| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
I just had this case on a recent Ubuntu...
|
| |
|
| |
|
|
|
|
| |
The code is still quite experimental and sometimes generate errors while it could still continue.
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
| |
names.load() calls names.update() without an argument if no font
database is found, but the later will call the former if invoked without
an argument. Now calling names.update() with an empty names database.
fixes #23
|
|
|
|
| |
This was a debugging leftover that went unnoticed.
|
|
|
|
|
|
|
|
|
|
|
| |
* --database-dir is useless anyway since we don't support loading names
database from arbitrary path.
* --sys (and mkluatexfontdb-sys) are dangerous since we will load user
specific fonts (in ~/.fonts for example) which can not be loaded by
other users (can be security concern as well). If there is real need,
we should then implement it properly.
+ some clean up and less code duplication in mkluatexfontdb.
|
|
|
|
|
|
|
| |
Fonts should always have a names table, I suppose, so if not we should
reject the font.
Reported by Philipp Stephani.
|
|
|
|
|
|
| |
Move code that does not belong to otfl-font-nms.lua to luaotfload.dtx.
Please test.
|
| |
|
|
|
|
|
| |
The database building/updating that the, unusual, progress bar is not
needed.
|
|
|
|
|
| |
This way it is easier for people to actually find the broken ones so we
can blacklist them explicitly.
|
|
|
|
|
| |
io.flush() is needed, otherwise nothing gets printed until the loading
is finished.
|
|
|
|
| |
This turned out to be easier that I first thought.
|
|
|
|
|
| |
I didn't know how to check if a table contains a certain value so I had
this value=true hack, now I know better so removed it.
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Blacklisted fonts are listed one per line in a file named
"otfl-blacklist.cnf", there can be as much copies of this file as kpse
can find (we look for "tex" format despite of the file extension, so it
can exist any where a .tex file can) and all found files will be loaded
an read.
The parser is rather dump, it assumes a good formated file; one font
per line, no trailing spaces, no inline comments, lines starting with
"%" and empty lines will be ignored, any other lines will end in the
blacklist table without any further processing.
|
|
|
|
|
|
| |
This reverts commit 1473edf932ec729f4a41736fb7ee52654c848d59.
This broke mkluatexfontdab.
|
| |
|
| |
|
|
|
|
| |
The less locally changed ConTeXt files, the simpler my life, so I'm adapting ConTeXt suggested paths instead.
|
|
|
|
| |
Utilize lualibs functions when apporiate.
|
|
|
|
|
|
|
| |
If the requested style is not found, fallback to regular. I think this
broke in commit b9d377647240804d7d85429a18e1bdf0ff00823c
closes #13
|
|
|
|
|
| |
It takes a dir path as input and index its fonts temporarily in the names
database, changes are not saved to the disk.
|
| |
|
| |
|
|
|
|
|
| |
(so users know what's taking so long)
|
|
|
|
|
|
| |
* comments are handled correctly
* include are handled
* directories containing "texmf" are not explored... maybe it should be changed?
|
|
|
|
|
| |
This was a leftover from old code. This should fix Adobe Garamond Pro
issue.
|
|
|
|
|
|
|
|
| |
We now detect tfm fonts much earlier and force a file based lookup,
thus bypassing |names.resolve()| completely.
So, now |\font\tenrm=cmr10| is exactly the same as |\font\tenrm=file:cmr10|
as long as there is a tfm/ofm file named cmr10.
|
| |
|
|
|
|
|
| |
Make sure to check last file extension; ignoring files like
foo.ttf.bar-x.xyz.
|
| |
|
|
|
|
|
| |
kpse.readable_file() returns nil if given a directory, so disable this
check for now.
|
| |
|
| |
|
| |
|
|
|
|
| |
Patch to detect symlinks and follow them in path normalization. Reduces my database by 30 fonts. Please test (especially on Mac OSX).
|
| |
|
|
|
|
| |
This works fine on my system. Actually it works even better than fc-list, as it finds arial.ttf which fc-list doesn't... strange... please test.
|
| |
|
|
|
|
| |
Strangely, OSFONTDIR is set under windows... It works even if it's unset anyway.
|
|
|
|
| |
On my system a lot of fonts were indexed two times, this seems to fix it. It's not perfect but that's all we can do with the current database (please tell me if it introduces some regressions).
|