ae9e652f1e
Mostly implemented Simultaneous Declaration, only template instantation during pass 2 remains to be implemented
Nathan Braswell2014-05-24 14:04:32 -04:00
0f6b6c0c67
Posets now integrated into the CGenerator, so the CGenerator now chooses a valid order for the object definitions based on their dependence on each other. Another test now passing
Nathan Braswell2014-05-21 13:14:16 -04:00
d37a07201a
Created a Poset template class. It can store the relationships of a partially ordered set, then generate a topological ordering. It will be used to order the type_def declarations in the CGenerator, as they depend on each other in a poset fashion.
Nathan Braswell2014-05-21 12:01:45 -04:00
6350f93f24
Fixed the recursion bug. 8/10 tests pass now, remaining need to fix define order and CGenerator order
Nathan Braswell2014-05-21 01:20:39 -04:00
17b80d4102
Fixed a typo that caused indirection operators to do the reverse on types. Also cleaned up some comments. One more test working.
Nathan Braswell2014-05-20 23:53:19 -04:00
2566cbb67c
Added testing! call kraken like so ./kraken --test ../path/to/test/name_of_test_without_extention This will make kraken compile and run name_of_test_without_extention.krak and compare the output it generates on stdout to name_of_test_without_extention.expected_results. If they pass, then it records the pass, if not, it records the failure and saves the intermediate files generated. It has revealed some bugs which I will fix in upcoming commits.
Nathan Braswell2014-05-20 22:21:07 -04:00
39f945940d
Fixed a ton of stuff with function templates. Works well now. Next up: multiple template types and fixing object definition ordering (both where templates should go and objects with other pointers)
Nathan Braswell2014-05-19 20:00:35 -04:00
2a4edf9afd
Simple Templates work! Even Templates in other files work. Happy day
Nathan Braswell2014-05-10 19:28:36 -04:00
5022fc0802
Significant work on templates. They should be close to working now (for single replacement). However, they do not show up in the global scope the the C generator can't see them. Plus, their names will have to be Cified.
Nathan Braswell2014-05-09 02:56:55 -04:00
209985310e
Fixed type propagation for dereference and address-of, needs work for all the other operators also. Starting ground work for array notation
Nathan Braswell2014-05-05 13:52:12 -04:00
9a4507a0f5
Added passthroughs and small fix for malloc/free
Nathan Braswell2014-05-03 20:46:10 -04:00
6a75832b59
Added trivial standard library and search paths.
Nathan Braswell2014-05-01 01:18:01 -04:00
1997ba49d2
Fixed operator overloading so that both method and non-method operator overloads work.
Nathan Braswell2014-04-27 02:48:57 -04:00
57976beb40
Just got paranoid about saving all this work. Almost finished operator overloading, but everything is slightly broken right now.
Nathan Braswell2014-03-06 13:13:40 -05:00
7f902880c5
Cleaned up and added line number to error messages
Nathan Braswell2014-02-20 18:24:04 -05:00
53b45f360d
Now generates for all files that have been imported. CGenerator uses this to generate all files AND a shell script with the compile command to compile the generated C file.
Nathan Braswell2014-01-01 17:29:19 -06:00
dbf1820428
Can import other files now. (in same directory, use filename without .krak file extention) Right now, still need to compile both files with kraken.
Nathan Braswell2013-12-31 23:43:49 -06:00
005659b7b7
Added caching of the RNGLR table. It is automatically regenerated whenever the grammer changes. Right now it has dropped compiling the test file from 30 seconds to less than one second.
Nathan Braswell2013-12-31 02:53:52 -06:00
15674fec2a
Added Type class, bettered types a bit, made address of and dereference operators work.
Nathan Braswell2013-12-23 01:26:24 -06:00
935cc6f968
Added generator-dependent compilation and simple passthrough that allows us to have non-cheated emitted, printing c-code for the first time! (no typechecking or anything yet, but we'll get there). It's also still rough.
Nathan Braswell2013-12-22 01:34:59 -06:00
6ad406e42d
Small fixes to the grammer, ASTTransformation and CGenerator. Should now be ready to begin implementation of multiple files, conditional inclusion, and code passthrough.
Nathan Braswell2013-12-19 10:39:36 -06:00
f273deaedc
FINALLY fixed an error that took weeks. Turned out that the ParseRule was shallow copied, and the lookahead was not copied correctly. So it got extended and thus skipped over the state when it should have been redone.
Nathan Braswell2013-12-18 18:05:21 -06:00
9336193aaf
Allow RegEx matches of length 0. This seems more correct to me. (However, this is not super important in practice. Grammar files should not contain regular expressions that could match the empty string; if such a RegEx matched 0 characters once, it would match again and again forever, since it wouldn't consume any input.)
Jason Orendorff2013-10-26 23:52:54 -07:00
d2d38e2516
Fix what looks like an off-by-one error in RegEx::longestMatch()'s lastMatch calculation, and a corresponding +1 in code using longestMatch, and add a test.
Jason Orendorff2013-10-26 23:29:23 -07:00
7859b29725
Fix a minor bug in the lexer (it would not match the last character of the input) and add Lexer tests.
Jason Orendorff2013-10-26 23:05:25 -07:00
ebe6b3520a
More AST work. Getting close to fully transforming language as specified by current grammer. (which is a subset of what it will become)
Nathan Braswell2013-10-26 15:05:42 -04:00
e818236e90
Some more AST work. Functions no longer retain an identifer for their name.
Nathan Braswell2013-10-26 11:47:34 -04:00
83af1b1c5a
Use std::string::substr() instead of adding one character at a time. This patch also removes a few unused util functions that are easily written using substr().
