Commit Graph

577 Commits

Author SHA1 Message Date
krangelov
d4ccd2848c Take a step back and do bottom up prediction only when needed. This solves the problem that some sentences were impossible to parse. 2020-11-17 10:26:00 +01:00
krangelov
90fc1d750e remove the deprecated pgf_print_expr_tuple 2020-08-14 21:03:48 +02:00
krangelov
23edeec5a9 added an API for cloning expressions/types/literals 2020-08-14 20:54:15 +02:00
krangelov
a2d7f1369c filter out empty brackets 2020-07-26 15:55:18 +02:00
krangelov
77a2630ed9 revert to using unconditional probabilities in the different lookup functions 2020-07-11 09:52:43 +02:00
krangelov
33818076ff drop the SG library completely. 2020-07-08 21:12:01 +02:00
krangelov
165c5a6d9d bugfix in parsing literals 2020-05-20 19:57:33 +02:00
krangelov
48d3973daa fix the uggly typo that broke the literals. 2020-05-20 10:15:53 +02:00
krangelov
9a1f982b14 split unknown words by spaces 2020-05-16 08:32:00 +02:00
krangelov
62bc78380e lookupCohorts now detects and reports unknown words. Also:
- added added two filtering functions: filterLongest and filterBest
 - updated the PGF service to work with the new API
2020-05-14 15:03:30 +02:00
krangelov
762d83c1f0 switch off the debugger 2020-03-25 20:03:06 +01:00
krangelov
00e25d0ccb an API to access the names of all fields withing a category 2020-03-06 12:29:08 +01:00
krangelov
88f76ef671 lookup_morpho, lookup_cohorts and fullform_lexicon now report the usual conditional probability. This turns out to be more useful for part of speech tagging 2020-03-05 12:04:42 +01:00
krangelov
f22bd70585 The APIs for callbacks and the bracketed strings now use a string for the analysis intead of an integer. This is now consistent with lookupMorpho and friends 2020-03-05 11:58:21 +01:00
krangelov
14f394c9e9 a version of the parser which returns a chart rather than a list of expressions 2019-12-07 22:00:39 +01:00
krangelov
95b3fb306f forgot that debugging is on 2019-10-09 14:34:04 +02:00
krangelov
5b790b82c5 fix chunk extraction when there are literals 2019-10-09 14:32:20 +02:00
krangelov
26361b3692 fix the parsing for literals after the latest changes in the parser 2019-10-09 14:18:05 +02:00
krangelov
e2ddea6c7d first version of a parser which returns chunks in case of failure 2019-08-30 13:31:57 +02:00
krangelov
59a6e3cfdd fix gu_map_next 2019-08-30 13:31:19 +02:00
krangelov
72cfc1f48a a more reasonable API to iterate over a map 2019-08-30 08:12:15 +02:00
krangelov
394d033d19 added gu_map_delete 2019-08-20 18:55:36 +02:00
krangelov
3e662475ee bugfix in the parser's scanner 2019-08-09 15:15:30 +02:00
krangelov
32379a8d11 fully supported case-insensitive parsing/lookup 2019-06-30 08:48:23 +02:00
krangelov
b56591c6b6 the parser now ensures that all word senses are in the chart 2019-06-25 12:58:28 +02:00
krangelov
b94bb50ec9 fix in gu_buf_heap_pop 2019-06-24 12:27:36 +02:00
krangelov
2d9478b973 share some code for printing 2019-06-24 12:13:14 +02:00
krangelov
ac039ec74f filter out empty cohorts 2019-05-28 12:42:59 +02:00
krangelov
9f0ea19a1c API for scanning for cohorts in an arbitrary text 2019-05-28 12:26:00 +02:00
krangelov
8b9719bd2d remove duplicate definition for RecordCompare 2019-05-27 09:05:41 +02:00
krangelov
1d1e65185a fixes in the headers for use with Microsoft Visual C++ 2019-02-18 11:39:21 +01:00
krangelov
c32cd7133f fixes in the headers for compilation on Windows 2019-02-18 11:23:01 +01:00
krangelov
409731413e disable the parser debugger which was left enabled by accident 2019-02-14 15:33:21 +01:00
krangelov
8a5e7fa25d fix the reader for patterns in a def rule 2019-02-14 15:32:12 +01:00
krangelov
e05c79a751 different definition for GF_ARRAY_LEN 2019-02-14 15:31:36 +01:00
krangelov
ef21d08225 bugfix in computing record lengths 2019-02-14 15:29:33 +01:00
Krasimir Angelov
26dabeab9b save the original concrete category in BracketedString 2018-12-20 10:52:45 +01:00
krangelov
eb22112178 a pool where the smallest chunks are memory pages 2018-12-16 07:46:14 +01:00
Krasimir Angelov
6c4c22ff7a undo the partial fix for word completion since it breaks normal parsing 2018-06-27 11:53:40 +02:00
Krasimir Angelov
68db7695f5 a simple fix for the linearizer 2018-06-20 22:21:10 +02:00
Krasimir Angelov
b86373438b Revert "when linearizing missing functions with brackets, the name of the function should still be reported"
This reverts commit 18204bdd25bd460904ac475f3ea340daa96589df.
2018-06-20 21:16:19 +02:00
Krasimir Angelov
4cdd4dd8a7 Revert "forgot to switch off the debugger"
This reverts commit 5919dfa3366dfd2f2af8c3ce7749d066a2033f0d.
2018-06-20 21:13:58 +02:00
Krasimir Angelov
427f8d8478 forgot to switch off the debugger 2018-06-20 16:36:45 +02:00
Krasimir Angelov
aaefe4310d when linearizing missing functions with brackets, the name of the function should still be reported 2018-06-20 16:35:28 +02:00
Krasimir Angelov
9d39648dfb fixed typo which broke the compilation on Windows 2018-05-21 16:03:30 +02:00
Krasimir Angelov
a6779486c5 fix the compilation when the byte order cannot be detected statically 2018-05-21 09:36:39 +02:00
Krasimir Angelov
2d898e78c0 bugfix in the C parser 2018-04-12 14:45:02 +02:00
Krasimir Angelov
bb4218433f - tweak the tokenizer in pgf_lookup_sentence to threat .!?,: as separate tokens
+ bugfix which causes crashes
2018-02-22 11:35:54 +01:00
Krasimir Angelov
a16fe3415a partial fix for word completion in the C runtime 2018-02-04 17:21:15 +01:00
Krasimir Angelov
fa8530add1 silence some warnings on MINGW32 2017-12-09 23:11:04 +01:00