple times
for header in (mime-message-headers message)
;; MIME Headers should be case insensitive
;; https://stackoverflow.com/a/6143644
when (string-equal (car header) name)
collect (if decode
(decode-RFC2047 (cdr header))
(cdr header))))
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
(defvar *lazy-mime-decode* t
"If true don't decode mime bodies in memory.")
(defgeneric decode-mime-body (part input-stream))
refactor(3p/lisp/mime4cl): use flexi-streams and binary input This refactor is driven by the following (ultimate) aims: - Get rid of as much of the custom stream code in mime4cl which makes less code to maintain in the future. - Lay the groundwork for correct handling of 8bit transfer encoding: The mime4cl we inherited assumes that any MIME message can be decoded completely by the CL implementation (in SBCL's case using latin1) into CHARACTERs. This is not necessarily the case. flexi-streams allows changing how the stream is decoded on the fly and also has support for reading the underlying bytes which is perfect for the requirements decoding MIME has. - Since flexi-streams uses trivial-gray-streams, it supports READ-SEQUENCE. Taking advantage of this may improve decoding performance significantly in the future. This incurs the following changes: - Naturally we now open given files as binary files in MIME-MESSAGE. Given strings are encoded using STRING-TO-OCTETS and then passed on to a new octet vector method. Instead of MY-STRING-INPUT-STREAM this now uses flexi-streams' WITH-INPUT-FROM-SEQUENCE. - OPEN-FILE-PORTION and OPEN-DECODED-FILE-PORTION need to be merged, since the transfer encoding not only implies an extra decoder stream that needs to be attached after file portion stream, but also imply a certain encoding of the stream itself (mostly binary vs. ASCII). As flexi-streams can change their encoding on the fly this could be untangled again, but it is not strictly necessary. As before, we use the DATA slot of the file portion to create a fresh stream if possible. Instead of strings we now use an vector of octets to match MIME-MESSAGE. The actual portioned stream relies on POSITIONED-FLEXI-INPUT-STREAM, a subclass of the stock FLEXI-INPUT-STREAM class, described below. - POSITIONED-FLEXI-INPUT-STREAM replaces DELIMITED-INPUT-STREAM. It is created using MAKE-POSITIONED-FLEXI-INPUT-STREAM which accepts the same arguments as MAKE-FLEXI-STREAMS and, additionally, :IGNORE-CLOSE. A POSITIONED-FLEXI-INPUT-STREAM works the same as an FLEXI-INPUT-STREAM, but upon creation, the underlying stream is rewinded or forwarded to the argument given by :POSITION using FILE-POSITION. If :IGNORE-CLOSE is T, a call to CLOSE is not forwarded to the underlying stream. Change-Id: I2d48c769bb110ca0b7cf52441bd63c1e1c2ccd04 Reviewed-on: https://cl.tvl.fyi/c/depot/+/8559 Reviewed-by: sterni <sternenseemann@systemli.org> Autosubmit: sterni <sternenseemann@systemli.org> Tested-by: BuildkiteCI
2023-05-03 18:19:41 +02:00
(defmethod decode-mime-body ((part mime-part) (stream flexi-stream))
(let ((base (flexi-stream-root-stream stream)))
(if *lazy-mime-decode*
(setf (mime-body part)
(make-file-portion :data (etypecase base
(vector-stream
(flexi-streams::vector-stream-vector base))
(file-stream
(pathname base)))
:encoding (mime-encoding part)
:start (flexi-stream-position stream)
:end (flexi-stream-bound stream)))
(call-next-method))))
(defmethod decode-mime-body ((part mime-part) (stream file-stream))
(if *lazy-mime-decode*
(setf (mime-body part)
(make-file-portion :data (pathname stream)
:encoding (mime-encoding part)
:start (file-position stream)))
(call-next-method)))
