TMW->Wylie conversions with the new-and-improved TMW->Wylie
algorithm faulty.
Now I'm using it a little more than you need to, e.g. b.lha instead of blha is
generated because bla and b.la are ambiguous.
like \bullet, \emdash, etc., and this fix only works for Windows or OS/2 RTF
files, not for Mac RTF files. So if you want a TM->TMW conversion to work,
use MS Word for Windows, not for the Mac.
'<' and '>'. The current keyboard implementation makes this an either-or
proposition, when fundamentally it need not be.
Added a <?Numbers?> command and an <?Input:Numbers?> command to
tibwn.ini; broke the numbers apart from the consonants. This facilitates the
new-and-improved Tibetan->Wylie conversion.
Tibetan->Wylie is now done by forming legal tsheg-bars. A legal tsheg bar
is converted into perfect THDL Wylie. See code comments to learn what
it thinks is a legal tsheg-bar, but it inlcudes bskyUMbsH minus the trailing
punctuation (H), e.g.
Illegal sequences, such as runs of transliterated Sanskrit, are turned into
unambiguous Wylie; each glyph is followed by a vowel or a disambiguator
('.').
I've made it so that the illegal sequences are as beautiful as possible. You
get 'pad+me', for example, not the equivalent but uglier 'pad+m.e.'.
which means that the command-line tool can finally function with a headless
graphics device. Hopefully it will speed things up, too. It also means that
entering Roman text into the TMW->Unicode conversion and TMW->TM
conversion will be easy.
Added support for two more oddballs.
Deprecated the oddball lookup method because it drops up to 30 glyphs in
TibetanMachine. The correct solution is to transform the RTF before Java's
busted RTF readers ever see it. \'97 becomes \u151, e.g.
beginning of the document as they should and as they are documented to.
They now do, and they bracket the bad characters with the TM or TMW for
U+0F3C on the left and the TM or TMW for U+0F3D on the right.
Some cleanup.
the troublesome glyphs are now put at the beginning of the document
AFTER AN ACHEN. This makes a glyph like \tmw7095 visible atop the
achen.
Major fix to the handling of paragraphs in conversion; we were (for
whatever reason) dropping paragraphs before.
inserting 5 characters at a time and then skipping ahead just one
position. I don't think this affected correctness.
I believe there's still a terrible (exponential?) slowdown as the
input file gets bigger, however. Perhaps not -- but we run through
the first 1000 TMW glyphs in 6 seconds, the 20th thousand takes at
least 60 seconds. Is TMW->Wylie faster than TMW->Unicode? If so,
why?
Thought: don't use a DuffPane within TibetanConverter -- it can only
add overhead, right? My hprof profile said that the conversion was
taking just a couple of percent of the work; the rest was going to
display-related stuff that you should only see if you were displaying
the document. I'm not!
Is the EWTS '_' to be represented as U+0020, or is it a wider space?
Does TMW9.42, Dza, map to U+0F5F,U+0F39?
Does TMW6.60, r+y, map to U+0F62,U+0FBB or to U+0F6A,U+0FBB? (Likewise with r+w, TMW6.61, TMW6.62, etc.)
Is U+0F7E a bindu? What Unicode does TMW7.96 map to, for example? What does TMW7.91 map to?
Should TMW8.97 and TMW8.98 map to swastiskas elsewhere in Unicode? If so, which codepoints? Likewise with TMW9.60, a Chinese character.
Does TMW7.68 map to U+0F39?
Does TMW7.74, the ITHI secret sign, have a Unicode mapping? f68,fa0,f80,f72 comes close, but fa0 would be too large, wouldn't it?
What Unicode does TMW9.61 map to? Is it for sequences like f40,f7c,f60,f72? Or is it for f60,f72,f7c?
and then tried to convert that TMW to Wylie. I swear it's Java's
problem (see the ugly stack trace in the code and decide for
yourself), and I tried replacing rather than
inserting-and-then-removing, but it didn't work. I've left these
things as options.
default, all oddballs to appear once in the resulting document.
