python-pillow/Pillow
1
1
Fork 0
mirror of https://github.com/python-pillow/Pillow.git synced 2024-12-29 11:26:17 +03:00
Code Issues Packages Projects Releases Wiki Activity
6,405 Commits 48 Branches 91 Tags 380 MiB
67459795db
Commit Graph

517 Commits

Author SHA1 Message Date
homm
1ba4e9e5ba remove end of file banner from all files 2016-07-10 14:11:28 +03:00
homm
66715246c1 remove mitchell filter. Has no advantages over bicubic 2016-07-02 01:54:39 +03:00
homm
a67e5453d5 add tests, fix errors ) 2016-07-02 01:54:39 +03:00
homm
c3d702538d add BOX, HAMMING, MITCHELL filters to resample.c 2016-07-02 01:54:39 +03:00
homm
0e9beed76d add constants for filters: BOX, HAMMING, MITCHELL 
rearrange filters everywhere
 2016-07-02 01:54:39 +03:00
Andrew Murray
ebf2121338 Removed trailing whitespace 2016-07-01 22:05:18 +10:00
Christoph Gohlke
525ba37722 Fix msvc9/10 compile error 2016-06-30 14:04:53 -07:00
Andrew Murray
5e5bda95e0 Removed trailing whitespace 2016-06-30 20:12:50 +01:00
Alexander Karpinsky
92660a1a05 Merge pull request #1986 from didrix/fix-1979-p2la 
Fix issue converting P mode to LA
 2016-06-30 17:43:53 +04:00
wiredfool
a5dde79068 Merge pull request #1934 from wiredfool/incremental_removal 
Rewrite of Jpeg2k data handling
 2016-06-30 12:28:10 +01:00
wiredfool
291c00c58a Merge pull request #1954 from uploadcare/skip-resampling-passes 
Do not do unnecessary passes on resizing
 2016-06-30 12:23:07 +01:00
wiredfool
b152d99d66 refactor to different file 2016-06-27 00:58:06 -07:00
wiredfool
857ee63436 cleanup unused variables 2016-06-27 00:58:06 -07:00
wiredfool
96907b9158 Comments 2016-06-27 00:58:06 -07:00
wiredfool
d3b0413780 Initial take for removing the Incremental decoder due to threading issues. J2k tests pass, other tests fail 2016-06-27 00:58:05 -07:00
Diederik Veeze
ca69de192b Assign to unused layers as well, jtbs 2016-06-26 13:23:53 +02:00
Diederik Veeze
12685ef7c1 Convert P to LA without deforming image 2016-06-25 23:14:21 +02:00
wiredfool
bdd0a6a4e4 Merge pull request #1781 from wiredfool/malloc_check 
Integer overflow checks on malloc
 2016-06-21 12:09:19 +01:00
wiredfool
95f464f827 removed redundant check 2016-06-16 01:09:38 -07:00
wiredfool
8aedf8b1d8 added comment closer to malloc 2016-06-16 01:07:28 -07:00
wiredfool
ce57e6adfb We're not actually multiplying out the bytes, only the indexes 2016-06-16 01:06:28 -07:00
wiredfool
d0ae5bc75e added check to prevent arcs > 360 degrees 2016-06-16 01:05:41 -07:00
homm
78fad0f819 Do not make unnecessary passes on resizing 2016-06-15 01:04:02 +03:00
wiredfool
92a13d975d Malloc check merge/rebase 2016-06-08 06:25:46 -07:00
wiredfool
76605634e6 MSVC doesn't define UINT32_MAX 2016-06-08 06:25:46 -07:00
wiredfool
5369d8edd7 Replace SIZE_MAX with type specific _MAX 2016-06-08 06:21:38 -07:00
wiredfool
54a979749c Rework block allocator 2016-06-08 06:21:38 -07:00
wiredfool
768936fa33 overflow check for im->linesize 2016-06-08 06:21:38 -07:00
wiredfool
