extmod/moddeflate.c: Add deflate.DeflateIO providing compression and decompression. by jimmo · Pull Request #11905 · micropython/micropython

jimmo · 2023-06-30T03:41:44Z

This replaces #11879 -- because there was no way to provide a gzip.GzipFile that exactly matched CPython's (e.g. because we need the wbits parameter), instead we will put this functionality in a separate module/class. I have updated micropython/micropython-lib#694 to also provide a CPython-compatible gzip module.

Updated copy of the #11879 description below:

Summary: Adds gzip/zlib/raw-deflate compression support. (See issues #7228, #5590, and earlier PRs #8195, #5613).

Historically MicroPython provided a CPython-compatible implementation of zlib.decompress as well as a MicroPython-specific zlib.DecompIO which is similar to CPython's zlib.decompressobj except more useful as a stream wrapper, for example to decompress a file or socket, because you pull (read) decompressed data out of it, rather than pushing (write) compressed data into it.

However, CPython also provides gzip.GzipFile which works exactly like MicroPython's DecompIO, the only limitation is that it only supports gzip-format data (whereas DecompIO supported the same wbits argument that zlib.decompress supported).

This PR removes the zlib module, and replaces it with a deflate module providing a stream wrapper class DeflateIO that is essentially DeflateIO with compression support, but also suitable as a building block for implementing gzip.GzipFile. Because it no longer part of the zlib module, it also has a simpler API for specifying format and window size.

For backwards compatibility, I have written a pure-Python implementation of the functionality of the former zlib module (including now zlib.compress), as well as the gzip module (including compress, decompress, open, and GzipFile), which will be published to micropython-lib (micropython/micropython-lib#694) and can be optionally installed via mip (or frozen). Even though they are very small (~450 bytes each), they are not frozen by default, with the justification:

(De)compression isn't a feature that everyone uses.
Even in applications that use (de)compression, DeflateIO provides a much more memory-efficient way to work with reading/writing compressed data to/from a stream (which is the typical use case).
GzipFile not supporting the wbits argument makes it awkward to use in a general-purpose way.

This PR includes the following changes:

Add a shared mp_stream___exit___obj that avoids each file/stream-like object needing to implement this themselves. (This is a code size optimisation saving ~200 bytes)
Add a memory-efficient implementation of LZ77 compression to lib/uzlib (Written by @dpgeorge ).
General code size optimisations for lib/uzlib. (Saves around 650 bytes)
Code cleanup and simplication for lib/uzlib.
Remove the zlib module.
Add the deflate module.
Add tests for deflate.DeflateIO.
Update docs for zlib and add docs for gzip to describe the micropython-lib versions.
Add docs for deflate.

Overall this PR adds +1156 bytes to PYBV11 (i.e. adding compression support). This is made up of -2840 (remove zlib), +2928 (add decompression-only DeflateIO), +1068 (add compression support to DeflateIO).

Currently compression support (via MICROPY_PY_DEFLATE_COMPRESS) is enabled at the same level ("extra features") as decompression (via MICROPY_PY_DEFLATE), but I think we should consider making it either "full features" or "everything". I do think we should enable it on Unix/Windows though (but really I think that means we should make it "full" and move Unix/Windows to "full").

Compared to #8195 & #5613 which are quite similar (cc @harbaum, @andrewleech) this PR adds support for streaming as well as non-gzip (i.e. zlib/raw) formats and a code size reduction as well as reduced memory usage in the compressor (see #11879 (comment) for more notes from @dpgeorge).

This work was funded through GitHub Sponsors, by specific request from a supporter.

github-actions · 2023-06-30T03:52:04Z

Code size report:

   bare-arm:    +0 +0.000% 
minimal x86:    +0 +0.000% 
   unix x64:  -384 -0.048% standard[incl -32(data)]
      stm32:   -96 -0.025% PYBV10
     mimxrt:  +988 +0.276% TEENSY40
        rp2:   -88 -0.027% PICO
       samd:  +992 +0.382% ADAFRUIT_ITSYBITSY_M4_EXPRESS

jimmo · 2023-06-30T05:42:21Z

Some quick performance tests:
On PYBV11, decompression is unchanged 781ms (from 783ms) to decompress a 7.4kiB gzip file (into 14kiB) 25 times (using DeflateIO vs DecompIO from BytesIO using readinto to avoid allocations).
On unix, decompression on MicroPython (DeflateIO) is roughly 7 times slower than in Python (gzip.GzipFile). 465ms vs 65ms for 1000 iterations of the same test.

For compression (on the unix port), MicroPython (wbits=8) is about 8x slower than Python (wbits=15) to compress the same file. (4x if you use wbits=6).