This'll help me find the correct glyphs for the oddballs, and it'll
prevent the average user from converting a document with oddballs.
I've fixed that.
I've also added a couple of Unicode mappings to give a flavor for how
multi-codepoint mappings will be represented.
TM->TMW conversion takes about 1 second per thousand glyphs on my
PIII-550.
noticed that formatting is mostly OK but sometimes gets bungled slightly.
I tried everything I could think of, and now I'm passing the buck to Java's
RTF support.
TMW_RTF_TO_THDL_WYLIE (now misnamed) support TMW->TM
conversion (but not TM->TMW). There is an automated test case for a
TMW->TM conversion.
I have full confidence in this conversion. Even the smallest glitch in the core
functionality (not formatting) would surprise me.
Note that the JUnit test TMW_RTF_TO_THDL_WYLIETest sometimes fails
due to one- or two-line diffs between the actual and expected outputs. This
is because Java's RTF support is not deterministic, I'm guessing, and is not
a real failure. I'm too lazy to make a more elaborate sed/diff mechanism
that works on all platforms, and that would complicate the build anyway.
verified this extensively and have full confidence that these mappings
agree with Tony Duff's Tibetan! 5.1 documentation (except as described
below).
To get them, I had to disregard Tony Duff's tables for a few glyphs: the
characters with ordinal 32 and 45 (space and hyphen in Roman ASCII,
space and tsheg in Tibetan). For these glyphs, we must have mappings
from TibetanMachineSkt4.32 to something, etc., and those mappings were
not present. I've normalized the mapping for these glyphs, as it is arbitrary
because the same two glyphs just appear fifteen times each.
brace problem upon opening any RTF document.
The TMW_RTF_TO_THDL_WYLIE test baselines changed because
I fixed (a while ago) some inconsistencies between the EWTS standard and
Jskad.
Conversion of TibetanMachineWeb8.40, @#, to Wylie now works correctly.
Unfortunately, though, typing @# doesn't produce 8.40, it still produces
8.38 and 8.39, two glyphs.
org.thdl.tib.input.TMW_RTF_TO_THDL_WYLIE. It converts RTF files
consisting of TMW characters to the corresponding THDL Extended Wylie.
It supports --find-some-non-tmw mode, which allows you to ensure that no
unusual characters will spoil the conversion. The converter has built-in
intelligence that allows it to handle Tahoma '{', '}', and '\\' characters
properly.
The converter works on mixed Roman/TMW also, but --find-some-non-tmw
and --find-all-non-tmw modes are not as useful.
Invoke org.thdl.tib.input.TMW_RTF_TO_THDL_WYLIE, which resides in
Jskad's jar, with no command-line options to see usage information.
Noted some failures. "Fixed" the code to do what I want it to do for
the (no sanskrit stacking, tibetan stacking) case [which is exercised
by this keyboard only].
clean check'. Right now there are tests to ensure that typing certain
sequences of keys in the Extended Wylie keyboard gives the expected
Extended Wylie back when "Tools/Convert Tibetan to Wylie" is invoked.
The syntactically illegal d.wa now converts to Tibetan and then back
to d.wa (not dwa, as it did); likewise with the illegal g.wa. wa
doesn't take any prefixes, but I prefer clean end-to-end
behavior. (jeskd doesn't go end-to-end, though.)
Note that you cannot successfully run the DuffPane tests on a Linux
box unless your DISPLAY variable is set correctly. Thus, my nightly
builds will fail with an Error (as opposed to a Failure).
Better tests. As part of that, I had to break TibetanMachineWeb into
TibetanMachineWeb+THDLWylieConstants, because I don't want the
class-wide initialization code from TibetanMachineWeb causing errors
in LegalTshegBarTest.
converts TibetanMachineWeb glyphs to THDL Wylie. Three-glyph and
four-glyph sequences with implicit "a" vowels are now handled
correctly, except for disambiguation w.r.t. things like b-la-g
vs. bla-g and d-wa vs. dwa.
pa'am, pa'ang etc. now work too.