d48e5cd500 Convert xsize/ysize to ints in function declarations to match all other uses of xsize/ysize 2016-06-08 06:21:38 -07:00
wiredfool
49566b287e added cleanup to free dictionary memory in ZipEncode, fixes old comment about leaking memory from prior to when we had the cleanup mechanisim 2016-06-08 06:21:38 -07:00
wiredfool
4b4ef5f1e2 Malloc check, realloc, #1715 2016-06-08 06:21:38 -07:00
wiredfool
52d60cd096 Malloc check, #1715 2016-06-08 06:21:38 -07:00
wiredfool
b1a190ad78 mixed 8c tabs+spaces -> spaces 2016-06-08 06:21:38 -07:00
homm
3d622d60cf make perspecive transform arguments order more common 2016-06-08 16:17:41 +03:00
homm
2b77b1cec7 make affine transform arguments order more common 2016-06-08 16:17:41 +03:00
homm
4cfc791a5b remove unused filter_data 2016-06-08 16:17:41 +03:00
homm
bc51928171 move ImagingCopyInfo after affine_fixed 2016-06-08 16:17:41 +03:00
homm
9902d2e1c5 move transform dispatcher into libImaging 2016-06-08 16:17:41 +03:00
homm
2e914af758 unused 2016-06-08 16:17:41 +03:00
homm
6be3df2a43 remove internal ImagingTransform from headers 
rename to ImagingGenericTransform
 2016-06-08 16:17:41 +03:00
homm
1f8c2527ba internal rotating functions not used anymore 2016-06-08 16:17:41 +03:00
homm
3fa3b48ae9 ImagingTransposeToNew not used anymore 2016-06-08 16:17:41 +03:00
homm
e5b2cbe4ad undef local defines 
add section cookie for affine_fixed
 2016-06-08 16:17:41 +03:00
homm
a18325afe1 remove WITH_FILTERS 2016-06-08 16:17:41 +03:00
wiredfool
62551a8b49 Merge pull request #1933 from uploadcare/resample-vertical-pass 
Resample horizontal + vertical pass
 2016-06-08 14:14:41 +01:00
homm
12c8cf94f2 replace calloc with remaining values emptying 2016-06-08 03:45:08 +03:00
homm
3b7923c09f add calloc explanation 2016-06-03 12:51:58 +03:00
homm
34afabbf18 return malloc where clearing is not required 2016-05-28 10:47:07 +03:00
homm
4ea0c83450 do not rotate image between passes 2016-05-27 07:24:22 +03:00
homm
c051f1d79d ImagingResampleVertical_32bpc implementation 2016-05-27 07:21:34 +03:00
homm
5cd8c35c82 complete ImagingResampleVertical_8bpc implementation 2016-05-27 07:12:01 +03:00
homm
8180b0f4c3 ResampleVertical first implementation 2016-05-27 06:52:19 +03:00
homm
edc9880b61 accept RGBa and La modes for blur 2016-05-27 03:16:48 +03:00
wiredfool
3657bc10a6 Merge pull request #1889 from rr-/libpngquant 
Add libimagequant support in quantize()
 2016-05-26 21:12:25 +01:00
homm
b56d8f3f1f speedup La -> LA conversion in most cases 2016-05-26 15:22:01 +03:00
homm
d1272b9b8a speedup RGBa -> RGBA conversion in most cases 2016-05-26 15:15:08 +03:00
homm
c4b92d09b7 support for La mode 2016-05-26 15:15:08 +03:00
homm
5ffd9e53bc use calloc and INT_MAX 2016-05-26 02:28:35 +03:00
homm
1c3def1d5a return malloc 2016-05-26 02:07:56 +03:00
homm
eec17c0c7a change xmax meaning to avoid substructions 2016-05-26 02:07:56 +03:00
homm
7722d00daf simpler coefficients calculation in ImagingResampleHorizontal_8bpc 2016-05-26 02:07:56 +03:00
homm
d3749ccc7c possible increase precision and avoid compiler warnings 2016-05-26 02:07:56 +03:00
