gnustep/libs-base
0
0
Fork 0
mirror of https://github.com/gnustep/libs-base.git synced 2025-04-22 08:26:27 +00:00
Code Issues Projects Releases Packages Wiki Activity

chunked encoding optimisations

Browse source 
git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/base/trunk@35331 72102866-910b-0410-8b05-ffd578937521
This commit is contained in:
Richard Frith-MacDonald 2012-07-31 06:53:02 +00:00
parent 8a605b360a
commit c58ccdb72c
2 changed files with 37 additions and 24 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

9
ChangeLog
View file

@ -1,3 +1,12 @@
2012-07-31 Richard Frith-Macdonald <rfm@gnu.org>
* Source/Additions/GSMime.m: Optimisations for chunked encoding.
a. Don't resize the buffer for each chunk ... that's very inefficient
if the other end sends tiny chunks. Instead, make the buffer big
enough to handle all the available data, then shrink back to the
decoded size.
b. Move method call out of the loop handling individual bytes.
2012-07-30 Richard Frith-Macdonald <rfm@gnu.org>
* Source/NSData.m: ([setLength:]) increase size of buffer in chunks

52
Source/Additions/GSMime.m
View file

@ -1071,31 +1071,36 @@ wordData(NSString *word)
{
GSMimeChunkedDecoderContext *ctxt;
NSUInteger size = [dData length];
unsigned char *beg;
unsigned char *dst;
const char *src;
const char *end;
const char *footers;
unsigned char *buf;
unsigned char *beg;
unsigned char *dst;
const char *src;
const char *end;
const char *footers;
BOOL atEnd;
ctxt = (GSMimeChunkedDecoderContext*)con;
/*
* Get pointers into source data buffer.
/* Get pointers into source data buffer.
*/
src = (const char *)[sData bytes];
footers = src;
src += aRange.location;
end = src + aRange.length;
beg = 0;
/*
* Make sure buffer is big enough, and set up output pointers.
/* Make sure buffer is big enough, and set up output pointers.
* ctxt->size is the amount of data laready decoded, so if we
* add the size of the chunked source data we are guaranteed
* to have enough space for the unchunked data.
*/
[dData setLength: ctxt->size];
dst = (unsigned char*)[dData mutableBytes];
dst = dst + size;
[dData setLength: ctxt->size + [sData length]];
buf = (unsigned char*)[dData mutableBytes];
dst = buf + size;
beg = dst;
while ([ctxt atEnd] == NO && src < end)
atEnd = [ctxt atEnd];
while (NO == atEnd && src < end)
{
switch (ctxt->state)
{
@ -1139,16 +1144,13 @@ wordData(NSString *word)
}
}
ctxt->pos = val;
/*
* If we have read a chunk already, make sure that our
* destination size is updated correctly before growing
* the buffer for another chunk.
/* If we have read a chunk already, make sure that our
* destination size stored in context is updated before
* resetting the destination pointer for a new chunk.
*/
size += (dst - beg);
ctxt->size = size + val;
[dData setLength: ctxt->size];
dst = (unsigned char*)[dData mutableBytes];
dst += size;
beg = dst;
}
break;
@ -1208,7 +1210,8 @@ wordData(NSString *word)
}
else if (*src == '\n')
{
[ctxt setAtEnd: YES];
atEnd = YES;
[ctxt setAtEnd: atEnd];
}
else
{
@ -1228,7 +1231,7 @@ wordData(NSString *word)
if (ctxt->state == ChunkFoot || ctxt->state == ChunkFootA)
{
[ctxt->data appendBytes: footers length: src - footers];
if ([ctxt atEnd] == YES)
if (YES == atEnd)
{
NSMutableData *old;
@ -1270,8 +1273,9 @@ wordData(NSString *word)
flags.inBody = 1;
}
}
/*
* Correct size of output buffer.
/** Correct the size of the output buffer (shrink back from the
* original allocation to the actual unchunked size).
*/
[dData setLength: size + dst - beg];
}