/usr/share/doc/libgnu-regexp-java/TODO is in libgnu-regexp-java 1.1.4-4.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 | gnu.regexp TODO
---------------
These are items that have yet to be implemented or are part of the
general wish list for gnu.regexp. You can contribute to this program
and get your name permanently embedded in the distribution if you care
to take on any of these tasks. See the contact info in the README.
Bugs / Problems with the engine:
Using a ')' in a character class which is within a () grouping
triggers an REException while parsing. Workaround is to put a backslash
in front of the literal ')'.
The leftmost longest rule is not strictly adhered to. For example,
the expression ".*?z?", given input "xyz", should match the whole string,
not the empty string.
Missing operators:
[.symbol.] -- collating symbol (POSIX)
[=class=] -- equivalence class (POSIX)
I18n support:
Some POSIX char classes ([:alpha:]) are locale sensitive, others need work
Ranges (e.g. [a-z]) should be locale sensitive
Should use collation elements, not chars, for things like Spanish "ll"
REG_ICASE is not very friendly with Cyrillic and other locales.
Messages used for exceptions should be localizable.
Performance:
Engine speed needs some major work.
Miscellaneous/Requests for Enhancements:
"Free software needs free manuals" -- you heard the man...
Some features have been added to Perl between 5.0 and 5.6. Should the
PERL5 syntax support these, or should a new syntax be added? Which should
be the default?
There could be a flag to enforce strict POSIX leftmost-longest matching.
gnu.regexp.util.Tests should be fleshed out, or perhaps changed to use JUnit.
It might make sense for getMatchEnumeration() to optionally return matches
in reverse order, so that the indices were valid all the way through,
even if you were doing text replacement on the way.
Support J2ME CLDC limited APIs as optional target:
Should be straightforward, but need to avoid use of
REFilter(InputStream|Reader) classes, as well as all references to
java.io.Serializable. RESyntax needs to replace java.util.BitSet with
a home-grown version. Other than that, it should work. May want to
package separately (e.g. gnu.regexp-1.0.8-j2me-cldc.jar).
Add support for named subgroups: "(?P<name>...)" (match) and "(?P=name)"
(recall, like \n). Add the ability to get at an enumeration of names.
Support associative-array style interface that allows users to define
their own substitution behavior on matches. Perhaps a Substitutor interface
with String substitute(REMatch match), and then RE.substituteAll(input,
Substitutor) variations.
Support excluded subrange syntax as in Unicode recommendations, e.g.
"[a-z-[m-p]]" being a through z except for m through p.
Add a split() method.
Provide a build.xml file for Ant.
|