refactor(3p/lisp/mime4cl): use flexi-streams and binary input This refactor is driven by the following (ultimate) aims: - Get rid of as much of the custom stream code in mime4cl which makes less code to maintain in the future. - Lay the groundwork for correct handling of 8bit transfer encoding: The mime4cl we inherited assumes that any MIME message can be decoded completely by the CL implementation (in SBCL's case using latin1) into CHARACTERs. This is not necessarily the case. flexi-streams allows changing how the stream is decoded on the fly and also has support for reading the underlying bytes which is perfect for the requirements decoding MIME has. - Since flexi-streams uses trivial-gray-streams, it supports READ-SEQUENCE. Taking advantage of this may improve decoding performance significantly in the future. This incurs the following changes: - Naturally we now open given files as binary files in MIME-MESSAGE. Given strings are encoded using STRING-TO-OCTETS and then passed on to a new octet vector method. Instead of MY-STRING-INPUT-STREAM this now uses flexi-streams' WITH-INPUT-FROM-SEQUENCE. - OPEN-FILE-PORTION and OPEN-DECODED-FILE-PORTION need to be merged, since the transfer encoding not only implies an extra decoder stream that needs to be attached after file portion stream, but also imply a certain encoding of the stream itself (mostly binary vs. ASCII). As flexi-streams can change their encoding on the fly this could be untangled again, but it is not strictly necessary. As before, we use the DATA slot of the file portion to create a fresh stream if possible. Instead of strings we now use an vector of octets to match MIME-MESSAGE. The actual portioned stream relies on POSITIONED-FLEXI-INPUT-STREAM, a subclass of the stock FLEXI-INPUT-STREAM class, described below. - POSITIONED-FLEXI-INPUT-STREAM replaces DELIMITED-INPUT-STREAM. It is created using MAKE-POSITIONED-FLEXI-INPUT-STREAM which accepts the same arguments as MAKE-FLEXI-STREAMS and, additionally, :IGNORE-CLOSE. A POSITIONED-FLEXI-INPUT-STREAM works the same as an FLEXI-INPUT-STREAM, but upon creation, the underlying stream is rewinded or forwarded to the argument given by :POSITION using FILE-POSITION. If :IGNORE-CLOSE is T, a call to CLOSE is not forwarded to the underlying stream. Change-Id: I2d48c769bb110ca0b7cf52441bd63c1e1c2ccd04 Reviewed-on: https://cl.tvl.fyi/c/depot/+/8559 Reviewed-by: sterni <sternenseemann@systemli.org> Autosubmit: sterni <sternenseemann@systemli.org> Tested-by: BuildkiteCI
2023-05-03 18:19:41 +02:00
(defmethod decode-mime-body ((part mime-part) (stream vector-stream))
(if *lazy-mime-decode*
(setf (mime-body part)
refactor(3p/lisp/mime4cl): use flexi-streams and binary input This refactor is driven by the following (ultimate) aims: - Get rid of as much of the custom stream code in mime4cl which makes less code to maintain in the future. - Lay the groundwork for correct handling of 8bit transfer encoding: The mime4cl we inherited assumes that any MIME message can be decoded completely by the CL implementation (in SBCL's case using latin1) into CHARACTERs. This is not necessarily the case. flexi-streams allows changing how the stream is decoded on the fly and also has support for reading the underlying bytes which is perfect for the requirements decoding MIME has. - Since flexi-streams uses trivial-gray-streams, it supports READ-SEQUENCE. Taking advantage of this may improve decoding performance significantly in the future. This incurs the following changes: - Naturally we now open given files as binary files in MIME-MESSAGE. Given strings are encoded using STRING-TO-OCTETS and then passed on to a new octet vector method. Instead of MY-STRING-INPUT-STREAM this now uses flexi-streams' WITH-INPUT-FROM-SEQUENCE. - OPEN-FILE-PORTION and OPEN-DECODED-FILE-PORTION need to be merged, since the transfer encoding not only implies an extra decoder stream that needs to be attached after file portion stream, but also imply a certain encoding of the stream itself (mostly binary vs. ASCII). As flexi-streams can change their encoding on the fly this could be untangled again, but it is not strictly necessary. As before, we use the DATA slot of the file portion to create a fresh stream if possible. Instead of strings we now use an vector of octets to match MIME-MESSAGE. The actual portioned stream relies on POSITIONED-FLEXI-INPUT-STREAM, a subclass of the stock FLEXI-INPUT-STREAM class, described below. - POSITIONED-FLEXI-INPUT-STREAM replaces DELIMITED-INPUT-STREAM. It is created using MAKE-POSITIONED-FLEXI-INPUT-STREAM which accepts the same arguments as MAKE-FLEXI-STREAMS and, additionally, :IGNORE-CLOSE. A POSITIONED-FLEXI-INPUT-STREAM works the same as an FLEXI-INPUT-STREAM, but upon creation, the underlying stream is rewinded or forwarded to the argument given by :POSITION using FILE-POSITION. If :IGNORE-CLOSE is T, a call to CLOSE is not forwarded to the underlying stream. Change-Id: I2d48c769bb110ca0b7cf52441bd63c1e1c2ccd04 Reviewed-on: https://cl.tvl.fyi/c/depot/+/8559 Reviewed-by: sterni <sternenseemann@systemli.org> Autosubmit: sterni <sternenseemann@systemli.org> Tested-by: BuildkiteCI