homm
f44a7f8b11 implement round 
fix windows build
 2016-05-26 02:07:56 +03:00
homm
babaaf9bbe raise on special cases even if imIn->image8 2016-05-26 02:07:56 +03:00
homm
d48324bd21 unroll 8bpc version 2016-05-26 02:07:56 +03:00
homm
126c3151b5 use ImagingPrecompute in ImagingResampleHorizontal_8bpc 2016-05-26 02:07:56 +03:00
homm
f3f03bcd67 unroll 32bpc version 2016-05-26 02:07:56 +03:00
homm
04552b015d use ImagingPrecompute in ImagingResampleHorizontal_32bpc 2016-05-26 02:07:56 +03:00
homm
023f7adcf6 precompute function 2016-05-26 02:07:56 +03:00
homm
7693da9287 separate 8 and 32 bpc implementations 2016-05-26 02:07:56 +03:00
homm
9618ec98b8 test for consistency 2016-05-26 02:07:56 +03:00
homm
e13297b396 make coefficients ints 2016-05-26 02:07:03 +03:00
homm
342e6f213b extra line 2016-05-26 02:07:03 +03:00
homm
3ee407e859 pass struct filter to ImagingResampleHorizontal 2016-05-26 02:07:03 +03:00
homm
0e2a8e4323 intermediate not normalized buffer for coefficients 2016-05-26 02:07:03 +03:00
homm
a5b99dd302 move checks before mallocs to prevent memory leaks 2016-05-26 02:07:03 +03:00
Stephen Arthur
0bc4423487 Move the declaration to the beginning of scope to make vs happy 2016-05-22 19:13:19 -07:00
Stephen Arthur
67a243d82d Removing the one-line variable declare / assign to make appveyor happy 2016-05-22 18:50:42 -07:00
Stephen Arthur
8b572ade81 Found edge cases with custom qtables 
And there was a lingering bug since the previous qtable unsigned char fix
(#1814) since the call to array.array was in another place, the roundtrip was no
longer equivalent.

That was a minor change, but the revised test wouldn't pass because saving an image
with one custom qtable automatically adds a second to it by the call to
jpeg_set_defaults.
With 1 or >2 custom tables, there is extra work we have to do due to that call.
 2016-05-22 17:54:31 -07:00
rr-
0430a4a834 Fix libimagequant facade header 2016-05-21 21:11:54 +02:00
rr-
bc0e896b35 Change libimagequant facade signature 2016-05-21 21:01:57 +02:00
rr-
8b7fb44c6b Make libimagequant visibility detectable 2016-05-06 21:06:54 +02:00
rr-
0b4a089321 Fix C99 errors inside libimagequant quantizer 2016-05-06 20:08:35 +02:00
rr-
cb4ded4929 Minor fixes 2016-05-06 19:29:42 +02:00
rr-
eb354be7c4 Add method=Image.LIBIMAGEQUANT for quantize() 2016-05-05 21:39:16 +02:00
wiredfool
faeb269e30 Revert "Fix for DIB/BMP images" 2016-04-19 16:37:29 +01:00
Hugo
a7c678b1af Merge pull request #1813 from wiredfool/DIB_1293 
Fix for DIB/BMP images
 2016-04-06 19:58:25 +03:00
wiredfool
0e3caec317 Added Transparency mask unpacker for BMP files 2016-04-06 07:40:49 -07:00
wiredfool
14f3b45f46 Merge pull request #1789 from zwhfly/tiff 
Add Support for 2/4 bpp Tiff Grayscale Images
 2016-04-03 12:07:48 -07:00
wiredfool
a1f244343d Fixed j2k integer overflow error on encode - CVE-2016-3076 2016-04-01 05:11:54 -07:00
zwhfly
4d147c16d3 Add Some Unpackers for 2/4 bpp Grayscale Format 
As seen in some tiff image files.
