If a gamma encoded file that has a gamma not matching that of sRGB is passed to
the simplified API the previous code simply interpreted it as a power law
encoding. However old Mac files had a power law correction of 1.45 built in to
the encoding, even though the display devices were consistent with sRGB.
Assuming a power law encoding results in substantial differences in the
interpretation of low 8-bit values; below 10. For example an Apple '5' which is
equivalent to an sRGB '17' ends up as the value '8'.
This patch provides some measure of correction for this by making the gamma
correction done within the simplified API assume that any encoded data is
encoded relative to an sRGB-like transfer function; the data is corrected back
to the PNG-nominal 2.2 value then decoded to linear (if required) using the sRGB
transfer function.
This reduces the errors reported by pngstest for such files (colormapped ones)
but still leaves the issue with files where the standard libpng code does the
gamma decoding. To cope with the latter cases the patch also includes a new
pngstest-errors which allows the result, however this is still a
work-in-progress; a better solution is possible.
Signed-off-by: John Bowler <jbowler@acm.org>
This implements a new chunk parse implementation that can be shared, it
is currently shared by the progressive reader and the sequential one
(not, yet, the writer).
The patch also implements shared transform handling that is used
throughout.
Signed-off-by: John Bowler <jbowler@acm.org>
depend on READ_TRANSFORMS and moves all the transform gamma (READ_GAMMA)
code from elsewhere to png_tran.c. There are no code changes.
There is one remaining use of the gamma (16-bit) code in the simplified
API in pngread.c but that is because of a long-standing libpng bug,
namely that the gamma corrected palette is no produced by
png_read_update_info (John Bowler).
internal struct (png_transform_control) to replace row_info and uses
that to implement affirms correctly. The change also adds checks on
the rowbytes calculation and additional checks on most transform
implementations.
Added png_uint_16 range checking, pngvalid tRNS, fixed png_uint_16:
review of previous checks, removal of some where SAFE. pngvalid: add
testing of tRNS for better code coverage pngvalid: correct rgb-to-gray
error calculations. Code coverage is still incomplete: see /*UNTESTED*/
in pngrtran.c
added RELEASE/!RELEASE convenience macros. png_muldiv_warn was used in
only one place, and the overflow condition is a genuine warning not
an internal error. Four macros allow code or function parameters to be
condition on RELEASE (or not) builds and tidy up the #ifdef handling of
functions.
where the code depended on the build base type and can be defined on
the command line, allowing testing in beta builds. Overflows previously
handled by png_warning in png.c have been changed to use new
macros/affirm functions so that beta builds will abort on overflow and
release builds will quietly ignore it. This avoids release builds
producing warnings that are of no use to end users.
Added 'assert' based overflow checking for debug builds for char
and short assignments. Simply ignore the error in release builds
(a truncated value will be used without warning). Controlled by
PNG_RANGE_CHECK_SUPPORTED.
seem to generate warnings when an unsigned value is implicitly
converted to double. This is probably a GCC bug but this change
avoids the issue by explicitly converting to (int) where safe.
Free all allocated memory in pngimage. The file buffer cache was left
allocated at the end of the program, harmless but it causes memory
leak reports from clang.
Fixed array size calculations to avoid warnings. At various points
in the code the number of elements in an array is calculated using
sizeof. This generates a compile time constant of type (size_t) which
is then typically assigned to an (unsigned int) or (int). Some versions
of GCC on 64-bit systems warn about the apparent narrowing, even though
the same compiler does apparently generate the correct, in-range,
numeric constant. This adds appropriate, safe, casts to make the
warnings go away.
Removed "option READ_COMPRESSED_TEXT enables READ_TEXT" from pnglibconf.dfa,
to make it possible to configure a libpng that supports iCCP but not TEXT.
Removed "option WRITE_COMPRESSED_TEXT enables WRITE_TEXT" from pnglibconf.dfa
Only mark text chunks as written after successfully writing them.
to run all the png_handle_*() and depend on them to return if info_ptr == NULL.
This improves the performance of png_read_end(png_ptr, NULL) and makes
it more robust against future programming errors.
PNG_TRANSFORM_* values are always defined in png.h and, because they
are used for both read and write in some cases, it is not reliable
to #if out ones that are totally unsupported. This change adds error
detection in png_read_image() and png_write_image() to do a
png_app_error() if the app requests something that cannot be done
and it adds corresponding code to pngimage.c to handle such options
by not attempting to test them.
function definitions before the place where they are called so that
they can be masde static. Move the intrapixel functions and the
grayscale palette builder out of the png?tran.c files. The latter
isn't a transform function and is no longer used internally, and the
former MNG specific functions are better placed in pngread/pngwrite.c
Fixed 'minimal' builds. Various obviously useful minimal configurations
don't build because of missing contrib/libtests test programs and overly
complex dependencies in scripts/pnglibconf.dfa. This change adds
contrib/conftest/*.dfa files that can be used in automatic build
scripts to ensure that these configurations continue to build.
Enabled WRITE_INVERT and WRITE_PACK in contrib/pngminim/encoder.
a flag to change default. In 1.6.0 when the simplified API was used
to produce color-mapped output from an input image with no gamma
information the gamma assumed for the input could be different from
that assumed for non-color-mapped output. In particular 16-bit depth
input files were assumed to be sRGB encoded, whereas in the 'direct'
case they were assumed to have linear data. This was an error. The
fix makes the simplified API treat all input files the same way and
adds a new flag to the png_image::flags member to allow the
application/user to specify that 16-bit files contain sRGB data
rather than the default linear.
Fixed bugs in the pngpixel and makepng test programs.
fixed Cygwin+MSVC build errors. The ICC profile handling now includes more
checking. Several errors that caused rejection of the profile are now
handled with a warning in such a way that the invalid profiles will be
read by default in release (but not pre-RC) builds but will not be
written by default. The easy part of handling a cHRM chunk is written,
where the ICC profile contains the required data. The more difficult
part plus guessing a gAMA value requires code to pass selected RGB values
through the profile.
control of the unknown handling, corrects the pre-existing bug where
the per-chunk 'keep' setting is ignored and makes it possible to skip
IDAT chunks in the sequential reader (broken in earlier 1.6 versions).
There is a new test program, test-unknown.c, which is a work in progress
(not currently part of the test suite). Comments in the header files now
explain how the unknown handling works.
ignore all unknown chunks and all known chunks except for IHDR, PLTE, tRNS,
IDAT, and IEND. Previously it only meant ignore all unknown chunks, the
same as num_chunks == 0. Revised png_image_skip_unused_chunks() to
provide a list of chunks to be processed instead of a list of chunks to
ignore. Revised contrib/gregbook/readpng2.c accordingly.
on alignment increasing pointer casts when -Wcast-align is passed. This
fixes the cases clang warns about either (pngread.c) by eliminating the
casts from png_bytep to png_uint_16p or, for pngrutil.c where the cast
is previously verified or pngstest.c where it is OK by introducing new
png_aligncast macros to do the cast in a way that clang accepts.
checks. The code now does a png_error if an attempt is made to do the
row initialization twice; this is an application error and it has
serious consequences because the transform data in png_struct is
changed by each call.
This turns on PNG_FLAG_DETECT_UNINITIALIZED to detect app code that
fails to call one of the 'start' routines (not enabled in libpng-1.5
because it is technically an API change, since it did normally work
before.) It also makes duplicate calls to png_read_start_row (an
internal function called at the start of the image read) benign, as
they were before changes to use png_inflate_claim. Somehow webkit is
causing this to happen; this is probably a mis-feature in the zlib
changes so this commit is only a work-round.