2023-05-03 18:19:41 +02:00
(make-file-portion :data (flexi-streams::vector-stream-vector stream)
:encoding (mime-encoding part)
refactor(3p/lisp/mime4cl): use flexi-streams and binary input This refactor is driven by the following (ultimate) aims: - Get rid of as much of the custom stream code in mime4cl which makes less code to maintain in the future. - Lay the groundwork for correct handling of 8bit transfer encoding: The mime4cl we inherited assumes that any MIME message can be decoded completely by the CL implementation (in SBCL's case using latin1) into CHARACTERs. This is not necessarily the case. flexi-streams allows changing how the stream is decoded on the fly and also has support for reading the underlying bytes which is perfect for the requirements decoding MIME has. - Since flexi-streams uses trivial-gray-streams, it supports READ-SEQUENCE. Taking advantage of this may improve decoding performance significantly in the future. This incurs the following changes: - Naturally we now open given files as binary files in MIME-MESSAGE. Given strings are encoded using STRING-TO-OCTETS and then passed on to a new octet vector method. Instead of MY-STRING-INPUT-STREAM this now uses flexi-streams' WITH-INPUT-FROM-SEQUENCE. - OPEN-FILE-PORTION and OPEN-DECODED-FILE-PORTION need to be merged, since the transfer encoding not only implies an extra decoder stream that needs to be attached after file portion stream, but also imply a certain encoding of the stream itself (mostly binary vs. ASCII). As flexi-streams can change their encoding on the fly this could be untangled again, but it is not strictly necessary. As before, we use the DATA slot of the file portion to create a fresh stream if possible. Instead of strings we now use an vector of octets to match MIME-MESSAGE. The actual portioned stream relies on POSITIONED-FLEXI-INPUT-STREAM, a subclass of the stock FLEXI-INPUT-STREAM class, described below. - POSITIONED-FLEXI-INPUT-STREAM replaces DELIMITED-INPUT-STREAM. It is created using MAKE-POSITIONED-FLEXI-INPUT-STREAM which accepts the same arguments as MAKE-FLEXI-STREAMS and, additionally, :IGNORE-CLOSE. A POSITIONED-FLEXI-INPUT-STREAM works the same as an FLEXI-INPUT-STREAM, but upon creation, the underlying stream is rewinded or forwarded to the argument given by :POSITION using FILE-POSITION. If :IGNORE-CLOSE is T, a call to CLOSE is not forwarded to the underlying stream. Change-Id: I2d48c769bb110ca0b7cf52441bd63c1e1c2ccd04 Reviewed-on: https://cl.tvl.fyi/c/depot/+/8559 Reviewed-by: sterni <sternenseemann@systemli.org> Autosubmit: sterni <sternenseemann@systemli.org> Tested-by: BuildkiteCI
2023-05-03 18:19:41 +02:00
:start (flexi-streams::vector-stream-index stream)))
(call-next-method)))
(defmethod decode-mime-body ((part mime-part) stream)
(setf (mime-body part)
(decode-stream-to-sequence stream (mime-encoding part))))
(defmethod decode-mime-body ((part mime-multipart) stream)
"Decode STREAM according to PART characteristics and return a
list of MIME parts."
(save-file-excursion (stream)
(let ((offsets (index-multipart-parts stream (get-mime-type-parameter part :boundary))))
(setf (mime-parts part)
(mapcar #'(lambda (p)
(destructuring-bind (start . end) p
(let ((*default-type* (if (eq :digest (mime-subtype part))
'("message" "rfc822" ())
'("text" "plain" (("charset" . "us-ascii")))))
(in (make-positioned-flexi-input-stream stream
:position start
:bound end
:ignore-close t)))
(read-mime-part in))))
offsets)))))
(defmethod decode-mime-body ((part mime-message) stream)
"Read from STREAM the body of PART. Return the decoded MIME
body."
