2c4c039056
add "unlimited" to jpegload
...
To disable DoS limits for JPEG loading. Adding API on a stable branch is
bad, but this fixes a regression, so I think it's necessary,
unfortunately.
See https://github.com/libvips/libvips/issues/2973
2022-08-02 13:50:09 +01:00
5221df224f
fix ssize_t / ssize_t confusion
2022-05-11 22:28:34 +01:00
460a19b78e
revise arg names
...
use _ everywhere (we have a few uses of - as a separator)
2022-05-07 14:50:05 +01:00
6c03343581
Add comments to `setjmp` invocations ( #2751 )
...
Also, pre-increment `num_warnings` to ensure that we really stop
at 100 warnings.
2022-04-06 12:01:18 +01:00
df26bd1e46
Stop JPEG load after 100 warnings ( #2749 )
...
See: https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=24383
2022-04-05 11:07:29 +02:00
2dc319b1b1
Revert "stop JPEG load after 20 warnings"
...
This reverts commit 89bd46d1c4
2022-04-05 08:34:45 +01:00
1aebd95387
Revert "improve fail on too many warings"
...
This reverts commit ae7e5e7a54
2022-04-05 08:34:20 +01:00
ae7e5e7a54
improve fail on too many warings
...
thanks lovell, see 89bd46d1c4 (commitcomment-70409015)
2022-04-04 15:44:37 +01:00
89bd46d1c4
stop JPEG load after 20 warnings
...
mitigates some DoS attacks somewhat
see https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=24383
2022-04-03 19:21:06 +01:00
0337c09700
Cleanup gettext handling ( #2695 )
...
Use GLib's i18n support instead of copying and pasting that
logic into its own header. This deprecates the vips/intl.h
header in favour of glib/gi18n.h.
2022-02-27 15:27:28 +00:00
260b2ddf92
Fix read from pipe with variable chunk sizes
...
free the header cache only once it's exhausted
2022-01-24 11:27:22 +00:00
0c70f3dc7d
add fail-on : better control over loader error handling ( #2360 )
...
Instead of a simple fail/don't-fail boolean switch, add fail-on, an enum which sets the sensitivity of loaders to errors.
There's a new sensitivity level which tries to detect truncated images, but ignores other types of error.
2021-10-31 14:13:18 +00:00
757d031005
always check the return of vips_image_pipeline()
...
For consistency. Although I don't think it can meaningfully fail.
2021-06-29 16:12:17 +01:00
7a1fb2a627
slight formatting improvement
2020-12-29 13:29:27 +00:00
0982d0efbb
signal error on EOF in jpegload more reliably
...
we were only warning on EOF in the read stub, even if fail was set
thanks bozaro
see https://github.com/libvips/libvips/issues/1946
2020-12-28 22:46:33 +00:00
4227606f5a
fix tiff thumbnail from buffer and source
...
We were missing the new tiff thumbnail logic on the source and buffer
paths.
see https://github.com/libvips/libvips/issues/1815
2020-09-14 17:38:20 +01:00
a88061dfec
fix jpegload autorotate
...
thanks chregu
see https://github.com/libvips/php-vips/issues/105
2020-09-03 19:31:56 +01:00
9246094033
Fix function pointer cast issues
...
It is undefined behavior in C and C++ to cast a function pointer
to another type and call it that way. This does work in most native
platforms, however, despite it being UB, but in WASM it can fail.
See:
https://emscripten.org/docs/porting/guidelines/function_pointer_issues.html
2020-06-30 13:35:41 +02:00
981d5c4b16
revise autorot system
...
- deprecate vips_autorot_get_angle() since orientation is no longer a
simple rotate
- add vips_image_get_orientation() and vips_image_get_orientation_swap()
- revise tiff and jpeg loader autorotate to just call vips_autorot(),
but only if necessary
- revise thumbnail autorotate too
2020-06-06 17:25:46 +01:00
ffa10eb148
default JFIF res to 72dpi
...
