We essentially just end up moving `release_value_but_fixme_...` one
layer down, but it makes adding more fallible operations in the
serialization function more comfortable.
The seac command provides the base and accented character that are
needed to create an accented character glyph. Storing these values is
all that was left to properly support these composed glyphs.
Type1 accented character glyphs are composed of two other glyphs in the
same font: a base glyph and an accent glyph, given as char codes in the
standard encoding. These two glyphs are then composed together to form
the accented character.
This commit adds the data structures to hold the information for
accented characters, and also the routine that composes the final glyph
path out of the two individual components. All glyphs must have been
loaded by the time this composition takes place, and thus a new
protected consolidate_glyphs() routine has been added to perform this
calculation.
Glyph was a simple structure, but even now it's become more complex that
it was initially. Turning it into a class hides some of that complexity,
and make sit easier to understand to external eyes.
While doing this I also decided to remove the float + bool combo for
keeping track of the glyph's width, and replaced it with an Optional
instead.
Storing glyphs indexed by char code in a Type1 Font Program binds a Font
Program instance to the particular Encoding that was used at Font
Program construction time. This makes it difficult to reuse Font Program
instances against different Encodings, which would be otherwise
possible.
This commit changes how we store the glyphs on Type1 Font Programs.
Instead of storing them on a map indexed by char code, the map is now
indexed by glyph name. In turn, when rendering a glyph we use the
Encoding object to turn the char code into a glyph name, which in turn
is used to index into the map of glyphs.
This is the first step towards reusability of Type1 Font Programs. It
also unlocks the ability to render glyphs that are described via the
"seac" command (standard encoding accented character), which requires
accessing the base and accent glyphs by name.
This is useful in general (I'd imagine), but in particular having this
will allow us to implement accented PDF Type1 Font glyphs, which consist
of two separate glyphs that are composed into a single one.
When parsing streams we rely on a /Length item being defined in the
stream's dictionary to know how much data comprises the stream. Its
value is usually a direct value, but it can be indirect. There was
however a contradiction in the code: the condition that allowed it to
read and use the /Length value required it to be a direct value, but the
actual code using the value would have worked with indirect ones. This
meant that indirect /Length values triggered the fallback, "manual"
stream parsing code.
On the other hand, this latter code was also buggy, because it relied on
the "endstream" keyword to appear on a separate line, which isn't always
the case.
This commit both fixes the bug in the manual stream parsing scenario,
while also allowing for indirect /Length values to be used to parse
streams more directly and avoid the manual approach. The main caveat to
this second change is that for a brief period of time the Document is
not able to resolve references (i.e., before the xref table itself is
not parsed). Any parsing happening before that (e..g, the linearization
dictionary) must therefore use the manual stream parsing approach.
Functionality for checking image size in advance has been removed, as
this would require a SeekableStream and we now check for read errors
everywhere anyways.
This is a very new tag used for HDR content. The only files I know that
use it are the jpegs on https://ccameron-chromium.github.io/hdr-jpeg/
But they have an invalid ICC creation date, so `icc` can't process them.
(Commenting out the check for that does allow to print them.)
If the CIPC tag is present, it takes precedence about the actual data
in the profile and from what I understand, the ICC profile is
basically ignored. See https://www.color.org/events/HDR_experts.xalter
for background, in particular
https://www.color.org/hdr/02-Luke_Wallis.pdf (but the other talks
are very interesting too).
(PNG also has a cICP chunk that's supposed to take precedence over
iCCP.)
This isn't used by any mandatory tags, and it's not terribly useful.
But jpegs exported by Lightroom Classic write the 'tech' tag, and
it seems nice to be able to dump its contents.
signatureType stores a single u32 which for different tags with this
type means different things.
In each case, the value is one from a short table of valid values,
suggesting this should be a per-tag enum class instead of a
per-tag DistinctFourCC, per the comment at the top of DistincFourCC.h.
On the other hand, 3 of the 4 tables have an explicit "It is possible
that the ICC will define other signature values in the future" note,
which suggests the FourCC might actually be the way to go.
For now, just punt on that and manually dump the u32 in fourcc style
in icc.cpp and don't add any to_string() methods that return a readable
string based on the contents of these tables.
This is the type of namedColor2Tag, which is a required tag in
NamedColor profiles.
The implementation is pretty basic for now and only exposes the
numbers stored in the file directly (after endian conversion).
Though table wrappers are anonymous block containers (because
TableWrapper is inherited from BlockContainer) with no lines they
should not be skipped in block auto height calculation.
This was called from LibCore and passed raw StringView data that may
not be null terminated, then incorrectly passed those strings to
getenv() and also tried printing them with just the %s format
specifier.