(setf (mime-body part)
(read-mime-message stream)))
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
(defvar +known-encodings+ '(:7BIT :8BIT :BINARY :QUOTED-PRINTABLE :BASE64)
"List of known content encodings.")
(defun keywordify-encoding (string)
"Return a keyword for a content transfer encoding string.
Return STRING itself if STRING is an unkown encoding."
(aif (member string +known-encodings+ :test #'string-equal)
(car it)
string))
(defun header (name headers)
(let ((elt (assoc name headers :test #'string-equal)))
(values (cdr elt) (car elt))))
(defun (setf header) (value name headers)
(let ((entry (assoc name headers :test #'string-equal)))
(unless entry
(error "missing header ~A can't be set" name))
(setf (cdr entry) value)))
(defun make-mime-part (headers stream)
"Create a MIME-PART object based on HEADERS and a body which
has to be read from STREAM. If the mime part type can't be
guessed from the headers, use the *DEFAULT-TYPE*."
(flet ((hdr (what)
(header what headers)))
(destructuring-bind (type subtype parms)
(or
(aand (hdr :content-type)
(parse-content-type it))
*default-type*)
(let* ((class (or (cadr (assoc type *mime-types* :test #'string-equal))
'mime-unknown-part))
(disp (aif (hdr :content-disposition)
(parse-content-disposition it)
(values nil nil)))
(part (make-instance class
:type (hdr :content-type)
:subtype subtype
:type-parameters parms
:disposition (car disp)
:disposition-parameters (cdr disp)
:mime-version (hdr :mime-version)
:encoding (keywordify-encoding
(hdr :content-transfer-encoding))
:description (hdr :content-description)
:id (hdr :content-id)
:allow-other-keys t)))
(decode-mime-body part stream)
part))))
(defun read-mime-part (stream)
"Read mime part from STREAM. Return a MIME-PART object."
(let ((headers (read-rfc822-headers stream
'(:mime-version :content-transfer-encoding :content-type
:content-disposition :content-description :content-id))))
(make-mime-part headers stream)))
(defun read-mime-message (stream)
"Main function to read a MIME message from a stream. It
returns a MIME-MESSAGE object."
(let ((headers (read-rfc822-headers stream))
(*default-type* '("text" "plain" (("charset" . "us-ascii")))))
(flet ((hdr (what)
(header what headers)))
(destructuring-bind (type subtype parms)
(or (aand (hdr :content-type)
(parse-content-type it))
*default-type*)
(declare (ignore type subtype))
(make-instance 'mime-message
:headers headers
;; this is just for easy access
:type-parameters parms
:body (make-mime-part headers stream))))))
(defmethod mime-message ((msg mime-message))
msg)
(defmethod mime-message ((msg string))
refactor(3p/lisp/mime4cl): use flexi-streams and binary input This refactor is driven by the following (ultimate) aims: - Get rid of as much of the custom stream code in mime4cl which makes less code to maintain in the future. - Lay the groundwork for correct handling of 8bit transfer encoding: The mime4cl we inherited assumes that any MIME message can be decoded completely by the CL implementation (in SBCL's case using latin1) into CHARACTERs. This is not necessarily the case. flexi-streams allows changing how the stream is decoded on the fly and also has support for reading the underlying bytes which is perfect for the requirements decoding MIME has. - Since flexi-streams uses trivial-gray-streams, it supports READ-SEQUENCE. Taking advantage of this may improve decoding performance significantly in the future. This incurs the following changes: - Naturally we now open given files as binary files in MIME-MESSAGE. Given strings are encoded using STRING-TO-OCTETS and then passed on to a new octet vector method. Instead of MY-STRING-INPUT-STREAM this now uses flexi-streams' WITH-INPUT-FROM-SEQUENCE. - OPEN-FILE-PORTION and OPEN-DECODED-FILE-PORTION need to be merged, since the transfer encoding not only implies an extra decoder stream that needs to be attached after file portion stream, but also imply a certain encoding of the stream itself (mostly binary vs. ASCII). As flexi-streams can change their encoding on the fly this could be untangled again, but it is not strictly necessary. As before, we use the DATA slot of the file portion to create a fresh stream if possible. Instead of strings we now use an vector of octets to match MIME-MESSAGE. The actual portioned stream