Jason Orendorff2013-10-25 02:42:12 -07:00
727529fe0b
Remove one unnecessary backslash-check in StringReader::getTokens(). (This patch looks like it's removing the last backslash-counting loop, but really it removes the first one and then reindents everything.)
Jason Orendorff2013-10-25 02:32:39 -07:00
64a405cab1
Replace some code in StringReader::getTokens() with std::string::find_first_of(). Since this makes found_pos unsigned, a few comparisons involving found_pos must be adjusted (unsigned integers are always >= 0).
Jason Orendorff2013-10-25 02:26:31 -07:00
d21f521266
Have a nullableParts generator now. It works for simple cases and may work for advanced cases, but more testing is required to ensure full usability.
Nathan Braswell2013-08-13 14:01:53 -04:00
3a5d94caaa
Started adding the nullable parts machinery. Also started to fix addStates for RNGLR and null reductions so it only truncated the rule to what had already been parsed, but now it looks like it maybe shouldn't truncate at all and use pointer position for length. However, I have to go to bed and will pick up later.
Nathan Braswell2013-08-12 00:02:50 -04:00
ea42cb5bc7
The out of orderness was actually the fault of the dot graphing program, not Kraken, so that's good news. DOT generation has also been modified so that it properly inserts \n's (actually \n's) where line returns are so that dot can parse the \n and not the line return.
Nathan Braswell2013-08-11 00:37:12 -04:00
b292bd15ae
Now parses into parse trees! Still does not add the null-reducable trees, and there are a few ordering mistakes, but it works.
Nathan Braswell2013-08-10 18:24:37 -04:00
9f3370481e
Fixed all syntax errors, but NULL crops up where it shouldn't. Checks for it as stop-gap, creates very truncated trees. Does run, and still correctly accepts. Calling it a night.
Nathan Braswell2013-08-09 04:39:43 -04:00
8570b8f641
Well, just before I went to bed I fixed the errors in NodeTree and GraphStructuredStack. To tackle RNGLRParser tommrow.
Nathan Braswell2013-08-08 03:06:28 -04:00
31d7c02ef9
Finished writing first code for turning the RNGLR recognizer into a parser. Doesn't compile yet, needs lots of bugfixing, but I'm calling it a night.
Nathan Braswell2013-08-08 02:44:17 -04:00
680d978dcb
Problem was actually in grammer, fixed it. Also made identical rules with different lookahead merge. Now just started on creating parse trees. Stopping for night.
Nathan Braswell2013-08-06 01:49:45 -04:00
9460bacf1c
Fixed some problems with grammer loading of null rules and rules non-terminals and terminals that had the same name
Nathan Braswell2013-08-04 14:24:20 -04:00
d5b33efb22
Null rules work in RNGLR now, but there seems to be a bug where somehow some states/rules get no lookahead.
Nathan Braswell2013-08-02 15:21:42 -04:00
49d149bc1f
RNGLR Recognizer now works for grammers without null rules.
Nathan Braswell2013-08-02 02:47:01 -04:00
9887555dd5
Have almost got the RNGLR recognizer. Something is wrong with the GSS, I think when it's built. It seems to sometimes have parents as children, or something.
Nathan Braswell2013-07-31 23:51:05 -04:00
3fba45591b
Changed Parser to be a virtual base class, inherited by LALRParser
Nathan Braswell2013-07-30 01:42:51 -04:00
733fe1c08d
Changed Table to store vectors of actions, changed Parser appropriately.
Nathan Braswell2013-07-28 23:48:10 -04:00
6d7b38a03b
More work towards RNGLR. First, NodeTree is now a template. Second, I've started writing the actual GLR parser and GSS and other things, but am still in the first write process.
Nathan Braswell2013-07-28 19:45:08 -04:00
726ead0455
Pull out table to it's own Table class in prep for adding RNGLR algorithm.
Nathan Braswell2013-07-16 11:15:58 -04:00
f84657f1ed
Fixed a lot of bugs. Actually gets through the entire experimental grammer. (A largeish experimental grammer for Kraken written to continue testing and to really start language development.
Nathan Braswell2013-07-10 23:50:53 -04:00
4c2fd967f0
Perens now work fully! The RegEx part of Kraken should now be fully legitimate. The only problem is not gracefully letting the user know about faulty input, but that goes for all of Kraken.
Nathan Braswell2013-07-07 02:13:05 -04:00
502929963c
Added alternation and fixed a bug with the repetition operators exposed by the addition of the alternation. (for some reason the repetition operators used a loop making endless circles)
Nathan Braswell2013-07-04 15:10:32 -04:00
84566c4ff6
Extended the Regular Expression class to now support * and ? as well as +. Next up, perens and alternation
Nathan Braswell2013-07-03 23:40:36 -04:00
cc6ff21986
Fixed some bugs, including modifing StringReader to treat quoted strings as whole words.
Nathan Braswell2013-07-02 13:14:40 -04:00
85da0bf646
Lexer/RegExs now work on a very basic level. Regular expressions have not been fully implemented, but the structure is there. It creates trivial regexs that only accept a specified sequence, no actual regular expression power.
Nathan Braswell2013-07-02 01:47:42 -04:00
94a7739bd9
Added structure for Regular Expressions, saving work as a backup because of software upgrade.
Nathan Braswell2013-07-01 22:45:33 -04:00
c2520ec2c4
NodeTree now also has a link to the Symbol it represents.
Nathan Braswell2013-06-27 23:45:38 -04:00
Nathan Braswell