 2016-03-29 20:17:39 +08:00
Andrew Murray
a7c7aa171d Updated redirected URLs 2016-02-11 18:51:36 +11:00
Eric Soroos
07a42363c4 change ifd entry to use uint following libtiff interface 2016-02-04 08:13:17 -08:00
Eric Soroos
717720b051 Fix for buffer overflow in TiffDecode.c CVE-2016-0740 2016-02-04 08:13:11 -08:00
wiredfool
893a40850c FLI overflow error fix and testcase CVE-2016-0775 2016-02-04 08:13:02 -08:00
wiredfool
1723dc2d07 Merge pull request #1714 from wiredfool/resample-overflow 
Overflow in resample.c, with tests
 2016-02-04 07:52:59 -08:00
Ned Williamson
4e0d9b0b97 fix integer overflow in Resample.c 2016-02-04 07:19:54 -08:00
wiredfool
ae453aa18b PCD decoder overruns the shuffle buffer, Fixes #568 2016-02-02 05:46:26 -08:00
Andrew Murray
8f9b3a7372 Changed arcs, chords and pie slices to use floats 2015-12-09 00:28:52 +11:00
Andrew Murray
9a2903bf5b Changed 32 bit to 32-bit and 64 bit to 64-bit 2015-10-15 00:49:03 +11:00
Hugo van Kemenade
00615df3b5 Merge pull request #1487 from radarhere/fix 
Removed trailing whitespace
 2015-10-12 11:06:11 +03:00
Andrew Murray
89fb9965ef Capitalised Python and Pillow 2015-10-11 21:27:43 +11:00
Andrew Murray
6bd7e7f02d Removed trailing whitespace 2015-10-11 21:24:35 +11:00
nu774
2be12dec2b Don't use int to handle type 
Couldn't accept handle values greater than 0x7FFFFFFF, which wasn't enough even on 32bit system, and completely wrong for 64bit.
 2015-10-01 14:54:32 +01:00
Christoph Gohlke
e6733c813a Disable compiler optimizations for topalette and tobilevel functions for all msvc versions 2015-08-28 23:15:56 -07:00
wiredfool
a92d737177 Merge pull request #1278 from wiredfool/winbuild 
Windows Build/CI
 2015-06-18 09:03:38 -07:00
wiredfool
d2e54d7973 Merge pull request #1255 from radarhere/polygon 
Fixed polygon edge drawing
 2015-06-17 18:18:36 -07:00
wiredfool
fd55099ffc merge a year of master into winbuild 2015-06-10 12:49:07 -07:00
Andrew Murray
9e24ae023d Fixed various typos 2015-06-08 01:17:49 +10:00
Andrew Murray
16261beeba Fixed polygon edge drawing 2015-06-05 11:16:33 +10:00
Andrew Murray
3cd7f0e6f2 Removed unused variable in Jpeg2KEncode 2015-04-01 15:14:56 +11:00
wiredfool
4566a4dcbe Merge pull request #1128 from hugovk/noise 
Fix ImagingEffectNoise
 2015-03-08 18:29:44 -07:00
hugovk
f1333617d5 Replace harcoded assumption of RAND_MAX with constant. It caused an infinite loop if RAND_MAX > 32767 2015-03-05 23:14:36 +02:00
hugovk
0b79c6e24a Remove unreachable code (if 0) 2015-03-05 21:07:16 +02:00
wiredfool
b384a7f992 Merge pull request #1088 from bgilbert/opaque-RGBa 
Fast path for opaque pixels in RGBa unpacker
 2015-02-05 22:32:58 -08:00
immerrr
d41fed6732 Enable basic support for 'RGBa' raw encoding/decoding 2015-02-02 13:06:56 +03:00
Benjamin Gilbert
77b020b374 Special-case opaque pixels in RGBa unpacker 
Avoid the expensive multiply and divide when the pixel is opaque.