We were 1ppm before. 72dpi is the EXIF default.
2020-05-13 14:34:59 +01:00
b9eddecca3
prevent /0 in jfif res decode
2020-05-06 20:08:36 +01:00
77365a8a6f
better handling of JFIF res unit 0
...
res unit 0 sets the aspect ratio, rather than being unitless
thanks angelmixu
see https://github.com/libvips/libvips/issues/1641
2020-05-05 11:47:31 +01:00
1338def535
fix some autorot issues
...
and a problem with vipsthumbnail of tif
see https://github.com/libvips/libvips/issues/1515
2020-01-07 09:00:14 +00:00
119dd2cb72
pytest passes again after stream -> source/target
2019-12-29 23:08:33 +00:00
4c5873809f
experiment with renaming stream
...
rename as VipsConnection, VipsSource, VipsTarget etc.
see https://github.com/libvips/libvips/issues/1494#issuecomment-569498619
renamed with this script:
```
set -e
edit() {
sed -i -E "$1" rename
}
for i in $*; do
cp $i rename
edit s/VIPS_STREAMOU/VIPS_TARGET_CUSTOM/g
edit s/VIPS_STREAMO/VIPS_TARGET/g
edit s/VIPS_STREAMIU/VIPS_SOURCE_CUSTOM/g
edit s/VIPS_STREAMI/VIPS_SOURCE/g
edit s/VIPS_STREAM/VIPS_CONNECTION/g
edit s/vips_streamou/vips_target_custom/g
edit s/vips_streamo/vips_target/g
edit s/vips_streamiu/vips_source_custom/g
edit s/vips_streami/vips_source/g
edit s/vips_stream/vips_connection/g
edit s/VipsStreamou/VipsTargetCustom/g
edit s/VipsStreamo/VipsTarget/g
edit s/VipsStreamiu/VipsSourceCustom/g
edit s/VipsStreami/VipsSource/g
edit s/VipsStream/VipsConnection/g
# eg. VIPS_TYPE_STREAM or VIPS_IS_STREAM
edit "s/VIPS_([A-Z]+)_STREAMOU/VIPS_\1_TARGET_CUSTOM/g"
edit "s/VIPS_([A-Z]+)_STREAMO/VIPS_\1_TARGET/g"
edit "s/VIPS_([A-Z]+)_STREAMIU/VIPS_\1_SOURCE_CUSTOM/g"
edit "s/VIPS_([A-Z]+)_STREAMI/VIPS_\1_SOURCE/g"
edit "s/VIPS_([A-Z]+)_STREAM/VIPS_\1_CONNECTION/g"
edit s/streamou/target_custom/g
edit s/streamo/target/g
edit s/streamiu/source_custom/g
edit s/streami/source/g
# various identifiers which also change
edit s/is_a_stream/is_a_source/g
edit s/find_load_stream/find_load_source/g
edit s/find_save_stream/find_save_target/g
edit s/new_from_stream/new_from_source/g
edit s/write_to_stream/write_to_target/g
edit s/vips_thumbnail_stream/vips_thumbnail_source/g
# eg. vips_webpload_stream
edit "s/vips_([a-z]+)load_stream/vips_\1load_source/g"
# eg. vips_webpsave_stream
edit "s/vips_([a-z]+)save_stream/vips_\1save_target/g"
mv rename $i
done
```
2019-12-29 21:40:21 +00:00
f5a4f41edf
rename VipsStream* variables
...
The convention is now:
VipStreami *streami;
We had `input` in many places, a left-over from the old VipStreamInput
name.