relies on POSITIONED-FLEXI-INPUT-STREAM, a subclass of the stock FLEXI-INPUT-STREAM class, described below. - POSITIONED-FLEXI-INPUT-STREAM replaces DELIMITED-INPUT-STREAM. It is created using MAKE-POSITIONED-FLEXI-INPUT-STREAM which accepts the same arguments as MAKE-FLEXI-STREAMS and, additionally, :IGNORE-CLOSE. A POSITIONED-FLEXI-INPUT-STREAM works the same as an FLEXI-INPUT-STREAM, but upon creation, the underlying stream is rewinded or forwarded to the argument given by :POSITION using FILE-POSITION. If :IGNORE-CLOSE is T, a call to CLOSE is not forwarded to the underlying stream. Change-Id: I2d48c769bb110ca0b7cf52441bd63c1e1c2ccd04 Reviewed-on: https://cl.tvl.fyi/c/depot/+/8559 Reviewed-by: sterni <sternenseemann@systemli.org> Autosubmit: sterni <sternenseemann@systemli.org> Tested-by: BuildkiteCI
2023-05-03 18:19:41 +02:00
(mime-message (flexi-streams:string-to-octets msg)))
refactor(3p/lisp/mime4cl): use flexi-streams and binary input This refactor is driven by the following (ultimate) aims: - Get rid of as much of the custom stream code in mime4cl which makes less code to maintain in the future. - Lay the groundwork for correct handling of 8bit transfer encoding: The mime4cl we inherited assumes that any MIME message can be decoded completely by the CL implementation (in SBCL's case using latin1) into CHARACTERs. This is not necessarily the case. flexi-streams allows changing how the stream is decoded on the fly and also has support for reading the underlying bytes which is perfect for the requirements decoding MIME has. - Since flexi-streams uses trivial-gray-streams, it supports READ-SEQUENCE. Taking advantage of this may improve decoding performance significantly in the future. This incurs the following changes: - Naturally we now open given files as binary files in MIME-MESSAGE. Given strings are encoded using STRING-TO-OCTETS and then passed on to a new octet vector method. Instead of MY-STRING-INPUT-STREAM this now uses flexi-streams' WITH-INPUT-FROM-SEQUENCE. - OPEN-FILE-PORTION and OPEN-DECODED-FILE-PORTION need to be merged, since the transfer encoding not only implies an extra decoder stream that needs to be attached after file portion stream, but also imply a certain encoding of the stream itself (mostly binary vs. ASCII). As flexi-streams can change their encoding on the fly this could be untangled again, but it is not strictly necessary. As before, we use the DATA slot of the file portion to create a fresh stream if possible. Instead of strings we now use an vector of octets to match MIME-MESSAGE. The actual portioned stream relies on POSITIONED-FLEXI-INPUT-STREAM, a subclass of the stock FLEXI-INPUT-STREAM class, described below. - POSITIONED-FLEXI-INPUT-STREAM replaces DELIMITED-INPUT-STREAM. It is created using MAKE-POSITIONED-FLEXI-INPUT-STREAM which accepts the same arguments as MAKE-FLEXI-STREAMS and, additionally, :IGNORE-CLOSE. A POSITIONED-FLEXI-INPUT-STREAM works the same as an FLEXI-INPUT-STREAM, but upon creation, the underlying stream is rewinded or forwarded to the argument given by :POSITION using FILE-POSITION. If :IGNORE-CLOSE is T, a call to CLOSE is not forwarded to the underlying stream. Change-Id: I2d48c769bb110ca0b7cf52441bd63c1e1c2ccd04 Reviewed-on: https://cl.tvl.fyi/c/depot/+/8559 Reviewed-by: sterni <sternenseemann@systemli.org> Autosubmit: sterni <sternenseemann@systemli.org> Tested-by: BuildkiteCI
2023-05-03 18:19:41 +02:00
(defmethod mime-message ((msg vector))
(with-input-from-sequence (in msg)
(mime-message in)))
(defmethod mime-message ((msg pathname))
refactor(3p/lisp/mime4cl): use flexi-streams and binary input This refactor is driven by the following (ultimate) aims: - Get rid of as much of the custom stream code in mime4cl which makes less code to maintain in the future. - Lay the groundwork for correct handling of 8bit transfer encoding: The mime4cl we inherited assumes that any MIME message can be decoded completely by the CL implementation (in SBCL's case using latin1) into CHARACTERs. This is not necessarily the case. flexi-streams allows changing how the stream is decoded on the fly and also has support for reading the underlying bytes which is perfect for the requirements decoding MIME has. - Since flexi-streams uses trivial-gray-streams, it supports