| Age | Commit message (Collapse) | Author |
|---|
|
Faster, shinier, better.
|
|
I broke this as part of the PDF Name refactoring.
|
|
Ensure that %010d works.
Ensure that we can output 64 bit values (%ll{d,u,x}).
Ensure that we can output size_t and fz_off_t (%z{d,u,x} and %Z{d,u,x}).
fz_off_t isn't defined yet (it will be introduced by a commit that
depends on this one), so for now, we put a stub definition in printf.c
that we will remove later.
|
|
Introduce the concept of 'packed' paths. These reduce the header
overhead for most common paths (ones with less than 256 commands
and 256 coords) to a single 32bit int once stored in the
display list.
The previous commit reduces the torture-test.pdf from 95 to 87Meg.
This commit futher reduces it to 70Meg.
|
|
Firstly, we make the definition of the path structures local to
path.c. This is achieved by using an fz_path_processor function to
step through paths enumerating each section using callback functions.
Next, we extend the internal path representation to include other
section types, including quads, beziers with common control points
rectangles, horizontal, vertical and degenerate lines. We also roll
close path sections up into the previous sections commands.
The hairiest part of this is that fz_transform_path has to cope with
changing the path commands depending on the matrix. This is a
relatively rare operation though.
|
|
Historically pdf_obj was a structure with a header and a union in it.
As time has gone by more stuff has been put into the header, and the
different arms of the union have changed in size. We've even adopted
the idea of different 'kinds' of pdf_obj's being different sizes
(names and strings for examples).
Here we rework the system slightly; we minimise the header, and split
out everything into different structures. Every different 'kind' of
pdf_obj is now it's own structure, just as big as it needs to be.
Key changes:
* refs is now a short rather than an int. We are never going
to need more than 32767 refs (indeed, if we ever need more
than about 3 (10 at the outside), something has gone very
wrong!). This aids structure packing.
* Only arrays, dicts and refs actually need the pdf_document
pointer.
* Only arrays and dicts need the parent_num pointer.
|
|
Currently, every PDF name is allocated in a pdf_obj structure, and
comparisons are done using strcmp. Given that we can predict most
of the PDF names we'll use in a given file, this seems wasteful.
The pdf_obj type is opaque outside the pdf-object.c file, so we can
abuse it slightly without anyone outside knowing.
We collect a sorted list of names used in PDF (resources/pdf/names.txt),
and we add a utility (namedump) that preprocesses this into 2 header
files.
The first (include/mupdf/pdf/pdf-names-table.h, included as part of
include/mupdf/pdf/object.h), defines a set of "PDF_NAME_xxxx"
entries. These are pdf_obj *'s that callers can use to mean "A PDF
object that means literal name 'xxxx'"
The second (source/pdf/pdf-name-impl.h) is a C array of names.
We therefore update the code so that rather than passing "xxxx" to
functions (such as pdf_dict_gets(...)) we now pass PDF_NAME_xxxx (to
pdf_dict_get(...)). This is a fairly natural (if widespread) change.
The pdf_dict_getp (and sibling) functions that take a path (e.g.
"foo/bar/baz") are therefore supplemented with equivalents that
take a list (pdf_dict_getl(... , PDF_NAME_foo, PDF_NAME_bar,
PDF_NAME_baz, NULL)).
The actual implementation of this relies on the fact that small
pointer values are never valid values. For a given pdf_obj *p,
if NULL < (intptr_t)p < PDF_NAME__LIMIT then p is a literal
entry in the name table.
This enables us to do fast pointer compares and to skip expensive
strcmps.
Also, bring "null", "true" and "false" into the same style as PDF names.
Rather than using full pdf_obj structures for null/true/false, use
special pointer values just above the PDF_NAME_ table. This saves
memory and makes comparisons easier.
|
|
Update buffer and filter processors.
Filter both colors and stroke states.
Move OCG hiding logic into interpreter.
|
|
|
|
|
|
|
|
|
|
|
|
We were failing to allow for the change in length of the hint
stream caused by the ascii encoding when calculating offsets.
|
|
MuPDF (and other PDF readers) treat invalid references as 'null'
objects. For instance, in the supplied file, object 239 is supposedly
free, but a reference is made to it.
When cleaning (or linearising) a file, we renumber objects; such
illegal refs then end up pointing somewhere else.
The workaround here is simply to spot the invalid refs during the
mark phase, and to set the referencing to null.
|
|
|
|
pdf_parse_file_spec sometimes extracts the wrong path from a FileSpec:
E.g. the /DOS path should never be returned under Unix systems, neither
should be the old /Mac paths.