2019-11-11 09:09:34 +00:00
3d344b4b5f
add Kleis's patches
...
and add minimise support to radiance load
2019-11-06 17:04:13 +00:00
ba0100eb13
revise decode/unminimise use
2019-10-28 05:04:04 +00:00
ee3270f8e9
more refactoring
2019-10-22 11:08:54 +01:00
696ff2b24a
fix up jpeg load
...
and revise descriptor test
2019-10-17 13:16:12 +01:00
1bdadeed61
add png stream load
2019-10-14 18:03:45 +01:00
c01c3b9100
add minimise support to VipsStreamInput
2019-10-14 10:46:17 +01:00
20cb0da247
test suite passes
...
with all jpegload coming via VipsStreamInput
2019-10-11 14:03:10 +01:00
6c1d7db31b
remove old jpeg load stuff
...
it's just stream now
works on buffer, but seems to fail on
vipsheader йцук.jpg
curiously
2019-10-11 09:43:12 +01:00
ad822109d4
input works
...
this works now:
cat k2.jpg | vips invert - x.jpg
output next
2019-10-11 07:51:43 +01:00
442720a0a1
add magic "-" stdin filename
...
this almost works!
$ vips invert - x.jpg < k2.jpg
2019-10-11 05:38:58 +01:00
387cafa738
add jpegload_stream
2019-10-10 20:42:39 +01:00
efcd31b498
add something to test the new restart system
...
do multiple renders from one seq iage, check fds are opened and closed
as expected
see https://github.com/libvips/libvips/issues/1370
2019-10-06 07:25:09 +01:00
c1921b0ad1
add a test for file descriptors
...
verify new loader behaviour with minimise
2019-10-04 17:44:25 +01:00
c328b089b1
jpegload restart after minimise
...
after minimise, we need to reopen the underlying file
passes pytest but a proper test is still to come
https://github.com/libvips/libvips/issues/1370
2019-10-03 16:40:52 +01:00
5d19bcf955
Merge branch '8.8'
2019-08-30 11:03:24 +01:00
6ea76f9632
improve data_length handling in jpg load
...
libjpeg uses unsigned ints for data length, so we must use size_t
everywhere.
2019-08-30 11:00:25 +01:00
19a80e4efb
Merge branch '8.8'
2019-08-21 17:37:16 +01:00
b5e8e99746
fix a read-one-byte-beyond issue in jpeg load
...
libvips could harmlessly read beyond the end of a string with a crafted jpg
file
2019-08-21 17:17:54 +01:00
0e5447e537
final cleanup
2019-07-28 17:15:54 +01:00
5ef14db544
put close-on-last-line back for jpg/tif/png
...
consider something like:
$ vips arrayjoin "$(echo *.jpg)" x.tif --across 10 --vips-progress
close on minimise won't close until the whole pipeline finishes, so
we'll need to keep every input file open
close on last line will shut down inputs as we are done with them, so we
save a lot of file descriptors
this patch puts close-on-last-line back for jpg/tif/png
2019-07-28 11:34:40 +01:00
2c654060f9
experiment with a different early-close strategy
...
We close loaders early in order to save file handles, and on Windows to
make sure that files can be deleted as soon as possible.
Currently loaders do this by watching the Y coordinate of requests and
freeing the fd when the final line of the file is fetched. This is messy
and does not always work, since there are cases when the final line is
not fetched.
Instead, this patch gets loaders to listen for "minimise" on their
output and close on that. This signal is emitted on all upstream images
whenever a threadpool finishes a scan of an image and is usually used to
trim caches after computation.
See https://github.com/libvips/libvips/issues/1370
2019-07-20 16:31:30 +01:00
120c3b62d9
more cmyk jpg tweaks
...
passes tes_formats.sh now
2019-04-27 13:27:18 +01:00
e081ae2d99
experiment with app14 decode
2019-04-15 10:50:42 +01:00
37613c196d
fix jpeg xmp read/write
...
it was not removing/reattaching the magic XMP URL
see https://github.com/libvips/libvips/issues/1229
2019-02-11 18:16:20 +00:00
John Cupitt
Kleis Auke Wolthuizen