Illegal Tibetan sequences now become very ugly, but "correct" Wylie.
Correct in the sense that converting it back to glyphs should get you
the glyphs you started with.
I also made a change to TibetanMachineWeb.java that I hope will clear
up problems with this feature when keyboards other than "Extended
Wylie" are selected.
Took nga out of the farRightSet [postsuffixes]; only da and sa belong
there, right?
I tried to get the system in a state such that I could run automated
tests of this stuff, but I ran into difficulties. I have some manual
test cases; ask if you're interested.
I'm committing in order to sync with my laptop, really. This stuff will disappear
and reappear in better form later, after a holiday of coding and eggless,
alcohol-free nog.
and for this package only.
I'm committing in order to sync with my laptop, really. This stuff will disappear
and reappear in better form later, after a holiday of coding and eggless,
alcohol-free nog.
UnicodeCodepointToThdlWylie.java.
Added a new class, UnicodeGraphemeCluster, that can tell you
the components of a grapheme cluster from top to bottom. It does not
yet have good error checking; it is not yet finished.
Next is to parse clean Unicode into GraphemeClusters. After that comes
scanning dirty Unicode into best-guess GraphemeClusters, and scanning
dirty Unicode to get nice error messages.
because a Japanese scholar has an "Extended Wylie" also.
NFKD and NFD have a new brother, NFTHDL. I wish there weren't a need,
but as my yet-to-be-put-into-CVS break-unicode-into-grapheme-clusters code
demonstrates, the-need-is-there. forgive-me for the hyphens, it's late.
characters, for example.
Normalization forms NFKD and NFD are supported for the Tibetan Unicode
range. I don't like either, actually. I've tested NFKD, but I've not yet
committed the tests.
and the build system is not yet aware of them.
I'm adding some classes for representing legal tsheg-bars (syllables, for the
most part) in Unicode. These classes were designed bottom-up (OK, OK --
they weren't designed designed, but I had to write down everything I knew
about Tibetan syntax somewhere). The classes are aware of extended
wylie. I doubt the Javadocs work yet, and I'm still testing (and am not
committing my testing code with these as it is not yet ready).
Next on my list--fix these up to reflect my new awareness of suffix particles
(like le'u'i'o) add classes to support syntactically incorrect Unicode
sequences. Then add a UnicodeReader, and we've got the back end of
a Tibetan Unicode shaping system (like half of MS's Uniscribe or Apple's
Worldscript or FreeType Layout or Omega's OTPs).
A top-down design would not have included LegalTshegBar. But now that
my itch has been scratched, potential uses are lingering about. For example,
it would be nice to scan some input and break it into LegalTshegBars,
punctuation/marks/signs, and illegal stacks. Then we could alert the client
of the illegality, its precise form, and its precise location.
The real system for turning a Unicode stream into an internal representation
suitable for conversion to EWTS/ACIP/XHTML/what-have-you need not be
aware of Tibetan syntax. But to make the very best conversion from
Unicode to, e.g., EWTS, it is necessary to konw that gaskad is better
represented as gskad, but that jaskad is not the same as jskad.
It does not have many methods for determining the root letter, suffix,
and so on, but these should be easy to add. David, please use this
class to the extent that it and your new work overlap.
'Fonts' module inside the 'Jskad' module. I.e., you must now have the
tree like so:
Jskad/
source/
dist/
Fonts/
TibetanMachineWeb/
.
.
.
This is because the THDL tools now optionally (and by default) load
the TibetanMachineWeb fonts automatically.
Updated the build system so that the 'web-start-releases' and
'self-contained-dist' targets JAR up optional JARs to create
double-clickable, self-contained joy. Even the TMW fonts are in the
JARs now.
Changed the strings describing two Jskad keyboards so that "keyboard"
is no longer in the description. It's in the label next to the combo
box.
Jskad now saves preferences on exit or when the user selects a menu
item (that is there for debugging mainly) to ~/my_thdl_preferences.txt
on *nix or C:\my_thdl_preferences.txt on Win32. I don't know the
correct Mac location.