READ-SEQUENCE. Taking advantage of this may improve decoding performance significantly in the future. This incurs the following changes: - Naturally we now open given files as binary files in MIME-MESSAGE. Given strings are encoded using STRING-TO-OCTETS and then passed on to a new octet vector method. Instead of MY-STRING-INPUT-STREAM this now uses flexi-streams' WITH-INPUT-FROM-SEQUENCE. - OPEN-FILE-PORTION and OPEN-DECODED-FILE-PORTION need to be merged, since the transfer encoding not only implies an extra decoder stream that needs to be attached after file portion stream, but also imply a certain encoding of the stream itself (mostly binary vs. ASCII). As flexi-streams can change their encoding on the fly this could be untangled again, but it is not strictly necessary. As before, we use the DATA slot of the file portion to create a fresh stream if possible. Instead of strings we now use an vector of octets to match MIME-MESSAGE. The actual portioned stream relies on POSITIONED-FLEXI-INPUT-STREAM, a subclass of the stock FLEXI-INPUT-STREAM class, described below. - POSITIONED-FLEXI-INPUT-STREAM replaces DELIMITED-INPUT-STREAM. It is created using MAKE-POSITIONED-FLEXI-INPUT-STREAM which accepts the same arguments as MAKE-FLEXI-STREAMS and, additionally, :IGNORE-CLOSE. A POSITIONED-FLEXI-INPUT-STREAM works the same as an FLEXI-INPUT-STREAM, but upon creation, the underlying stream is rewinded or forwarded to the argument given by :POSITION using FILE-POSITION. If :IGNORE-CLOSE is T, a call to CLOSE is not forwarded to the underlying stream. Change-Id: I2d48c769bb110ca0b7cf52441bd63c1e1c2ccd04 Reviewed-on: https://cl.tvl.fyi/c/depot/+/8559 Reviewed-by: sterni <sternenseemann@systemli.org> Autosubmit: sterni <sternenseemann@systemli.org> Tested-by: BuildkiteCI
2023-05-03 18:19:41 +02:00
(with-open-file (in msg :element-type '(unsigned-byte 8))
(mime-message in)))
(defmethod mime-message ((msg flexi-stream))
(read-mime-message msg))
(defmethod mime-message ((msg stream))
(read-mime-message (make-flexi-stream msg)))
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
(defgeneric mime-part (object)
(:documentation
"Promote object, if necessary, to MIME-PART."))
(defmethod mime-part ((object string))
(make-instance 'mime-text :subtype "plain" :body object))
(defmethod mime-part ((object pathname))
(make-instance 'mime-application
:subtype "octect-stream"
:content-transfer-encoding :base64
:body (read-file object :element-type '(unsigned-byte 8))))
(defmethod mime-part ((object mime-part))
object)
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
(defmethod make-encoded-body-stream ((part mime-bodily-part))
(let ((body (mime-body part)))
(make-instance (case (mime-encoding part)
(:base64
'base64-encoder-input-stream)
(:quoted-printable
'quoted-printable-encoder-input-stream)
(otherwise
'8bit-encoder-input-stream))
:underlying-stream
refactor(mime4cl): replace *-input-adapter-stream with flexi-streams The input adapter streams were input streams yielding either binary or character data that could be constructed from a variable data source. The stream would take care not to destroy the underlying data source (i.e. not close it if it was a stream), so similar to with FILE-PORTIONs, but simpler. Unfortunately, the implementation was quite inefficient: They are ultimately defined in terms of a function that retrieves the next character in the source. This only allows for an implementation of READ-CHAR (and READ-BYTE). Thanks to cl/8559, READ-SEQUENCE can be used on e.g. FILE-PORTION, but this was still negated by a input adapter based on one—then, READ-SEQUENCE would need to fall back on READ-CHAR or READ-BYTE again. Luckily, we can replace BINARY-INPUT-ADAPTER-STREAM and CHARACTER-INPUT-ADAPTER-STREAM with a much simpler abstraction: Instead of extra stream classes, we have a function, MAKE-INPUT-ADAPTER, which returns an appropriate instance of FLEXI-STREAM based on a given source. This way, the need for a distinction between binary and character input adapter is eliminated, since FLEXI-STREAMS supports both binary and character reads (external format is not yet handled, though). Consequently, the :binary keyword argument to MIME-BODY-STREAM can be dropped. flexi-streams provides stream classes for everything