ZLib module

Looks like it still available from this link here: http://riscos.com/support/developers/riscos6/programmer/index.html

It’s still on the site, they just moved everything around without setting up any redirects.

Thanks. It hasn’t helped figure out how to decompress a stream extracted from a ZIP file through! I’ve coded the inflate routine based on ZLib’s C unzip example, which is along the lines of:

DIM stream 56
...traul through <zipfile> and find the position of the stream to extract, its compression method and compressed/uncompressed sizes
SYS "ZLib_InflateInit2", stream, -1, "1.1.4", 56 : REM initialise stream with no GZIP header
SYS "ZLib_GZOpen", "<zipfile>", "rb" TO F%
SYS "ZLib_GZSeek", F%, position_of_stream, 0 : REM seek to the start of the compressed data
SYS "ZLib_GZRead", F%, data%, uncompressed_size% : REM inflate the stream
SYS "ZLib_GZClose", F%

There’s several issues with this code:

1. ZLib_InflateInit2 may not be interpreting a windowBits parameter of <0 as meaning “there’s no GZIP header”
2. There’s no SWI to seek to a physical position in the file opened with ZLib_GZOpen, although ZLib_GZSeek appears in this instance to be seeking to the correct position
3. I can’t see how you specify the compression method used, so ZLib_GZRead deflates it correctly

That mini unzip code doesn’t use the gzip methods to decompress the data. If you look at unzReadCurrentFile you’ll see that it uses the standard inflate function (i.e. ZLib_Inflate SWI) to decompress the data, using buffers that are filled/emptied manually (ZREAD64 & ZSEEK64 macros used to fill the input buffer, which I expect are just wrappers around the C file I/O functions).

I can’t see how you specify the compression method used, so ZLib_GZRead deflates it correctly

That’s because the GZip functions are only good for reading/writing GZip files, and GZip files have their own header which indicates the compression method (which according to RFC 1952, will only be “deflate” (i.e. ZLib)).

Zip != GZip.

it uses the standard inflate function (i.e. ZLib_Inflate SWI) to decompress the data

I’ve already tried that method, with the following code:

SYS "ZLib_InflateInit2", stream, -1, "1.1.4", 56
stream!0=compressed_data_address%
stream!4=compressed_size%
stream!12=uncompressed_data_address%
stream!16=uncompressed_size%
SYS "ZLib_Inflate", stream, 4
SYS "ZLib_InflateEnd", stream

It returns “The stream structure was inconsistent” for ZLib_Inflate

Zip != GZip

Are you saying your ZLib Module can’t be used to inflate streams contained within a ZIP file?

EDIT: Working code example. After scanning the ZIP file and extracting the compressed files stream to memory:

DIM streamCB% 56
FOR A%=0 TO 52 STEP 4:stream!A%=0:NEXT
SYS "ZLib_InflateInit2", streamCB%, -15, "1.1.4", 56
stream!0=compressed_data_address%
stream!4=compressed_size%
stream!12=uncompressed_data_address%
stream!16=uncompressed_size%
SYS "ZLib_Inflate", streamCB%, 4
SYS "ZLib_InflateEnd", streamCB%

Zip != GZip.

Most zips are type 8 “Deflate” which is supposed to be compatible with the GZip compression so long as you manage the file headers yourself.
Any other type of Zip (there are many, but they are rare) can’t be used with ZLib.

Are you saying your ZLib Module can’t be used to inflate streams contained within a ZIP file?

Can you get access to raw deflate? There are three types of header that ZLib is supposed to support: None (you’ll want to use this with Zip files and manage the header and blocks yourself), ZLib headers (usually used by PNG), and GZip headers (used by <drumroll> GZip!).

It is worth noting a primary difference between GZip and Zip is that GZip typically only compresses one file (a throwover from the compress behaviour). That’s why everything is typically stuffed into a tar file within the Gzip. Zip, on the other hand, has a complex structure. However so long as the compression type is 8, the data – once you have a pointer to it, ought to be decompressible by ZLib in its raw deflate mode (where it won’t look for any sort of headers). This is why you can’t use the GZip commands to deal with Zip files.

Quote (but Textile mucks it up so not in a quote block):

SYS "ZLib_InflateInit2", stream, -1, "1.1.4", 56
steam!0=compressed_data_address%
stream!4=compressed_size%
stream!12=uncompressed_data_address%
stream!16=uncompressed_size%
SYS "ZLib_Inflate", stream, 4
SYS "ZLib_InflateEnd", stream

Shouldn’t “stream” be set up prior to InflateInit2? I see it as Init, [Repeat]Inflate[Until Done], End.
Has the array been set to zero (except those words set otherwise)?
Why flush type 4 to ZLib_Inflate – doesn’t that flag that the decompression is finished?

Shouldn’t “stream” be set up prior to InflateInit2?

It’s a 56 byte block that’s zeroed, I just didn’t include it in the paste.

ZLib can decode the compression in ZIP, but it will only work if you can tell it, the stream is headerless. This is done on the original ZLib by initialising windowBits<0

Why flush type 4 to ZLib_Inflate

I copied example ZLib code line for line.

-1 isn’t a valid value for windowBits. Try -15 instead (i.e. -MAX_WBITS).

https://www.riscosopen.org/viewer/view/mixed/RiscOS/Sources/Lib/zlib/h/zlib?rev=1.2#l759

Try -15 instead (i.e. -MAX_WBITS).

Yay, it now works. Would never have guessed that as the documentation just says <0