For consistency, this patch also converts filesystem paths under
Windows into a format applications will expect (e.g. from "/C/path/..."
to "C:\path\...").
Finally, pdf_parse_file_spec is exposed to callers (SumatraPDF requires
that for manually processing FZ_ANNOT_FILEATTACHMENT and embedded files).
|
|
When watermarking, we may want to use the PDF device on an
existing buffer. In this case, we have no 'contents' object.
|
|
In order to be able to watermark etc, we want the ability to
add more operators/resources after page cleaning.
Add a post processing hook to enable this to be done more
easily.
|
|
Add missing initialisation of glo.ctx required due to API
change.
|
|
A few casts are required within the code, along with a few #ifdef
changes.
Some tweaks to curl are required too.
|
|
|
|
Silly oversight.
|
|
We were not filling in the matrix and bbox fields for images
collected as part of the text extraction device. Fixed here.
|
|
When we meet cached tiles when rendering the display list, we need to
skip over their contents. Previously we did this by skipping
display list nodes in their entirety.
With the new display list scheme however, we cannot simply skip
nodes completely as the graphic state changes must be remembered.
We therefore update the list playback routine to keep track of the
clip depth and to skip the function calls as required.
|
|
I was testing an untransformed rectangle. This was not being picked up
as our cluster tests use the identity matrix.
|
|
Add the new source files to the solution.
Windows builds whinge about float->double conversions. Fix these
with explicit casts.
Avoid calling strtof and strcasecmp.
|
|
The example file for this bug has an invalid font bbox. The current
code uses this bbox (or some multiple of it) to clip the glyphs
size.
In the new code, when we convert the glyphs to display lists we
watch for the bbox given in any d1 operator used. If we find one,
we gather the rectangle specified and store it as the glyph rectangle
in the fz_font.
If we then attempt to bound a glyph that used d1, it happens instantly
without needing to run the list. This seems to match acrobats behaviour.
Tests indicate that Acrobat never clips d0 glyphs, so our behaviour
is still different here, but I am not changing this at the moment.
Also, I note that t3flags should be a un unsigned short but are currently
just a char. Fix that too.
Also fix some missing code in fz_new_font that would cause leaks if mallocs
failed.
|
|
Move the logic in pdf_show_char to use the same idiom as used
elsewhere. Specifically this ensures that empty rects are
handled correctly.
|
|
This avoids up problems with a forthcoming commit.
|
|
|
|
Conflicts:
Makefile
|
|
|
|
|
|
Whenever we fz_keep a path, it's an indication that we're going to be
keeping the path around for a while (anod not changing it any more).
We therefore take the opportunity to trim the path buffers down.
|
|
Rather than a linked list of display nodes, we use solid block of
serialised data. We send a 32bit word, which contains various bitfields.
These bitfields indicate the command type, and the presence or absence
of various fields (such as paths, colorspaces, colors etc).
If these fields are not present, they are held to be the same as the
previous values.
|
|
Thanks to malc for pointing out the problem.
|
|
|
|
|
|
|
|
|
|
Add locks around fz_path and fz_text reference counting.
|
|
fz_open_file does not return NULL on failure -- it throws an exception!
|
|
Purge several embedded contexts:
Remove embedded context in fz_output.
Remove embedded context in fz_stream.
Remove embedded context in fz_device.
Remove fz_rebind_stream (since it is no longer necessary).
Remove embedded context in svg_device.
Remove embedded context in XML parser.
Add ctx argument to fz_document functions.
Remove embedded context in fz_document.
Remove embedded context in pdf_document.
Remove embedded context in pdf_obj.
Make fz_page independent of fz_document in the interface.
We shouldn't need to pass the document to all functions handling a page.
If a page is tied to the source document, it's redundant; otherwise it's
just pointless.
Fix reference counting oddity in fz_new_image_from_pixmap.
|
|
Rename fz_close to fz_drop_stream.
Rename fz_close_archive to fz_drop_archive.
Rename fz_close_output to fz_drop_output.
Rename fz_free_* to fz_drop_*.
Rename pdf_free_* to pdf_drop_*.
Rename xps_free_* to xps_drop_*.
|
|
|
|
Disallow modification of shared fz_path and fz_text objects.
They should follow a create once, consume often pattern, and as such should
be immutable once created.
|
|
|
|
We end up trying to scale the JPEG up 72 times and fail a malloc.
A better plan is to make the image handler disbelieve any xres or
yres values less than 72dpi. We take care to still preserve aspect
ratios etc.
|
|
Add the new URW base fonts that include greek and cyrillic scripts.
These new fonts remove the need for DroidSans as a generic fallback font.
|
