summaryrefslogtreecommitdiff
path: root/src
Commit message (Collapse)AuthorAge
* expr: apparently strings are streams tooJari Vetoniemi2018-09-27
| | | | rename collapsestr and collapse to just stream
* compiler: implement enum countingJari Vetoniemi2018-09-27
|
* implement the binary number typeJari Vetoniemi2018-09-27
|
* share lex blocks between functions and referencesJari Vetoniemi2018-09-26
|
* types.lm: use lex blocksJari Vetoniemi2018-09-26
| | | | So the matcher wont include the tokens used by other tokens
* New makefileJari Vetoniemi2018-09-26
|
* fix some parsing bugsJari Vetoniemi2018-09-26
|
* remove type::name add endianess supportJari Vetoniemi2018-09-26
| | | | | type::name is too agressive as it clashes with keywords use reference::variable::type and reference::function::type instead
* Indicate fields with slices in outputJari Vetoniemi2018-09-26
|
* More parsing fixes for the fspec files I haveJari Vetoniemi2018-09-26
|
* Print out select expression informationJari Vetoniemi2018-09-26
|
* Add some commentsJari Vetoniemi2018-09-26
|
* Fix some parsing mistakesJari Vetoniemi2018-09-26
|
* Goodbye C compiler, hello colm compilerJari Vetoniemi2018-09-26
|
* tmpJari Vetoniemi2018-02-23
|
* xi2path: Cast before snprintfJari Vetoniemi2018-02-23
| | | | | Make sure results are unsigned Found by cppcheck
* wipJari Vetoniemi2018-02-23
|
* lexer.rl: hex can be as long as we wantJari Vetoniemi2018-02-23
| | | | | | | | hex expression shared the esc_hex expression before. But esc_hex is limited to 2 hexdecimal characters, and thus poor choice for parsing hexdecimal numbers in source. (This is different from C which allows "n" length hex escapes causing problems when you mix hex escapes with strings for example)
* ragel.rl: mark anything as errorJari Vetoniemi2018-02-23
| | | | | | | | The word label is mainly meant to define printable and conceated characters. In print_mark we use it to print tail in case we hit the valid* kleene star expression. Otherwise we will just show '^' mark on whatever byte we failed on. This allows to catch semantical failures such as having newlines at wrong place etc..
* ragel.rl: better error matchingJari Vetoniemi2018-02-23
| | | | | | Combine tail action into mark. Tokenize on 'space <valid>' or on punctuation. Treat single printable or valid* as word.
* dump.c: Postcond that we actually read somethingJari Vetoniemi2017-12-07
|
* Add assert to tell compiler wrong behaviour is UDHEADmasterJari Vetoniemi2017-10-01
|
* s/s32/s64/Jari Vetoniemi2017-08-15
|
* ragel.rl: Add some asserts to confirm inputJari Vetoniemi2017-05-07
| | | | Just some hardening: p should always be equal or behind pe.
* ragel.rl: Make search_err read "p" before stopJari Vetoniemi2017-05-07
| | | | | | | | | If we stop before reading the "p" false positive may be returned since "p" may be the start of new token. Thus case like `type: 3u8[4] str;` would fail to mark the real error since '3' is not read. Thus start marking from the last valid token that is ':'.
* validator.rl: Unused memberJari Vetoniemi2017-05-03
|
* validator.rl: Allow multiple filtersJari Vetoniemi2017-05-02
|
* dump.c: Align to_hex to at least 2 charsJari Vetoniemi2017-05-01
| | | | | This is to have nice aligned output when outputting byte arrays for example.
* dump.c: Fix print_array when data ends on 8 columnJari Vetoniemi2017-05-01
|
* dump.c: More compact hex dumpingJari Vetoniemi2017-05-01
|
* work in progressJari Vetoniemi2017-05-01
|
* Initial commitJari Vetoniemi2017-04-13