Age | Commit message (Collapse) | Author |
|
It is not very useful.
|
|
|
|
Support some common FB2 encodings as well as UTF-8 and UTF-16.
|
|
|
|
Add tables for Windows-1250, Windows-1251, and ISO-8859-1.
Also add unicode_from_encoding tables.
Move encodings from PDF namespace to Fitz.
|
|
Otherwise we get unsightly [0 0 0 0] bounding boxes for lines with
a single zero area character.
|
|
Only use the PDF font metrics for stretched substitute fonts, in all
other cases, trust the embedded font file instead.
|
|
Rendering modes like fill+stroke or fill+stroke+clip send the same
text object multiple times. Only extract the text from the first instance.
|
|
The core library doesn't know where a document comes from, since we can
open it as a stream, etc. Let the viewer handle relative file URLs.
|
|
A '+' prefix indicates an open branch.
A '-' prefix indicates a closed branch.
A ' ' prefix indicates a leaf node.
|
|
Don't pass the information as a separate argument.
|
|
|
|
|
|
|
|
This doesn't cover all the PDF options, but it's better than
nothing.
This does mean we have to "enable-background" for all SVG files
whether they use blending or not.
|
|
When generating a softmask, which is to be rendered with a blend,
don't pass that blend down into the rendering of the softmask.
Otherwise the blend gets used twice. In the example given this is
causing the content to disappear.
|
|
Extracting embedded files can be trivially done with 'mutool show'
or a simple mutool run script.
The portfolio creation code is incomplete, and would require a lot of
work and testing with Adobe in order to be useful.
|
|
Remove pointless paranoia checks.
Move the check whether it has already been loaded into load_portfolio().
|
|
|
|
|
|
What's the difference between PDF_OBJ_FLAG_MARK and PDF_FLAGS_MARKED?
PDF_OBJ_FLAG_MARK is used to flag an xref entry as one that we want
to keep cached with the pdf_mark_xref/pdf_clear_xref_to_mark functions.
PDF_FLAGS_MARKED is used to detect unwanted recursion in PDF structures.
Renaming the xref_entry.flags field to xref_entry.marked should remove this
source of potential confusion.
|
|
|
|
15 years have passed since we first set the sampling to a 32x32 mesh.
Quadrupling it to 64x64 sounds about right to me.
|
|
|
|
Useful if printing text or traces to stdout.
Be quiet automatically if printing to stdout.
|
|
Since we call end_layer without distinction, we should also call
begin_layer the same way.
Also change the XML to emit separate tags for beginning and ending
a layer, since there is no guarantee the layer calls will be neatly
nested with clipping pushes and pops; nor is there a guarantee that
the PDF will even balance the BMC/BDC and EMC operators.
|
|
|
|
Make sure that the ICC colorspace has the same number of components
as the photometric interpretation.
|
|
|
|
|
|
|
|
A proper appearance stream is written when signing with a certificate.
This is just to create a placeholder appearance when the original document
did not write one.
|
|
|
|
|
|
|
|
Also fiddle the #ifdeffery to get sane defaults.
|
|
|
|
|
|
Only use the PDF character widths when also stretching glyphs to match
the PDF metrics.
|
|
Ignore space-sized backward motions. Assume that these motions are
either extreme levels of kerning, or something else fishy going on.
|
|
This option is exposed as the -D flag to mutool clean.
Allow saving a document without encryption.
The next step is to allow saving a document with new encryption, but this
may require a fair amount of rejigging since the PDF write code to keep
separate doc->crypt objects for decryption and encryption.
|
|
|
|
|
|
Update to match Adobe Reader's behaviour. This fixes bug #700128.
|
|
These formats are all almost identical to GNU tar format.
|
|
Without guards the device calls might end up with a device pointer
being NULL, causing segfaults.
|
|
|
|
|
|
Hide fz_stack_slot and exception handling details too.
Also make sure we have an initialized jmp_buf so we can safely throw from
the always block even in the exception stack overflow case.
|
|
Now the image size limit is 131072 x 131072 instead of 32768 x 32768.
|