In terms of bytes/second:
On Unix, CPython compresses at about 45MiB/s, whereas MicroPython at about 2.8MiB/s. CPython decompresses at about 110MiB/s, whereas MicroPython is about 18MiB/s.
On PYBV11, MicroPython compresses at 63kiB/s and decompresses at 245kiB/s.

MicroPython's decompression performance is impacted slightly by not buffering its input as it reads byte at a time from the underlying stream. On PYBV11 it can be increased to 280kiB/s by adding a 4 byte buffer. Diminishing marginal returns for a bigger buffer. Similar relative gain on Unix (22MiB/s).

MicroPython's compression performance is limited by the linear searching in the history window (it's not meant to be fast, it's meant to use less memory).

andrewleech · 2023-06-30T17:31:32Z

I'm really excited to see this coming together, thanks @jimmo !

codecov · 2023-07-03T07:17:15Z

Codecov Report

Merging #11905 (ea1a5e4) into master (14c2b64) will increase coverage by 0.00%.
The diff coverage is 98.33%.

@@           Coverage Diff            @@
##           master   #11905    +/-   ##
========================================
  Coverage   98.41%   98.41%            
========================================
  Files         155      156     +1     
  Lines       20564    20696   +132     
========================================
+ Hits        20238    20368   +130     
- Misses        326      328     +2

Impacted Files	Coverage Δ
extmod/modbinascii.c	`100.00% <ø> (ø)`
extmod/vfs_fat_file.c	`95.50% <ø> (-0.15%)`	⬇️
extmod/vfs_lfsx_file.c	`94.39% <ø> (-0.16%)`	⬇️
extmod/vfs_posix_file.c	`91.95% <ø> (-0.27%)`	⬇️
lib/uzlib/adler32.c	`100.00% <ø> (ø)`
lib/uzlib/crc32.c	`100.00% <ø> (ø)`
py/builtinimport.c	`100.00% <ø> (ø)`
py/objstringio.c	`92.38% <ø> (-0.22%)`	⬇️
py/stream.h	`100.00% <ø> (ø)`
lib/uzlib/tinflate.c	`93.75% <93.75%> (-0.94%)`	⬇️
... and 6 more

... and 1 file with indirect coverage changes

dpgeorge · 2023-07-11T01:13:19Z

Re performance: compiling relevant lib/uzlib files with -O2 will probably go a long way to improve performance.

dpgeorge · 2023-07-13T06:21:35Z

+        // -----CMF------  ----------FLG---------------
+        // CINFO(5) CM(3)  FLEVEL(2) FDICT(1) FCHECK(5)
+        uint8_t buf[] = { 0x08, 0x80 };     // CM=2 (deflate), FLEVEL=2 (default), FDICT=0 (no dictionary)
+        buf[0] |= MAX(self->window_bits - 8, 1) << 4;     // base-2 logarithm of the LZ77 window size, minus eight.


Should this use wbits? In case self->window_bits==0?

Good catch. Fixed.

dpgeorge · 2023-07-13T06:22:09Z

+    static const mp_arg_t allowed_args[] = {
+        { MP_QSTR_stream, MP_ARG_REQUIRED | MP_ARG_OBJ, {.u_obj = MP_OBJ_NULL} },
+        { MP_QSTR_format, MP_ARG_INT, {.u_int = DEFLATEIO_FORMAT_NONE} },
+        { MP_QSTR_wbits, MP_ARG_INT, {.u_int = 0} },


Perhaps the default here should be DEFLATEIO_DEFAULT_WBITS, and then the code that checks for wbits==0 elsewhere in this file can be removed?

Edit: or does wbits=0 mean something special?

Edit: or does wbits=0 mean something special?

Yes.

dpgeorge · 2023-07-13T06:22:53Z

+        { MP_QSTR_close, MP_ARG_KW_ONLY | MP_ARG_OBJ, {.u_rom_obj = MP_ROM_FALSE} },
+    };
+    mp_arg_val_t args[MP_ARRAY_SIZE(allowed_args)];
+    mp_arg_parse_all_kw_array(n_args, n_kw, all_args, MP_ARRAY_SIZE(allowed_args), allowed_args, args);


Possible make these positional-only args, to reduce code size and allow a natmod version.

dpgeorge · 2023-07-13T06:56:54Z

I added a new test for stream errors.

dpgeorge · 2023-07-14T05:57:40Z

+STATIC mp_uint_t deflateio_read(mp_obj_t o_in, void *buf, mp_uint_t size, int *errcode) {
+    mp_obj_deflateio_t *self = MP_OBJ_TO_PTR(o_in);
+
+    if (!deflateio_init_read(self) || self->stream == MP_OBJ_NULL) {


These checks should be swapped.

jimmo · 2023-07-14T07:26:26Z

Added deflate as a dynamic native module example (to replace the old zlib version).

jimmo · 2023-07-14T07:32:57Z

Made compression enabled at the "full" level by default, and updated docs to match.

jimmo · 2023-07-21T03:34:00Z

Updated to address comments. Added AUTO as a module constant.

