Commit Graph

60 Commits

Author SHA1 Message Date
kr.angelov
4ac5902fc9 replace the hash maps in the abstract syntax with binary search tables 2014-10-09 08:38:55 +00:00
kr.angelov
23642fbb90 now we statically allocate closures for all top-level functions and all nullary constructors. closures are dynamically allocated only for CAFs. this reduces memory use and time to allocate dynamic closures 2014-10-08 12:57:29 +00:00
kr.angelov
8d09f62efc using instruction RET was wrong; now use EVAL 2014-10-06 15:57:42 +00:00
kr.angelov
698329f469 bugfix in the pattern matching compiler and a number of other fixes that I somehow did not push before 2014-09-29 15:00:04 +00:00
kr.angelov
25e8e4ce5a now a complete JIT compiler and ByteCode compiler for the def rules in the abstract syntax. there might be some bugs yet to be found, meta variables and computation under lambda is only partially supported 2014-09-25 10:35:06 +00:00
kr.angelov
923ad6b3c0 a major revision of the bytecode generator and JIT compiler. the effect is that now we can compute with lambda functions and with true tail recursion 2014-09-11 15:39:39 +00:00
kr.angelov
d7dc541f74 the code for def rules now uses proper graph update to preserve lazyness 2014-09-05 11:53:02 +00:00
kr.angelov
5e5ad8f2db full support for recursive def rules in the C runtime 2014-09-05 10:09:43 +00:00
kr.angelov
442dadf100 partial implementation for recursive def rules 2014-09-01 14:51:20 +00:00
kr.angelov
03b067782c a partial support for def rules in the C runtime
The def rules are now compiled to byte code by the compiler and then to
native code by the JIT compiler in the runtime. Not all constructions
are implemented yet. The partial implementation is now in the repository
but it is not activated by default since this requires changes in the
PGF format. I will enable it only after it is complete.
2014-08-11 10:59:10 +00:00
kr.angelov
a9ebeaea44 now fully functional Java API for custom literals 2014-04-10 14:14:31 +00:00
kr.angelov
7c66e438e1 GuVariant is now plain uintptr_t instead of a structure to make it easier to write Haskell bindings 2014-02-09 19:07:15 +00:00
kr.angelov
6028a63b42 loading and unloading of languages in the C runtime and in the Python bindings 2013-12-10 15:00:52 +00:00
kr.angelov
1d7b6f44b6 added Predef.SOFT_BIND. This special token allows zero or more spaces between ordinary tokens. It is also used in the English RGL to attach the commas to the previous word. 2013-11-12 09:54:57 +00:00
kr.angelov
475f213c99 the content of ParseEngAbs3.probs is now merged with ParseEngAbs.probs. The later is now retrained. Once the grammar is compiled with the .probs file now it doesn't need anything more to do robust parsing. The robustness itself is controlled by the flags 'heuristic_search_factor', 'meta_prob' and 'meta_token_prob' in ParseEngAbs.gf 2013-11-06 10:21:46 +00:00
kr.angelov
43fb9b3b7a added the linref construction in GF. The PGF version number is now bumped 2013-10-30 12:53:36 +00:00
kr.angelov
841880144d fix the handling of 'pre' in the C runtime 2013-10-28 12:35:37 +00:00
kr.angelov
c98a9a3266 tabular linearization in the C and Java runtimes 2013-10-23 14:49:28 +00:00
kr.angelov
b058fdd17d a major redesign in the C runtime. The parser and the linearizer now fully support BIND. The following things are still broken: parseval, word completion, handling 'pre', the robust mode 2013-10-09 12:08:51 +00:00
kr.angelov
2f78333e28 GuString is now an ordinary C string - it makes live easier. In addition PgfSymbolKS, PgfExprFun and PgfLiteralStr now keep their strings as embedded flexible arrays. The latest change gives us the same compactness as the old representation but it is a lot easier to use. 2013-10-04 12:04:39 +00:00
kr.angelov
357ddfb5d3 the symbol for nonExist in the GF runtime should be the last. this simplifies the binary search in the C runtime 2013-10-03 08:21:31 +00:00
kr.angelov
efa4bc4d62 a major refactoring in the C and the Haskell runtimes. Note incompatible change in the PGF format!!!
The following are the outcomes:

   - Predef.nonExist is fully supported by both the Haskell and the C runtimes

   - Predef.BIND is now an internal compiler defined token. For now
     it behaves just as usual for the Haskell runtime, i.e. it generates &+.
     However, the special treatment will let us to handle it properly in 
     the C runtime.