except a stream that doesn't close the underlying one. Since we have already implemented this in POSITIONED-FLEXI-INPUT-STREAM, we can split this functionality into a new superclass ADAPTER-FLEXI-INPUT-STREAM. This change also allows addressing the performance regression encountered in cl/8559: It seems that flexi-streams performs worse when we are reading byte by byte or char by char. (After this change mblog is still two times slower than on r/6150.) By eliminating the adapter streams, we can start utilizing READ-SEQUENCE via decoding code that supports it (i.e. qbase64) and bring performance on par with r/6150 again. Surely there are also ways to gain back even more performance which has to be determined using profiling. Buffering more aggressively seems like a sure bet, though. Switching to flexi-streams still seems like a no-brainer, as it allows us to drop a lot of code that was quite hacky (e.g. DELIMITED-INPUT- STREAM) and implements en/decoding handling we did not support before, but would need for improved correctness. Change-Id: Ie2d1f4e42b47512a5660a1ccc0deeec2bff9788d Reviewed-on: https://cl.tvl.fyi/c/depot/+/8581 Autosubmit: sterni <sternenseemann@systemli.org> Reviewed-by: sterni <sternenseemann@systemli.org> Tested-by: BuildkiteCI
2023-05-15 23:36:40 +02:00
(make-input-adapter body))))
(defun choose-boundary (parts &optional default)
(labels ((match-in-parts (boundary parts)
(loop
for p in parts
thereis (typecase p
(mime-multipart
(match-in-parts boundary (mime-parts p)))
(mime-bodily-part
(match-in-body p boundary)))))
(match-in-body (part boundary)
(with-open-stream (in (make-encoded-body-stream part))
(loop
for line = (read-line in nil)
while line
when (string= line boundary)
return t
finally (return nil)))))
(do ((boundary (if default
(format nil "--~A" default)
#1=(format nil "--~{~36R~}"
(loop
for i from 0 below 20
collect (random 36))))
#1#))
((not (match-in-parts boundary parts)) (subseq boundary 2)))))
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;; fall back method
(defmethod mime-part-size ((part mime-part))
(let ((body (mime-body part)))
(typecase body
(pathname
(file-size body))
(string
(length body))
(vector
(length body))
(t nil))))
(defmethod mime-part-size ((part mime-multipart))
(loop
for p in (mime-parts part)
for size = (mime-part-size p)
unless size
return nil
sum size))
(defmethod mime-part-size ((part mime-message))
(mime-part-size (mime-body part)))
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
(defmethod print-mime-part ((part mime-multipart) (out stream))
(case (mime-subtype part)
(:alternative
;; try to choose something simple to print or the first thing
(let ((parts (mime-parts part)))
(print-mime-part (or (find-if #'(lambda (part)
(and (eq (class-of part) (find-class 'mime-text))
(eq (mime-subtype part) :plain)))
parts)
(car parts)) out)))
(otherwise
(dolist (subpart (mime-parts part))
(print-mime-part subpart out)))))
;; This is WRONG. Here we don't use any special character encoding
;; because we don't know which one we should use. Messages written in
;; anything but ASCII will likely be unreadable -wcp11/10/07.
(defmethod print-mime-part ((part mime-text) (out stream))
(let ((body (mime-body part)))
(etypecase body
(string
(write-string body out))
(vector
(loop
for byte across body
do (write-char (code-char byte) out)))
(pathname
(with-open-file (in body)
(loop
for c = (read-char in nil)
while c
do (write-char c out)))))))
(defmethod print-mime-part ((part mime-message) (out stream))
(flet ((hdr (name)
(multiple-value-bind (value tag)
(header name (mime-message-headers part))
(cons tag value))))
(dolist (h (mapcar #'hdr '("from" "subject" "to" "date" "x-march-archive-id")))
(when h
(format out "~&~A: ~A" (car h) (cdr h))))
(format out "~2%")
(print-mime-part (mime-body part) out)))
(defmethod print-mime-part ((part mime-part) (out stream))
(format out "~&[ ~A subtype=~A ~@[description=~S ~]~@[size=~A~] ]~%"
(type-of part) (mime-subtype part) (mime-description part) (mime-part-size part)))
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
(defgeneric find-mime-part-by-path (mime path)
(:documentation
"Return a subpart of MIME identified by PATH, which is a list of
integers. For example '(2 3 1) is the first part of the third of the
second in MIME."))