This will be replaced with a new deflate module providing the same functionality, with an optional frozen Python wrapper providing a replacement zlib module. binascii.crc32 is temporarily disabled. This work was funded through GitHub Sponsors. Signed-off-by: Jim Mussared <[email protected]>

There are enough places that implement __exit__ by forwarding directly to mp_stream_close that this saves code size. For the cases where __exit__ is a no-op, additionally make their MP_STREAM_CLOSE ioctl handled as a no-op. This work was funded through GitHub Sponsors. Signed-off-by: Jim Mussared <[email protected]>

The compression algorithm implemented in this commit uses much less memory compared to the standard way of implementing it using a hash table and large look-back window. In particular the algorithm here doesn't allocate hash table to store indices into the history of the previously seen text. Instead it simply does a brute-force-search of the history text to find a match for the compressor. This is slower (linear search vs hash table lookup) but with a small enough history (eg 512 bytes) it's not that slow. And a small history does not impact the compression too much. To give some more concrete numbers comparing memory use between the approaches: - Standard approach: inplace compression, all text to compress must be in RAM (or at least memory addressable), and then an additional 16k bytes RAM of hash table pointers, pointing into the text - The approach in this commit: streaming compression, only a limited amount of previous text must be in RAM (user selectable, defaults to 512 bytes). To compress, say, 1k of data, the standard approach requires all that data to be in RAM, plus an additional 16k of RAM for the hash table pointers. With this commit, you only need the 1k of data in RAM. Or if it's streaming from a file (or elsewhere), you could get away with only 256 bytes of RAM for the sliding history and still get very decent compression. In summary: because compression takes such a large amount of RAM (in the standard algorithm) and it's not really suitable for microcontrollers, the approach taken in this commit is to minimise RAM usage as much as possible, and still have acceptable performance (speed and compression ratio). Signed-off-by: Damien George <[email protected]>

Because we only use the streaming source, this is just extra code size. Saves 64 bytes on PYBV11. Signed-off-by: Jim Mussared <[email protected]>

This commit makes the following changes: - Replace 256-byte reverse-bits-in-byte lookup table with computation. - Replace length and distance code lookup tables with computation. - Remove comp_disabled check (it's unused). - Make the dest_write_cb take the data pointer directly, rather than the Outbuf. Saves 500 bytes on PYBV11. Signed-off-by: Jim Mussared <[email protected]>

This library used a mix of "tinf" and "uzlib" to refer to itself. Remove all use of "tinf" in the public API. This work was funded through GitHub Sponsors. Signed-off-by: Jim Mussared <[email protected]>

This supports `wbits` values between +40 to +47. This work was funded through GitHub Sponsors. Signed-off-by: Jim Mussared <[email protected]>

Saves 68 bytes on PYBV11. This work was funded through GitHub Sponsors. Signed-off-by: Jim Mussared <[email protected]>

Collapsing the two adjacent calls to outbits saves 32 bytes. Bringing defl_static.c into lz77.c allows better inlining, saves 24 bytes. Merge the Outbuf/uzlib_lz77_state_t structs, a minor simplification that doesn't change code size. This work was funded through GitHub Sponsors. Signed-off-by: Jim Mussared <[email protected]>

For better abstraction for users of this API. Signed-off-by: Jim Mussared <[email protected]>

This provides similar functionality to the former zlib.DecompIO and especially CPython's gzip.GzipFile for both compression and decompression. This class can be used directly, and also can be used from Python to implement (via io.BytesIO) zlib.decompress and zlib.compress, as well as gzip.GzipFile. Enable/disable this on all ports/boards that zlib was previously configured for. This work was funded through GitHub Sponsors. Signed-off-by: Jim Mussared <[email protected]>

Signed-off-by: Jim Mussared <[email protected]>

Also update zlib & gzip docs to describe the micropython-lib modules. This work was funded through GitHub Sponsors. Signed-off-by: Jim Mussared <[email protected]>

This replaces the previous zlib version. This work was funded through GitHub Sponsors. Signed-off-by: Jim Mussared <[email protected]>

dpgeorge · 2023-07-21T13:31:27Z

Now merged! Thanks @jimmo for all the hard work on this.