   - This required a major change in the PGF format since both 
     nonExist and BIND may appear inside 'pre' and this was not supported
     before.
2013-09-27 15:09:48 +00:00
kr.angelov
f984bfeb24 a major refactoring in the C runtime. GuList is now removed and replaced with GuSeq. The GuSeq/GuBuf API is simplified 2013-09-17 12:45:00 +00:00
kr.angelov
d5438439dc remove leftcorner_cat_idx which is now redundant 2013-09-13 13:49:17 +00:00
kr.angelov
288bcafb79 nonExist now does the expected thing 2013-08-23 13:17:45 +00:00
kr.angelov
03690ccc4b we no longer maintain an explicit list of functions per category since now it is implicitly kept in the JIT compiled code 2013-06-26 09:35:50 +00:00
kr.angelov
16584d4368 Now there is a just-in-time compiler which generates native code for proof search. This is already used by the exhaustive generator. The time to generate 10000 abstract trees with ParseEng went down from 4.43 sec to 0.29 sec. 2013-06-25 19:22:42 +00:00
kr.angelov
6cc44193b8 finally the statistical parser is able to return all possible abstract trees 2013-04-26 20:44:01 +00:00
kr.angelov
2a0c69a412 added API for computing bracketed strings from Python and C 2013-04-18 13:37:09 +00:00
kr.angelov
0b7b939aca refactoring: now all named objects in the C runtime have an explicit name field 2013-02-11 14:10:54 +00:00
kr.angelov
56c8f91d19 remove the pgf2yaml tool which was both broken and redundant. The declarations for generic programming from data.c are removed as well 2013-02-11 13:51:12 +00:00
kr.angelov
ff25ba8f90 the grammar reader in the C runtime is completely rewritten and it doesn't use the generic programming API 2013-02-11 10:16:58 +00:00
kr.angelov
3be31c62e9 a new reasoner in the C runtime. It supports tabling which makes it decideable for propositional logic. dependent types and high-order types are not supported yet. The generation is still in decreasing probability order 2013-01-07 12:50:32 +00:00
kr.angelov
5c9ee467a9 a major reimplementation of the linearizer in the C runtime 2012-12-19 09:07:05 +00:00
kr.angelov
2ba632dc9f a top-level API for parsing in the C runtime 2012-12-13 14:44:33 +00:00
kr.angelov
3182e382dc bugfix for robust parsing with multi-word units 2012-12-11 12:57:22 +00:00
kr.angelov
9721833680 a major refactoring in the robust parser: bottom-up filtering and garbage collection for the chart 2012-10-25 14:42:53 +00:00
kr.angelov
44df7a33cf the C runtime now has a type prob_t which is used only for probability values 2012-09-18 09:18:48 +00:00
kr.angelov
cd3cca4aa2 bugfix in the C parser 2012-09-06 14:52:19 +00:00
kr.angelov
3a352a953f Use a separated tag for meta productions in the robust parser. This cleans up the code a lot 2012-06-13 05:49:30 +00:00
kr.angelov
7549a4876d now there is a limit of 2000000 items in the chart of the robust parser. This prevents from explosion in the memory size but it will also prevent us from parsing some sentences. 2012-06-12 11:30:01 +00:00
kr.angelov
b765b0c054 now the robust parser is purely top-down and the meta rules compete on a fair basis with the grammar rules 2012-06-12 09:29:51 +00:00
kr.angelov
bd8046f23d another attempt to port the robust parser to MacOS 2012-05-16 15:18:44 +00:00
kr.angelov
4aca965109 a new unbiased statistical parser. it is still far from perfect use it on your own risk. 2012-05-08 12:13:28 +00:00
kr.angelov
c6c54f8815 some fixes in the robust parser and a new API for literals 2012-04-12 06:55:25 +00:00
kr.angelov
2bf3f22fac libpgf: a new implementation for literals which also allows custom literals. the same mechanism is now used for the metavariables 2012-03-12 14:25:51 +00:00
kr.angelov
1da464a4cc libpgf: implementation for built in literal categories 2012-03-07 16:39:29 +00:00
kr.angelov
791a1a17b0 libpgf: now all concrete functions and categories are explicitly linked to their abstract counter parts 2012-03-05 12:59:31 +00:00
kr.angelov
aca0bd5ee5 libpgf: the first prototype for the robust parser 2012-02-29 14:43:08 +00:00
kr.angelov
f1d2852c4d libpgf: now we have both complete bottom up index for robust parsing and fast lexical lookup from the same index 2012-02-22 21:27:54 +00:00