On my system, this change gives a 5.76x speedup loading an opaque image
with this call:

    PIL.Image.frombuffer('RGBA', (1000, 1000), buf, 'raw', 'RGBa', 0, 1)
 2015-01-24 01:13:58 -05:00
Christoph Gohlke
9b8202203a Use Py_ssize_t instead of ssize_t 2014-12-19 01:30:16 -08:00
Christoph Gohlke
b8fafdef69 Use Py_ssize_t instead of ssize_t 2014-12-19 01:28:21 -08:00
Christoph Gohlke
8fb8d2ec3a Fix msvc compiler error 2014-12-15 14:44:12 -08:00
wiredfool
2d2293ba3b Merge pull request #1028 from homm/resize-constants 
Rename Image.ANTIALIAS to Image.LANCZOS
 2014-12-03 16:14:29 -08:00
wiredfool
05b578e7f4 Fix for building on OSX/Clang. 2014-12-01 20:40:10 -08:00
homm
43d5d06ce4 rename Image.ANTIALIAS to Image.LANCZOS 2014-11-28 01:41:56 +03:00
wiredfool
e16ee15f2c Merge pull request #997 from homm/replace-resize 
Replace resize method
 2014-11-27 10:26:48 -08:00
wiredfool
8a3302ba5d Merge pull request #961 from homm/fast-box-blur 
Merge Fast Gaussian Blur
 2014-11-27 10:20:11 -08:00
homm
107b4a8b07 sharp alpha channel 
resolve sharp endianness issues
 2014-11-22 01:58:31 +03:00
homm
d0c419632e resolve endianness issues 2014-11-21 21:53:13 +03:00
homm
bc0f896a47 rename Antialias and stretch to resample 2014-11-19 14:36:00 +03:00
homm
1a097d2e02 Merge commit 'e0b94d65bedb4a2124228aa579686a3ba2974d11' into replace-resize 2014-11-19 13:45:04 +03:00
homm
6078e6e1a7 this bug actually was in previous implementation too: 
type switch default case was never achieved because special
images is in image8 pointers, not in image32
 2014-11-19 03:15:24 +03:00
wiredfool
9d1fde0fff Use __NO_INLINE__ instead of Py_DEBUG 2014-11-13 21:35:06 -08:00
wiredfool
112a4815f0 Fix compilation on debug versions of python 2014-11-13 15:53:51 -08:00
homm
aba798af25 replace resize implementation 2014-11-09 21:52:16 +03:00
homm
90ee223305 Move ImagingTransposeToNew from Antialias.c to Geometry.c 2014-11-09 03:15:54 +03:00
homm
7a64f7be80 Replace UINT32 assignment with per-channel UINT8 assignment 2014-11-09 03:08:25 +03:00
homm
abc5e11371 two bands case 2014-11-09 03:08:25 +03:00
Eric Soroos
af02f2b29e Fix for endianness issues in stretch 2014-11-09 03:07:19 +03:00
homm
b8a2b5b907 typo. Free mem after ModeError. 2014-11-09 03:05:57 +03:00
homm
1cd6da4a49 fix performance regression on 64 bit GCC 4.8. 2014-11-09 03:05:57 +03:00
homm
42967dd1a6 speedup by unrolling loops 2014-11-09 03:05:57 +03:00
homm
c22af89ef0 optimize memory usage 2014-11-09 03:05:56 +03:00
homm
e9fc720709 faster float to 8bit convertion 2014-11-09 03:05:56 +03:00
homm
a484d28d1f make x indexes int 2014-11-09 03:05:56 +03:00
homm
e276e6a535 move ww into coefficients 2014-11-09 03:05:56 +03:00
homm
01b947c602 Iterate pixels in native order 2014-11-09 03:05:56 +03:00
homm
b77521bd6c Precompute coefficients for all x 2014-11-09 03:05:56 +03:00
homm
40f9f48680 two ImagingStretchHorizaontal pass with two transposes 2014-11-09 03:05:56 +03:00
homm
c8471bcbda Hide stretch implementation detail in Antialias.c 2014-11-09 03:05:56 +03:00
wiredfool
cfbe49f124 Merge pull request #994 from homm/add-transpose 
Add transpose and cache aware rotation
 2014-11-07 13:58:18 -08:00
homm
4cf2f158b6 add test 
update comment
 2014-11-07 11:48:47 +03:00
homm
a51b176644 make ROTATE_CHUNK constant 
add comment
 2014-11-07 04:46:16 +03:00
homm
cfa20f1004 rotate 90 and 270 by chunks 2014-11-07 04:15:18 +03:00
homm
a960d440db less operations in loop 2014-11-07 03:57:00 +03:00
homm
d41bc4fbfc convert tabs into spaces 2014-11-07 03:21:38 +03:00
wiredfool
20be641329 Merge pull request #970 from homm/fix-bicubic-stretch 
