zzkt/osc
1
0
Fork 0
Code Issues 7 Pull requests Projects Releases Activity Actions

Merge pull request #1 from erikronstrom/master

Browse source 
Encoding and decoding blobs now follows the spec with regards to size
This commit is contained in:
nik gaffney 2015-08-26 11:18:09 +02:00
commit 815ed4683f
1 changed files with 79 additions and 28 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

107
osc.lisp
View file

@ -96,6 +96,7 @@
f => #(102) => float
s => #(115) => string
b => #(98) => blob
h => #(104) => int64
and considers non int/float/string data to be a blob."
(let ((lump (make-array 0 :adjustable t
@ -106,7 +107,7 @@
(write-to-vector #\,)
(dolist (x data)
(typecase x
(integer (write-to-vector #\i))
(integer (if (>= x 4294967296) (write-to-vector #\h) (write-to-vector #\i)))
(float (write-to-vector #\f))
(simple-string (write-to-vector #\s))
(t (write-to-vector #\b)))))
@ -120,8 +121,8 @@
`(setf lump (cat lump (,f x)))))
(dolist (x data)
(typecase x
(integer (enc encode-int32))
(float (enc encode-float32))
(integer (if (>= x 4294967296) (enc encode-int64) (enc encode-int32)))
(float (enc encode-float32))
(simple-string (enc encode-string))
(t (enc encode-blob))))
lump)))
@ -174,37 +175,43 @@
i => #(105) => int32
f => #(102) => float
s => #(115) => string
b => #(98) => blob"
b => #(98) => blob
h => #(104) => int64"
(let ((div (position 0 data)))
(let ((tags (subseq data 1 div))
(acc (subseq data (padded-length div)))
(result '()))
(acc (subseq data (padded-length div)))
(result '()))
(map 'vector
#'(lambda (x)
(cond
((eq x (char-code #\i))
(push (decode-int32 (subseq acc 0 4))
result)
(setf acc (subseq acc 4)))
((eq x (char-code #\f))
(push (decode-float32 (subseq acc 0 4))
result)
(setf acc (subseq acc 4)))
((eq x (char-code #\s))
(let ((pointer (padded-length (position 0 acc))))
(push (decode-string
(subseq acc 0 pointer))
result)
(setf acc (subseq acc pointer))))
((eq x (char-code #\b))
(let* ((size (decode-int32 (subseq acc 0 4)))
(end (padded-length (+ 4 size))))
#'(lambda (x)
(cond
((eq x (char-code #\i))
(push (decode-int32 (subseq acc 0 4))
result)
(setf acc (subseq acc 4)))
((eq x (char-code #\h))
(push (decode-uint64 (subseq acc 0 8))
result)
(setf acc (subseq acc 8)))
((eq x (char-code #\f))
(push (decode-float32 (subseq acc 0 4))
result)
(setf acc (subseq acc 4)))
((eq x (char-code #\s))
(let ((pointer (padded-length (position 0 acc))))
(push (decode-string
(subseq acc 0 pointer))
result)
(setf acc (subseq acc pointer))))
((eq x (char-code #\b))
(let* ((size (decode-int32 (subseq acc 0 4)))
(bl (+ 4 size))
(end (+ bl (mod (- 4 bl) 4)))) ; NOTE: cannot use (padded-length bl), as it is not the same algorithm. Blobs of 4, 8, 12 etc bytes should not be padded!
(push (decode-blob (subseq acc 0 end))
result)
(setf acc (subseq acc end))))
(t (error "unrecognised typetag"))))
tags)
(t (error "unrecognised typetag ~a" x))))
tags)
(nreverse result))))
@ -316,6 +323,50 @@
(set-byte 3))
buf))
(defun encode-int64 (i)
"convert an integer into a sequence of 8 bytes in network byte order."
(declare (type integer i))
(let ((buf (make-sequence
'(vector (unsigned-byte 8)) 8)))
(macrolet ((set-byte (n)
`(setf (elt buf ,n)
(logand #xff (ash i ,(* 8 (- n 7)))))))
(set-byte 0)
(set-byte 1)
(set-byte 2)
(set-byte 3)
(set-byte 4)
(set-byte 5)
(set-byte 6)
(set-byte 7))
buf))
(defun decode-uint64 (s)
"8 byte -> 64 bit unsigned int"
(let ((i (+ (ash (elt s 0) 56)
(ash (elt s 1) 48)
(ash (elt s 2) 40)
(ash (elt s 3) 32)
(ash (elt s 4) 24)
(ash (elt s 5) 16)
(ash (elt s 6) 8)
(elt s 7))))
i))
(defun decode-int64 (s)
"8 byte -> 64 bit int -> two's compliment (in network byte order)"
(let ((i (+ (ash (elt s 0) 56)
(ash (elt s 1) 48)
(ash (elt s 2) 40)
(ash (elt s 3) 32)
(ash (elt s 4) 24)
(ash (elt s 5) 16)
(ash (elt s 6) 8)
(elt s 7))))
(if (>= i #x7fffffffffffffff)
(- 0 (- #x10000000000000000 i))
i)))
;; osc-strings are unsigned bytes, padded to a 4 byte boundary
(defun decode-string (data)
@ -340,7 +391,7 @@
"encodes a blob from a given vector"
(let ((bl (length blob)))
(cat (encode-int32 bl) blob
(pad (padding-length bl)))))
(pad (mod (- 4 bl) 4))))) ; NOTE: cannot use (padding-length bl), as it is not the same algorithm. Blobs of 4, 8, 12 etc bytes should not be padded!
;; utility functions for osc-string/padding slonking