(defmethod find-mime-part-by-path ((part mime-part) path)
(if (null path)
part
(error "~S doesn't have subparts" part)))
(defmethod find-mime-part-by-path ((part mime-message) path)
(if (null path)
part
(if (= 1 (car path))
(find-mime-part-by-path (mime-body part) (cdr path))
(error "~S may have just one subpart, but part ~D was requested (parts are enumerated base 1)."
part (car path)))))
(defmethod find-mime-part-by-path ((part mime-multipart) path)
(if (null path)
part
(let ((parts (mime-parts part))
(part-number (car path)))
(if (<= 1 part-number (length parts))
(find-mime-part-by-path (nth (1- (car path)) (mime-parts part)) (cdr path))
(error "~S has just ~D subparts, but part ~D was requested (parts are enumerated base 1)."
part (length parts) part-number)))))
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
(defgeneric find-mime-part-by-id (part id)
(:documentation
"Return a subpart of PAR, whose Content-ID is the same as ID, which
is a string."))
(defmethod find-mime-part-by-id ((part mime-part) id)
(when (string= id (mime-id part))
part))
(defmethod find-mime-part-by-id ((part mime-message) id)
(find-mime-part-by-id (mime-body part) id))
(defmethod find-mime-part-by-id ((part mime-multipart) id)
(or (call-next-method)
(some #'(lambda (p)
(find-mime-part-by-id p id))
(mime-parts part))))
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
(defgeneric find-mime-text-part (msg)
(:documentation
"Return message if it is a text message or first text part.
If no suitable text part is found, return NIL."))
(defmethod find-mime-text-part ((part mime-text))
part) ; found our target
(defmethod find-mime-text-part ((msg mime-message))
;; mime-body is either a mime-part or mime-multipart
(find-mime-text-part (mime-body msg)))
(defmethod find-mime-text-part ((parts mime-multipart))
;; multipart messages may have a body, otherwise we
;; search for the first text part
(or (call-next-method)
(find-if #'find-mime-text-part (mime-parts parts))))
(defmethod find-mime-text-part ((part mime-part))
nil) ; default case
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
(defgeneric mime-type-string (mime-part)
(:documentation
"Return the string describing the MIME part."))
(defmethod mime-type-string ((part mime-unknown-part))
(mime-type part))
(defmethod mime-type-string ((part mime-text))
(format nil "text/~A" (mime-subtype part)))
(defmethod mime-type-string ((part mime-image))
(format nil "image/~A" (mime-subtype part)))
(defmethod mime-type-string ((part mime-audio))
(format nil "audio/~A" (mime-subtype part)))
(defmethod mime-type-string ((part mime-video))
(format nil "video/~A" (mime-subtype part)))
(defmethod mime-type-string ((part mime-application))
(format nil "application/~A" (mime-subtype part)))
(defmethod mime-type-string ((part mime-multipart))
(format nil "multipart/~A" (mime-subtype part)))
(defmethod mime-type-string ((part mime-message))
(format nil "message/~A" (mime-subtype part)))
(defmethod mime-type-string ((part mime-unknown-part))
(mime-type part))
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
(defgeneric map-parts (function mime-part)
(:documentation
"Recursively map FUNCTION to MIME-PART or its components."))
;; Here we wrongly assume that we'll never want to replace messages
;; and multiparts altogether. If you need to do so you have to write
;; your own mapping functions.
(defmethod map-parts ((function function) (part mime-part))
(funcall function part))
(defmethod map-parts ((function function) (part mime-message))
(setf (mime-body part) (map-parts function (mime-body part)))
part)
(defmethod map-parts ((function function) (part mime-multipart))
(setf (mime-parts part) (mapcar #'(lambda (p)
(map-parts function p))
(mime-parts part)))
part)
;; apply-on-parts is like map-parts but doesn't modify the parts (at least
;; not implicitly)
(defgeneric apply-on-parts (function part))
(defmethod apply-on-parts ((function function) (part mime-part))
(funcall function part))
(defmethod apply-on-parts ((function function) (part mime-multipart))
(dolist (p (mime-parts part))
(apply-on-parts function p)))
(defmethod apply-on-parts ((function function) (part mime-message))
(apply-on-parts function (mime-body part)))
(defmacro do-parts ((var mime-part) &body body)
`(apply-on-parts #'(lambda (,var) ,@body) ,mime-part))