There's a new paradigm for telling org.thdl.util.ThdlOptions that a
user preference has been changed. If, for example, a combo box is
manipulated so that the ACIP keyboard is selected, then you must call
a certain method in ThdlOptions.
but does line breaks correctly. I.e., I refactored DuffPane into two classes.
I did this trying to track down a subtle bug in line breaking: 'gye ' breaks
after 'gy' sometimes, with the dreng bo on the next line, but only when you
resize the window certain ways, and only in Savant (and maybe QD and the
translation tool, I don't know) but not in Jskad.
I was not successful in finding the bug, but it still exists when I use
TibetanPanes instead of DuffPanes in org.thdl.savant.tib.*.
ACIP's 'WA' represents Wylie's 'wa', but ACIP's 'ZHVA' represents Wylie's
'zhwa'. The key for wasur is the same as the key for the twentieth
consonant in extended Wylie, but not in ACIP.
ACIP or 'ShSm' in Extended Wylie to see the new behavior.
We use a trie to store valid input sequences. In the future, we could use
the same trie as a replacement for the more inefficient HashSets we use to
store characters, vowels, and punctuation. For example, we'd use
'validInputSequences.put("K", new Pair("consonant", "k"))' when reading
in the ACIP keyboard's description of the first consonant of the Tibetan
alphabet in 'TibetanKeyboard.java'.
Note that the current trie implementation is only useful for 7- or 8-bit
transcription systems, and works best for tries with low average depth, which
describes a transcription system's trie very well. If you used arbitrary
Unicode in your keyboard, you'd need a different trie implementation.
Improved the optional keyboard input mode status messages.
Added a JUnit test for the new Trie that fails at present since the Trie is
case-insensitive. Running JUnit tests is not something our build system
knows about at present, but Eclipse 2.0 makes it very easy.
Fixed a few compiler errors due to imports I'd forgotten.
DefaultStyledDocument, and another consisting entirely of static utility
methods for processing Tibetan text. Moved TibetanDocument.DuffData
into its own class.
I think this makes things a bit more transparent, and gets us a little closer to
making clean use of Swing.
DefaultStyledDocument, and another consisting entirely of static utility
methods for processing Tibetan text. Moved TibetanDocument.DuffData
into its own class.
I think this makes things a bit more transparent, and gets us a little closer to
making clean use of Swing.
I cleaned things up a bit, and I've made logging optional since I don't yet
trust the code fully.
A Wylie underscore at the end of a line is worth looking into further, at the
very least.
is basically that we use our own special ViewFactory, with a new
subclass of LabelView (the view RTFEditorKit uses for the nitty
gritty) that is aware of Tibetan.
There are a couple of nasty hacks still here, and Swing's
documentation for doing what I did was quite poor. I searched the web
for hours, read the Javadocs and the tutorials, and consulted a Swing
reference book, but I still don't have tremendous confidence in this
solution. If it fundamentally doesn't work, though, we have to define
our own first-class Document, Element hierarchy, ViewFactory, Views,
and EditorKit. So let's hope it *does* work fundamentally.
I can't say for sure if this even works, as I have yet to run this
code on a machine where Jskad works properly. I had major trouble
installing the TMW fonts on Linux, and have yet to resolve it, even
after verifying via xlsfonts that the fonts were installed and then
changing TibetanMachineWeb.java to look for them. Because I haven't
tested this yet, a lot of nasty code is tagged 'DLC' and commented
out.
Added a "Quit" option to Savant's File menu. Factored out the Close
option in doing so.
Exceptions in many action listeners are now handled by
org.thdl.util.ThdlActionListener or org.thdl.util.ThdlAbstractAction.
Many exceptions that we used to just log now optionally cause aborts.
This option is on by default for developers using 'ant savant-run'-style
targets, but it is off for users.
An erroneous CLASSPATH now causes a useful error message in almost
all situations.
Fixed some typos and bad links in Javadoc comments.
Added a simple assertion facility, but the overhead is suffered even in
release builds.
Factored out the code that sets up log files like savant.log and jskad.log.