Fix bicubic stretch interpolation
 2014-11-05 17:45:47 -08:00
homm
4007a78d53 rename temp to imTransposed 
actualize comments
 2014-10-30 14:48:29 +03:00
homm
c99f05cd6e change arguments order and names in LineBoxBlurXX and HorizontalBoxBlur 2014-10-30 14:32:41 +03:00
homm
36a35e7e27 update box_blur help 
fix code style for `!`
fix spelling
 2014-10-30 14:19:12 +03:00
homm
4323683a63 move ImagingGaussianBlur to BoxBlur.c 2014-10-25 17:17:53 +04:00
homm
274f0a14ba change arguments order and names to match other functions 2014-10-25 17:16:14 +04:00
homm
65173f946b replace gaussian blur with extended box blur implementation 2014-10-25 16:36:41 +04:00
homm
d89c9ab750 Merge branch 'gaussian-refactor' into fast-box-blur 2014-10-25 15:50:57 +04:00
homm
36cbb16bf6 Add tests. Rise for "P" and "1". 2014-10-25 14:23:07 +04:00
homm
acdcdd487f convert tabs to spaces 2014-10-25 14:19:29 +04:00
homm
12ba034dc8 unlock upscale filters 2014-10-25 12:39:03 +04:00
homm
d7d92275f1 Change arguments order and names 2014-10-25 04:28:05 +04:00
homm
57cd15f9b1 Use Transpose from geometry. 2014-10-25 04:27:46 +04:00
homm
316ac5ad57 Merge branch 'add-transpose' into fast-box-blur 2014-10-25 04:18:43 +04:00
homm
0c90ce2504 Add new fast ImagingTranspose function 2014-10-25 04:17:59 +04:00
homm
c2d688c4b6 fix bicubic stretch interpolation 2014-10-24 12:57:32 +04:00
homm
c1eb77a4ed faster commit 2014-10-19 18:34:49 +04:00
homm
f55ea4c04c Do not use lineOut buffer if source and destination are different. 2014-10-19 13:54:18 +04:00
homm
516bd90cf7 support for multiple passes with only two transposes 2014-10-19 13:54:13 +04:00
homm
d91d7b9826 significant speed improvement (about 55%) 2014-10-19 00:01:38 +04:00
homm
22668de6db HorizontalBoxBlur unified 
LineBoxBlur8 & LineBoxBlur32
 2014-10-18 18:37:14 +04:00
homm
7d77db20ad Small optimization: replace division with multiply. 2014-10-18 04:02:10 +04:00
homm
d07c9c2e4e unused 2014-10-16 16:00:29 +04:00
homm
53c8352f39 highlevel api 2014-10-15 04:08:21 +04:00
homm
5861a46a54 check image mode 2014-10-14 23:29:50 +04:00
homm
a27baf90be fix large radius. tests 2014-10-14 03:26:46 +04:00
homm
da84cd7ea6 big radius special case 2014-10-14 02:17:53 +04:00
homm
d3d4ff69eb introduce corners 2014-10-13 23:45:21 +04:00
homm
b1cef839bd fix window calculation 2014-10-13 04:40:44 +04:00
homm
3906ff3c76 fix x bounds 2014-10-13 03:39:35 +04:00
homm
835db5d12f type check 2014-10-12 22:40:36 +04:00
homm
2c723abbca move HorizontalBoxBlur8 down and remove comments 2014-10-12 22:40:36 +04:00
homm
591fc67aae micro optimizations 2014-10-12 22:40:31 +04:00
homm
92635d0624 8bit 2014-10-12 22:26:00 +04:00
homm
e95382eb08 integer implementation for float radius 2014-10-12 20:30:46 +04:00
homm
f3af171dc4 fix round compensation 2014-10-12 20:09:28 +04:00
homm
f711ab3eb6 float radius 2014-10-12 20:01:49 +04:00
homm
23068e49be allow float radius (without implementation) 2014-10-12 19:24:38 +04:00
homm
b7b3c719fa blur in both directions 2014-10-12 17:59:20 +04:00
homm
cc9eff5c68 fast realization with moving window. 
linear speed against any radius
 2014-10-12 17:47:00 +04:00
homm
a6937d84d8 first Horizontal Blur for 32 2014-10-12 17:11:24 +04:00
homm
d49459f5b2 box blur dummy 2014-10-12 16:30:00 +04:00
homm
e38a64b21b round values with initial .5 
this improves performance and readabillity
 2014-10-12 15:13:37 +04:00
homm
bc24685344 do sum calculation in same loop as maskData calculation 2014-10-12 14:21:13 +04:00
homm
51984f4aa8 fix radius == 0 2014-10-12 14:16:09 +04:00
homm
84b9f16bfd add effective_scale to all APIs 2014-10-12 03:29:20 +04:00