This was referenced Jun 30, 2023

extmod/modgzip.c: Add gzip.GzipFile providing compression and decompression. #11879

Closed

python-stdlib/zlib: Add zlib module. micropython/micropython-lib#694

Merged

jimmo force-pushed the extmod-deflate branch from 6947a71 to f8e9ef2 Compare June 30, 2023 05:42

jimmo force-pushed the extmod-deflate branch from f8e9ef2 to 75d623e Compare June 30, 2023 05:49

jimmo force-pushed the extmod-deflate branch from 75d623e to ab0d0be Compare July 3, 2023 07:00

dpgeorge added the extmod Relates to extmod/ directory in source label Jul 11, 2023

dpgeorge added this to the release-1.21.0 milestone Jul 11, 2023

dpgeorge reviewed Jul 13, 2023

View reviewed changes

Comment thread extmod/moddeflate.c

dpgeorge reviewed Jul 14, 2023

View reviewed changes

jimmo force-pushed the extmod-deflate branch 2 times, most recently from ea972d4 to b67d30e Compare July 14, 2023 07:26

jimmo force-pushed the extmod-deflate branch from b67d30e to 1b1d3c9 Compare July 14, 2023 07:32

jimmo force-pushed the extmod-deflate branch 3 times, most recently from 680da47 to a78e5be Compare July 14, 2023 08:04

dpgeorge reviewed Jul 21, 2023

View reviewed changes

Comment thread docs/library/deflate.rst Outdated

Comment thread extmod/moddeflate.c Outdated

Comment thread ports/nrf/modules/os/microbitfs.c Outdated

Comment thread lib/uzlib/uzlib.h

jimmo force-pushed the extmod-deflate branch from a78e5be to 8cafd33 Compare July 21, 2023 03:33

jimmo and others added 3 commits July 21, 2023 18:48

jimmo added 2 commits July 21, 2023 18:57

lib/uzlib/lz77: Always use separate history buffer.

82db992

Because we only use the streaming source, this is just extra code size. Saves 64 bytes on PYBV11. Signed-off-by: Jim Mussared <[email protected]>

dpgeorge force-pushed the extmod-deflate branch from 8cafd33 to b7e4107 Compare July 21, 2023 08:59

jimmo added 9 commits July 21, 2023 19:29

lib/uzlib: Clean up tinf -> uzlib rename.

c2b8e6e

This library used a mix of "tinf" and "uzlib" to refer to itself. Remove all use of "tinf" in the public API. This work was funded through GitHub Sponsors. Signed-off-by: Jim Mussared <[email protected]>

lib/uzlib: Combine zlib/gzip header parsing to allow auto-detect.

d75a3cd

This supports `wbits` values between +40 to +47. This work was funded through GitHub Sponsors. Signed-off-by: Jim Mussared <[email protected]>

lib/uzlib/tinflate: Implement more compact lookup tables.

ef5061f

Saves 68 bytes on PYBV11. This work was funded through GitHub Sponsors. Signed-off-by: Jim Mussared <[email protected]>

lib/uzlib: Add a source_read_data var to pass to source_read_cb.

e6c290c

For better abstraction for users of this API. Signed-off-by: Jim Mussared <[email protected]>

tests/extmod: Add deflate.DeflateIO tests.

8b315ef

Signed-off-by: Jim Mussared <[email protected]>

docs/library/deflate: Add docs for deflate.DeflateIO.

b804443

Also update zlib & gzip docs to describe the micropython-lib modules. This work was funded through GitHub Sponsors. Signed-off-by: Jim Mussared <[email protected]>

examples/natmod/deflate: Add deflate as a dynamic native module.

ea1a5e4

This replaces the previous zlib version. This work was funded through GitHub Sponsors. Signed-off-by: Jim Mussared <[email protected]>

dpgeorge force-pushed the extmod-deflate branch from b7e4107 to ea1a5e4 Compare July 21, 2023 09:42

dpgeorge merged commit ea1a5e4 into micropython:master Jul 21, 2023

This was referenced Jul 21, 2023

extmod/uzlib: Add gzip compression support. #5613

Closed

extmod/uzlib: Add gzip compression support. (update to #5613 and #6972) #8195

Closed

Uh oh!

Conversation

jimmo commented Jun 30, 2023

Uh oh!

github-actions Bot commented Jun 30, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

jimmo commented Jun 30, 2023

Uh oh!

andrewleech commented Jun 30, 2023

Uh oh!

codecov Bot commented Jul 3, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

dpgeorge commented Jul 11, 2023

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

dpgeorge commented Jul 13, 2023

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jimmo commented Jul 14, 2023

Uh oh!

jimmo commented Jul 14, 2023

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

jimmo commented Jul 21, 2023

Uh oh!

dpgeorge commented Jul 21, 2023

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

github-actions Bot commented Jun 30, 2023 •

edited

Loading

codecov Bot commented Jul 3, 2023 •

edited

Loading