| Age | Commit message (Collapse) | Author |
|---|
|
|
|
Would be fine were it not for the finite number of local
refs we are allowed.
|
|
Previously an exception was thrown for documents without an outline.
Now document both with and without an outline are handled correctly.
|
|
I did debate about removing these, but it looks like they are
flagging up places where processing is incomplete, so I've
left them in, and they will just appear on stderr now.
|
|
Apparently, if MuPDF is invoked on a content stream that
comes from a URI, then is.available() can report 0 bytes
(as there is no data buffered). Use a modified formulation
that reads as much data as possible from the stream into
a BufferedOutputStream and then makes a byte array from that.
Would be nicer if the core could actually read from the
stream directly, perhaps, but that can wait for the new JNI
based version.
Thanks to Marc K for identifying the problem and supplying
the patch.
|
|
Broken by alpha changes.
|
|
|
|
|
|
|
|
This is for easy of use, in addition to accepting Buffer.
|
|
|
|
These are meant to make it easy to interact with mupdf's Buffer
object in a normal Java fashion.
|
|
|
|
|
|
|
|
zooming and scrolling.
|
|
|
|
|
|
|
|
|
|
|
|
fz_bound_page needs the correct 2nd argument
pdf_widget_get_type should be pdf_widget_type
|
|
|
|
|
|
|
|
|
|
Thanks again to Tamir Evan for pointing this out.
|
|
Also fix bug in sample java viewer where zoom caused an Exception.
|
|
Some objects get their interfaces expanded or the entire object
is added both corresponding to what was already present in the
mujs interface. Some interfaces, e.g. Buffer.writeLines() has
been adapted to suit Java.
|
|
* Make sure that arguments are checked for NULL before being
passed to ->Get JNI-functions (as these may crash on NULL).
* Remove unused argument to from_Device().
* Consistently use jint and jfloat in JNIEXPORTed functions.
* Check ctx for NULL in all relevant functions.
* Throw Java exception when constructors fail.
* to_*_safe() means that the function doesn't fz_throw().
* to_*_own() means that the function takes ownership of the object.
* Make sure that jprefixing of argument identifies does not
clash with jprefixing of types, specifically jstring.
* Rename (j)colorspace -> (j)cs, (j)shade -> (j)shd,
(j)device -> (j)dev.
* Always check that the context is available in bindings.
* Always check that required arguments are non-NULL or have
decent default values (e.g. for fz_matrix and fz_rect).
|
|
|
|
|
|
|
|
Reflect API change in example viewer.
|
|
Reflect the change in API in the ios code.
|
|
|
|
|
|
|
|
|
|
Since the removal of the begin_page device function, structured
text extraction has been unable to correctly establish the
mediabox for extracted pages.
Update the fz_new_stext_page call to take this mediabox
information. This is an API change, but hopefully most people
are calling fz_new_stext_page_from_page or
fz_new_stext_page_from_display_list which are updated here to
cope.
Update all the apps/tools to behave properly.
|
|
Adopt patch from Tamir Evan. Many thanks.
|
|
The code would SEGV if we were trying to synthesise an appearance
stream for an annotation, and the docs pdf resources table
had not been initialised.
We now intialise the pdf resource tables when we initialise a
pdf device. This is the earliest point we know we are going to
need them, and covers all cases.
|
|
This helps with Memento debugging, and looks neater.
|
|
Additional flags/compiler settings required for successful
building on MINGW due to harfbuzz's use of strdup.
|
|
|
|
This has been broken since fz_text was changed to be multiple
fz_text_spans.
|
|
When we are rendering a knockout group, we set the knockout flag.
Do NOT carry this forward into the renderings of any softmasks.
This fixes Bug 696870 and 696872.
|
|
|
|
|
|
|
